CURRENT TAXONOMIC RESEARCH BRITISH FLORA Edited by J.P. Bailey fklGCGC] SGTTJCa mtaCGf jSf ?CG' Jogcg\ yisttcn / A* AC &C ACe ItteGGG J TC G' fmi, aaGC&cftccTreaj^r CGACACGTGaGCGCaCCTTG/ f 5*TCft CGftyic&TftiiiCGcacccct^ (;gttcg ~f ftAaCGAACCTCGA \aaacGcaccccGGGM”«:a! Vaac gc acc^c ggg airc a CCaT&GaaATCGCTTTb Stem leaves Stem leaves Leaflets Leaflets hairy Leaflets hairs below = absent = present F = absent = present F = simple = compound F = glabrous = hairy S = below F = above = both sides = sparse = dense S Experience so far with the Stace Flora suggests about 1500 characters will be needed to separate the genera. The characters are being extracted from the family and generic keys and the text by using a specially written ‘data mining’ program which will be the subject of another paper. This data mining software is founded on a relational database of the flora and is already provided with all the scientific names and the hierarchy of their classification. It separates the characters into the structures e.g. stem, leaf, flower and the descriptors e.g. size, colour, roughness of which they are composed. So for example, ‘petal colour' is made up of the structure ‘petal’ (which is part of the structure ‘flower’) and the descriptor ‘colour’ of type ‘appearance’. This breakdown allows some standardisation of the characters, so that for example, the way that the hairs on the stem are described may draw from the same set of 12 CURRENT TAXONOMIC RESEARCH states as are used for the hairs on the fruit. Such standardisation must not of course sacrifice any of the descriptive power of the original text, nor ignore the actual diversity of the plants themselves. This is partly achieved by associating a list of possible descriptive states with a descriptor, and then picking out a subset of these for use with each particular structure. The work being done here will create a character set specifically for Clive’s Flora, and will only represent one possible way in which this might be done. In the future, other rival character sets for representing specific floras could emerge. CONVERSION OF THE STACE FLORA Clive’s Flora is a concise Flora, which is to say that it deliberately sets out to provide the essential minimum of information, so that every plant that you might encounter in Britain can be recognised reliably with as little information as possible. To have done this is a considerable achievement. A second motivation for conciseness is to reduce the size of the resulting book (so that you can cany it in a rucksack), and therefore to reduce its cost. Once the Flora is moved onto a computer, however, the limitations of space related to cost immediately vanish. Pennanent disc memory in modem desktop computers is now very spacious and cheap, and a blank DVD-ROM capable of holding 4 gigabytes costs about £2 sterling, regardless of what data it contains, and a CD-ROM of about 600 megabytes costs much less. There are still reasons why conciseness is a virtue and will be needed on occasion, but it should now be possible to include in the Flora all manner of useful data for which previously there was no room. Table 4 lists some of these possibilities. TABLE 4. POSSIBLE CONTENTS OF A PORTABLE FLORA DATABASE • National nomenclatural data (Kent list) • Computerised descriptions of all taxa and all characters; vegetative, flowers, fruits, seeds and seedlings • Ecological data • Images; colour photos and drawings of plants and characters, and lots of them! • Records; National (Atlas 2000) Census catalogue Individual records from VC databases VC distribution maps and old record cards • Maps; OS plus relief, climate, geology, vegetation • National Vegetation Classification (NVC) and identification of vegetation types • Built-in GPS In order to be able to select and retrieve such information, a stronger form of organisation is needed than simply the page-sequencing and indexing of a book. ETI are doing this by what is called ‘hypertext’ or the linking of sections of text with pointers to other sections of related text, which is a kind of text-oriented database. A stronger way of doing this is to use a properly organised relational database. There is no better known way of organising data so that you can navigate through it and retrieve relevant and related information, so one may expect computerised Floras to be organised in this way in the future. KEYS FOR THE FUTURE 13 HARDWARE The conventional desktop computer already has enough memory to store enough data, not just text and numerical data but coloured images as well. How about taking this into the field? Richard Pryce has already been experimenting with pocket computers for botanical recording, but he tells me that their memories are too small at present to contain a complete Flora. The alternative is to carry a wireless-enabled terminal so that the data that you need can be received from a geo-stationary satellite. That is not really practical just yet either. But perhaps within 10 years or so, one way or another, a computerised flora will be practical in the field, and not just indoors. It will be tremendously exciting. REFERENCES Boughey, A.S., Bridges, K.W. & Ikeda, A.G. (1968). An automated biological identification key. University of California, Irvine. Research Series no. 2. Clapham A.R., Tutin T.G. & Warburg E.F. (1962) Flora of the British Isles , edn 2. Cambridge University Press. DALLWITZ, M.J. (1980). User ’s guide to the DELTA System: a general system for encoding taxonomic descriptions. CSIRO Division of Entomology, Canberra. JORNA, S. (ed.) (2001). Flora Europaea on CD-ROM Cambridge University Press. Laubner, K. & Wagner, G. (1996). Flora Helvetica and CD-ROM. Paul Haupt, Bern. Lamarck, J.B.P. (1778). Flore Frangoise, first edition, Paris, Imprimerie Royale. PANKHURST, R.J. (1970). A computer program for the generation of diagnostic keys. Computer Journal 12: 145-151. PANKHURST, R.J. (1983). The construction of a floristic database. Taxon 32(2): 193-202. PANKHURST, R.J. (1989). A computer program with colour graphics to identify orchids. Orchid Review 97(1144): 53-55, 67. PANKHURST, R.J. (1991). Practical Taxonomic Computing. Cambridge University Press. PANKHURST, R.J. & ALLINSON, J.M. (1985). British Grasses, a punched-card key to grasses in the vegetative state. Field Studies Council & British Museum (Natural History) FSC Occasional Publication no. 10. PANKHURST, R.J. & Chater, A.O. (1988). Sedges of the British Isles. Computer Key No 1. Botanical Society of the British Isles. Stace, C.A. (1997). New Flora of the British Isles , edn 2. Cambridge University Press. Young, T., Yeates, D. & Thiele, K. (1997). LUCID player instruction manual, University of Queensland, Australia. Current taxonomic research on the British & European flora Bailey, J. & Ellis, R.G. (eds) 15-25 (2006), BSBI. London. 15 Population based approaches in the study of Pilosella Hill (Asteraceae): A new view of its taxonomy? Frantisek Krahulec and Anna KrahulcovA Institute of Botany, Academy of Sciences of the Czech Republic, CZ-252 43 Pruhonice, Czech Republic; krahulec@ibot.cas.cz ABSTRACT We present a structure of the Pilosella populations occurring in the Krkonose Mts, northern Czech Republic. Each basic species, hybridogenous species and recent hybrid is documented by its frequen¬ cy, cytotypes, breeding system, and chloroplast haplotypes. We compare this structure with the situation in another mountain range, the Sumava Mts in the south western Czech Republic. Both regions have the same structure of hybridising species, but the resulting population are different. We deduce that random phenomena in the past and the residual sexuality of apomictic species has influenced the present population composition. Our results are discussed in connection with existing approaches to the taxonomy of Pilosella. Keywords: breeding systems, cytotypes, chloroplast haplotypes, population structure, taxonomy INTRODUCTION The genera Hieracium and Pilosella (often considered as subgenera of Hieracium) belong to one of the most complicated groups of temperate flora. Their complexity is closely connect¬ ed with the number of recognized species, but reflects also the number of underlying mechanisms involved (for Pilosella see review by Krahulcova et al. 2000). In both genera, the pattern of morphological variation is clearly reticulate. Both genera differ by several morphological characters, but also in some evolutionary trends: one of the most important is that hybridisation is common in Pilosella , but recent hybridisation in Hieracium is extremely rare. Studies of hybridisation between Pilosella species began in the 19th century; one of the well-known papers being that by Mendel (1869); this era ended with the monumental book by Nageli & Peter (1 885). In spite of this there is a lack of population studies similar to those which elucidated processes involved in many agamic complexes, as summarized in Richards et al. (1996). In Pilosella , some studies were only carried out recently by Russian authors (Kashin et al. 1999, 2000a, 2000b) and also outside their native distribution area in New Zealand (e.g. Chapman et al. 2003). The probable reasons for this absence of deep insight are difficulties with the determination of plants and the complexity of populations. Recent methodological advances have allowed us to look at Pilosella populations using different approaches. In cooperation with German and other Czech colleagues we looked at Pilosella occurring in the Krkonose Mts, in the western part of the Sudeten Mts; this area was already extensively studied in 19th Century and many hybridogenous species were described from it (results were summed by Nageli & Peter 1885 and Schneider 1888-1895). 16 CURRENT TAXONOMIC RESEARCH We tried to explore the existing variation at the levels of the population and the mountain range. We also started to compare this situation with another region outside the Sudeten Mts, V the Sumava/Bohmerwald, a mountain range forming the border between Germany and the Czech Republic. Our approach included especially the determination of chromosome num¬ bers and breeding systems (summarized by Krahulcova et al. 2001), hybridisation experi¬ ments (Krahulcova, unpublished data), residual sexuality of facultative apomicts (Krahulcova et al. 2004 and unpublished data), haplotype determination (Fehrer et al. 2005, Krahulec et al. 2004) and clonal structure detected by isozyme spectra and DNA fingerprint¬ ing (Fehrer et al. 2005, Krahulec unpublished data). Knowledge of the detailed structure of the apomictic complexes within the two areas enabled us to search for common features. The aim of the present paper is to show these common features and to discuss the possibilities of some pragmatic, reasonable taxonomy of this complicated genus. THE STUDY AREAS The plants studied occur mainly in montane meadows in Central Europe. The montane meadows in the Krkonose Mts were established in late medieval times. They are situated at altitudes between 600 and 1300 m, being relatively species rich, especially on a fine scale. V The other region, the Sumava Mts, belongs to the Herzynian system. The habitats of the Pilosella species here are partly meadows and partly various successional stages with disturbed soil surfaces situated between 780 and 1 150 m. Both mountain areas are isolated, and are separated across the Czech basin by more than 250 km (for location see Figure 1). Figure 1. Location of the Krkonose Mts and Sumava Mts, where the populations were studied STUDY OF PILOSELLA 17 THE COMPLEXES STUDIED The whole complex occurring in the Krkonose Mts is rather complicated. For simplicity we divided it into two parts, the yellow-flowered and orange-flowered (red) groups. Both have P. officinarum (H. pilosella) as one parental species (a sexual one). This putative origin has been checked, especially by experimental hybridisation. Apomixis in Pilosella is of the aposporous type; it is facultative. When we say that the breeding system is apomictic, it is apomictic with some degree of residual sexuality. A complete list of species discussed is in the Appendix (p. 25), together with their equivalents in Hieracium. RESULTS THE KRKONOSE MOUNTAINS The whole complex is formed from six basic species, their hybrids behaving as hybridog- enous species, and secondary hybrids (Figs 2 & 3 (left)). Two of them are diploids (P. auricula, P. onegensis), the others are tetraploids to hexaploids, and we found one heptaploid plant. Some of the species are represented by one ploidy level, others by two ploidy levels and P. piloselliflora by three ploidy levels. With respect to the breeding. system, three species were found to be fully sexual: P. auricula, P. officinarum, P. onegensis. The tetraploid cytotypes of two other species, P. piloselliflora and P. schultesii, were recorded as both sexual and apomictic. All other species (and cytotypes at higher ploidy level) were apomictic (Krahulcova & Krahulec 1999, 2001 and unpublished results). Two main haplotype groups were distinguished, each of them with two subtypes differing in shorter sections of cpDNA (Fehrer et al. 2005). The following basic species share individual haplotype groups, whose presence in individual hybrids is indicated in Figures 2 & 3 (left). Haplotype group 1: P. auricula, P. aurantiaca (pentaploid). Subtype characterized by deletion of 5 bp was found in tetraploid P. aurantiaca. Haplotype group 2: P. onegensis, P. caespitosa, P. officinarum ; subtype characterized by 6bp insert is present in P. cymosa (here represented by subsp. cymigera). Besides the sexual P. officinarum , a key hybridogenous species P. floribunda is also involved in the formation of other hybrids (Fig. 2). It is tetraploid and always apomictic. The hybrid morphologically closer to P. floribunda (P. iserana), mostly tetraploid and apomictic, is evidently an n+n hybrid with P. floribunda as the mother plant. Artificial hybridisation between P. floribunda and P. officinarum was successful only in one direction, using apomictic P. floribunda as a mother plant, which is also supported by the identical haplotype of P. floribunda and P. iserana. P. iserana has low morphological as well as molecular (DNA fingerprinting) variation. The type morphologically closer to P. officinarum (P. piloselliflora) is variable in all respects, i.e. in morphology, ploidy levels and breeding systems. It is also variable in haplotypes, which indicates that both parents, P. iserana (or rarely P. floribunda ) and P. officinarum ) served as mother plants. Many isozyme phenotypes were found, even at the locality level. All these facts indicate that P. piloselliflora is a complex of genotypes repeatedly formed by hybridisation between P. iserana (or P. floribunda) and P. officinarum. The hybridisation between P. iserana and P. officinarum (in both directions) is easily achieved and was repeated several times during garden experiments. 18 CURRENT TAXONOMIC RESEARCH STUDY OF PILO SELLA \ 19 <4-* o t: cd a, a> su c cd •— O 2 cd > cd s 3 00 _c -a c cd 2 o Cd cd -C co S-h o s 4—* 2 CO o Ch o 5 so 4= 4— > c • F*=* X JO Hh £ o o t/3 ao Q- iX frame. Names of the basic species are given in bold; arrows going from their parents indicate origin of hybrids. For all species, hybridogenous species, and recent hybrids their cytotypes, breeding system and chlo- roplast haplotype (by different shading of box with chromosome num¬ bers) are given. 20 CURRENT TAXONOMIC RESEARCH Triploid hybrids between P. officinarum and P. auricula were not found in the field; we found tetraploid and pentaploid plants, the first being sexual or apomictic. These are similar to P. piloselliflora. The orange-flowered complex consists of hybrids between P. aurantiaca and the yellow-flowered species, P. caespitosa , P. auricula and especially P. officinarum (Fig. 3 left). Within the region, P. aurantiaca occurs as a common tetraploid (two isozyme pheno¬ types) and a rarely as a pentaploid, both apomicts. The frequency of its hybrids is variable: its hybrid with P. caespitosa (P. fuscoatra ) was found for the first time at the end of 1970s and has a limited distribution. Its hybrid with P. auricula (P. blyttiana ) has been known from the area for more than 100 years (and it was independently described as H. latibracteum Peter). It is rare at present, without any morphological or DNA (fingerprinting profiles) variation; evidently, there is only one clone present. The variation of hybrids between P. aurantiaca and P. officinarum is more complicated. One type closer to P. aurantiaca was described from this region as H. rubrum (P. rubra). It is apomictic, morphologically as well as cytologically uniform (hexaploid). Its cp-haplotype differs from that of the common tetraploid P. aurantiaca forms, but is the same as that of the pentaploid P. aurantiaca. A direct relationship with present tetraploid P. aurantiaca seems improbable. We have so far detected two clones by DNA fingerprinting and isozyme phenotypes. The types closer to P. officinarum (P. stoloniflora ) are more variable, both cytologically (pentaploids- and hexaploids) and morphologically. One pentaploid plant was found within the stand of P. rubra. It probably originated as a product of back-cross between P. rubra and P. officinarum , morphologically corresponding to hybrids of experimental origin. Hexa¬ ploids have the same cp-haplotype as P. rubra and pentaploid P. aurantiaca. So, in spite of our efforts the exact origin of these types remains unclear, as there are several possible routes. SUMMARY OF THE STRUCTURE OF THE COMPLEX IN THE KRKONOSE MTS 1 A number of hybrids where predominantly an apomictic plant was the mother. They were either formed by hybridisation between two apomictic types, or an apomictic type was the mother in those hybrids coming from the crosses between apomictic and sexual plants. This indicates that residual sexuality is an important factor shaping the structure of an agamic complex. 2 In both the yellow and orange-flowered parts of the complex, there are features common to the hybrids between sexual P. officinarum and apomictic species ( P . floribunda, P. aurantiaca ), forming morphologically different types. Hybrids closer to the apomictic parent are morphologically and karyologically uniform, while others closer to the sexual parent are more variable with respect to morphology, cytotypes, clonal structure and, in the case of P. piloselliflora, also in breeding system. 3 Hybrids which are morphologically closer to the apomictic parent have always had apomictic type as mother. Hybrids closer to the sexual parent ( P . officinarum) are variable with respect to cp-haplotype; it indicates that either parent served as a mother plant. STUDY OF PILOSELLA 21 COMPARISON WITH THE SUMAVA MOUNTAINS. The set of basic species involved is in general the same as in the Krkonose Mts; P. onegensis and P. vaillantii are absent, but both are so rare that they do not influence the shape of the complex. We present here only data about presence and frequency of hybrids (Figs 2 & 3 (right)). In spite of the fact that the basic species are the same, there are different hybrids occurring with different frequency. Some of them originated recently (we found only individual plants of P. rubra and P. stolonijlora ), and some others are common (P. scandi- navica). Some hybrids are present in only one mountain range. It seems that the hybridisa- V tion of some species in the Sumava Mts is rare, as for example that between P. floribunda and P. officinarum. The rise of the successful (with respect to its abundance and widespread distribution) P. iserana in the Krkonose area led to its subsequent hybridisation with - P. officinarum and formation of many cytotypes and genotypes of P. piloselliflora. In spite of the common occurrence of P. aurantiaca and P. auricula there is nothing resembling V P. blyttiana in the Sumava Mts. On the other hand, P. aurantiaca hybridises here with P. floribunda and P. scandinavica; these hybrids are absent in the Krkonose Mts. The comparison of the two mountains is summarized in the following points: 1 The same set of basic species forms a different pattern of hybrids in particular regions. 2 The different age of hybrids is reflected in their different distribution and frequency. 3 The same hybridogenous type (with regard to morphology, ploidy level, breeding system and genome constitution) can be an established and widespread in one region and a very rare recent hybrid in another region (e.g. P. rubra in the Krkonose Mts and P. scandinavi¬ ca in the Sumava Mts behave as established hybridogenous species). 4 Evidently, some rare hybridisation events in the past shaped the whole hybridogenous complex. 5 Some hybrids are usually unique, confined to individual localities. They represent with high probability recent hybrids. CONCLUSIONS FOR TAXONOMIC TREATMENT AND DETERMINATION OF PLANTS In our opinion the total population of the genus growing together in one geographic region should be considered as a whole. Firstly, the basic species should be recognized. The next step should be the recognition of the common plants of hybridogenous origin, which may shape the formation of other (secondary) hybrids. The determination of all plants collected in the field is often impossible without special analysis of their genetic markers, because progeny of some hybridogenous types is extremely variable and in different regions different types survive. The complex pattern within Pilosella has attracted many taxonomists in the past. In our opinion our approach might help to support some of the existing philosophies. There are at least four different approaches that are still alive in Europe (Schuhwerk 2002): One originated on the European continent and was summarized by the monumental book by Nageli & Peter (1885), followed and further developed by Zahn (1922-1930); it is in general followed in most national floras in Central Europe: e.g. by Nyarady (1965), Gottschlich (1998), Brautigam & Schuhwerk (2002), Schuhwerk & Fischer (2003), Chrtek in Kubat et al. (2002), Mirek et al. (2002). These authors distinguished basic species and intermediate ones, the latter being labelled by a formula indicating the origin and the quantitative influence of its particular parents. In addition, there is a rich structure of 22 CURRENT TAXONOMIC RESEARCH infraspecific taxa, treated as grex (group of subspecies), subspecies, and variety. Recently, some authors have stressed (e.g. Schuhwerk 2002) that some of the hybrids are old, fully established types behaving as independent species with their own distribution area, but others are recent hybrids. The second approach was developed in Scandinavia (Sweden, Finland) and Russia and is still used by Russian authors (e.g. Schlyakov 1989). They described each distinguishable entity as microspecies and paid little attention to possible relationships between these microspecies. The third approach, developed by British authors Sell and West was used e.g. in Flora Europaea (Sell & West 1976). They considered as species the main species of continental botanists (with three exceptions, they included also H. flagellare, H. sphaerocephalum and H. vahlianum usually considered as hybrids in Central Europe). All other types were evaluated as hybrids, those having the same parental combination being synonymized. This approach fully corresponds to the ICBN. The resulting structure is simple and understand¬ able, even for non-specialists. The main difficulty of this approach is the fact that in some regions only hybridogenous types occur. Tyler (2001) has recently developed the fourth approach. It is based on the knowledge of Scandinavian types and on the fact of common hybridisation within Pilosella species. He formulated several rules (cf. Tyler 2001 : 67, Schuhwerk 2002) to define levels of the species, subspecies and variety. Based on them he created systems involving both main and hybri¬ dogenous species under one species name. This approach substantially reduced the number of species, especially for Scandinavia, but complexity is still reflected at infraspecific level. Our results reject the possibility that all distinguishable types can be classified. The number is too high to produce any reasonable system. They are repeatedly formed, because even in apomicts the degree of residual sexuality is not negligible. The approach by Tyler (2001) reflects this high degree of hybridisation; however, our main objection to his system of subspecies is based on the facts demonstrated above. He overemphasized his Scandinavian experience and lumped together those species which often hybridise in Scandinavia. But if we use his approach consistently, it would be necessary to develop different and incompatible systems for different regions. The frequency of hybridi¬ sation is evidently influenced by the degree of sexuality (including the presence of fully sexual types), compatibility of co-occurring cytotypes, availability of suitable habitats, and some random events in the past. For those reasons we consider the Tyler's approach as not suitable. The differences of the last two approaches are not so big as might appear. Both approaches use the corresponding categories: the basic species of Zahn and his successors correspond to species used by Sell and West. The main difference is in the treatment of intermediates of hybrid origin. Our experience showed that at least in some complexes it is possible to distinguish two types of hybrids, each closer to one of the parents. Further splitting seems impossible, because of high number of coexisting types, either within one region or in different regions. In fact, this approach has recently been accepted by Brautigam & Schuhwerk (2002). Within the yellow-flowered group, whose structure we have shown, they included P. apatelia {H. apatelium ) within P. piloselliflora. This is recommended because of continuous variation between H. apatelium and H. piloselliflorum , due to the high number of genotypes and variable breeding systems producing new types. The same seems to hold for the hybrids between P. aurantiaca and P. officinarum : P. rubra has low variation and also in other regions it is present as a hexaploid type (Schuhwerk & Lippert 1997, our STUDY OF PILOSELLA 23 unpublished data from the Sumava Mts). On the other hand, P. stoloniflora represents a set of different hybrids with a higher influence from P. officinarum (both primary hybrids and backcrosses of P. rubra). We realize that this approach is not fully consistent with ICBN, because it is difficult to decide whether a particular type is a hybridogenous species or a hybrid. The main difficulty of this approach is that in other regions the hybridising species can differ in their ploidy level and other features and so the result of the same hybridisation can seem different. As an example we can use the occurrence of a hybrid similar to P. iserana in the Sumava Mts; this plant is a hexaploid and evidently has a different genomic constitution in comparison with the tetraploid P. iserana from the Sudeten. Evidently, the use of this approach would lead again to the formation of geographically limited systems, but these systems could be compatible provided they use the same basic species as a firm framework. Such systems are useful in some regions, as for Central Europe. Schuhwerk & ' Fischer (2002-2003) recently produced an excellent account for Austria. Our experience from experimental hybridisation supports the Sell & West approach: hybrids differing in the proportion of parental genomes originate even from one cross, combining reduced and unreduced gametes (Krahulcova & Krahulec 2000, Krahulcova et al. 2004). Consequently, both approaches may be combined provided that they use the same species as a stable framework. Evidently, we need more data about the structure of individual hybridogenous complex¬ es from different regions. The present situation, when only limited number of regions (and species complexes) was studied in detail, does not allow us to make any final generalization. However, we think that it is necessary to define the basic species and their distribution areas. Even this process does not seem to have reached finality: e.g. diploid P. onegensis, which has an extensive distribution area from Central and southeastern Europe to western Siberia, is usually considered as a subspecies of P. caespitosa. Similarly, there are no fully clear relationships between P. officinarum and its diploid relatives such as P. hoppeana , P. macrantha , P. peleteriana and especially diploids occurring in southern Europe. Even the relationships between these diploids are unclear: e.g. P. macrantha is sometimes evaluated as a species, and sometimes as a subspecies of P. hoppeana (subsp. testimonial is). This definition of a set of basic species can form a firm framework for the development of a consistent system of hybrids, which can be used by non-specialists. ACKNOWLEDGEMENTS We would like to express our thanks to the Grant Agency of the Academy of Sciences of the Czech Republic (grants No. A6005803 and IAA6005203 for A.K. and F.K.) for financing our research. The field research was supported especially by the administrations of the V Krkonose and Sumava National Parks, who allowed us to do detailed studies in their areas. Work in the area of the Sumava Mts was stimulated and supported by F. Prochazka; Judith Fehrer started and carried out molecular studies; J. Wild provided the map presented. Their help is deeply appreciated. Many participants at the annual Hieracium workshops are acknowledged for stimulating discussions, especially those with S. Brautigam (Gorlitz), J. Chrtek (Pruhonice), F. Schuhwerk (Munich) and Z. Skala (Praha) about plant identifica¬ tion and taxonomic concepts. 24 CURRENT TAXONOMIC RESEARCH REFERENCES BrAutigam, S. & SCHUHWERK, F. (2002). Hieracium L. - Habichtskraut. In: E.J. Jager & K. WERNER (eds), Exkursionsflora von Deutschland. Gefasspflanzen: Kr ids cher Band, p. 709- 734. Chapman, H., Houliston, G.J., Robson, B. & Iline I. (2003). A case of reversal: The evolution and maintenance of sexuals from parthenogenetic clones in Hieracium pilosella. Int. J. Plant. Sci. 164: 719-728. CHRTEK, J. JUN. (2002). Hieracium. In: K. KubAt et al. (eds), Klic ke kvetene Ceske republiky, Academia, Praha, pp. 206-232. Fehrer, J., Simek, R., KrahulcovA, A., Krahulec, F., Chrtek. J., BrAutigam, E. & BrAutigam, S. (2005). Evolution, hybridization, and clonal distribution of apo- and amphimictic species of Hieracium subgen. Pilosella (Asteraceae: Lactuceae) in a Central European mountain range. In: F.T. Barker, L.W. Chatrou, B. Gravendeel & P.B. Pelser: Plant species-level systematics: patterns, processes and new applications. Regnum Vegetabile 143: 175-201. Koeltz, Konigstein. Gottschlich, G. (1998). Hieracium. In: R. WlSSKlRCHEN & H. HAEUPLER, Standartliste der Farn und Bliitenpflanzen Deutschlands, pp. 245-263. Kashin, A.S. et al. (1999). Speciation potential of the Pilosella (Asteraceae) agamic complex. 1. Base species. Bot. Zhurn. 84(4): 25-38. Kashin, A.S., Zalesnaya, S.V., Titovets, V.V. & Kireev, E.A. (2000). Speciation potential of the Pilosella (Asteraceae) agamic complex. 2. Natural interspecies hybridisation. Bot. Zhurn. 85(3): 1-13. Kashin, A. S., Zalesnaya, S. V. & Titovets, V.V. (2000). Speciation potential of the Pilosella (Asteraceae) agamic complex. 3. Genomic variability in populations and in separate plant progenies, ito A Zhurn. 85(12): 13-28. KrahulcovA, A. & Krahulec, F. (1999). Chromosome numbers and reproductive systems in selected representatives of Hieracium subgen. Pilosella in the Krkonose Mts (the Sudeten Mts). Preslia , 71: 217-234. KrahulcovA, A. & Krahulec, F. (2000). Offspring diversity in Hieracium subgen. Pilosella (Asteraceae): New cytotypes from hybridisation experiments. 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. KrahulcovA, A., Krahulec, F. & Chrtek, J. (2001). Chromosome numbers and reproductive systems in selected representatives of Hieracium subgen. Pilosella in the Krkonose Mts (the Sudeten Mts) - 2. Preslia 73: 1 93-2 1 1 . KrahulcovA, A., PapouskovA, S. & Krahulec, F. (2004). Reproduction mode in the allopolyploid facultatively apomictic hawkweed Hieracium rubrum (Asteraceae, H. subgen. Pilosella). Hereditas 141: 1-12. Krahulec, F., KrahulcovA A., Fehrer J., BrAutigam S., PlackovA I. & Chrtek J. JUN. (2004). The sudetic group of Hieracium subgen. Pilosella in the Krkonose Mts: a synthetic view, Preslia 76: 223-243 Mendel, G. (1869). Ueber einige aus kunstliche Befruchtung gewonnenen Hi eraci wn-Bastarde. Verb. Naturf. Vereins Briinn 8: 26-31. Mirek, Z., Mirkowa-Pi^kotS, H., Zajac, A. & ZAJAC, M. (eds), (2002). Flowering plants and pteridophytes of Poland. A checklist , Krakow. STUDY OF PILOSELLA 25 NAGELI, C. VON & PETER, A. (1885). Die Hieracien Mittel-Europas. Monographische Bearbeitung der Piloselloiden mit besonderer Berucksichtigung der mitteleuropaischen Sippen. Miinchen. NyArAdy, E.I. (ed.) (1965). Flora Republicii populare Romine X. Bucure^ti. Richards, A.J., Kirschner, J., StepAnek, J. & Marhold, K., (eds) (1996). Apomixis and taxonomy. Opulus Press, Uppsala. SCHNEIDER, G. (1888-1895): Die Hieracien der Westsudeten. Das Riesengebirge in Wort u. Bild , Trautenau, Marschendorf, 8: 75-80, 110-114 (1888), 9: 18-24, 54-59, 83-100, 147-152 (1889), 10: 21-31, 69-71 (1890), 11(1-2): 30-35 (1891), 11(3-4): 21-28 (1891), 12: 23-25, 65-68 (1892), 13(1-2): 20-23 (1893), 13(3-4): 20-29 (1893), 14: 21-28, 65-69 (1894), 15: 17-21 (1895). SCHLYAKOV, R.N. (1989). Rod 33. Yastrabinocka — Pilosella Hill. In: N.N. TSVELEV (ed.), Flora Evropeiskoi casti SSSR 8, Nauka, Leningrad, pp. 300-377. SCHUHWERK, F. (2002). Some thoughts on the taxonomy of Hieracium. Ber. Bayer. Bot. Ges. 72: 193-198. SCHUHWERK, F. & FISCHER, M.A. (2003). Bestimmungsschliissel der Untergattung Hieracium subg. Pilosella in Osterreich und Siidtirol. Neilreichia 2-3: 13-58. SCHUHWERK, F. & LlPPERT, W. (1997). Chromosomenzahlen von Hieracium ( Compositae , Lactuceae) Teil 1. Sendtnera 4: 181-206. SELL, P.D. & WEST, C. (1976). Hieracium. In: T.G. TUTIN et al. (eds), Flora Europaea 4, Cambridge University Press, Cambridge, London, etc., pp. 358^110. Tyler, T. (2001). The genus Pilosella in the Nordic countries. Svensk. Bot. Tidskr. 95: 39-67. ZAHN, K.H. (1922-1930). Hieracium. In: P. ASCHERSON & P. Graebner (eds), Synopsis der mitteleuropaischen Flora 12(1), Leipzig. APPENDIX The Pilosella species studied with their equivalents in Hieracium P. aurantiaca (L.) L.W. & C.H. Schultz H. aurantiacum L. P. aurantiaca x P. floribunda H. aurantiacum x H. floribundum P. auricula (L.) F.W. & C.H. Schultz H. lactucella Wallr. P. blyttiana (Pries) F.W. & C.H. Schultz H. blyttianum Fries P. caespitosa (Dumort.) Sell & West H. caespitosum Dumort P. floribunda (Wimmer & Grabowski) Arv.-Touv. H. floribundum Wimmer & Grabowski P. fuscoatra (Nageli & Peter) Sojak P. glomerata (Froel. in DC) Arv.-Tour. P. macranthela (Nageli & Peter) Sojak P. officinarum F.W. & C.H. Schultz P. one gens is (Norrl.) Norrl. P. iserana (Uechtr.) Sojak P. piloselliflora (Nageli & Peter) Sojak P. rubra (Peter) Sojak H. fuscoatrum Nageli & Peter H. glomeratum Froel. in DC. H. macranthelum Nageli & Peter H. pilosella L. H. onegense Norrl. H. iseranum Uechtr. H. piloselliflorum Nageli & Peter H. rubrum Peter H. scandinavicum Dahlst. H. schultesii F. Schultz P. scandinavica (Dahlst.) R.N. Schlyakov P. schultesii (F. Schultz) F.W. & C.H. Schultz P. stoloniflora (Waldst. & Kit.) F.W. & C.H. Schultz H. stoloniflorum Waldst. & Kit. P. tubulascens Norrl. H. tubulascens Norrl. P. vaillantii (Tausch) Sojak H. cymosum L. subsp. cymigerum (Reichenb.) Peter Current taxonomic research on the British & European flora Bailey, J. & Ellis, R.G. (eds) 27^4 (2006), BSBI. London. 27 Taxonomic complexity, conservation and recurrent origins of self- pollination in Epipactis (Orchidaceae) P.M. Hollingsworth, J. Squirrell, M.L. Hollingsworth Royal Botanic Garden, 20A Inverleith Row, Edinburgh, EH3 5LR, UK and A.J. Richards School of Biology, University of Newcastle-upon-Tyne, NE1 7RU, UK and R.M. Bateman Jodrell Laboratory, Royal Botanic Gardens Kew, Richmond, Surrey, TW9 OED, UK ABSTRACT The recent post-glacial colonisation of Britain has given little time for the evolution of endemic plant species. The few British endemic species that have been recognised tend to occur in taxonomically complex groups that possess mechanisms promoting rapid diversi¬ fication. Such taxonomic complexity leads to problems for plant conservation because a species has to be circumscribed and recognised before its distribution, status and threats can be established. One classic example of the challenges for conservation in taxonomical¬ ly complex groups is the British endemic orchid Epipactis youngiana. This species is afforded full legal conservation protection but is one of a large number of taxonomically difficult species recently recognised in the genus; it is difficult to distinguish from the more common E. helleborine, and its species status has been questioned. We have used a combination of genetic markers (allozymes, chloroplast microsatellites and RFLPs) from a large sample set to establish the taxonomic and conservation status of E. youngiana and to place it in the wider context of patterns of breeding system variation and taxon differenti¬ ation in the genus. Our data provide evidence that recurrent breeding system transitions between cross-pollination and self-pollination are an important mechanism for diversifica¬ tion in the genus, and there are numerous genetically different taxa that are homozygous and uniform for different subsets of allelic diversity found in allogamous species. E. youngiana is the one major exception to this pattern. It has a floral morphology consistent with self-pollination, but has not achieved reproductive isolation from E. helleborine. The potential mechanisms underlying the recurrent evolution of self- pollination in Epipactis are discussed, as is the need for developing conservation strategies that reflect dynamic diversification in the genus, rather than the current heavily typological (is it distinct or not?) species-based approach. Keywords: allozymes, cpSSRs, endemic, Epipactis, genetic markers, RFLPs, self-pollination, taxo¬ nomic complexity. 28 CURRENT TAXONOMIC RESEARCH INTRODUCTION The British flora has benefited from a long history of floristic study (Clapham et al. 1989; Stace 1991, 1997), two major mapping exercises based on over 9 million records (Preston et al. 2002), various accounts of species of conservation importance (Stewart et al. 1994; Wigginton 1999), detailed biological floras of individual species, a comprehensive evalua¬ tion of chromosome number variation, a vice comital flora (Stace et al. 2003), and an overview of hybridisation in the flora (Stace 1975). This enormous resource base provides an excellent foundation for taxonomic, ecological, evolutionary and conservation research. Building on this resource base, the delimitation and identification of many species in the British flora is now considered routine. The recognition and description of clear-cut morpho¬ logical differences between species has effectively led to a consensus on the most appropri¬ ate taxonomic treatment for a large number of genera. There are, however, some persistently troublesome groups of plants that defy attempts to achieve a widely accepted taxonomic treatment. These taxonomically complex groups contain a large proportion of Britain’s endemic higher plant species (e.g. Sorbus , Epipactis, Euphrasia , Ulmus , Hieracium , Tarax¬ acum , Rubus , Limonium). Taxonomic complexities which affect the most appropriate delimitation of species cause problems for conservation. Making an assessment of the distribution and conservation status of any species first requires that the species can be delimited and recognised. If the unit to be conserved is in a taxonomically complex group, there can be major problems in assessing threats, devising conservation strategies and monitoring their success. As taxonomically complex groups account for almost half of the species on the UK Biodiversity Action Plan ‘short list’, there are real difficulties in implementing conservation actions for these species (Hollingsworth 2003). The association between endemism and taxonomic complexity in the British flora is at least partly attributable to recent ice ages. The vast majority of species in the British flora have achieved their current distributions via recent post-glacial colonisation. The limited time period since colonisation (within the last 12,000 years) has given little time for the evolution of endemic species. Those endemic species that are recognised within the British flora typically show a mechanism, or combination of mechanisms that promote rapid biological diversification: these include polyploidy, hybridisation, self-pollination, agamos- permy and clonal growth (Stace 1989, 1997). In a recent review of the history of the North Atlantic biota, Brochmann et al. (2003) noted that among the 43 hardy vascular plant species accepted as being endemic to the region, there was not a single sexual diploid indicative of long-term evolution. Similarly, mechanisms which promote rapid diversification have been central to the evolution of the endemic element in the British flora; this rapid diversification also results in taxonomic complexity and knock-on problems for conservation. MECHANISMS UNDERLYING TAXONOMIC COMPLEXITY One mechanism that promotes rapid diversification and also leads to taxonomic complexity is a change in breeding system. The evolution of self-pollination (autogamy) from outcross¬ ing (allogamy) is one. of the most frequent evolutionary transitions in flowering plants (Stebbins 1974). This change can lead to a neospecies achieving rapid reproductive isolation from its progenitor (Levin 2000), and can generate morphological differences between selfers and their allogamous progenitors due to an increased level of homozygosity. Morpho- TAXONOMIC COMPLEXITY IN EPIPACTIS 29 logical characters that include phenotypes encoded by recessive alleles can be selected for, as self-pollination can lead to the fixation of advantageous recessive mutations (Charlesworth 1992, Levin 2000). As the age of the self-pollinating lineage increases, novel mutations can result in further morphological differences. Self-pollinating species typically show higher levels of variation between populations than do outcrossing species (Hamrick & Godt 1996, Nybom 2004). This is attributable to reduced opportunities for gene flow and also potentially an association between self-pollina¬ tion and short-lived and/or colonist populations, which show rapid differentiation as a consequence of repeated founder events and genetic drift. There is thus the opportunity for morphological divergence between populations within a selfing species. In addition, there is the possibility that outcrossing progenitor species may generate multiple independent selfing lineages of morphologically similar appearance. Both of these scenarios yield sets of - populations separated only by subtle morphological differences which can result in difficul¬ ties in the allocation of the most appropriate taxonomic rank and affiliation. TAXONOMIC COMPLEXITY IN EPIPACTIS The genus Epipactis is a classic example of a plant group in which active diversification and breeding system transitions appear to have led to taxonomic complexity and subsequent problems for conservation. The genus consists of between 25 and 60 species. They exhibit a predominantly Eurasian distribution with outlying species in North America and North Africa (Bateman et ah 2005). The taxonomic complexity in the genus is reflected in the uncertainty regarding species numbers, and for instance the number of European species accepted by Delforge changed from 36 to 58 in seven years (cf. Delforge 1995, 2001), whereas Sundermann (1970) recognised only 14 species. anther cap covering pollinia Figure 1. Floral morphology of an allogamous Epipactis flower. Drawing by Mary Mendum 30 CURRENT TAXONOMIC RESEARCH The floral morphology of species in the genus varies considerably. The two major informal groups in the genus are the ‘ palustris ’ group (rhizomatous clonal species, with an open concave hypochile) and the ‘helleborine’ group which are typically more vegetatively discrete and possess a more cupped hypochile. The ‘helleborine’ group (the subject of this paper) encom¬ passes two contrasting floral morphologies. Many species, including the widespread E. helle- borine have a well-developed rostellum that serves as a barrier between the male and female parts of the flower (Fig. 1). The viscidium is exuded from the tip of the rostellum (Richards 1982) and serves as a glue to bind the pollinia to visiting insects. The vast majority of taxonomic problems in the genus are centred around a large number of named taxa in which the viscidium and rostellum are strongly reduced or are less persistent. Any reduction in the rostellum diminishes the physical barrier between the male and female parts of the flower, while any reduction in the viscidium lowers the likelihood of the pollinia being transferred to flowers on a different plant. Instead, the pollinia can remain in situ and fall downward onto the stigmatic surface, resulting in self-pollination. A series of changes such as a reduction in flower size, a more pendulous habit, and reduction in pigmentation have been invoked as secondary modifications which further promote self-pollination (Richards 1982). Taxa with a floral morphology consistent with self-pollination are typically of restricted distribution, and can show limited variation within populations but subtle and consistent morphological differences between populations. At the heart of taxonomic complexity in Epipactis is the extent to which populations with floral morphologies such as these reflect minor mutational variants of single self-pollinating lineages, or whether they reflect inde¬ pendent taxa resulting from iterative allogamy-to-autogamy transitions. In Britain seven species were recognised by Stace (1997). Four of these have an outcross¬ ing type of floral morphology: E. palustris, E. atrorubens, E. helleborine and E. purpurata and three have a floral morphology consistent with autogamy: E. phyllanthes, E. leptochila and E. youngiana. Molecular data have shown that within the variation encompassed by E. leptochila there are two clearly distinct taxa best treated separately as E. dunensis (T. Stephenson & T.A. Stephenson) Godfery and E. leptochila s.s. (Godfery) Godfery, and also a third lineage endemic to Lindisfame (Squirrell et al. 2002) which has subsequently been named E. sancta (P. Delforge) P. Delforge (Delforge & Gevaudan 2002; Bateman 2006). EPIPACTIS YOUNGIANA AS A CASE STUDY One British Epipactis species that remains enigmatic is the high conservation profile endem¬ ic E. youngiana A. J. Richards & D. F. Porter. This species was first described in 1982 and was found growing on some mine spoil heaps in Northumberland (Richards & Porter 1982). It has subsequently been recorded from some similar habitats in Scotland, centred around Glasgow and Falkirk (Dickson et al. 2000). It has a floral morphology consistent with self-pollination; in particular its viscidium withers very rapidly. Its presence predominantly in anthropogenic sites has led to suggestions of a recent evolutionary origin stemming from either (a) divergence from E. helleborine or (b) a hybrid origin between E. helleborine and E. dunensis. Self-pollination is considered to have led to rapid reproductive isolation from other Epipactis species. As an endemic British orchid species (one of very few endemic plant species recognised in Britain), E. youngiana has been afforded full conservation status under Schedule 8 of the Wildlife and Countryside Act (1981). The species also has a Species Action Plan as part of the UK Biodiversity Action plans (UK Biodiversity Group 1995), and associated allocation of conservation resources. However, populations of E. youngiana occur TAXONOMIC COMPLEXITY IN EPIPACTIS 31 sympatrically with plants of E. helleborine (and also at some sites with either E. phyllanthes or plants resembling E. dunensis). The mixture of taxa growing together and the presence of plants of intermediate morphologies has caused difficulties in identification of material and uncertainty over the taxonomic status of E. youngiana. Thus E. youngiana receives formal conservation protection, but conservation actions are extremely difficult to implement due to taxonomic uncertainty. To provide some insights into the taxonomic distinctness of E. youngiana , and more generally into the evolutionary processes underlying taxonomic complexity in Epipactis, we have carried out large-scale genetic surveys using a combination of sequencing of the internal transcribed spacers (ITS) of nuclear ribosomal DNA and the chloroplast trnL intron, allozyme analyses, and screening of RFLP variation and mononucleotide length variation in the trnL intron. The full details of this work will be published elsewhere; our goals here are - to explore the taxonomic status of E. youngiana and to infer the most appropriate conserva¬ tion treatment of this putative species. MATERIALS AND METHODS A total of 2828 individuals of 164 populations of 23 named species in the ‘helleborine' group have been examined using a range of genetic markers (Table 1). The species were selected to encompass a range of putative species including several with a floral morphology consist¬ ent with self-pollination and several with a floral morphology consistent with outcrossing. The sampling of E. youngiana involved two populations from Northumberland and two from Scotland. Sympatric plants of E. helleborine (and E. dunensis and E. phyllanthes where present) were also sampled from these sites. These samples have been examined using a range of molecular approaches (allozyme electrophoresis, chloroplast and ITS sequencing, chloroplast microsatellites and RFLPs). Between nine and ten allozyme loci were screened, eight of which were polymorphic. At the polymorphic loci there were between 2-4 alleles. The chloroplast data set considered here consists of information from the trnL intron. Based on sequences of 86 individuals, two variable markers were selected for widescale screening. Firstly, we screened for the presence or absence of a 10 bp duplication in the trnL intron via RFLPs, and secondly we examined length polymorphism in a chloroplast microsatellite that consisted of a mononucleotide poly-A repeat of between (A)9 and (A) 13 bp. Sequencing of the nuclear ITS regions from 63 individuals revealed phylogenetic resolution at deeper levels within the genus, but within the taxonomically complex E. helleborine group little variation was detected. Consequently information from ITS is not discussed further in this paper. RESULTS AND DISCUSSION PATTERNS OF ALLOZYME VARIATION The Epipactis species examined here show some marked differences in the amounts and organisation of genetic variation among individuals and populations (Table 2). With the exception of plants from the E. youngiana sites (which are discussed later), all of the species examined with a floral morphology consistent with self-pollination consisted of homozygous and genetically uniform lines. No heterozygous individuals were detected in the total dataset TABLE 1. EPIPA CTIS T AX A SAMPLED, DISTRIBUTIONS, FLORAL MORPHOLOGY, SAMPLE LOCATIONS, AND SAMPLE SIZES FOR DIFFERENT MOLECULAR ASSAYS. Floral morphology designation and geographical range is based on field observations and Delforge (2001). n = sample size for different assays. 32 CURRENT TAXONOMIC RESEARCH TAXONOMIC COMPLEXITY IN EPIPACTIS 33 for the examined loci. Although this uniformity precludes formal breeding system estimates, it is at the very least strongly congruent with self-pollination leading to homozygosity and uniformity. In contrast, all of the species with a floral morphology consistent with outcrossing exhibited at least some genetic variation and some heterozygosity (Table 2). The inbreeding coefficient in all of these species was not significantly different from zero, and thus was consistent with random mating. The one minor exception was E. atrorubens ; despite hetero¬ zygous individuals being detected, there was a small but significant deficit of heterozygotes compared with Hardy- Weinberg expectations (Fis = 0.180; Table 3). It is not clear why this species should show some slight heterozygote deficit when the other allogamous species do not. Possible explanations are that these populations experience higher levels of geitonogamy or lower pollinator availability than the other species. However, aside from this minor exception, there is a marked correlation between floral morphology and patterns of genetic variation, and these results confirm the importance of minor modifications in floral morpholo¬ gy as determinants of the organisation of genetic variation (Richards 1982). However, floral morphology is not the only determinant of patterns of allozyme varia¬ tion in these species. Although there is homogeneity in the variation patterns within the putative self-pollinating taxa (all taxa show zero within-species variation for these loci), the situation is more varied among the allogams (Table 3). Levels of variation ranged from 1 1% to 56% of allozyme loci being variable within populations, and although all populations of most species showed some variation, in E. purpurata only 47% of sampled populations were variable (Table 3). This suggests that other factors, such as population size, plant size, demographic history and pollinator activity are also likely to be important determinants of the amounts and partitioning of genetic variation. Considering the range of allelic variation between the species with outcrossing and selfmg floral morphologies, for the most part the different selfmg lineages are fixed for different character combinations that represent subsets of the alleles found in the allogams (Table 2). This pattern is consistent with multiple transitions from outcrossing to selfmg, giving rise to a series of genetically and morphologically discrete lineages, rather than the taxonomic complexity in the genus stemming from a single outcrossing-selfmg transition followed by mutational divergence. Thus the transition from allogamy to autogamy seems particularly labile in Epipactis. Of course, multiple origins of selfmg and mutational diver¬ gence of lineages are not mutually exclusive and following independent origins there is the possibility for further divergence. Although this aspect of the work is beyond the scope of this paper and will be explored elsewhere, it is worth noting that there is evidence for some alleles present in autogams that have not been found in allogams, and also for between- population divergence in autogamous taxa (Table 2). For example, the inland populations of E. dunensis differ genetically from the coastal populations (Table 2; see also Squirrell et al 2002) and there are further differences between populations of E. microphylla for the rapidly evolving cpSSRs locus (Table 2). In summary, genetic variability and heterozygosity are found within putative outcross¬ ing taxa (albeit with differences in absolute levels of diversity). In contrast, the autogamous lineages are united in their allozyme uniformity and complete lack of within-population variation. How does the British endemic E. youngiana match these patterns of variability, and does the evidence support the notion that it is a distinct, recently evolved species that has achieved reproductive isolation via the evolution of autogamy? 34 CURRENT TAXONOMIC RESEARCH TABLE 2. THE DISTRIBUTION OF ALLOZYME AND CHLOROPLAST VARIANTS WITHIN EUROPEAN EPIPACTIS SPECIES. (See top of following page for details) Epipactis species Allozyme loci Chloroplast ( trnL ) Autogams mdh-1 mdh-2 idh-1 pgm aat-1 pgi-1 Pgd idh-2 Duplication SSR no. present repeats confusa 1 c c a c a b a a No 9 phyllanthes 1 c c a c a b a a No 9 albensis b a a a a c a Yes 10 campeadorii 2 b c a b a c a No 10 dunensis (w coastal) b c a b a c a Yes 11 dunensis (inland) b c a b a c a Yes 10 fibri b b a a c a a Yes 10 futakii b b a a a c a No 9 leptochila b b a a a c a No 11 microphylla a d a a a b a No 12/133 muelleri a b b a b b a No 10 peitzii a b a a a c a Yes 12 placentina a b a a b b a Yes 9 provincialis b b a a a a a No 11 pseudopurpurata a b a a b a a No rhodanensis b c a a a c a No 11 sancta2 b c a b a c a No 10 Allogams atrorubens ab abc acd abc abc abc a No 9 distans be b a a a c a No 11 helleborine ab abc abc ab abed abc ab No & Yes 9/10/11 parviflora a ab ad a a ab a No puvpurata ab be ab ab abc abc a Yes 10/11 tremolsii ab abc abd ab abc abc a No 10 Settlingstones • helleborine ab abc a ab abc abc a No & Yes 10/11 youngiana ab abc a ab abc abc a No & Yes 10/11 phyllanthes a c a c a b a a No 9 Bardykes Bing helleborine ab be ab ab abc abc a Yes 9/10 youngiana ab be ab ab abc abc a No & Yes 9/10 Bothwell Castle helleborine ab be ab ab abc abc a No & Yes 10/11 youngiana ab be ab ab ab abc a No & Yes 9/10/11 dunenesis 4 ab be ab ab a be a Yes 10/11 Gosforth Park helleborine ab abc a ab ab abc a No & Yes 10/11 youngiana ab abc a ab a abc a No & Yes 10/11 TAXONOMIC COMPLEXITY IN EPIPACTIS 35 Table 2 cont. The different letters represent the allelic variants found within each taxon. ]Our genetic data were unable to distinguish between E. confusa and E. phyllanthes; the taxonomic implications of this will be discussed elsewhere. 2The data presented here do not distinguish E. sancta and E. campeadorii 2 but these two species can be distinguished based on their ITS sequences (data not shown). 3Different populations of E. microphylla are fixed for either 12 or 13 repeats at the cpSSR locus. 4Plants with the morphology of E. dunensis at the E. youngiana sites in Scotland do not show the classic homozygous and uniform E. dunensis allozyme profile. TABLE 3. WITHIN-POPULATION DIVERSITY MEASURES AND ESTIMATES OF THE INBREEDING COEFFI¬ CIENT IN ALLOGAMOUS EUROPEAN EPIPACTIS SPECIES BASED ON NINE ALLOZYME LOCI Species N pops Mean n P A F\s He PP E. atrorubens 9 17.8 29 1.37 0.180* 0.120 100 E. distans 1 19.0 11 1.11 0.390ns 0.038 100 E. parviflora 2 5.0 12 1.12 -0.148ns 0.056 100 E. purpurata 18 11.5 11 1.12 0.033ns 0.024 47 E. tremolsii 1 15.2 56 2.00 0.009ns 0.227 100 E. helleborine 47 23.5 56 1.81 0.003ns 0.231 100 N pops = number of populations; Mean n = mean sample size per population per locus; P = % polymorphic loci; A = mean number of alleles per locus; F\s — global inbreeding coefficient; * = significantly different from zero p < 0.05, ns = not significant; He = gene diversity; PP = % of populations that are polymorphic. TABLE 4. WITHIN-POPULATION DIVERSITY MEASURES AND ESTIMATES OF THE INBREEDING COEFFI¬ CIENT IN FOUR SYMPATRIC POPULATIONS OF E. YOUNGIANA AND E. HELLEBORINE IN BRITAIN BASED ON NINE ALLOZYME LOCI Species Region Location n P A Fis E. helleborine Northumberland Settlingstones 30 56 1.89 0.1 10ns Northumberland Gosforth Park 9 56 1.78 0.139ns Glasgow Bardykes Bing 52 67 1.89 0.035ns Glasgow Bothwell Castle 31 67 1.89 -0.059ns E. youngiana Northumberland Settlingstones 36 56 1.89 0.036ns N orthumberland Gosforth Park 8 44 1.67 0.099ns Glasgow Bardykes Bing 21 67 1.89 0.141* Glasgow Bothwell Castle 24 67 1.78 -0.048ns n = sample size, P = % polymorphic loci, A = mean number of alleles per locus, F\ s = inbreeding coefficient, * = significantly different from zero p < 0.05. 36 CURRENT TAXONOMIC RESEARCH IS E. YOUNGIANA AUTOGAMOUS? (NO) E. youngiana typically occurs in sympatry with E. helleborine and sometimes also with other Epipactis species. The mechanism proposed for the development of reproductive isolation and speciation is self-pollination. There is, however, no clear evidence that the plants ascribed to E. youngiana are undergoing self-pollination. Both our studies and those of Harris & Abbott (1997) have recovered high levels of heterozygosity in populations of this species. In three of the four populations examined here, the distribution of alleles among individuals is consistent with random sexual mating (Table 4). Only in the Bardykes Bing population was a (just) significant deviation from Hardy- Weinberg equilibrium detected (p = 0.05), and this was attributable to a very minor deficit of heterozygosity that might represent a sampling artefact (an allozyme survey of the same population by Harris & Abbott (1997) did not detect any significant deviation from Hardy- Weinberg expectations). If complete self-pollination was occurring, the rate of homozygosity should increase by 50% per generation. Even allowing for non-overlapping generations and a recent origin for the species, if self-pollination had been sufficiently extensive to lead to reproductive isolation from sympatric plants of E. helleborine, it would be expected to leave a much stronger signature on the partitioning of allelic variation within and among individuals. IS E. YOUNGIANA DISTINCT? (NO) If E. youngiana is a distinct cohesive species, we should expect allele frequencies in different populations of E. youngiana to be more similar to one another than to local populations of E. helleborine. However, this is not the pattern recovered from either the allozyme data or the cpDNA data. For both chloroplast and nuclear allozyme data, the allelic diversity in E. youngiana and E. helleborine shows greater similarities by site than by taxon (Figs 2, 3). Indeed, if one tests for random mating by pooling individuals of E. helleborine and E. youngiana at each site, at three of the four sites no significant deviation from random mating is detected among individuals between ‘species’ (Fis = 0.062 Bardykes Bing, Fis = -0.036 Bothwell Castle, Fis = 0.172 Gosforth Park; all non-significant). In the fourth population (Settlingstones), pooling individuals between E. helleborine and E. youngiana does result in a statistically significant departure from random mating, but the deviation from panmixia is again minor (Fis = 0.091,/? = 0.043). ARE POPULATIONS OF E. YOUNGIANA NORMAL POPULATIONS OF F. HELLEBORINE THAT HAVE BEEN TAXONOMIC ALLY OVER-SPLIT? (NO) Even the briefest of visits to the populations of F. youngiana in both Northumberland and Scotland reveals a pattern of morphological diversity outwith the norm. The classic ‘ youngiana ’ morphology is not a phenotype that is present in typical popula¬ tions of F. helleborine. There is undoubtedly something unusual about these popula¬ tions which contain atypical mixtures of floral morphologies, ranging from individuals whose floral morphology resembles autogamous plants, to those whose floral morphol¬ ogy resembles outcrossing plants. Given that our data suggests that self-pollination has TAXONOMIC COMPLEXITY IN EPIPACTIS 37 Idh-l b □ mdh- 2 Figure 2. The distribution of chloroplast haplotype variation and allelic variants for two allozyme loci in four sympatric populations of E. youngiana and E. helleborine in Britain. If E. youngiana was a distinct species, populations of this taxon should be more genetically similar than they are to sympatric populations of E. helleborine. Instead, greater generic similarities are found between E. youngiana and the local E. helleborine populations. H = E. helleborine, Y -E. youngiana 38 CURRENT TAXONOMIC RESEARCH youngiana helleborine youngiana helleborine youngiana helleborine youngiana helleborine Bardykes Bing, Glasgow Bardykes Bing, Glasgow Bothwell Castle, Glasgow Bothwell Castle, Glasgow Gosforth Park, Northumberland Gosforth Park, Northumberland Settlingstones, Northumberland Settlingstones, Northumberland Figure 3. UPGMA clustering based on pairwise estimates of population differentiation (Fst) for eight polymorphic allozyme loci from four sympatric populations of E. youngiana and E. helleborine in Britain. Populations of E. youngiana do not form a distinct cohesive genetic entity; instead, each population is more closely related to its local E. helleborine population. arisen from outcrossing taxa on numerous occasions in Epipactis , perhaps it is not surprising to find populations in which there is a mixture of floral morphologies, even if there is as yet no clear divergence of taxa. It is possible that these polymorphic populations represent the type of population from which future autogamous lineages may originate. The apparent absence of self-pollination in E. youngiana , despite the flowers having a selfing floral morphology, may reflect the presence of sympatric allogamous E. helleborine at these sites. Even if pollinia export is reduced in individuals with the E. youngiana morphology, these plants presumably can behave as functional females and receive pollinia import from neighbouring allogamous E. helleborine plants via visiting pollinators. To achieve reproductive isolation, plants with a selfing floral morphology may need to disperse to a site where no allogamous plants occur. ARE THE E. YOUNGIANA POPULATIONS TYPICAL OF TAXONOMIC COMPLEXITY IN EPIPACTIS! (NO) The populations of E. youngiana show a very different pattern of genetic variation from the consistent pattern seen in autogamous Epipactis species (all of which were homozygous and uniform for the allozyme loci considered here; Figure 4b). There is thus a clear difference between E. youngiana and the vast majority of other autogamous lineages recognised at the species level in Epipactis (Figs 4a, b; Table 2). The only close parallel we are aware of is E. renzii Robatsch, a taxon restricted to coastal dunes in Denmark. Based on allozyme electrophoresis, Pedersen & Ehlers (2000) concluded that it had arisen on multiple occasions from local populations of E. helleborine subsp. neerlandica (Verm.) Buttler. Like E. young¬ iana, E. renzii occurs sympatrically with populations of E. helleborine and at individual sites TAXONOMIC COMPLEXITY IN EPIPACTIS 39 the two putative taxa share the same alleles. Unlike E. youngiana, all three populations of E. renzii examined by Pedersen & Ehlers (2000) showed a strongly significant inbreeding coefficient (Fis = 0.486, Fis = 0.832, Fis = 1 .0, all p < 0.001). The authors concluded that the origin of self-pollination was recent and attributable to adaptation for reproductive assurance due to a short flowering season caused by water stress and early wilting of flowers. Based on the absence of extensive molecular and morphological divergence, Pedersen & Ehlers (2000) argued that F. renzii should be given varietal status rather than species status. This apparently intermediate phase represented by E. renzii, between allogamous populations of F. helle- borine and homozygous uniform autogamous segregates is interesting, and a parallel taxo¬ nomic treatment for F. youngiana (varietal status) may be appropriate. This approach was adopted informally by Lang (2004) and Bateman (2006), and a formal transfer was per¬ formed by Kreutz (2004). WHAT IS THE MECHANISM FOR THE RECURRENT EVOLUTION OF A SELFING FLORAL MORPHOLOGY WITHIN EPIPACTIS ? This assessment of patterns of genetic diversity in Epipactis has demonstrated the frequency of transitions in floral morphology from outcrossing to selfing types. Given the lability of this switch, it is worth evaluating the evolutionary processes hypothesised to underlie these transitions. Selection for reproductive assurance under conditions of poor pollinator availa¬ bility and/or a short flowering season provide some explanations for the advantages of self-pollination, but not the mechanism underlying the transitions. Pedersen & Ehlers (2000) argued that recurrent mutations may be responsible and noted that a mutation resulting in paedomorphosis via an arrested development of the rostellum may be the key step required in the evolution of autogamy. Whilst we consider this hypothesis plausible, we also believe that an alternative hypothesis is worth considering: that hybridisation between autogamous and allogamous Epipactis species provides a mechanism for the transfer of genes encoding the selfmg-floral morphology into novel heterozygous backgrounds from which new selfing lineages with new character combinations can arise. Under this scenario the evolution of autogamy could in some cases be considered as a cyclical process more akin to an ‘evolutionary detour’ than the ‘evolutionary dead-end’ proposed by Stebbins (1957). Selfing lineages evolve and differentiate, and at some future point occur in sympatry with alloga¬ mous taxa, hybridise and result in the production of further selfing lineages. As yet there is little evidence to support or refute this ‘evolutionary detour' hypothesis. Evidence from mixed populations of autogamous taxa (e.g. mixed populations of F phyllan- thes and F dunensis on the west coast of England) suggests that these taxa co-exist at high densities without undergoing any gene exchange (Fig. 5). However, where we have sampled populations of putative autogams occurring in sympatry with allogamous taxa in the com¬ plex F. youngiana sites, the pattern changes somewhat: 1 Plants with a morphology consistent with F. dunensis occur in the same complex sites that contain Scottish populations of F youngiana and F helleborine. At these sites, plants with the morphology of F. dunensis are genetically variable and heterozygous for the same alleles found in local populations of F. helleborine and F youngiana (Fig. 4, Table 2), and plants of intermediate morphologies occur. This is a marked contrast with 40 CURRENT TAXONOMIC RESEARCH Fig. 4a HHH HH Y Y Y Y Y H H Y Y Y YYYY 'D' HYDHY HY HY HYD 'D' HY 1c 2c Fig. 4b MDH 1" - ► . . i . i . . . . . . i . . . . . . . . . "i" . £ phySanthes E. rhodanensis £. mueSeri E. mbmphyitaE. leptochia la lb 2b 2d FIGURE 4. Patterns of allozyme diversity in a complex E. youngiana population compared with typical autogamous Epipactis taxa. (a) MDH variation at two loci from a mixed Scottish population (Bothwell Castle) of plants with the morphology of E. youngiana, E. helleborine and E. dunensis showing high levels of heterozygosity for the same set of alleles; H = E. helleborine, Y = E. youngiana, and ‘D’ = plants which resemble morphologically E. dunensis (no plants with the classic E. dunensis molecular genotype have yet been detected in Scotland). Combinations of these letters represent morphological intermediates, (b) MDH variation at two loci showing the classic autogamous genetic signature in Epipactis'. fixed homozygous and uniform allozyme genotypes within species, but fixed differences for different allelic combinations between species. Arrows represent locus (number) and allele (letter) designations as used in Table 2. Unlabelled bands were not scored and are assumed to represent heterodimers and breakdown products. MDH iMt iH 4$^ ^ in§ Up E. dunensis E. phyllanthes Figure 5. Representative MDH variation in plants of E. dunensis and E. phyllanthes from a mixed site on the north-west coast of England, demonstrating the maintenance of clear genetic differences between autogamous species growing in sympatry. TAXONOMIC COMPLEXITY IN EPIPACTIS 41 all other E. dunensis populations we have examined, in which the plants are all genetical¬ ly distinct, homozygous and uniform, and suggests that the plants of E. dunensis- type morphology in the Scottish sites are not ‘pure’. 2 Likewise, at Settlingstones, the local plants of E. phyllanthes possess an MDH allele that is also found in the local E. helleborine and E. youngiana populations. This was absent from a survey of 408 plants from 26 other populations of E. phyllanthes , all of which showed a single uniform allozyme profile. This again may indicate some past hybridisa¬ tion, although E. phyllanthes at this site appears to be morphologically uniform and typical. 3 One additional curious feature of the Settlingstones site is that for both E. helleborine and E. youngiana there is a high frequency of an unusual chloroplast type (82% and 87% respectively: Fig. 2). Although present at similar frequencies in some populations of E. helleborine introduced to North American (Squirrell et al. 2001) this haplotype is typically absent or occurs at a low frequency in British populations (in a survey of ten populations it ranged from a frequency of zero to 9.5%, mean = 1%). This may just be chance, although the coincidence of this atypical marker in a taxonomically complex population also may suggest a genetic signature of past hybridisation. However, if this pattern is due to hybridity, it is not obvious which species was involved (this haplotype is not found in geographically proximal species such as E. dunensis or E. phyllanthes). The only other British species that possesses this chloroplast haplotype is E. leptochila s.s., a species with a much more southerly distribution in the UK, so a hybrid explanation for the unusually high frequency of this marker would require a rather convoluted scenario. Thus the evidence for the evolutionary detour hypothesis is somewhat circumstantial and equivocal. Further research on mixed populations of allogamous and autogamous taxa is required to test the importance of hybridisation as a mechanism underlying the recurrent origins of self-pollination. However, it does at least seem plausible that a normally autoga¬ mous species such as E. dunensis may receive insect visits and pollinia import (cf. Richards 1986) if growing in sympatry with an outcrossing species, and thus potentially can serve as a conduit for the transfer of genes encoding selfing floral morphologies into novel hetero¬ zygous backgrounds. IS A SPECIES-BASED APPROACH TO CONSERVATION APPROPRIATE FOR DEALING WITH DIVERSITY IN TAXONOMICALLY COMPLEX GROUPS? Under current species-based conservation programmes, the conservation status of E. young¬ iana should be revised. The genetic data and the extreme difficulties of identifying morpho¬ logical discontinuities in the field all suggest that this does not represent a cohesive, distinct, reproductively isolated species that has stabilised by autogamy. Instead, it is best considered as a series of complex populations that have not currently achieved separate evolutionary trajectories from the sympatric populations of E. helleborine. Given the available evidence, it would in practice be exceedingly difficult to enforce the current legislative conservation protection of this ‘species’ under the Wildlife and Countryside Act. However, it is equally important to note that casually dismissing the conservation value of complex populations like E. youngiana is a simplistic view, and that it typifies a wide¬ spread problem regarding the conservation of taxonomically complex groups in the post¬ glacial flora of Britain. Work on Epipactis has revealed a range of genetically variable allogamous taxa, a range of uniform homozygous lineages with a floral morphology consist- 42 CURRENT TAXONOMIC RESEARCH ent with selfing, and some populations such as E. youngiana that fall someway between the two. This pattern of common variable species, local endemic entities, and morphologically complex populations containing individuals not readily assignable to any discrete taxon is paralleled in other actively diversifying taxonomically complex groups such as Euphrasia (French 2004, French et al. 2005) and Sorbus (Robertson et al. 2004). Taxonomic complex¬ ity, recent/ongoing diversification and endemism are all tightly associated in the British flora, and indeed in that of the broader North Atlantic region. If diversification is ongoing, one should not expect all diversity to fall conveniently into neat discrete packages. Therefore it seems appropriate to develop conservation strategies that encompass the broad range of diversity and evolutionary processes in these groups, rather than focusing attention and resources entirely on the taxonomic status of a fraction of this diversity. An alternative conservation goal for taxonomically complex groups would be to develop conservation strategies aimed at the diversification process itself, which recognise the value of all ele¬ ments in the system such as progenitor species, endemics and taxonomically complex sites (Hollingsworth 2003, Ennos et al. 2005). ACKNOWLEDGEMENTS We are delighted to contribute this paper to a publication marking Clive Stace’s retirement. PMH, JS and MLH all developed their interests in plant systematics and evolution during their time in the Botany Department at the University of Leicester and are extremely grateful to Clive for his extensive support and advice. We are also very grateful to the many people who have contributed to this research: M. di Antonio, J. Ayluso, A. Clark, H. Cameron, P. Delforge, J. Dickson, B. Ehlers, R. Ennos, O. Gerbaud, A. Gevaudan, M. Jenkinson, J. Lewin, M. Lowe, L. McAulay, H. Mathe, M. Mendum, P. Mered'a, J. Moingeon, M. Rohmer, F. Tausch, M. Tebbitt, W. Timpe, D. Tumer-Ettlinger, M. Vauthey, W. Voth; and particularly to Keith Watson for his advice and field support on numerous visits to the Glasgow populations of E. youngiana. This work was supported in part by the NERC Taxonomy Initiative. The Royal Botanic Garden Edinburgh is supported by the Scottish Executive Environment and Rural Affairs Department. REFERENCES Bateman, R.M. (2006). How many orchid species are currently native to the British Isles? In: J. Bailey & G. Ellis (eds), 2006. Current taxonomic research on the British & European flora, pp. 89-110. BSBI, London. Bateman, R.M., Hollingsworth, P.M., Squirrell, J. & Hollingsworth, M.L. (2005). Neottieae: Phylogenetics. In: A.M. Pridgeon, P.J. Cribb, M.W. Chase & F.N. RASMUSSEN (eds), Genera Orchidacearum, volume 4 , Epidendroideae I, pp. 487^195. Oxford University Press, Oxford. Brochmann, C., Gabrielsen, T.M., Nordal, I., Landvik, J.Y. & Elven, R. (2003). Glacial survival or tabula rasa ? The history of the North Atlantic biota revisited. Taxon 52: 417-450. Charlesworth, B. (1992). Evolutionary rates in partially self-fertilizing species. American Naturalistic. 126-148. CLAPHAM, A.R., TUTIN, T.G. & MOORE, D.M. (1989). Flora of the British Isles , edn 3. Cambridge University Press, Cambridge. Delforge, P. (1995). Orchids of Britain and Europe. Harper-Collins, London. Delforge, P. (2001). Guide des orchidees d' Europe. Delachaux et Niestle, Lausanne. TAXONOMIC COMPLEXITY IN EPIPACTIS 43 Delforge, P. & Gevaudan, A. (2002). Contribution taxonomique et nomenclaturale au groupe d' Epipactis leptochila. Les Naturalistes beiges, hors-serie - special Orchidees 83 (Orchid 15): 19-35. Dickson, J.H., Macpherson, P. & Watson, K. (2000). The changing flora of Glasgow: Urban and rural plants through the centuries. Edinburgh University Press, Edinburgh. Ennos, R.A., French, G.C. & Hollingsworth, P.M. (2005). Conserving taxonomic complexity. Trends in Ecology and Evolution 20: 164-168. French, G.C. (2004). Conservation genetics of the critical plant genus Euphrasia L. Unpublished PhD thesis, University of Edinburgh and the Royal Botanic Garden Edinburgh. French, G.C., Ennos, R.A., Silverside, A.J. & Hollingsworth, P.M. (2005). The relationship between flower size, inbreeding coefficient and inferred selfmg rate in British Euphrasia species. Heredity 94: 44-5 1 . Hamrick, J.L. & Godt, M.J.W. (1996). Effects of life history traits on genetic diversity in plant species. Philosophical Transactions of the Royal Society of London, series B 351: 1291-1298. Harris, S.A. & Abbott, R.J. (1997). Isozyme analysis of the reported origin of a new hybrid orchid species, Epipactis youngiana (Young’s helleborine) in the British Isles. Heredity 79: 402-407. Hollingsworth, P.M. (2003). Taxonomic complexity, population genetics, and plant conservation in Scotland. Botanical Journal of Scotland 55: 55-63. Kreutz, C.A.J. (2004). Kompendium der Europdischen Orchideen. Published by the author, Landgraaf. Lang, D. (2004). Britain ’s orchids. Wild Guides Ltd, Old Basing, Hampshire. Levin, D.A. (2000). The origin, expansion and demise of plant species. Oxford University Press, Oxford. Nybom, H. (2004). Comparison of different nuclear DNA markers for estimating intraspecific genetic diversity in plants. Molecular Ecology 13: 1 143-1155. Pedersen, H.7E. & Ehlers, B.K. (2000). Local evolution of obligate autogamy in Epipactis helleborine subsp. neerlandica (Orchidaceae). Plant Systematics and Evolution 223: 173-183. Preston, C.D., Pearman, D.A. & Dines, T.D. (2002). New atlas of the British & Irish flora. Oxford University Press, Oxford. Richards, A.J. (1982). The influence of minor structural changes in the flower on breeding systems and speciation in Epipactis Zinn. (Orchidaceae). In: J.A. Armstrong, J.M. Powell & A.J. Richards (eds), Pollination and evolution , 47-53. Royal Botanic Gardens, Sydney. Richards, A.J. (1986). Cross-pollination by wasps in Epipactis leptochila (Godf.) Godf. s.l. Watsonia 16: 180-182. Richards, A.J. & Porter, A.F. (1982). On the identity of a Northumberland Epipactis. Watsonia 14: 121-128. Robertson, A., Newton, A.C. & Ennos, R.A. (2004). Multiple hybrid origins, genetic diversity and population genetic structure of two endemic Sorbus taxa on the Isle of Arran, Scotland. Molecular Ecology 13: 123-134. Squirrell, J., Hollingsworth, P.M., Bateman, R.M., Dickson, J.H., Light, M.H.S., MacConaill, M. & Tebbitt, M.C. (2001). Partitioning and diversity of nuclear and 44 CURRENT TAXONOMIC RESEARCH organelle markers in native and introduced populations of Epipactis helleborine (Orchidaceae). American Journal of Botany 88: 1409-1418. Squirrell, J., Hollingsworth, P.M., Bateman, R.M., Tebbitt, M.C. & HOLLINGSWORTH, M.L. (2002). Taxonomic complexity and breeding system transitions: conservation genetics of the Epipactis leptochila complex (Orchidaceae). Molecular Ecology 11: 1957-1964. STACE, C.A. ( 1 975). Hybridization and the flora of the British Isles. Academic Press, London. Stace, C.A. (1989). Plant taxonomy and biosystematics , edn 3. Edward Arnold, London. STACE, C.A. (1991). New flora of the British Isles , edn 1. Cambridge University Press, Cambridge. STACE, C.A. (1997). New flora of the British Isles , edn 2. Cambridge University Press, Cambridge. Stace, C.A., Ellis, R.G., Kent, D.H. & McCosh, D.J. (eds) (2003). Vice-county census catalogue of the vascular plants of Great Britain, the Isle of Man and the Channel Islands. BSBI, London. Stebbins, G.L. (1957). Self fertilization and population variability in higher plants. American Naturalist 91: 337-354. Stebbins, G.L. (1974). Flowering plants: Evolution above the species level. Belknap Press, Cambridge, Mass. Stewart, A., Pearman, D.A. & Preston, C.D. (1994). Scarce plants in Britain. JNCC, Peterborough. SUNDERMANN, H. (1970). Europdische und Mediterrane orchideen: eine bestimmungsflora. Brucke-Verlag Kurt Schmerson, Hannover. U.K. Biodiversity Group (1995). Biodiversity : the UK steering group report. HMSO, London. WlGGINTON, M.J. (1999). British red data books: 1 Vascular plants. JNCC, Peterborough. Current taxonomic research on the British & European flora Bailey, J. & Ellis, R.G. (eds) 45-70 (2006). BSBI. London. 45 Molecular evolutionary rates shed new light on the relationships of Festuca, Lolium, Vulpia and related grasses (Loliinae, Pooideae, Poaceae) Pilar Catalan1, Pedro Torrecilla2, Jose Angel Lopez Rodriguez1, and Jochen Muller3 1 - Department of Agriculture (Botany), High Polytechnic School of Huesca, University of Zaragoza, Spain 2 - Department of Botany, Fac. Agronomy, Universidad Central de Venezuela, Venezuela 3 - Department of Botany, University of Jena, Germany ABSTRACT Festuca L., Vulpia Gmel. and Lolium L. are the three main genera of subtribe Loliinae, a group of temperate grasses characterized by their festucoid-type spikelet. Recent phylogenetic studies based on separate and combined analyses of nuclear ITS and chloroplast trnL-F nucleotide sequences have shown that the large genus Festuca , as traditionally circumscribed, is a para- phyletic assemblage of several Loliinae representatives. Monophyletic Festuca s.l. splits into two main diverging lineages, the ‘broad-leaved Festuca ’ and the ‘fine-leaved Festuca ’, each encom¬ passing different genera and with several morphologically-intermediate taxa placed between them or at the base of the ‘fme-leaved’ clade. All these lineages include both perennial and annual taxa that show significant and parallel differences in rates of nucleotide substitution for the two genomes analyzed. Comparative rate analysis indicate that the Tine-leaved Festuca ’ evolve, in general terms, faster than the ‘broad-leaved Festuca ’, and have presumably been derived from them. Relative rate ratios support the minimum-generation-time hypothesis in both lineages, implying the existence of stabilized selection processes for the slow evolving perennials and rapid adaptive mechanisms for the rapidly evolving annuals. High divergences in rate heterogeneity may be responsible for unsatisfactory resolution at the basal node of the festucoid ITS tree. These results, together with conflicts between the ITS and the r/7?L-F trees in the placement of several taxa suggests that the existence of past reticulation events, might have profound implications on their classification. Keywords: Festuca , Loliinae, Lolium , ITS, trn L-F, phylogeny, relative-rate-test, Vulpia. INTRODUCTION Festuca L. with c.500 species distributed in all five continents is the main genus of subtribe Loliinae, a group of temperate Pooid grasses characterized by their dorsally rounded lemma 46 CURRENT TAXONOMIC RESEARCH and linear hilum. Taxonomic circumscription of Festuca and its close allies has changed over the past centuries. Clayton & Renvoize (1986) considered that Lolium, Vulpia and other small genera ( Castellia , Cynosurus , Lamarckia , Micropyropsis , Micropyrum, Psilurus , and Wangenheimia , among others) were derived groups of Festuca , a proposal largely congruent with that of Tzvelev (1982) who restricted the festucoids to nine genera ( Bellardiochloa , Cutandia , Festuca , Loliolum , Lolium , Nardurus , Scleropoa , Sphenopus , and Vulpia). Festuca has been historically divided into several subgenera and sections (Hackel 1882, 1887, 1906; Piper 1906; Krechetovich & Bobrov 1934; Krivotulenko 1960; Tzvelev 1971; and Alexeev 1977, 1978, 1980, 1981, 1986). Several segregates of this genus that were first included within it ( Vulpia , Schedonorus , Diymochloa, and Leucopoa) have been recognized as independent genera at different times (Gmelin 1805; Cotton & Stace 1977; Stace 1981; Holub 1984, 1998; Tzvelev 1999, 2000; Soreng & Terrell 1998). Lolium has been always treated as a genus different from Festuca based on its typical inflorescence traits related to its excavated inflorescence axis with sunken spikelets covered by a single glume. Lolium was first classified in its own subtribe Loliinae (Dumortier 1823) but later transferred to subtribe Festucineae C. Presl by Tzvelev (1982) based on morpholo¬ gy, karyology and hybridization traits. However, the name Loliinae Dumort. has nomenclat- ural priority over Festucineae C. Presl. (Soreng & Davis 2000). Lolium encompasses c. 10-12 species that are mostly native to the Mediterranean region (Terrell 1968). Vulpia was erected as a new genus by Gmelin (1805) and has usually been separated from Festuca ever since, based on on their annual habit, long unequal glumes, and long- awned lemma (Cotton & Stace 1977). The circumscription of Vulpia has changed depending on the inclusion or exclusion of different genera and infrageneric taxa (Stace 1981); up to five different sections have been recognized within Vulpia. based on breeding system and inflorescence traits (Cotton & Stace 1977; Stace 1987). The Vulpia taxa account for c.22 species native to the Mediterranean region and to America (Cotton & Stace 1976). Lolium and Vulpia have been considered related to Festuca based on chromosome and breeding affinities (Jenkin 1933; Malik & Thomas 1966, Ainscough et al. 1986). Lolium hybridizes spontaneously with representatives of Festuca subgen. Schedonorus (a broad¬ leaved group) (Lewis 1975) and rarely with section A ulaxyper (a fine-leaved group), whereas Vulpia (sects. Vulpia and Monachne) intercrosses with those of Festuca sect. Aulaxyper (Barker & Stace 1982, 1984, 1986; Ainscough et al. 1986). Other minor genera of the subtribe Loliinae have mostly been treated as independent genera more or less related to Festuca (Stace 1981, Clayton & Renvoize 1986). Up to eleven annual genera ( Castellia , Catapodium , Ctenopsis, Cutandia , Desmazeria, Loliolum , Micro¬ pyrum, Narduroides , Sclerochloa , Vulpiella , Wangenheimia) were grouped in the Vulpia- Desmazeria complex by Stace (1981). Four of them ( Castellia , Ctenopsis , Micropyrum , and Wangenheimia) were characterized as close allies of Vulpia , two others (. Loliolum and Narduroides) as intermediate between Vulpia and Desmazeria , and the remainder as more related to Desmazeria. The short-perennial genus Micropyropsis was classified within Loliinae by Clayton & Renvize (1986). A series of molecular phylogenetic studies on Festuca and its close satellites (Darbyshire & Warwick 1992; Charmet et al. 1997; Gaut et al. 2000; Torrecilla & Catalan 2002; Torrecilla et al. 2003, 2004; Catalan et al. 2004) demonstrated that Festuca sensu lato is a large paraphyletic assemblage that encompasses not only Lolium and Vulpia but also several other related genera. The most exhaustive survey of festucoid taxa conducted by Catalan et al. (2004) based on simultaneous analyses of nuclear ITS and chloroplast trnL-F sequences RELATIONSHIPS OF FESTUCA , LOLIUM, VULPIA 47 found a likely evolutionary trend from more ancestral ‘broad-leaved’ Festuca lineages towards more recently derived ‘fine-leaved’ Festuca ones. In this study polyphyletic Vulpia and other Mediterranean ephemeral genera were nested within the ‘fine-leaved’ clade whereas Lolium and Micropyropsis were included within the ‘broad-leaved’ one. We also found the sister clades Dactylidinae and Cynosurinae/Parapholinae (in the sense of Soreng & Davis 2000) to be the closest relatives of Loliinae (cf. Catalan et al. 2004). However, evolutionary rates within the Loliinae vary enormously showing a general trend from slow-evolving perennial lineages towards rapidly-evolving annual ones in the ‘fine-leaved’ clade (Torrecilla et al 2004). An increasing number of investigations have demonstrated that most angiosperm groups exhibit strong differences in rates of nucleotide substitutions between closely related lineages in both nuclear and organellar genomes (Sanderson 1997; Muse 2000). The generation-time-effect hypothesis of the neutral theory connects differences in substitution rates with variable generation times; thus, organisms with shorter reproductive cycles exhibit higher molecular rates (Wu & Li 1985; Gaut et al 1992). Although this hypothesis has been widely utilized to explain the mechanisms involved in the evolution of sister groups in vertebrates, it has been contested in evolutionary studies of angiosperms due to the uncertainity of the number of germ line cells per replication (Gaut et al. 1992, 1997; Eyre-Walker & Gaut 1997). Because of this, Gaut et al. (1992) postulated an alternative hypothesis for the generation-time-effect in plants, the so-called minimum- generation-time (MGT) hypothesis, measured as the time from germination to first flower¬ ing. Grasses have shown to be one of the most rapidly evolving lineages within monocotyle¬ dons and to respond in most cases, though not always, to the MGT hypothesis (Gaut et al 1992, 1997). Different evolutionary trends in rate heterogeneity have been also associated with the speciation rate hypothesis which assumes that higher cladogenic events are derived from higher substitutional rates (Barraclough et al. 1996; Barraclough & Savolainen, 2001). In our recent evolutionary study of a restricted group of Loliinae grasses (FEVRE: fine-leaved Festuca and related ephemerals; Torrecilla et al. 2004) we found a strong correlation between substitution rates in the ITS and trriL-F genome regions and the life- cycle strategies shown by these groups . Most annual FEVRE lineages exhibited significant higher mutational rates than their congeneric or cosectional perennial counterparts. Differ¬ ences in rates of heterogeneity were found both within and between these annual and perennial ‘fine-leaved’ lineages. Further evidence suggested that other evolutionary process¬ es, like hybridisation and polyploidy, could also have increased the evolutionary pace in some perennial and annual FEVRE groups (Torrecilla et al. 2004). Because significant differences in nucleotide substitution rates within the ‘fine-leaved’ Festuca seem to be correlated with the generation-time-effect hypothesis (Torrecilla et al 2004), we speculate that similar processes would be likely to have operated in the ‘broad¬ leaved’ Festuca where evolutionary trends towards short-life cycle or perennial polyploid vigour might have evolved independently in several lineages. Therefore, one of the aims of our present study is to test different evolutionary scenarios (i.e. MGT hypothesis, speciation hypothesis, reticulation/polyploidization hypothesis) that may be associated with different substitution rates across the festucoids based on the largest molecular analysis of this group of grasses conducted by Catalan et al. (2004). Because highly heterogeneous sequences may also be prone to accumulation of higher rates of homoplasy which could lead to undesirable long-branch attraction and site saturation effects in phylogenetic reconstruction (Bousquet et 48 CURRENT TAXONOMIC RESEARCH al. 1992; Gaut et al. 1992, 1996, 1997), a second aim of our study is to estimate if highly heterogeneous rates could increase the risk of recovering potential artifactual relationships that may confound the classification proposals based on those phylogenies. MATERIAL AND METHODS PHYLOGENETIC ANALYSES DNA sequence data of Loliinae representatives and other close subtribes is based on the phylogenetic survey of Catalan et al. (2004). A total sample of 119 representatives of subtribe Loliinae, 5 of subtribe Poinae (. Poa , Puccinellia , Sclerochloa), 4 of subtribe Dacty- lidinae (. Dcictylis , Lamarckia ), 2 of subtribe Cynosurinae ( Cynosurus ), 5 of subtribe Para- pholinae (C at ap odium, Monerma , Parapholis , Sphenopus ), and one each of subtribes Sesleriinae ( Sesleria argentea ), and Psilurinae ( Psilurus incurvus), together with 5 outgroup representatives of tribes Aveneae ( Avena barbata , A. eriantha and Deschampsia cespitosa ), Triticeae ( Secale cereale ) and Brachypodieae ( Brachypodium distachyon ) were used to conduct the phylogenetic analyses as indicated in Catalan et al. (2004). Within the festucoids (Loliinae), 71 samples corresponded to taxa of Festuca sensu lato, representing 5 subgenera and 12 sections, 18 corresponded to taxa of Vulpia , representing all the 5 sections of this genus, 10 corresponded to taxa of Lolium , and 9 corresponded to genera that have been considered or demonstrated to be more or less related to Festuca ( Castellia , Ctenopsis, Hellerochloa , Micropyropsis, Micropyrum, Narduroides , Parafestuca , Wangenheimia). The list of taxa with authorities, localities, herbarium vouchers, ploidy levels and GenBank accessions is indicated in Catalan et al. (2004). The ITS and trnL-Y data matrices consisted of 1 1 7 and 1 1 1 sequences, respectively . The ITS data set was made of 644 aligned nucleotide characters of which 46% were informative, whereas the trnL-Y data set was made of 1089 aligned nucleotide positions of which 21% were parsimony informative. Phylogenetic analyses were perfonned on each separate data matrix and on the combined data set as stated in Catalan et al (2004). Bayesian inference approaches produced similar topologies to those obtained through parsimony based heuristic searches (Catalan et al. 2004). Bayesian trees allowed a preliminary estimation of the rates of evolution of the festucoid lineages based on their respective branch lengths. The Bayesian searches were performed using the program MRBAYES v. 3.0 (Huelsenbeck and Ronquist, 2002); the analysis of each separate data set was performed through 1,000,000 generations by the Markov Chain Monte Carlo (MCMC) sampling trees every 100 generations and allowing the program to estimate the respective likelihood parameters (nucleotide frequen¬ cies, nucleotide substitution rates, gamma shape, proportion of invariable sites). Sampled points from generations previous to stationarity were discarded using the bum-in option of Mr. Bayes 3.0. Brachypodium distachyon was used to root the trees. All trees sampled from these searches were used to constmct the respective 50% majority-rule consensus trees where the percentage of times a clade is recovered is interpreted as an estimation of robustness. RELATIVE RATE TESTS Relative rate tests were performed between the main lineages of Loliinae and close allies for the two genome regions analyzed (ITS, trnL-V) in the search for significant differences in nucleotide substitution rates among them and as a means to test evolutionary models related RELATIONSHIPS OF FESTUCA , LOLIUM, VULPIA 49 to: i) the minimum-generation-time (MGT) hypothesis, ii) the speciation rate hypothesis, and iii) the reticulation/polyploidization hypothesis within these grasses. The program RRTree (Robinson-Rechavi & Huchon 2000), which computes distance-based relative tests among pairs of groups of sequences (Robinson et al. 1998) based on a priori group covariances derived from the method of Li & Bousquet (1992), was used to conduct the tests. This program allows pairwise comparisons among groups taking into account a weighted topolo¬ gy or ignoring it. Comparisons were performed for 31 groups of Loliinae and close allies that were circumscribed according to current classifications and to their respective phylogenetic resolution obtained in the ITS and trnL-F trees (Table 1). The Kimura two-parameter distance was used for these non-coding regions and topological references were based on the consensus maximum likelihood trees obtained from the respective ITS and trnL-F Bayesian - analyses. As the sensitivity of the relative-rate test improves with higher taxonomic sampling (Robinson et al. 1998), the complete ITS and trnL-F data matrices were used to define the tested groups trimming in each case uncommon or incomplete sequences that could lead to anomalous results. To improve the accuracy of the tests, appropriate outgroups for the Loliinae and close allies Dactylidinae and Parapholiinae/Cynosurinae were chosen from the close genus Poa. Perennial representatives with moderate rates of mutation were selected, respectively, for the ITS (P. trivialis) and trnL-F ( P . bulbosa) tested analyses. Distance-based rate ratios were also computed among all pair grotips without imposing topological weights; the observed differences were minimum (or none) indicating that the topological constraints do not seriously affect the conservative estimates of group covariances. The MGT was tested according to the parameters established by Gaut et al. (1992, 1997). Annual festucoids flower within their single seasonal cycle (<1 yr) and have, therefore, shorter generation times than perennial festucoids, which usually flower after their first year (1-2 yr). Although some perennial species of Festuca exhibit clines for seedling establish¬ ment and longevity under different ecological conditions (Suzuki et al. 1999), patterns referred to when seed is first returned and for how many years seed is returned by an individual are practically unknown for most of the perennial fescues thus avoiding further perennial-class subdivision analysis in the present study. The speciation rate hypothesis cannot be tested with confidence at the species level within the Loliinae groups as the specific and infraspecific hierarchical ranks and the numbers of the taxa attributed to some assemblages vary greatly depending on different authors. This hypothesis has therefore been tentatively assessed with respect to the number of genera that have been recognized within the main lineages of Loliinae (Holub, 1984, 1988; Watson & Dallwitz 1992) (Table 1). We have developed here an estimate for the reticulation-and-polyploidization evolution¬ ary scenario, predicted by previous authors for these highly hybridized groups of festucoid grasses (Ainscough et al. 1986; Soreng & Davis 2000), that is measured in terms of correlation between higher susbtitutional rates and higher ploidy levels (Table 1). Even if the nature of polyploidy across Festuca has long been debated (i.e. autopolyploidy vs. allopoly¬ ploidy), meiotic behaviour of artificial hybrids and molecular cytogenetic evidences support an hybrid allopolyploid origin for most of the broad- and fine-leaved Festuca complexes investigated so far (Jahuar 1975, 1993; Xu & Sleper 1994; Humphreys et al. 1995; Harper et al. 2004). 50 CURRENT TAXONOMIC RESEARCH TABLE 1. BIOLOGICAL PARAMETERS FOR 31 GROUPS OF FESTUCOID GRASSES (SUBTRIBE LOLIINAE) AND ITS CLOSE ALLIES Group Life-cycle1 MGT (yrs)2 3 Ploidy range No. genera Loliinae: Broad-leaved taxa: 5 Drymanthele P 1-2 2x Leucopoa P 1-2 6x-8x Subbulbosae P 1-2 2x F. paniculata complex P 1-2 2x-4x Scariosae P 1-2 2x Pseudoscariosa P 1-2 2x Schedonorus (European) P 1-2 2x-6x Schedonorus (Maghrebian) P 1-2 4x-10x Micropyropsis P 1-2 - Lolium (perennial) P 1-2 2x Lolium (annual) A <1 2x Subulatae P 1-2 2x-4x Amphigenes P 1-2 2x-4x F. californica P 1-2 4x-8x Castellia A <1 - Fine-leaved taxa: 8 Eskia P 1-2 2x (4x) Exaratae P 1-2 2x (4x) Festuca P 1-2 2x-6x HeUerochloa P 1-2 - Aulaxyper P 1-2 (2x) 6x-10x Micropyrum A <1 2x Wangenheimia A <1 2x Narduroides A <1 2x Ctenopsis A <1 2x Apalochloa A <1 2x Vulpia 2x A <1 2x Psilurus / Vulpia 4x-6x A <1 4x-6x Loretia complex A <1 2x (4x) Dactylidinae A-P <1 - 1-2 2x-4x Cynosurinae A-P <1 - 1-2 2x Parapholiinae complex A <1 2x-4x 1 A= annuals; P = perennials 2 MGT = Minimum-generation-time (since germination to first flowering) 3 Estimative values of the genera recognized by Holub (1984, 1988) and Watson and Dallwitz (1992) for the groups included in the present study. RELATIONSHIPS OF FESTUCA, LOLIUM, VULPIA 51 RESULTS THE ITS AND TRNL-F TREES Bayesian trees obtained from the separate analyses of the ITS and trnL-F data matrices correspond to those indicated in Catalan et al. (2004). A test of goodness of fit for 56 nucleotide substitutions models that was previously conducted on each individual dataset using the likelihood ratio test statistic included in the program Model Test ver. 3.06 (Posada & Crandall 1998) showed the same optimal model (GTR+G+I, 4 gamma rate categories) for the two independent datasets. The Bayesian search conducted on the ITS data set sampled 9931 trees which reached a stable likelihood value after the bum-in of 537 trees; the 50% majority rule consensus tree - of all sampled trees is shown in Figure 1. The analysis of the trnL-F data set sampled 9681 trees, which reached a stable likelihood value after the bum-in of 300 trees; the 50% majority mle consensus tree of all sampled trees is shown in Figure 2. The two topologies are congment in the resolution of a well to moderately supported clade of fine-leaved Festuca + Vulpia + Related Ephemerals (FEVRE group, cf. Torrecilla et al. 2004) in which the strongest support is for subclades Aulaxyper + Vulpia (2x), Festuca , and Psilurus / Vulpia (4x-6x). Representatives of Eskia and Amphigenes p.p. are resolved as basal paraphyletic assemblages of the FEVRE clade in both trees; Wangenheimia is resolved as the well supported sister taxon of the Festuca clade in the trnL-F tree, whereas Micropyrum is unexpectedly resolved as sister to the Aulaxyper clade in the ITS tree (Figs 2 & 1, respective¬ ly). A fourth resolved but differently supported lineage is that of representatives of Vulpia sects. Monachne and Loretia plus F. plicata ; the trnL-F tree also incorporates Apalochloa and Ctenopsis within it and shows an unresolved basal placement for representatives of Festuca subsect. Exaratae and close allies. The broad-leaved group is resolved as monophyletic in the trnL-F tree (Fig. 2) whereas a series of broad-leaved lineages collapse with the FEVRE clade and with clades of other close subtribes at the basal node of the ITS tree (Fig. 1). The two topologies resolve a well-supported clade of Folium + Micropyropsis + Schedonorus. Other resolved clades in the ITS tree ( Subbulbosae pp, F. paniculata gr., Leucopoa , Subulatae , and Scariosae + Pseudo- scariosa + Drymanthele) fonn a series of successive polytomies in the trnL-F tree. Some ‘intermediate’ taxa between the broad-leaved and the fme-leaved groups are nested in an odd position in one or the other tree (i.e F. californica ) (Figs 1 & 2). Castellia is differently resolved in each topology whereas Parafestuca falls apart from the festucoid clade. Folium , which is strongly resolved as monophyletic in the ITS tree, is shown to be paraphyletic in an intermingled clade of representatives of Micropyropsis and European Schedonorus in the trnL-F tree (Figs 1 & 2). The combined analysis provides a better resolution than the separate analyses; Loliinae is resolved as monophyletic and separated into two diverging lineages, a well-supported clade of fine-leaved Festuca and a poorly supported clade of broad-leaved Festuca (Fig. 3). The presence of ‘intermediate’ taxa at the base of or close to the fme-leaved clade indicates a trend from more ancestral broad-leaved Festuca lineages towards the more recently evolved FEVRE lineages, a finding that is correlated with the high mutational rates observed in most of the annual lineages of the fme-leaved group (cf. Torrecilla et al. 2004). The combined analysis also resolved the sister clades Dactylidinae and Cynosurinae/Parapholinae as the closest relatives of subtribe Loliinae. 52 CURRENT TAXONOMIC RESEARCH — — Festuca heterophyllal Festuca heterophylla2 _i Festuca rotnmalen V." — Festuca nevadensis Ffesufcamf)Crl 1 Aulaxyper J r + Micropyrum . + Vulpia pp(2x) r Festuca riyularis „ ' ' Micropyrum tenellum MicroDvrum patens , . . 100 p-|_ , r * . ri ,-Wlpia murahsl Mulpia.muralis'T \/n nia hmmninpc i Vulpia breyis 1. . vulpia sicula _ Vulpia gemculata Festuca nlicata ° , . _ -y uj^ia foptquerana Loretia Monachne Spirachne vulpia memuranacea F i- t I XU.UU al.opecuros ______ V , , , )^itfpi^Funilatpralis2 Apalochloa -KT , W.angeiinemna .lima , F , ■ N arduro ldes^alzm ann 1 1 + ephemerals ’. Q tencysji ddicattua . . . . Festuca pyrenaica stuca borderei . duca clementei ffrestuca.pvina :esprca FiTifarmis . lestuca Iongiaunculata estuca arat tuca gfa 'pFe ----- Festuca a. “ resruc nensisl is tor iaclali nna . . nvstrix Jvestuca frigida Festuca capiiulpua , - - Festuca Heltew rrosl iras2 , tulpia cipatal la dhata^ Lurus incurvus Vulpia pp (4x-6x) + Psilurus Amphigenes" " "s"s". ““ Festuca cauiomica Festuca, quadnflora2 i quadrmdrar Eskia raniculata erica , . . L o H u m “r Jgi dun" J qlium temuTentturL ilium multiftorurnlZ olium multirlorumF olium perenne “fLolium canariense Loliupr rigidum2 subulauim . ^olium renjutuuu ....... ~ Micropyropsis tuberosa Festuca pratensisl ina , eal aceaZ > a gigantea lesrlica fontquen . .. Festuca Ietourneuxiana estuca auantigena ltien aSlC13 ca oumatii Hestuca.olOgaJtS 100 1X1 "Dactyl Lslusp'an "ca “ """"""" ““TDactytis glomerata Dactylidinae F. paniculata gr. Lolium + Mycropyropsis European’ + Schedonorus SWSir 'Mahgrebian' *™ Festuca tenas _ ... _ _ ' M onerm ac yli n dnca , pnohs mcurva + r„n^F,v^HUtV,?^inatus Cynosurinae pmcnerla . T ana Festuca pulclle _tuca sqeaabiTis estuca altaica , . Festuca subulata Leucopoa + Subulatae Festuca kingii " ^estuca scaripsa . Dseudesia „ . r^„ca altissn „ Festuca drymeja „ . . Festuca lasto _ J , + Drymanthele Castelha tuberculosa . - ... ............ la issima + Scariosae Pseudoscariosa o.-J?e®c!tctelescens . . „ . .S.u.b.h.l>lbo.s^ . . PP, tvz - ■ " . . Poa miirrna . ■ Poa tnyiahs Poinae lerocnjpa dura . uccmelltfuijstaus. barbata gentea Avena + Parafestuca + Sesleriinae "fine- leaved" (FEVRE) "broad¬ leaved" + close subtribes Figure 1. Bayesian ITS tree: 50% MR consensus tree of 9331 trees. Posterior probability percentages are indicated on the corresponding branches. RELATIONSHIPS OF FESTUCA, LOLIUM, VULPIA 53 Festuca rubra2 esfuca a esfu ustyiii Festuca querana othmaleri estuga ampla Aulaxyper Festuca rivuians Vulpia bromoides Vulpia murahsl lis2 stuca nevadensis + Vulpia pp (2x) j^utpia muraiisi 1 S. American gr. aZ^l£Bierachlaa fragilis I , Festuca elviae + — ■ Jr estuca tngitla Festuca a - Festuca 0 _ JFestuca aragonensis2 Festuca ovirfa Festuca loahoensis Wanuenheimiadima .. Narffuroiaes salzmann “ Psilums mgurvus .yulpia cihatal Vulma cmafa2 , , , . .Vulpia myurosl Vulpta cihata3 " Jxlicropyrum tenellunr - Micropyrum pateus. «... Festuca mdigp 77 Festuca Festuca fr lpina ta ystrix da Festuca + Wangenheimia Vulpiapp (4x-6x) Psilums + Vulpia alopecuros Vulpia brevis i— T”, Vulpia unilateralisl H 1 Vulpia unilateral ir2 ■ _ Vulma sicufa ■T* Vulpia membranacea Vuluia fasciculatal . - - — vulpia Loretia + Monachne + Spirachne Vulpia fontquerana + Apalochloa :ulata Festuca borderei “Vestuca pyrenaica " Festuca cammioTia “ Eestuca clementei “™ Festuca gautien _ * Eestuca eskia " Festuca elegans Festuca Dumatn ffKpSfctuIa r Festuca pncata + Ctenopsis F. plicata Exaratae 100 51 Festuca carpatica Festuca altaica , , ■festuca dirnarphat Festuca quadmioral Festuca. fluadriflora2 . . . Vastellra tuberculosa restuca pulchella jurana stuca pulchella pulene Festuca califomica. Festuca subulata Eskia Subulatae Amphigenes s.s. "fine- leaved" (FEVRE) 77 96 53 99 99 90 99 - iStut lauonqueri Lolium .canariense „ Lolium ngidum2 ,, Festuc Festuca fenas. f j Fe — Lolium ngid “ Festuca. mairei ■ Festuca glaucescens. estuca pamculata estuca uurandoi J estuca spectabilis estuca sbadicea leroi inaceal Lolium + Mycropyropsis + Schedonoms - Festuca m estuca baetica estuca Ringu Leucopoa pp + F. paniculata gr. 100 EVestuea, drymeia Fesfucalasto ™ 1 Festuca scanosa Festuca coerulescens " Festuca tnfTora “ Festuca, pseudeskia Festuca alnssima ■ ,Dactylis his Drymanthele Scariosae Pseudoscariosa Subbulbosae pp Elactylis hispanica Dactylis glomerata , ™ dmdrrFici cnirpQ "broad¬ leaved" Lamaixkia aureal _ Deschampsia cespitosa.. - ... Monerma cylmdnca Earaphqhs incurva 0 , CatapQdium rigid Sphenopu Dactylidinae rum n lvarica .um cie«!dura-- Puccmema.distans oa inrirma igi - - tus Cvnosprus echmatus - C\ ,ynosurus cnstatus Parapholiinae + Cynosurinae Poa bulbosq Avena eoantha „ . , araiestuca albida Poinae 77™", Secale cereale , Brachypodium distachyon Avena + Parafestuca FIGURE 2. Bayesian trnL-Y tree: 50% MR consensus tree of 9681 trees. Posterior probability percentages are indicated on the corresponding branches. 54 CURRENT TAXONOMIC RESEARCH 100 querana i rothmaleri Aulaxyper + Vulpia pp (2x) Festuca rubral Festuca agustinii “ Festuca nevadensis Festuca rivplaris Festuca ltjqnca Vulpia muralisl Vulpia pluralist Vulpia bromoides Helrerochloa fragilis - . . " Festuca fngida Festuca alpm.a Festuca glacialis . Festuca nystrix Festuca longiauncplata Festuca aragonensisl Festuca ovina Wangenheirma lirna. es salzmannii _ , ™ Vulpia alopecuros Loretta -LUL1 - 1 1 1 Vulpia fontquerana ■ 1 1 1 r\i q m Ptnlaran qppci i / i Monachne Festuca + Wangenheimia Vulpia membranacea Vulpia brevis Vulpia fasciculataj 1 Vulpia sicpla — “ Vulpia gemculata Festuca pncata Vulpia unuaterahs 1 Spirachne Apalochloa F. plicata ia unuateralis 1 x • FlltOLa 'V«ffSSa Ctenopsis “ Micropyrum tenell — Micropvrum patens Festuca borderei Festuca pyrenaica Festuca capillifolia ” Festuca clementei 921 Vulpia. cur 100 Vulpia ciliata2 * Vulpia myurosl Psilurus mcurvus lum Exaratae iatal Vulpia pp (4x-6x) + Psilurus 100 Festuca elegans — Festuca gautien Festuca eskia Fes ■F1 — FestucTauaffl ora. JLJ Festuca carpatica T1— F< a califomica t-estuca carpatic Festuca dhnorphal Festuca quadn floral Festuca burnatn Castellia tuberculijsa 100 Eskia Amphigenes s.s. "fine- leaved" (FEVRE) olium canariense Lolium rigidumd "oruml _ Lohum r Lolium multi Festuca pratensisl Festuca gigantea “ Festuca forjtquer Micropyropsig rub Festuca apennma Festuca arundinacqal Festuca marrei * Festuca glaucescens Festuca fenas n erosa Lolium + Mycropyropsis + Schedonorus Festuca pamculata Festuca dqrandoi 100 _ ' Festuca baetica Festuca spectabilis . Festuca kingii " ' Festuca drymeja - - Festuca lasto Festuca scariosa Festuca pseudeskia Festuca altissima . ... Festuca coerulescens Festuca triflora — Festuca pulchella pulchella Festuca pulchella jurana Festuca altaica ' Festuca subulata - - - - ““ Monerrna cylmdrica Parap tolis lpcurva F. paniculata Leucopoa gr- PP Drymanthele + Scariosae + Pseudoscariosa Subbulbosae + . PP Leucopoa pp Subulatae "broad- leaved" ynosurus crtstatus , Dactyhs hispamca - glomerata ia aureal . . . . , Parapholiinae + Cynosurinae Deschampsia cespitosa _Sesleria argentea A vena Dactyli .amarc oa dura ia distans Poa mfirma Dactylidinae Poinae 0.1 - Avena spp Parafestuca albida “ Secale cereale Brachypodium distachyon Avena + Parafestuca + Sesleriinae FIGURE 3. Bayesian combined ITS/fr7?L-F tree: 50% MR consensus tree of 6216 trees. Posterior probability percentages are indicated on the corresponding branches. RELATIONSHIPS OF FESTUCA , LOLIUM, VULPIA 55 RELATIVE RATE TESTS Of the 3 1 groups tested, three corresponded to the close subtribes Dactylidinae, Cynosurinae and Parapholiinae, and the remaining 28 to subtribe Loliinae (Table 1). Within this last lineage, 14 groups belong to the ‘broad-leaved Festuca', another 13 to the ‘fine-leaved Festuca' , and 1 to the single unresolved taxon Castellia. The ‘broad-leaved’ lineage encom¬ passes the groups Amphigenes , Drymanthele , F. californica , F. paniculata complex, Leuco- poa ( F . kingii, F. spectabilis), Lolium, Micropyropsis , Pseudoscariosa , Scariosae, Schedonorus , Subbulbosae , and Subulatae. Schedonorus was further subdivided into two subgroups, the ‘European’ Schedonorus and the ‘Maghrebian’ Schedonorus , based on their divergence in the ITS tree. Lolium was also halved into two subgroups, each of them including perennial/biannual (L. perenne , L. multiflorum) and annual (L. rigidum , L. canar- iense) species, respectively, for both ITS and trnL-F regions. The ‘fine-leaved’ lineage presents a mixed array of perennial ( Aulaxyper , Eskia , Exara- tae , Festuca , Hellerochloa ) and the mostly annual (. Apalochloa , Ctenopsis , Loretia complex (Spirachne, Monachne, Loretia , F. plicata, Micropyrum , Narduroides , Psilurus complex ( Psilurus , Vulpia 4x-6x), Vulpia (2x), Wangenheimia ) groups. Most of the chosen groups were monophyletic or corresponded to terminal branches; however, in a few cases some paraphyletic assemblages of closely related taxa were also taken into account. The minimum-generation-time (MGT) hypothesis, the speciation hypoth¬ esis, and the reticulation/polyploidization hypothesis were assessed for different groupings of festucoids. The results obtained from the conducted relative-rate tests are presented in Table 2 (A-F). Compared to its closest relatives, Loliinae evolves slightly faster than Cynosuriinae and Parapholiinae but significantly slower than Dactylidinae for the ITS region (Table 2, E). The accelerated rate of mutation of the Dactylidinae ITS sequences is contrastingly higher than those presented by most assemblages of Loliinae and with respect to those of Cynosurinae and Parapholiinae (Table 2, A-D). Conversely, the more conserved substitutional rate of the chloroplast trnL-F region does not detect significant differences between the Loliinae and its closest relatives (Table 2, E) though levels of significance are manifested in more detailed comparisons among Loliinae groups and with respect to Cynosurinae (Table 2, A, B, D). Surprisingly, the Cynosurinae taxa show some of the fastest rates of trnL-F substitutions of all groups studied and are significantly higher than those of Dactylidinae, Parapholiinae, and the slowly-evolving Loliinae assemblages (Table 2, A, B). Within Loliinae, the ‘broad-leaved’ groups evolve in general terms at a lower pace than the ‘fme-leaved’ ones for both genome regions (Table 2, E) though differences are more pronounced in the ITS region. The broad-leaved perennial assemblages of Leucopoa , Dry¬ manthele, and Subbulbosae , which incude some of the tallest and most robust representatives of Festuca , show the lowest rates of substitution in all the Loliinae studied. Their rate differences are highly significant with respect to the rapidly evolving ephemeral groups of the fme-leaved Festuca clade (Table 2, A-D). Broad-leaved assemblages with slighly higher mutational rates are those of Scariosae , Pseudoscariosa , and the F. paniculata complex, which are close to those presented by the slow-evolving fme-leaved groups Eskia and Exaratae (Table 2, A-D). All these groups show significant differences from the fastest evolving ephemeral lineages of the fme-leaved clade. The most rapidly evolving lineage within the broad-leaved clade is that of Schedonorus! Micropyropsis/ Lolium. The ‘European’ Schedonorus evolve at a lower pace, TABLE ,2A. RELATIVE RATE RATIOS OBTAINED FROM RELATIVE-RATE-TESTS. THE RATIO IS GIVEN AS THE RATIO OF ROW OVER COLUMN. BOLD CARACTERS ARE SIGNIFICANT AT DIFFERENT P VALUES (* = P<0.05; ** = P<0.001; *** = PO.OOOl). 56 RELATIONSHIPS OF FESTUCA, LOLIUM, VULPIA Pairwise comparisons between Dactylidinae, Cynosurinae, Parapholiinae, and broad-leaved Loliinae groups (ITS: below diagonal; //77L-F: above diagonal). Abbreviations: Dac= Dactylidinae; Cyn=Cynosurinae; Par= Parapholiinae; Dry= Drymanthele; Leu= Leucopoa ; Subbul= Subbulbosae\ Fpan= F.paniculata complex; Sea = Scariosae; Pse= Pseudoscar iosa\ ScheE= ‘European’ Schedonorus\ ScheMN ‘Maghrebian ' Schedonorus', Microsi= Micropyropsis\ Lolium(p)= Lo/ium (perennials); Lolium(a)= Lolium (annuals); Subu= Subbulatae ; Amp= Amphigenes ; Fcal= F.californica; Cas=Castellia. TABLE. 2B. RELATIVE RATE RATIOS OBTAINED FROM RELATIVE-RATE-TESTS. THE RATIO IS GIVEN AS THE RATIO OF ROW OVER COLUMN. BOLD CARACTERS ARE SIGNIFICANT AT DIFFERENT P VALUES (* = P<0.05; ** = P<0.001; *** = PO.OOOl). RELATIONSHIPS OF FESTUCA , LOLIUM , VULP1A 57 I* o * n* ir, G © © o oo 00 * * * ITi G * * * G 'A d — 1 rn O CO d — * * * ~t r- oc © G G i i r~- oo © d * * * ""T m O' r~~ on Oh * © G (N VO © * * o oo © © G G ■ i © © © © © © © t— r- 4c t"- m © © in 1 © G © d © © c4 S § S5 © oo oo SO rn rn rn 4c rn n}- © OS r- >/-) SO SO 1 i c4 -- © — - © i o 1 4c Os so ’t oc nT C3 r- so m © r- O- in m oc © < o i — O c4 G — * © o’ i 4c 4c 4c 4c 4c 4c 4c 4c rn sO 4c 4c (N r- © OO rn oc Os rn in TO" rn rn l C> u ni i o "T i rn © i o 1 1 nJ 4c 4c 4c 4c 4C i n in '-t r~ oc J— c3 ri rn p fS © p © p r4 o i rn 1 nJ i © i o © ■ 1 4c c in so in m rn 4c sO © rn sc G rn (N r- rn 00 © TO" m ■ (N © © © fN fN © * © © G * © oo m cn G G * * r- o © oc G G * * © OC G * * * oc G nt" © oc © © © > U o CO Q 00 oo * * * Tt G © © m (N © r~ CN sO m in o 0 * sc nr © in in W-i p p rn © © i o’ i © © G o o’ o - 4c ■3: 4c 4c 4c * 4c 4c 4c 4c * 4c OC o r~- rn O O . — 1 r- Tf «r. — p © r- o in O © G rn G G — © c4 rn rn © TO a> c3 > c ’o •«> © O') * ^ C/3 o u JO C3 — O Cu •— O- o TO GO H o C3 a CL c3 O Cl Eskia c3 C3 X H i— 13 X < o Vh CJ i CD r ^ C3 o U! c^J z o C/3 Oh .2 "S >— o 03 4c 4c 4c 4c 4C 4c 4c 4c 4c 4c 4c 4c 4C — r 4c * * 4c Ov o IT4 vO T lO CO CO CO CO 1 CO i CO i 4c -r O' to o O' 04 04 • tq oo o^ vq 04 l 1 1 ^7 i 4c ■3? * 4c 4C 4c * 4c •3: Ov OI O' OI O' *=* O' O' v~i CO CO 04 04 oi 04 oi i 4c * 4c 4c * -r OV or ov 'T CO rq vq O' O' oi , — i , — ^ , — i 04 oi * O 104 CO O' vo CO O' oo O' VO VO • • • • • • OI H O— < 1 i i i i i •3: •3c •3c •3c CO . — < V© •O oo vq ® ® oi ~7 04 q" oi oi •3c •3c •3c 4c •3c •3c 4c 4c •3c •3c 4c •3c 4c SC 0© ® ® ■cr OO ® ® ® CO oi CO i CO co CO i 4c •3c •3c 4c 4c 4c -r 04 ® V© vo ® co OV co oi oi oi , — i oi oi 4c 4c co i Ov oi 4c v© co C4 oo (N 'vJ- cq co tO 4C ov — r r4 4c 4c 4c 4c * 4c t Ov lO tOl OO V© ^T ON OO co O cq * Nf ® oi oi 04 , — i . — - oi . — ^ i i i i i i i i to vq CO Of 04 O' CO 104 04 or vo o 04 O' o O i O o o’ o’ i o 1 o tO> vq to O' vo O'- o tq or 04 o oo or to i ~7 i O o’ 1 i o o -0.66 CO O' i -0.58 0.64 -0.02 -0.006 -0.002 0.85 vo 04 O' CO oo o or 04 cq oo 04 Ov CO o vo o o o i o o i O i o 1 o OV CO oo to o 04 CO O' tq vq or - — j 04 Ov ’ — J T— i . — I 1— H o o’ . — i O o i i i i i i i i X S 3 < * O (N (N tq VO tq or co o OO 04 O' VO O o CO o CO o OV CO o O' 04 oo ’ — J o o> o o 1— H CD r' O 1^ ■'cT O' O' oo to ^sO O' 04 oo VD 04 to 3 04 • rq • Ov oo oo tq o Ov Ov 3 04 1 r^i 0 —p 1 ^7 i 1 o 1 o o’ o 1 O o C/3 04 o- r" o r~- i oo OV o or OV o’ o> Ov I VO VO o tOi cq co 04 o I o- C4 o’ 0 4 o oo o 04 oT O 04 CO Ov iq Pairwise comparisons between broad-leaved and fine-leaved Loliinae groups (ITS). Abbreviations as indicated in A) and B). TABLE. 2D. RELATIVE RATE RATIOS OBTAINED FROM RELATIVE-RATE-TESTS. THE RATIO IS GIVEN AS THE RATIO OF ROW OVER COLUMN. BOLD CARACTERS ARE SIGNIFICANT AT DIFFERENT P VALUES (* = P<0.05; ** = P<0.001; *** = PO.OOOl). RELATIONSHIPS OF FESTUCA, LOLIUM, VULPIA 59 o P • HH C/5 CU 3 > 4s 4s 4s 4s 4s 4s 4s 4s 4s 4s 4s 4s 4s 4s 04 NO rf ® rf CO 04 m 04 O' NO in m 04 ® ® 04 ON m ON © O' r- O' •• m • • • • • • • O' • • • co 04 co 04 04 CO © © r—l o • 04 CO 04 i i i i i i i i i i i i i i 4s O' rt 4s 04 O or 4s ON O' 04 m in 4s 4s 4s 00 4s O' 4s CO • rt NO • ON oo in • p rr o CO o 04 ® 04 ® • ® • • 04 i 04 o’ 1 © i 0|4 ® i o i o 1 o 1 o’ co 04 04 04 4s o- t"- 4s in CO O' 4s 04 ON CO CO 04 rt 4s 4s ,£3Bk m ON 04 04 • P rr • p oo Tf • © 04 p 04 O' OC in in NO 04 — 04 1 o 04 o o’ o o • — • — i *— oc 04 OJ o C" o o o o O' CO O' oo o IT) CO o © no © 04 © 4s * rf 04 On ZT O' no oo O' • o o o 04 i i i 4s 4s 4s 4s 4s 4s 4s 4s 4s 45 r" in in 00 ON ON ”t in ON 04 04 *— 1 1 O' in in OH 0-> O' • ® • NO • OO in 00 • oo CO on © p o Tf • in • co • u 04 04 l 04 ! ' i 04 p ' p p p o 1 04 04 04 l 4s 4s 4s 4s 4s 4s 4s 4s 4s 'Ct CO 00 in 'St in CO oo 04 — oo ON OC O' NO NO NO co CO O' O' © OO O) ® 04 CO co • • • • • OC • • • z 04 i 04 i 1 04 l o 1 1 o i o’ 1 o 1 — 04 04 4 04 4s 4s 4s 4s NO ON co m *r-H T— 04 o- rr — oo OO 04 OO ^H C3 rr • or m • co 04 NO • rt r' co NO o O' 04 oo ON p £ 04 i 04 i p 04 ® i p © 1 p> o 1 04* 1 1 p o 3 Cu O 5 * O' PO 04 4s ON ON CO CO — 04 i i O 4s On O t O- 04 © © 04 o’ i i i ' — i OC o© O — < 04 o o o’ I I I ON co o’ 4s 4S rt oc 04* 4s O' m ® oo 04 -2 •3s ® 04 3 m 3 'A < 3 m oo o’ 4s ON O- O' rt — o NO in © oc oo © o’ m 04 o m o © NO NO o> 4S 4s 4s rt CO 04 X g a 2x) -0.1493 0.8813 -2.3280 0.0193* Loliinae(p)(2x) Loliinae(p)(>2x) -0.1658 0.0971 -3.0005 0.0027** Loliinae(a)(2x) Loliinae(a)(>2x) 2.3739 0.0176* 0.5959 0.5511 Broad-leaved(2x) Broad-leaved(>2x) 0.1515 0.8795 -0.5496 0.5825 Broad-leaved(p)(2x) Broad-leaved(p)(>2x) -0.6807 0.4960 -0.6793 0.4968 Fine-leaved(2x) Fine-leaved(>2x) 0.0788 0.9371 -2.4615 0.0138* Fine-leaved(p)(2x) F ine-leaved(p)(>2x) -1.1162 0.2643 -3.1404 0.0017** F ine-leaved( a)(2x) Fine-leaved(a)(>2x) 2.4504 0.0143* 2.1081 0.0350* Dactylidinae(2x) Dactylidinae(>2x) -1.5016 0.1332 3.2813 0.00103** Cynosurinae(2x) Parapholiinae(4x) 0.2275 0.8199 1.3381 0.1808 Relative-rate-tests ratios for the Minimum Generation Time (MGT) hypothesis (1), the speciation-rate hypothesis (2), and the reticulation/polyploidization hypothesis (3). Abbreviations: (a) = annuals; (p) = perennials; (2x) = diploids; (>2x) = polyploids. RELATIONSHIPS OF FESTUCA, LOLIUM , VULPIA 61 followed by Micropyropsis and the ‘Maghrebian' Schedonorus group, and then by Folium. Further differences within the last group indicate that the perennial/biennial Lolium taxa evolve at a rate similar to or slightly lower than Micropyropsis , whereas the annual Lolium taxa show accelerated substitution rates at both ITS and trnL-F regions, being significantly different from most of the remaining broad-leaved groups and as fast (or faster) than some of the fine-leaved groups (i.e. Apalochloa , Aulaxyper, Festuca , Hellerochloa , Psilurus ) (Table 2, A, C, D). Groups of ‘intermediate’ placement (i.e. Amphigenes, Castellia, F. californica) in the combined Bayesian tree also show intermediate rates of mutation between the slow-evolving ones of the ‘broad-leaved’ groups and the fast-evolving ones of the ‘fine-leaved' groups (Table 2, A, C, D). The more advanced ‘fine-leaved' Festuca groups show a trend towards higher substitu¬ tional rates in both ITS and trnL-F sequences. Rate heterogeneity ranges from the lowest ones corresponding to old relict perennial groups (. Eskia , Exaratae ), through the intermediate ones of more recently evolved perennials (. Aulaxyper , Festuca , Hellerochloa ), to the fastest ones acquired by the newly derived annuals lineages. Within the last group different assem¬ blages of the genus Vulpia and its close allies show significant differences from most of the remaining ‘fme-leaved' groups for the ITS ( Loretia ) and the trnL-F ( Loretia , Ctenopsis ) sequences (Table 2, B-D). Relative rate tests designed to test the MGT hypothesis indicate that annual lineages evolve faster than the perennial ones within the studied groups of Poeae grasses, though differences are only significant within the more largely sampled Loliinae lineage at both ITS and trnL-F regions (Table 2, F). Broad-leaved annual taxa ( Lolium spp.) show significant differences in ITS and trnL-F sequences from the rest of the broad-leaved groups whereas fme-leaved annual taxa ( Vulpia and other ephemerals) are significantly different from other fme-leaved groups only in ITS sequences. The close subtribes Dactylidinae and Cynosurinae/Parapholiinae also show higher but non-significant mutational rates in annuals than in perennials for the ITS sequences; however, the reverse pattern is observed for the chloroplast trnL-F sequences where perennials show higher (but non-significant) rates than the annuals. The limited taxon sampling of these last groups avoids further speculation on the lack of correlation detected for the trnL-F data set with respect to the MGT hypothesis. The speciation rate hypothesis was only estimated between the ‘fme-leaved’ and the ‘broad-leaved’ lineages of Festuca based on the recognized genera included in our present survey (Table 2, F). The more diverse ‘fme-leaved’ Festuca evolve faster than the less diverse ‘broad-leaved’ ones, though those differences were not significant, indicating a relative support for the cladogenetic process. A new evolutionary scenario has been confirmed for the perennial groups of Loliinae with respect to the reticulation/polyploidization hypothesis (Table 2, F). Highly polyploid taxa show higher substitutional rates than their respective ingroup diploid counterparts at both ITS and trnL-F regions, though those differences are not always significant. Conversely, diploid annual taxa tend to show higher mutational rates than the ephemeral polyploids (Table 2, F). Although hybridization may have affected equally diploid and polyploid lineages, it is a more common phenomenon within the latter groups, which sorted out the new sterility barriers via recurrent introgression and polyploidization (Stebbins 1956; Stace 1987). Polyploidization is expected to increase the rate of variability of the nuclear genome concordantly with the accumulation of more gene copies (Soltis & Soltis 1999), but should leave the chloroplast genome less affected. However, concurrent rates of nucleotide substi- 62 CURRENT TAXONOMIC RESEARCH tutions in the two genomes, detected for either the diploid and the polyploid lineages of the Loliinae, may be indicative of other concerted nuclear and cytoplasm replication mecha¬ nisms that could be operating in these plant cells (Gaut et al. 1997). DISCUSSION HETEROGENEITY RATES AND THE PHYLOGENY OF THE FESTUCOIDS (SUBTRIBE LOLIINAE) A more comprehensive phylogenetic framework for Festuca s. 1. was drawn after the molec¬ ular survey of Catalan et al. (2004) and the present study. It has been demonstrated that the paraphyletic Festuca lineage encompasses not only Lolium and Vulpia , but also other genera that are nested either within the Tine-leaved’ Festuca ( Ctenopsis , Hellerochloa , Micro- pyrum , Narduroides , Psilurus , Wangenheimia ), within the ‘broad-leaved’ Festuca (. Micropyropsis ), or in an ‘intermediate’ position between them ( Castellia ). The festucoid lineage emerges as a natural group. Our studies have also shown that subtribe Loliineae and its close relatives Dactylidinae and Cynosurinae/Parapholiinae form a monophyletic and well supported group within the Poeae (Fig. 3). Relative rate tests have demonstrated that the Dactylidinae taxa show extremely acceler¬ ated substitution rates in ITS sequences compared to other analyzed groups indicating a release from stabilized selection (Bousquet et al. 1992; Muse, 2000). High mutational rates may have negative consequences on phylogenetic reconstructions due to the loss of deep phylogenetic signal which conveys undesirable results like the long-branching attraction and site saturation effects (Wendel & Doyle 1998; Hillis & Wiens 2000). The lack of resolution observed at the basal node of the festucoid ITS Bayesian tree (Fig. 1) and the unexpected sister relationship of Dactylidinae to the fine-leaved clade can be associated with disturbing effects caused by increasing levels of homoplasy displayed by this rapidly evolving group. Potential arctifactual placements of Dactylis with respect to the fine-leaved Festuca were also observed in the phylogenetic trees of Charmet et al. (1997) and Lehvaslaiho et al. (1987); however, our chloroplast trnL-Y data does not indicate an accelerated rate of nucleotide substitution for the Dactylidinae, and the relationships recovered here indicate that Dactylidinae is a close relative but not part of the Loliinae assemblage. Relative rate ratios show that the Cynosurinae/Parapholinae ITS sequences evolve at an intermediate pace between the slow and fast evolving Loliinae lineages, though the Cyno- surinae taxa show significant differences in rate substitution for the chloroplast trn L-F region with respect to most of the groups studied. These differences, however, do not seem to have altered the topology of the Loliinae clade neither the basal position of Cynosurinae with respect to the Parapholiinae group, which is congruent with the resolution obtained from the ITS data. THE ‘BROAD-LEAVED’ FESTUCA , MICROPYROPSIS AND LOLIUM The typical dichotomy between ‘broad-leaved’ Festuca and ‘fine-leaved’ Festuca lineages observed in previous studies of Festuca based on more limited taxon sampling (Torrecilla & Catalan 2002) became blurred when sampling of festucoid representatives was increased (Catalan et al. 2004). The presence of a series of unresolved clades and polytomies in the ITS and trnL-F based trees, together with the doubtful position of several ‘intermediate’ taxa RELATIONSHIPS OF FESTUCA, LOLIUM , VULPIA 63 between the two lineages, suggests a basal paraphyly of the ‘broad-leaved’ lineages of Festuca s. 1. with respect to the more recently evolved ‘fine-leaved’ ones. However, a poorly supported clade of broad-leaved taxa is still recovered in the combined ITS/fr7?L-F analyses (Fig. 3). Lolium is nested within a paraphyletic Schedonorus clade; evidence from the two separate analyses (Figs 1 & 2) suggests that Lolium is of recent origin and probably evolved from 'a European Schedonorus ancestor. Nucleotide substitution rates parallel the evolutionary relationships recovered in the phylogenetic trees for the broad-leaved Festuca lineages. The slowest rates shown by Drymanthele , Subbulbosae and Leucopoa s.s. reflect their basal positions in the Loliinae trees, whereas the ‘intermediately’ placed Amphigenes , Breviaristatae (F. californica) and Subulatae (F. subulata ) also show overall intermediate rates of nucleotide mutation. The trnL-F sequence of F. subulata constitutes a singular case as this taxon presents significant ■ differences in chloroplast rate mutation with respect to all remaining lineages except Cyno- surinae. The accelerated rate of mutation shown by F. subulata could also have affected its phylogenetic placement in the trnL-F trees due to long-branch attraction or site saturation effects. Relative rate ratios corroborate the recent origin of the Schedonorus /Micropyr op sis /Lolium lineage within the broad-leaved Festuca clade. Our analyses indicate that this group has the fastest mutational rates for both ITS and trnL-F sequences in a trend ranging from slow-evolving ‘European’ Schedonorus through intermediate-evolving Micro- pyropsis and ‘Maghrebian’ Schedonorus towards rapidly-evolving Lolium. ‘Maghrebian’ Schedonorus forms a clade of highly polyploid taxa presumably derived from more ancestral diploid and low polyploid ‘European’ Schedonorus ancestors (Borrill et al. 1977). Within Lolium, the annual taxa show highly accelerated mutational rates significantly different from most of the slow-evolving perennial lineages. The high levels of morphological variability detected in L. rigidum and L. canariensis , which moved some authors to describe different infraspecific and specific taxa out of those complexes (Terrell 1968; Scholz et al. 2000), are thus correlated with their higher susbtitutional rates. THE ‘FINE-LEAVED’ FESTUCA , VULPIA AND RELATED EPHEMERAL S The ‘fine-leaved’ Festuca are resolved as monophyletic in all separate and combined analyses (Figs 1-3). Resolution of the main groups correspond to that described by Torrecilla et al. (2004) and Catalan et al. (2004) for the Festuca + Vulpia + Related Ephemerals (FEVRE) group with the addition of a group of S. American taxa (. Festuca coromotensis, F. elviae, Hellerochloa fragilis) that belong to the clade of red fescues (. Festuca sect. Aulaxyper ) in the trnL-F based tree (Fig. 2). The four best supported clades correspond to those of the Festuca (+ Wangenheimia ) clade, the Aulaxyper + Vulpia p.p. (2x) clade, the Psilurus + Vulpia p.p. (4x-6x) clade, and the Loretia-Monachne-Spirachne + Festuca plicata clade (Figs 1-3). Surprisingly, Vulpia is resolved as polyphyletic in both nuclear and chloroplast trees. This striking finding especially affects the divergent and robust resolution obtained for representatives of typical section Vulpia : a clade of diploid taxa which is closely related to the Aulaxyper group whereas higher ploidy-level taxa link with Psilurus in an unrelated polyploid clade (Figs 1 & 2). Relative rates of nucleotide substitutions for the ‘fine-leaved’ Festuca groups also correlate well with their recovered phylogenetic relationships. The basal Eskia and Exaratae 64 CURRENT TAXONOMIC RESEARCH assemblages are shown to be the slowest-evolving groups within the clade followed by the intermediate-evolving Festuca , Aulaxyper and Hellerochloa lineages, and then by the rapid¬ ly evolving annual groups. Changes in rate heterogeneity vary greatly within the perennial fine-leaved groups, being more pronounced in the highly heterogeneous Aulaxyper group. Within this lineage, the high polyploid taxa (8x-10x) show the highest subtitutional rates in their ITS sequences (Results not shown). The ephemeral fine-leaved groups are the most rapidly evolving festucoids as indicated by their relative rate ratios; rate differences are highly significant for the Loretia assemblage. Homoplasy may be enhanced in these highly evolving lineages altering, therefore, the phylogenetic inference. However, the consistent Loretia /Monachne/ Spirachne /Apalochloa +F. plicata clade in both ITS and trn L-F analyses support a common ancestor for all Vulpia lineages (the 4 Loretia assemblage’) except for typical sect. Vulpia. The odd position of diploid and polyploid taxa of sect. Vulpia in the ITS and trnL-Y trees cannot be explained in terms of significant differences in rate heterogeneity between them (Table 2 B). MINIMUM-GENERATION-TIME AND RETICULATION/POLYPLOIDIZATION HYPOTHESES ARE THE LIKELY EVOLUTIONAY SCENARIOS FOR THE FESTUCOIDS The relative rate analyses designed to test the MGT hypothesis within the festucoids and their closest allies have demonstrated that there are significant differences between molecu¬ lar evolutionary rates of annual and perennial groups in the broadly sampled Loliinae. Thus, we can conclude that MGT mechanisms are operating in the rapidly evolving ephemeral groups of these grass lineages even if the biological factors that regulate these processes have not yet been deciphered (Gaut et al. 1997; Muse 2000). Changes in the evolutionary rates between slowly-evolving perennials and fast-evolving annuals are likely to be a consequence of a release in stabilized selection followed by the ephemeral groups which also facilitated the acquisition by them of rapid adaptive changes to new environmental habitats (Givnish 1997). Our study suggests that this evolutionary scenario has occurred in parallel along the two main lineages of broad-leaved and fine-leaved Festuca. Most interestingly, the substitu¬ tional rates of the annual representatives of the broad-leaved clade ( Lolium ) are similar to some substitutional rates of annual representatives of the fine-leaved clade ( Apalochloa , Psilurus/Vulpia 4x-6x, Vulpia 2x) although most of the fine-leaved ephemerals evolve at a higher pace {Loretia assemblage, Wangenheimia). The extended presence of annual lineages derived from the fine-leaved fescues and their higher evolutionary rates are indicative of their more recent origin. Transilience is also manifested along the perennial groups that show a shift from low to high rates of mutation. The increased evolutionary rates are correlated with increased levels of ploidy supporting the reticulation/polyploidization hypothesis tested here. Again, this evolutionary scenario has been developed independently along the two main clades of Loliinae, as exemplified by the Aulaxyper and the ‘Maghrebian’ Schedonorus groups within the ‘fine-leaved’ and ‘broad-leaved’ Festuca lineages, respectively. These two groups encompass highly polyploid taxa (8x-10x) that are presumably derived from their respective lower-ploidy-level relatives. Diploid perennial lineages display the lowest mutational rates in both ‘broad-leaved’ and ‘fine-leaved’ Festuca lineages, though the broad-leaved ones evolve significantly more slowly. Thus, according to previous evolutionary predictions by RELATIONSHIPS OF FESTUCA, LOLIUM , VULPIA 65 Hackel (1882) and Tzvelez (1971), the nemoral and broad-leaved taxa of subgen. Diyman- thele and sect. Subbulbosae probably constitute some of the oldest relict lineages of Festuca. The minimum-generation-time hypothesis was not significantly supported in the close subtribes Dactylidinae and Cynosurinae/Parapholiinae, though it might be caused by the lower taxon sampling within these groups or by other phenomena related to hybridisation and polyploidy that might have fostered the acquisition of the overall accelerated rates shown by these taxa. In general terms, our results agree with the conclusions drawn from the study of Gaut et al. (1997) who found relative support for the MGT hypothesis within the grasses and with respect to the fastest evolving subfamily Pooideae. Preliminary analyses also favour the speciation rate hypothesis within Loliinae, as indicated by the higher diversifying rates shown by the highly accelerated fine-leaved Loliinae lineages compared to the less-diversified rates of the slowly mutational broad¬ leaved ones. Diversification has been measured here with respect to the number of genera traditionally recognized within the ‘fme-leaved’ clade ( Ctenopsis , Festuca p.p., Helleroch- loa, Micropyrum, Narduroides , Ps Hunts, Vulpia p.p., Wangenheimia) and ‘broad-leaved’ clade (Dtymochloa (= Festuca subgen. Drymanthele), Festuca p.p., Lolium, Micropyropsis , Schedonorus ) that have been studied by us (cf. Holub 1984, 1998; Watson & Dallwitz 1992). Despite the lack in taxon sampling for a few more Loliinae genera not included in the present survey (i.e. Loliolum , Vulpiella ; cf. Clayton & Renvoize, 1986; Watson & Dallwitz 1992), it is still predictable that the fme-leaved clade would accumulate more traditionally recognized genera than the broad-leaved one. Torrecilla et al. (2004) interpreted the wide array of monotypic and small-sized genera described within the fme-leaved Festuca clade as a direct consequence of the higher speciation rates developed by these ephemeral groups. However, Gaut et al. (1997) found scant support for the speciation-rate hypothesis within the grass family except for the rapidly evolving Pooideae lineage that adjusted better to it. Our evolutionary test within Loliinae agrees with this partial result of Gaut et al. (1997). The implications that the acquisition of the annual habit and the increase in ploidy might have had in the evolution of the festucoids cannot be conclusively established yet as taxon sampling is still incomplete for several groups of the large genus Festuca and its satellite genera. However, from the data analyzed here, it can be hypothetized with confidence that the two evolutionary scenarios (the MGT hypothesis and the reticulation/polyploidization hypothesis) are likely to have occurred at different evolutionary times within the subtribe Loliinae. The switch from perennial towards annual life-cycle probably represents a more ances¬ tral evolutionary phenomenon, independently experienced within the two main lineages of the festucoids as manifested in the mostly diploid nature of the ephemeral lineages. On the other hand, reticulate processes involving recurrent hybridization and polyploidy constitute secondary evolutionary events that have affected some of the most recently evolved perenni¬ al lineages of broad-leaved and fme-leaved Festuca and a few annual fme-leaved ones. Although accelerated rates are significantly different across the annual lineages, the poly¬ ploid complexes also show relatively higher rates, indicating that reticulation has fostered the subtitutional rates of these groups through the addition of new gene-pools. Both scenarios agree, partly, with the speciation-rate hypothesis, as annuals and highly polyploid taxa show a wider array of taxa within them than their congeneric or cosectional relatives. However, whereas the annual lineages show distinctive traits that were used to classify them as different genera (Clayton & Renvoize 1986; Watson & Dallwitz 1992), the polyploid complexes are formed by a series of microtaxa that could hardly be morphologi- 66 CURRENT TAXONOMIC RESEARCH cally differentiated from each other (Markgraf-Dannenberg 1980). These evidences add support for an older diversification of the annual lineages from their respective common ancestors and for a secondary and recent divergence of the polyploid lineages. As a concluding remark, there is scope for speculation on future refinement of the present findings by characterising and subdividing the festucoid polyploid lineages into polyploid like-complexes when more detailed data become available. There is also increas¬ ing scope for the application of new chloroplast and nuclear genes to calculations of differences in substitution rates among Loliinae lineages in order to put the observed rate heterogeneity into a genome map context. ACKNOWLEDGEMENTS We thank Clive A. Stace and an anonymous reviewer for their valuable comments on this manuscript, to Amoldo Santos, Carlos Romero-Zarco, and Robert Soreng for providing us fresh and silicagel dried materials of some taxa, and to the curatorial staffs of BC, COLO, G, JACA, MA, ORT, PRC, SEV, W, and WA for sending us herbarium vouchers of the taxa under study for analysis. This work has been subsidized by a Spanish Ministry of Science and Technology grant (REN2003-02818/GLO project) to PC and supported by a Central University of Venezuela (CDCH) doctorate fellowship to PT. REFERENCES Ainscough, M.M., Barker C.M. & Stace C.A. (1986). Natural hybrids between Festuca and species of Vulpia section Vulpia. Watsonia 16: 143-151. Alexeev, E.B. (1977). K sistematike asiatskich ovsjaniz (Festuca). I. Podrod Drymanthele, Subulatae, Schedonorus, Leucopoa (To the systematics of Asian fescues (Festuca). I. Subgenera Drymanthele , Subulatae , Schedonorus , Leucopoa). Bjulleten ' Moskovskogo Obshchestva Ispytatelei Prirody, Otdel Biologicheskii 82(3): 95-102. Alexeev, E.B. (1978). K sistematike asiatskich ovsjaniz (Festuca). 2. Podrod Festuca (To the systematics of Asian fescues (Festuca). 2. Subgenus Festuca). Bjulleten' Moskovskogo Obshchestva Ispytatelei Prirody, Otdel Biologicheskii 83(4): 109-122. Alexeev, E.B. (1980). Novye podrody i sekzii ovsjaniz (Festuca L.) Sevemoj Ameriki i Meksiki (Festuca L. Subgenera et sectiones novae ex America boreali et Mexica). Novosti Sistematiki Nizshikh Rastenii 17: 42-53. Alexeev, E.B. (1981). Novye taksoni roda Festuca (Poaceae) is Meksiki i Zentrafnoj Ameriki (The new taxa of the genus Festuca (Poaceae) from Mexico and Central America). Botanicheskii 'Zhurnal (Moscow and Leningrad) 66: 1492-1501. Alexeev, E.B. (1986). Ovsjanizy (Festuca L., Poaceae) Venezuely, Kolumbii i Ekvadora (Festuca L. (Poaceae) in Venezuela, Colombia et Ecuador). Novosti Sistematiki Nizshikh Rastenii 23: 5-23. Barker, C.M. & Stace, C.A. (1982). Hybridization in the genera Vulpia and Festuca : the production of artificial Fi plants. Nordic J. Bot. 2: 435^444. Barker, C.M. & STACE, C.A. (1984). Hybridization in the genera Vulpia and Festuca (Poaceae): The characteristics of artificial hybrids. Nordic J. Bot . 4(3): 289-302. Barker, C.M. & Stace, C.A. (1986). Hybridization in the genera Vulpia and Festuca (Poaceae): meiotic behaviours of artificial hybrids. Nordic J. Bot. 6: 1-10. RELATIONSHIPS OF FESTUCA, LOLIUM, VULPIA 67 Barracough, T.G. & Savolainen, V. (2001). Evolutionary rates and species diversity in flowering plants. Evolution 55: 677-683. Barraclough, T.G., Harvey, P. & Nee, S. (1996). Rate of rbcL gene sequence evolution and species diversification in flowering plants (angiosperms). Proc. R. Soc. Lond. B 263: 589-591. Borrill, M., Kirby, M. & Morgan, W.G. (1977). Studies in Festuca. 11. Inter¬ relationships of some diploid ancestors of the polyploid broad-leaved fescues. New Phytol. 78: 661-674. Bousquet, J., Strauss, S.H., Doerksen, A.H. & Price, R.A. (1992). Extensive variation in evolutionary rate of rbcL gene sequences among seed plants. Proc. Natl. Acad. Sci. USA 89: 7844-7848. Catalan P., Torrecilla P., Lopez-Rodriguez J.A. & Olsmtead R.G. (2004). Phylogeny of the festucoid grasses of subtribe Loliinae and allies (Poeae, Pooideae) inferred from ITS and trnL_F sequences. Mol. Phylogenet. Evol. 31: 517-541. Charmet, G., Ravel, C. & Balfourier, F. (1997). Phylogenetic analysis in the Festuca- Lolinm complex using molecular markers and ITS rDNA. Theor. Appl. Genet 94: 1038-1046 Clayton, W.D. & Renvoize, S.A. (1986). Genera Graminum: Grasses of the World, Kew Bull. Addit. series XIII, Royal Botanic Gardens, Kew. Cotton, R. & Stace, C.A. (1976). Taxonomy of the genus Vulpia (Gramineae). L Chromosomes numbers and 'geographical distribution of the world species. Genetica 46: 235-255. Cotton, R. & Stace, C.A. (1977). Morphological and anatomical variation of Vulpia (Gramineae). Bot. Notiser 130: 173-187. Darbyshire, S.J. & Warwick, S.I. (1992). Phylogeny of North American Festuca (Poaceae) and related genera using chloroplast DNA retriction site variation. Canad. J. Bot. 70: 2415-2429. DUMORTIER B.C.J. (1823). Observations sur les Graminees de la flore Belgique. J. Casterman. Tournay. Belgium. Eyre-Walker, A. & Gaut, B.S. (1997). Correlated rates of synonymous site evolution across plant genomes. Mol. Biol. Evol. 14: 455-460. Gaut, B.S., Clark, L.G., Wendel, J.F. & Muse, S.V. (1997). Comparisons of molecular evolutionary process at rbcL and ndhF in the grass family (Poaceae). Molec. Biol. Evol. 14: 769-777. Gaut, B.S., Muse, S.V., Clark, W.D. & Clegg, M.T. (1992). Relative rates of nucleotide substitutions at the rbcL locus of monocotyledonous plants. J. Mol. Evol. 35: 292-303. Gaut, B.S., Morton, B.R., McCaig, B.C. & Clegg, M.T. (1996). Substitution rate comparisons between grasses and palms: synonymous rate differences at the nuclear gene Adh parallel rate differences at the plastid gene rbcL. Proc. Natl. Acad. Sci. USA 93: 10274-10279. Gaut, B.S., Tredway, L.P., Kubik, C., Gaut, R.L. & Meyer, W. (2000). Phylogenetic relationships and genetic diversity among members of the Festuca-Lolium complex (Poaceae) based on ITS sequence data. PI. Syst. Evol. 224(1-2): 33-53. Givnish T.J. (1997). Adaptative radiation and molecular systematics: Issues and approaches. In: Givnish T.J. & Systma K.J. (eds), Molecular evolution and adaptive radiation. Pp. 1-54. Cambridge University Press. Cambridge. UK. Gmelin, C.C. (1805). Flora badensis alsatica . Vol. 1, Muller, Karlsruhe. 68 CURRENT TAXONOMIC RESEARCH Hackel, E. (1882). Monographia Festucarum Europearum , Theodor Fischer, Kassel und Berlin, pp. 216, 4 lam. Hackel, E. (1887). Gramineae. In: H.G.A. Engler & K.A.E. Prantl, (eds), Die Natiirlichen Pflanzenfam ilien , Teil 2, Abteilung 2, Engelmann, Leipzig, pp. 1-97. HACKEL, E. (1906). Gramineae novae. Repertorium Novarum Speciarum Regni Vegetabilis 2: 69-72. Harper, J.A., Thomas, I.D., Lovatt, J.A. & Thomas, H.M. (2004). Physical mapping of rDNA sites in posible diploid progenitors of polyploid Festuca species. PI Syst Evol 245: 163-168. Hillis, D.M. & WIENS, J.J. (2000). Molecules versus Morphology in Systematics. Conflicts, Artifacts, and Misconceptions. In: J.J. Wiens (ed.), Phylogenetic Analysis of Morphological Data , Smithsonian Series in Comparative Evolutionary Biology. Smithsonian Institution Press, Washington., pp. 1-19. Holub, J. (1984). New Genera in Phanerogamae [1-3]. Folia Geobotanica et Phytotaxonomica 19: 95-99. Holub, J. (1998). Reclassifications and new names in vascular plants 1 . Preslia 70: 97-122. HUELSENBECK, J. P. & RONQUIST, F. (2002). MRBAYES. Bayesian inference of phylogenetic trees. V. 3.0. Bioinformatics 17(8): 754-755. Humphreys, M.W., Thomas, H.M., Morgan, W.G., Meredith, M.R., Harper, J.A., Thomas, H., Zwierzykowski, Z. & Guesuure, M. (1995). Discriminating the ancestral progenitors of hexaploid Festuca arundinacea using genomic in situ hybridisation. Heredity 75: 171-174. JAHUAR, P.P. (1975). Genetic regulation of diploid-like chromosome pairing in the hexaploid species Festuca arundinacea Schreb. and F. rubra L. (Gramineae). Chromosoma 52: 363-382. JAHUAR, P.P. (1993). Cytogenetics of the Festuca-Lolium complex 1993; Berlin Heidelberg: Springer. Jenkin, T.J. (1933). Interspecific and intergeneric hybrids in herbage grasses. Initial crosses. Journal of Genetics 28: 205-264. Krechetovich, V.I. & Bobrov, E.G. (1934). Festuca L. s. str. In: V.L. Komarov, R.Y. Rozhevits & B.K. Shishkin (eds), Flora SSSR. Vol. 2, Akademija Nauk SSSR, Leningrad, pp. 497-535. Krivotulenko, U. (1960). Novye sekzii roda Festuca L. (Generis Festuca L, sectiones novae). Bot. Mat. (Leningrad) 20: 48-67. Lehvaslaiho, H., Saura, A. & Lokki, J. (1987). Chloroplast DNA variation in the grass tribe Festuceae. Theor. Appl. Genet. 74: 298-302. Lewis, E.J. (1975). Festuca L. x Lolium L. = x Festulolium Aschers. & Graebner. In: C.A. STACE (ed.), Hybridization and the flora of the British Isles , pp. 547-552. Li, P. & BOUSQUET, J. (1992). Relative-rate test for nucleotide substitutions between two lineages. Molec. Biol. Evol. 9: 1 185-1 189. Malik, C.P. & Thomas, P.T. (1966). Karyotypic studies in some Lolium and Festuca species. Caryologia 19: 167-196. Markgraf-Dannenberg, I. von. (1980). Festuca L. Pp. 125-153. In: T.G. Tutin et al. (eds), Flora Europaea , Vol. 5. Cambridge University Press, Cambridge, U.K. Muse, S.V. (2000). Examining rates and patterns of nucleotide substitutions in plants. PI. Mol. Biol 42: 25—43. RELATIONSHIPS OF FESTUCA, LOLIUM, VULPIA 69 PIPER, C.V. (1906). North American species of Festuca. Contributions from the United States National Herbarium 10: 1-48. Posada, D. & Crandall, K.A. (1998). Modeltest: testing the model of DNA substitution. Bioinformatics 14(9): 817-818. Robinson, M., Gouy, M., Gautier, C. & Mouchiroud, D. (1998). Sensitivity of the relative-rate test to taxonomic sampling. Mol Biol Evol. 15: 1091-1098. Robinson-Rechavi, M. & Huchon, D. (2000). RRTree: Relative-Rate Test between group of sequences on a phylogenetic tree. Bioinformatics 16: 296-297. Sanderson, M.J. (1997). A nonparametric approach to estimating divergence times in the absence of rate constancy. Mol Biol Evol 14: 1218-1231. SCHOLZ, H., STIERSTORFER, C. & Gaisberg, M.V. (2000). Folium edwardii sp. nova (Gramineae) and its relationship with Schedonorus sect. Plantynia Dumort. Feddes Repertorium 111: 561-565. Soltis, D.E. & Soltis, P.S. (1999). Polyploidy: recurrent formation and genome evolution. TREE 14: 348-352. Soreng, R.J. & Davis, J.L (2000). Phylogenetic structure in Poaceae subfamily Pooideae as inferred from molecular and morphological characters: misclassification versus reticulation. In: S.W.L. JACOBS & J. Everett (eds), Grasses: Systematics and Evolution , CSIRO, Melbourne, pp. 61-74. Soreng, R.J. & Terrell, E.E. (1998). Taxonomic notes on Schedonorus , a segregate genus from Festuca ox Folium, with a new nothogenus, xSchedololium , and new combinations. Phytologia 83: 85-88. Stace, C.A. (1981). Generic and infrageneric nomenclature of annual Poaceae: Poeae related to Vulpia and Desmazeria. Nordic J. Bot. 1: 17-26 Stace, C.A. (1987). Hybridization and the plant species. In: K.M. Urbanska (ed.), Differentiation patterns in higher plants, pp. 115-127. Academic Press, London. Stebbins, G.L. (1956). Cytogenetics and evolution of the grass family. Am. J. Bot. 43: 890-905. Suzuki, J.l, Herben, T., Krahulec, F. & Hara, T. (1999). Size and spatial pattern of Festuca rubra genets in a mountain grassland: its relevance to genet establishment and r dynamics. J. Ecol. 87: 942-954. Terrell, E.E. (1968). A taxonomic revision of the genus Folium. U.S. Dept. Agric. Tech. Bull. 1392. Torrecilla, P. & CatalAn, P. (2002). Phylogeny of broad-leaved and fine-leaved Festuca linages (Poaceae) based on nuclear ITS sequences. Syst. Bot. 27(2): 241-251. Torrecilla, P., Lopez-Rodrjguez, J.A., Stancik, D. & Catalan, P. (2003). Systematics of Festuca sects. Eskia Willk., Pseudatropis Kriv., Amphigenes (Janka) Tzvel., Pseudoscariosa Kriv., and Scariosae Hack, based on analysis of morphological characters and DNA sequences. PI. Syst. Evol. 239: 1 13-139. Torrecilla, P., Lopez-RodrIguez, J.A. & CatalAn, P. (2004). Phylogenetic relationships of Vulpia and related genera (Poeae, Poaceae) based on analisis of ITS and trnL-V sequences. Ann. Missouri Bot. Gard. 91: 124-158. Tzvelev, N.N. (1971). K sistematike i filogenii ovsjaniz ( Festuca L.) flory SSSR. I. Sistema roda i oshoriye naprav’enija evoljuzii.. (On the taxonomy and phylogeny of genus Festuca L. of the U.S.S.R. flora. I. The system of the genus and main trends of evolution). Botanicheskii Zhurnal (Moscow and Feningrad) 56: 1252-1262. Tzvelev, N.N. (1982). Poryadok zlaki (Poales). Zhizn Rast. 6: 341-378. 70 CURRENT TAXONOMIC RESEARCH Tzvelev, N.N. (1999). Ob obeme I nomenklatura nekotorykh rodov sosudistykh rastenii Evropeiskoi Rossii. (On the size and nomenclature of some genera of the vascular plants of European Russia). Bot. Zhurn. 84(7): 109-118. Tzvelev, N.N. (2000). Combinationes novae taxorum plantarum vascularium. Nov it. Syst. Pl. Vase. 32: 181-183. WATSON, L. & Dallwitz, M.J. (1992). The Grass Genera of the World. C.A.B. International. Cambridge University Press. Cambridge. Wendel, J.F. & Doyle, J.J. (1998). Phylogenetic incongruence: window into genome history and molecular evolution. In: D.E. Soltis, P.S. Soltis & J.J. Doyle (eds), Molecular Systematics of Plants II. DNA Sequencing , Kluwer Academic, Boston, pp. 265-296. Wu, C.I. & Li, W.H. (1985). Evidence for higher rates of nucleotide substitution in rodents than in man. Proc. Acad. Natl. Sci. USA 82: 1741-1745. Xu, W.W. & Sleper, D.A. (1994). Phylogeny of tall fescue and related species using RFLPs. Theor Appl Gen 88: 685-690. Current taxonomic research on the British & European flora Bailey, J. & Ellis, R.G. (eds) 71-88 (2006), BSBI. London. 71 What use is sex? Rock sea-lavenders ( Limonium binervosum agg.) revisited Martin J. Ingrouille School of Biological and Chemical Sciences, Birkbeck University of London, Malet Street, London WC1E 7HX ABSTRACT Analysis of morphometric variation within and between populations of Limonium binervosum agg., rock sea-lavender has revealed extensive, statistically significant, differentiation between populations of the same infra-specific taxon, even those growing in close geographical vicinity. Morphological groups are largely supported by patterns of amplified DNA fragment length polymorphisms but genetical distances contrast markedly from phenetic/morphological distances. It is recommended that a more informal and simplified classification of the group is utilised. The evolution of the obligately asexual group is discussed and it is suggested that much of the observed pattern of variation may be due to heritable epigenetic differences. Keywords: Geographical variation, apomict, epigenetic, evolution. INTRODUCTION Limonium binervosum agg., the rock sea-lavender, is a complex of agamospermous taxa that grows on the coastal cliffs and at the margins of salt-marshes and shingle banks in the British Isles and western Europe from northern France to southern Portugal. In the British Isles there are four widespread and five narrow endemic species (Table 1). In addition infra-specific taxa at subspecific and varietal rank have been described, each also confined to a greater or lesser extent to clearly defined stretches of coastline (Ingrouille & Stace 1986), for example the subspecies of Z. recurvum illustrated in Fig. 1. The narrow endemics are confined to a short stretch of coast or even a single headland but have a markedly different morphology from adjacent populations, such as the clustered spike of L. paradoxum (Fig. 2). This complex hierarchical classification was designed to summarise a hierarchical pattern of relationships between major and minor variants (Ingrouille & Stace 1985) that was discovered from a taxometric survey of 153 native populations carried out between 1979 and 1981 (Ingrouille 1982). The relationships were discovered by multivariate statistical analysis of morphological characters measured mainly in plants cultivated under uniform conditions in the University of Leicester Botanic Gardens. Morphological characters from both the vegetative and flowering parts of the plant were measured. Cluster analysis and principal components analysis of populations was used to discover groups of populations and clusters were validated by comparison with other characters not included in the multivariate analyses, such as pollen morphology, ecology, chromosome number and most importantly geographi¬ cal distribution. 72 CURRENT TAXONOMIC RESEARCH TABLE 1. TAX A OF LIMONIUM BINERVOSUM AGG. NATIVE TO THE BRITISH ISLES Widespread species L. binervosum (G.E.Sm.) C.E. Salmon L. procerum (C.E. Salmon) Ingr. subsp. binervosum subsp. procerum Ingr. subsp. mutatum Ingr. var. procerum subsp. anglicum Ingr. var. medium Ingr. subsp. saxonicum Ingr. var. hibernicum Ingr. subsp. cantianum Ingr. var. cornubiense Ingr. subsp. sarniense Ingr. var. paramedium Ingr. var. sarniense var. wessexense Ingr. var. aurigniense Ingr. subsp. devoniense Ingr. var. sercquense Ingr. subsp. cambrense Ingr. L. britannicum Ingr. L. recurvum C.E. Salmon subsp. britannicum subsp. recurvum var. britannicum subsp. portlandicum Ingr. var. kelseyanum Ingr. var. portlandicum subsp. coombense Ingr. var. recurviforme Ingr. var. coombense var. kerryense Ingr. var. grandicaule Ingr. subsp. pseudotranswallianum subsp. transcanalis Ingr. subsp. humile (Girard) Ingr. subsp. celticum Ingr. var. humile var. celticum var. donegalense Ingr. var. pharense Ingr. var. pseudoparadoxum Ingr. Narrow endemics L. transwallianum (Pugsley) Pugsley L. loganicum Ingr. L. paradoxum Pugsley L. parvum Ingr. var. paradoxum L. dodartiforme Ingr. var. mutabile Ingr. FIGURE 1. Variants of L. recurvum : A. subsp. recurvum , B. subsp. portlandicum\ C. subsp . pseudotranswal- lianum; D. subsp scoticum It is now nearly 30 years since this work and it is timely to evaluate the success or failure of the classification. 1 Although the classification has been incorporated in the New Flora of the British Isles (Stace 1997) to subspecific rank it has not proved practicable. Botanists have found it difficult to identify plants, especially if they are from populations not included originally in the 1979-81 survey. 2 In an unpublished survey of L. binervosum agg. on mainland of Europe carried out by the author in the 1990s it proved difficult to incorporate European taxa (Z. dodartii and L. multiflorum) and other distinct but undescribed variants within the classification that had been designed for the British Isles. 3 Although only a small amount of molecular data has been obtained it has in some cases indicated quite different patterns of variation (Cowan et al. 1 998). Analysis of amplified fragment polymorphisms has shown some taxa to be more and other taxa to be less genetically distinct than expected from their morphology. For example the subspecies Z. binervosum subsp. anglicum is as distinct as any species within the aggregate and the narrow endemic species Z. parvum , Z. paradoxum and Z. transwallianum that are very distinct morphologically are very close to other taxa (Fig. 3). 74 CURRENT TAXONOMIC RESEARCH Figure 2. Variants of the narrow endemic L. paradoxum from Saint David’s Head (v.c. 45) in Pembroke¬ shire differing in spike morphology and chromosome number These difficulties raise several questions about the original study and the way the results were translated into a revised classification. At that time the computer based methods for multivariate analysis of large datasets were limited and populations rather than individual plants were utilised as the smallest operational taxonomic unit. The use of population means obscured any within population variation present, though it was thought at the time from direct observation in nature, that there was little intra-population variation. This impression was exacerbated by the apparent homogeneity of a small sub-sample of wild populations (no more that 10 plants sampled as cuttings from rosettes) that was cultivated in the botanic garden. WHAT USE IS SEX? 75 bin anglicum Figure 3. Tree (created by unweighted paired group method analysis (UPGMA) of genetic distances measured from amplified DNA fragment polymorphisms (AFLPs)): acronyms relate to taxonomic names (see Table 1) except for the names of sites of populations on continental Europe. Note the separation of L. binervosum subsp. anglicum (‘bin anglicum’) from other taxa, and the inclusion of narrow endemics ( L . loganicum, L. paradoxum with other L. binervosum and L. parvum and L. transwallianum with L. procerum/britannicum ) Originally, as well as 41 metric characters a large number of ‘shape’ characters calculat¬ ed from ratios of the metric characters were utilised. This is no longer recommended because of the possibility of biasing the results if significant correlations between measurements exist. In the original analysis great weight was given to the results of a cluster analysis but this may have exaggerated the distinction between variants. Plants were cultivated before they were scored and while analysis of cultivated plants reduced potentially confounding environmentally caused variation, the results may not have properly represented genotypic variation as expressed in nature. 76 CURRENT TAXONOMIC RESEARCH A number of more fundamental questions are also relevant. • What is the relationship of morphological data to other kinds of data and which, if any, should have primacy in developing classifications? • How can taxonomic rank be determined where data on breeding behaviour is not available, as in an asexual group like L. binervosum agg.? • Is it always possible to create a workable classification for the non-expert that properly reflects natural relationships? • What taxonomic rank, if any should be used for agamospermous variants? • What is the value of taxonomically naming asexual variants? With the questions listed above in mind it was thought worthwhile to re-analyse data collected in the original survey. MATERIALS AND METHODS 146 populations sampled (81 from both cultivated and wild plants, 57 from cultivated only and 8 from wild only) with up to 10 individuals from each population. 41 metric characters measured. More than 2240 individuals were measured for 18 floral and spikelet characters and about half that number also for scape and leaf characters (an additional 23 characters) (Ingrouille 1982). Multivariate analysis was carried out utilising SPSS for Windows. Principal Components Analysis and Discriminant Analysis were carried out both on popula¬ tion means and individuals of all species and separately on individuals of pairs of species. RESULTS The results of the principal components analyses emphasise the close relationship of species. The narrow endemic species form distinct clusters at the margins of the distribution of individuals of the widespread taxa (Fig. 4). Principal components analysis of individuals of the widespread species provides some support for the existence of the four major clusters/species (Fig. 5). Discriminant analysis, with species as the discriminator, effectively separates the species even when plants are analysed rather than populations (Fig. 6). Discri¬ minant analysis of the four widespread species including each pair of species one at a time is very effective at separating the species. If only 18 floral characters are used the degree of overlap between species is greater than if all 41 possible characters are utilised (Table 2) especially in a discriminant analysis of L. procerum and L. britannicum (Fig. 7). From the percentage of miss-classified individuals shown in Table 2 it is clear that the two closest widespread species are L. britannicum and L. procerum. They are close in floral and spike characters but L. britannicum differs from L. procerum in its smaller stature and leaf shape. Similarly discriminant analysis separates intraspecific taxa. For example the subspecies of L. binervosum sensu stricto in SE England are clearly separated (Fig. 8). L. binervosum subsp. saxonicum is the most distinct andZ. binervosum subsp. anglicum and L. binervosum subsp. binervosum are closest to each other. Analysis of individuals reveals unsuspected inter-population variation even within named infra-specific taxa. For example a discriminant analysis of the populations of L. britannicum subsp. britannicum from the N Cornwall coast (Fig. 9) separates several of these. Only one distinct population at Kelsey Head has been previously recognised as a WHAT USE IS SEX? 77 SPECIES * L. transvuallianum O I. recurvum "I" L. procerum ▼ L. p a n/u m L. paradoxum * L. loganicum * L. d o d a rtif c rm e % L. britannicum x L. binervosum Figure 4. Three dimensional scatter-plot of the first three factor scores from a principal components analysis of individuals of all species. Note the position of L. loganicum and L. transwallianum on the margins of the main cluster. Other views and plots of different combinations of factors similarly isolates the other narrow endemic species. c ■2 factor 3 Species ^ L. recurvum ° L. britannicum + L. procerum x L. binervosum Figure 5. Three dimensional scatter-plot of the first three factor scores from a principal components analysis of individuals of the four widespread species with each occupying a different area of the main plot. 78 CURRENT TAXONOMIC RESEARCH 6 □ 4 . SPECIES * L. recurvum 0 L. britannicum D L. procerum + L. binervosum Score 1 6 2 - 0 O □ -2 - □ -4 • co CD O o CO -6 -4 SPECIES * L. recurvum 0 L. britannicum D L. procerum + L. binervosum Scores 2 FIGURE 6. Two-dimensional scatterplots of discriminate scores from an analysis of the four widespread species : A (upper). Scores 1 and 2 showing the separation of L. procerum , L. binervosum and L. britannicum! L. recurvum', B (lower). Scores 2 & 3 showing separation of L. recurvum, L. britannicum and L. procerum/L. binervosum. WHAT USE IS SEX? 79 L. britannicum Sid. Dev = .93 Mean = -2.3 N = 260.00 -4.5 -2.5 . 5 1.5 3.5 L. procerum Std. Dev = 1 .05 Mean = 1 .0 N = 508.00 -4.0 -2.0 0.0 2.0 4.0 FIGURE 7. Histograms of the discriminant score from a discrimant analysis of L. britannicum and L. procerum. 80 CURRENT TAXONOMIC RESEARCH FIGURE 8. L. binervosum sensu stricto from SE England: distribution of infraspecific taxa A. subsp. binervosum , B. subsp. cantianum, C. subsp. anglicum, D. subsp. saxonicum, and separation of taxa by a discriminant analysis (overleaf). WHAT USE IS SEX? 81 L. binervosum subsp. D anglicum saxonicum ■ cantianum 3* bi nervosum FIGURE 8. L. binervosum sensu stricto from SE England: separation of infraspecific taxa by a discriminant analysis. distinct variety because of its smaller flowers. Another example is provided by the plants from populations of L. binervosum subsp. sarniense in the Channel Isles (Fig. 10). Again several populations form distinct clusters in a scatter plot of discriminant scores even though only two populations were formally recognised as distinct (SACC with decumbent scapes = var. aurigiense , and the dwarf SSVP = var. sercquense.). Analysis of variance can find many significant differences between these and other populations. Other similar examples could have been provided from L. recurvum on the Isle of Portland in Dorset or western Ireland and L. procerum in N Wales. DISCUSSION Reanalysis of the data utilising in individuals as operational taxonomic units (OTUs) chal¬ lenges what was previously understood about the L. binervosum aggregate in two ways: firstly the taxonomic treatment; and secondly what is understood about the evolution of the group. 82 CURRENT TAXONOMIC RESEARCH SITE * Rump's Point 0 New Poize ath □ Boscastle O’ T revose Head A Tubby's Head V Pentire Point < Pothoothan Beach > Outer Trevone + Kelsey Head n Gunver Head Chapel Porth ■ BoobyJs Bay Figure 9. Discriminant analysis of individuals from N. Cornwall of L. britannicum subsp. britannicum A. populations (Key to code in map 1 - Chapel Porth, 2 - Tubby’s Head, 3 -Kelsey Head, 4 - Pentire Point, 5 - Porthcothan Beach, 6 - Booby’s Bay, 7 - Trevose Head, 8 - Gunver Head, 9 - Outer Trevone, 10 - New Polzeath, 1 1 - Rumps Point, 12 - Boscastle). WHAT USE IS SEX? 83 SITE □ SSVP ■ SSPG + SJPP ^ SJPM * SJEC A SGP8 V SGMP < SGLC > SGJB A SGCM ’ SGBE 0 SACP ■ SACC Figure 10. Discriminant analysis of individuals from the Channel Islands: Sark, SSVP - Venus’ Pool, SSPG - Port Gorey; Jersey SJPP - Pldmont Point, SJPM - Point La Moye, SJEC - Elizabeth Castle; Guernsey, SGPB - Petit Bot, SGMP - La Moye Point, SGLC - La Corbiere, SGJB - Jerbourg, SGCM - Creux Mahie, SGBE - Belle Elizabeth; Alderney, SACP - Cachaliere Point, SACC - Castle de Clonque. TABLE 2. MISALLOCATED INDIVIDUALS IN DISCRIMINANT ANALYSES BETWEEN THE FOUR WIDE¬ SPREAD SPECIES OF L. BINERVOSUM AGGREGATE INCLUDING EITHER INDIVIDUALS FOR WHICH BOTH FLORAL AND SCAPE AND LEAF CHARACTERS HAD BEEN MEASURED OR. FOR A GREATER NUMBER OF INDIVIDUALS, THOSE IN WHICH FLORAL CHARACTERS WERE MEASURED. | £ g 2 vJ * £ o > sj o s: ’Ni >« > o -S 'JJ 3< <3 *4 <3 <3 Discriminant analysis utilizing all available characters No. of 180 530 260 67 individuals L. binervosum N 320 6 4 0 ' =3 iL r— 0.7% 0.7% 0.0% L. procerum 1 ° 990 78 36 1 c o 6.0% 2.9% 0.1% a c3 L. britannicum — P3 C J= C3 o c ■a cs a j- 490 62 260 0 7.7% 17.6% 0.0% L. recurvum •a o s- i — 330 34 63 30 .22 to Q .£ 5.2% 4.8% 3.7% 84 CURRENT TAXONOMIC RESEARCH TAXONOMIC TREATMENT This analysis supports the conclusions of the original 1980s analysis that utilised population means as OTUs; there is a pattern of variation that is correlated to geographical distribution that can be apportioned hierarchically to species and infra-specific taxa. It was this hierar¬ chical pattern that was translated into a classification. However the analysis of individuals emphasises the closeness of species. Either popula¬ tions are more mixed than previously suspected or, and this is more likely, the normal range of expression of some taxa overlaps with others. In the latter case in order to correctly name an individual it is necessary to examine the whole range of variation in a population. The two morphologically closest widespread species are L. procerum and L. britanni¬ cum. Neighbouring populations of both these species are found in western Britain and in places it is difficult to discriminate between them. At the least the analysis reported here provides support for the combination of these species perhaps with a relegation to the rank of subspecies. The other widespread species are more distinct but it does require a very wide range of characters to be scored to ensure correct identification. L. recurvum is clearly evolutionarily distinct; it is triploid compared to the tetraploid status of the other taxa, but in practise it may be confused in the field with small stature L .britannicum. The key character of the roughness of the scape has proved rather variable in nature and difficult to apply if the alternatives (rough and smooth) are not available for comparison. Nevertheless its phyloge¬ netic distinctness is supported by genetic data from an analysis of AFLPs. L. binervosum can be more confidently identified though this is largely because of it is mainly allopatric distribution in SE England to the other species in western Britain. However in Sussex and S. Devon there are two populations that disturb this simplicity. A population from Lanna- combe in S. Devon has the long spike and distant spikelets of L. binervosum and was described as L. binervosum subsp. mutatum but it has other features such as leaf shape that relate it to the neighbouring populations of L. procerum. In addition at Rottingdene (v.c. 14) in Sussex there is a possibly introduced population of L. procerum on the margin of the range of L. binervosum and very isolated from the nearest L. procerum populations in S. Devon. The analysis reported here indicates that perhaps L. binervosum agg. has been over- taxonomised, with taxa recognised at too high a rank, not an unusual situation in agamosper- mous groups where it has been practise to describe myriad microspecies. However there is a more fundamental outcome of these results that may undercut the whole classification. The results of an analysis of individuals reveals much greater variation within populations than previously noted and places in doubt the narrow endemic species that have previously been described. Within the widespread species even geographically adjacent populations of the same taxonomic variety can be discriminated by multivariate statistical techniques. Previ¬ ously some populations that had a readily utilisable character that enabled their identifica¬ tion, such as the smaller floral parts of L. britannicum var. kelseyanum and the lax scapes of L. binervosum subsp. sarniense var. aurigniense , were described, but the analysis reported here shows that these are no more distinct and no more deserving of taxonomic recognition than other populations. In this context the narrowly endemic species, two of which have been recognised for decades, L. transwallianum (Pugsley 1924) and L. paradoxum (Pugsley 1931) and L. parvum, L. dodartiforme and L. loganicum described in the 1980s (Ingrouille & Stace 1986), may be only variant populations with readily utilisable identifying characters. In contrast it was the discovery of a shared triploid condition in several other populations, WHAT USE IS SEX? 85 especially in W. Ireland, that expanded the narrow concept of L. recurvum of Salmon (1903) perhaps making this widespread species rather difficult to comprehend because of the morphological variation it then contained. The morphometric analysis of populations (Ingrouille & Stace 1985) and of individuals reported here is broadly supported by the results of the analysis of AFLPs but the genetic distance varies quite considerably from the morphometric distance. The AFLP analysis indicates L. parvum and L. transwallianum are genetically close to L. procerum! britannicum\ L. paradoxum and L. loganicum are close to L. binervosum sensu stricto; and A. dodartiforme is close to L. binervosum agg. from mainland Europe. The greatest genetic distances are found within L. binervosum sensu stricto , in which the subspecies are genetically more distant from each other than many other species are in the aggregate despite their morphological closeness. In particular L. binervosum subsp. angli- cum is as genetically distant from other L. binervosum subspecies as is L. recurvum , though it is morphologically close to L. binervosum subsp. binervosum. Leaving aside the question of which should have primacy in determining taxonomic delimitation and rank, genetic distance or morphological distance, this analysis indicates strongly that the formal recognition of many specific and infra-specific taxa in L. binervosum agg. in the British Isles may be of limited value. It may be more accurate and useful to adopt a more informal naming system, incorporating the sample site, a kind of polynomial, such as ‘the L. binervosum variant with a lax spike from Lannacombe Beach (v.c. 3), S. Devon' rather than L. binervosum subsp. mutatum. This may seem a more cumbersome approach but it is less misleading than trying to lever a complex nexus of relationships into the straitjacket of a hierarchical classification. THE EVOLUTION OF THE L. BINERVOSUM AGGREGATE IN THE BRITISH ISLES These results pose a different kind of problem. L. binervosum is an obligate agamosperm. This is indicated by the male sterility of plants (lack of pollen or poorly stained and distorted pollen) and by the presence of a pollen and stigma combination (A/Cob) that prevents self-pollination (Ingrouille 1982). The A pollen/Cob stigma is half of the dimorphic self¬ incompatibility system found in sexual species of Limonium and requires the alternative B pollen/papillate stigma combination for successful pollination (Baker 1966), but the latter is entirely absent in the British Isles. As a consequence the evolution of the L. binervosum aggregate has been entirely asexual. Nevertheless a complex hierarchical pattern of variation has arisen and the geo¬ graphical correlation of this pattern indicates strongly that it has occurred in situ within the last 10,000 years since the end of the last glacial period. How has it evolved so rapidly? Where is the source of variation that has provided the opportunity of evolutionary diver¬ gence? This has been a significant problem in trying to understand the evolution of this group. In part the results reported here, that much more variation is present within populations than was previously suspected, goes some of the way to answering this conundrum. In fact variability within taxa has been previously reported, for example in pollen morphology and flower colour, but this was previously thought not to be significant. In a small number of cases, where it included a clearly marked variant and it was so clear that it could not be ignored, it led to the description of varieties within the same population. For example at St David’s Head (v.c. 45) and in its vicinity three different variants, differing in spike morphol- 86 CURRENT TAXONOMIC RESEARCH ogy, were described (X. paradoxum var. paradoxum and var. mntabile and L. procerum var. paramedium). Multiple variants of L. recurvum from the isle of Portland were described. Now it is clear that these cases are just examples of a more widespread phenomenon in L. binervosum aggregate. Here is the material that natural selection can act upon. However where is the source of this variation? Genetic mutation may have played a large part. Released from a normal meiosis by agamospermy the rate of mutation may have been enhanced. Chromosomal variation provides evidence of one kind of mutation with differenc¬ es in ploidy level (triploid and tetraploid) and aneuploidy between plants and populations, but most tetraploids have 2n = 35 (4x-l) and only L. paradoxum has 2n = 33 (4x-3). However, mutation rates recorded from a range of plants are generally too low (Gustafsson, 195 1). A direct indication of rates of mutation in L. binervosum agg. is provided from the results of the study of AFLPs (Cowan et al. 1998). AFLPs are polymorphic markers in the most rapidly evolving part of the genome, in the repetitive DNA, but the level of AFLP variation between L. binervosum taxa is low in comparison to sexual species. It therefore seems unlikely that mutation rates are enhanced in L. binemosum aggregate. Especially in agamospermous taxa hybridisation has been invoked as a potential source of variation. The complex pattern of variation observed is seen merely as the consequence of residual sexuality and occasional sexual crossing, with asexual reproduction merely multi¬ plying and fixing new hybrid variants. Hybridisation has been suggested in Limonium among Mediterranean taxa (Erben 1978, 1979) and in the origin of sister species differing in pollen/stigma morph in the L. ovalifo/ium/L. auriculae-ursifolium group (Ingrouille 1985). However the complete lack of sexuality in L. binervosum agg. in the British Isles makes this hypothesis very unlikely, at least here. An alternative source of variation would be if the differences between taxa are not genetic but epigenetic, arising from differences in patterns of gene expression in develop¬ ment. For example different patterns of methylation of the genome modulate gene expression and are potentially heritable. There is considerable evidence for this Non- Mendelian/Lamarckian possibility (Jablonka & Lamb 1998). Data from an analysis of AFLP variation within and between L. binervosum aggregate populations (R. Cowan personal communication) indicates extremely low levels of variation, even between the most distinct morphological microspecies. The level of variation is an order of magnitude different from that detected in other agamospermous groups where residual sexuality is present ( Taraxacum , Sorbus) (Hughes & Richards 1989, Asker 1980) and provides strong support that morphological differences detected are epigenetic. It is becoming more clear that the relationship between genotype and phenotype is not direct but is modulated by a complex nexus of relationship between gene and environment in the course of development. An analogy is with the BIOS (basic input/output system) of computer systems that provides the link between the hardware and the software (Fig. 11). Some of the BIOS is burned or flashed into a ROM chip that is both non-volatile and read-only, some of the BIOS is included on ROM chips installed on adapter cards, and some of the BIOS are additional drivers loaded when the system boots up. Potentially the BIOS of L. binervosum plants can evolve much more rapidly. Plants can look more similar than their genetic distance indicates or more different than their genetic closeness indicates. How significant are developmental differences (the ‘BIOS’) in the evolution of organisms? It may allow organisms to rapidly adapt to different circumstances and it may alter the rate of genetic change by shielding genetic variation from selection. There is direct evidence that some variation detected is potentially of ecological significance, WHAT USE IS SEX? 87 system a system b nonstandard^ interface hardware a. hv^ hardware standard interface — k standard interface operating system operating system (API) application application program program genotype genotype genotype kvN development development development phenotype phenotype phenotype environment environment environment Figure 11. An analogy for the relationship between genotype and phenotype from the operating systems of personal computers: A. the BIOSIS mediates the link between the computer hardware and software allowing different hardwares to run the same software or the same software to run different kinds of software; B. the biological BIOSIS, the developmental nexus, permits plants with different genotypes to survive in similar circumstances, or with the same genotype to survive in different circumstances. i.e. has been selected (Ingrouille 1982). When grown together variants differ in time of flowering; those with a natural western distribution flower earlier than those with an eastern distribution. Variants differ in their rate of seed germination. Variants normally found above the margins of salt-marshes are more sensitive to sea-water than those normally found growing on cliffs. If developmental differences are inherited in a significant degree then neither genetic distance nor morphological distance (as a measure of phenotypic distance) is an entirely reliable indicator of species limits, bringing us back to the question of what is more important in classification at species rank and below - the phenotype or the genotype? The phenotype is more fully and more readily measurable as a totality - it also connects the plant to its environment - whereas only a miniscule part of the genotype can be measured and the part that is measured is likely to have no relevance to the plant in its environment. However the genetic evidence provides a strong break on the impulse to over-taxonomise a variable species group, as has happened in the Limonium in general and in the L. binervosum aggregate in particular. 88 CURRENT TAXONOMIC RESEARCH REFERENCES Asker, S. (1980). Gametophytic apomixes: elements and genetic regulation. Hereditas 93: 277-293. Baker, H.G. (1966). The evolution, functioning and breakdown of heteromorphic incompatibility systems I. The Plumbaginaceae. Evolution 20: 349-368. Cowan, R., Ingrouille, M.J. & Dolores Lledo, M. (1998). The taxonomic treatment of agamosperms in the genus Limonium Mill., (Plumbaginaceae). Folia Geobotanica 33: 353-366. Erben, M. (1978). Die Gattung Limonium im Siidwestmediterranen Raum. Mitteilungen Botanische Staatsammlung Miinchen 14: 361-631. Erben, M. (1979). Karyotype differentiation and its consequences in Mediterranean “Limonium”. Webb ia 34: 409-417. GUSTAFFSON, A. (1951). Induction of changes in genes and chromosomes. II. Mutations, environment and evolution. Cold Spring Harbour Symposium of Quantitative Biology' 16:263-281. Hughes, J. & Richards A.J. (1989). Isozymes, and the status of Taraxacum (ASTERACEAE) Agamospecies. Botanical Journal of the Linnean Society > 99(4): 365- 376. INGROUILLE, M.J. (1982). The colony structure and taxonomic characterisation of apomictic Limonium (Plumbaginaceae) in the British Isles. PhD thesis, University of Leicester. INGROUILLE, M.J. (1985). The Limonium auriculae-ursifolium (Pourret) Druce group (Plumbaginaceae) in the Channel Isles. Watsonia 15: 221-229. Ingrouille, M.J. & Stace, C.A. (1985). Pattern of variation of agamospermous Limonium (Plumbaginaceae) in the British Isles. Nordic Journal of Botany 5: 1 13-125. Ingrouille, M.J. & Stace, C.A. (1986). The Limonium binervosum aggregate (Plumbaginaceae) in the British Isles. Botanical Journal of the Linnean Society 92: 177-217. Jablonka, E. & Lamb, M. (1998). Epigenetic inheritance in evolution. Journal of Evolutionary Biology 11: 159-183. PUGSLEY, H.W. (1924). A new Statice in Britain. Journal of Botany 24: 129-134. PUGSLEY, H.W. (1931). A further new Limonium in Britain. Journal of Botany 69: 44^-7. SALMON, C.E. (1903). Notes on Limonium. Journal of Botany, 41: 64-74. Stace, C.A. (1997). New Flora of the British Isles (second edition). Cambridge University Press, Cambridge. Current taxonomic research on the British & European flora Bailey, J. & Ellis, R.G. (eds) 89-1 10 (2006), BSBI. London. 89 How many orchid species are currently native to the British Isles? Richard M. Bateman Jodrell Laboratory, Royal Botanic Gardens Kew, Richmond, Surrey, TW9 OED, UK ABSTRACT The British Isles probably host the most intensively studied flora in the world, and within that flora the - Orchidaceae has long been the most intensively studied family. Nonetheless, molecular phylogenetic studies performed only during the last decade have revolutionised our understanding of species relationships among European orchids, eliminating former monotypic genera such as Aceras , Coelo- glosswn and Hammarbya, apportioning many former Orchis species to expanded concepts of Anacamptis and Neotinea , and combining ‘ Listera ’ into Neottia. Emphasis has now switched from species comparison to species delimitation, integrating morphometric approaches with population genetic techniques via simultaneous ordinations. Early results suggest that several distinct speciation mechanisms operate within the British and Irish orchid flora, and challenge the validity of several ‘Schedule 8’ species. No meaningful differences exist between British Dactylorhiza ‘ lapponica ’ and D. ‘ traunsteinerV , and neither represents the same allopolyploid speciation event as D. traunsteineri from the type locality in Austria. Also, contrary to the recent Atlas of the British & Irish flora , D. majalis s.s. does not occur in the British Isles. By contrast, three habitat ‘races’ within Gymnadenia merit species status. Epipactis ‘ youngiana ’ is not reliably distinct from E. helleborine, whereas the autogamous E. leptochila and E. dunensis both warrant species status, alongside E. sancta recently described from Lindisfame. Controversial taxa are either widely recognised but lack biological cohesion (Emperor’s New Clothes species), rarely if ever recognised but possess biological cohesion (Cinderella species, including Robinson Crusoe species recently diverged on islands such as Dactylo¬ rhiza ebudensis ), or are migrating northward, presumably in response to climate change (Bleriot species, such as Serapias parviflora). Recent arrivals by origination or migration are partly negated by extirpation of longer established species, notably Spiranthes aestivalis and arguably Epipogium aphyllum. Present evidence suggests that the orchid flora of the British Isles (excluding the biogeo- graphically French Channel Islands) currently consists of 52 species in 20 genera; these taxa are herein reclassified in anticipation of the third edition of Stace’s Flora. Keywords: classification, critical groups, DNA, Europe, morphometries, Orchidaceae, ordination, phylogenetics, population genetics, speciation. INTRODUCTION I have chosen to mark the notable occasion of Clive Stace’s retirement on my home turf, by preparing a benchmark overview of recent insights gained mainly from DNA-based studies into the biology, systematics and evolution of the British and Irish orchid flora. This text connects with my Presidential address to the UK Hardy Orchid Society (Bateman 2004). Here, I will focus especially on how best to integrate a wide range of morphological and molecular data in order to circumscribe species and infraspecific taxa, highlighting the most species-rich orchid genera in the British Isles, Epipactis and Dactylorhiza. 90 CURRENT TAXONOMIC RESEARCH PERSPECTIVES ON EVOLUTION Evolution can be viewed in two ways. We are most familiar with seeing it viewed laterally, morphological divergence among species being plotted against time as the vertical axis to generate the familiar tree motif (Fig. 1, bottom). However, we can also usefully view the present-day products of evolution from above and at ‘higher magnification’, seeking mor¬ phological gaps among sets of individuals representing particular populations (Fig. 1, top); these gaps should in theory reflect barriers to gene exchange between species. Seen from above, species are less frequently visualised as a tree but more often as clusters of points on two-dimensional ‘ordination’ plots; these resemble the simple graphs of the schoolroom but are based on more sophisticated statistical methods that summarise many more variables. PRESENT DAY Figure 1 . Comparison of perspectives of evolution as viewed laterally through time, dominantly repre¬ sented by the tree motif, and viewed perpendicular to the present time-plane, when taxa are best represented as clusters of individuals separated by discontinuities Evolution inevitably involves changes in the appearance or biological ‘behaviour’ of a species that are caused by one or more genetic modifications. Thus, in order to understand evolution, we need to describe both the morphology and the molecular genetics of each organism, and then compare the patterns revealed by these complementary sets of tech¬ niques. When comparing species by generating evolutionary trees (phylogenetics) we tend to represent each species with a single supposedly typical individual (an approach to sampling that can be termed typology, analogous to the type specimens of traditional taxonomy: e.g. Stace 1989). Consequently, the morphological data most appropriately ORCHID SPECIES NATIVE TO THE BRITISH ISLES 91 employed in such studies reflect discrete characters that are either present or absent, while corresponding genetic data are sequences of bases from particular regions of the DNA of the plants (both reviewed in detail by Bateman 1999, 2001). However, when attempting to circumscribe species, we need information from much larger numbers of individuals sampled from across the geographical range of each suspected species. Morphological data are diverse, including counts and measurements of particular structures; similarly, molecular genetic approaches are more diverse, including not only sequences but also techniques that dissect the DNA into comparable fragments and then measure with great precision their contrasting sizes. Such population-level studies are more labour-intensive, and because of the necessity to sample extensively within species they cannot encompass nearly as many species as tree-building approaches. In practice, each of these approaches feeds relevant information into the other; indeed, this feedback is reflected in most of the case-studies described below. We are fortunate that the orchids native to the British Isles have in the last decade probably been analysed more extensively using these techniques than any other plant family anywhere else in the world. This collective success reflects strong research collaborations between the three main plant systematics laboratories in Britain (Royal Botanic Gardens Kew, Royal Botanic Garden Edinburgh, and latterly Natural History Museum), working together with other universities located further afield (e.g. Lund, Naples, Estonia, Beijing). The initiative was also driven in part by materials supplied to the research team by ‘amateur’ orchidologists, particularly members of the UK Hardy Orchid Society and the Botanical Society of the British Isles. This review unashamedly focuses on these UK-based successes. MOLECULAR AND MORPHOLOGICAL PHYLOGENIES: SPECIES COMPARISON The results of studies constructing evolutionary trees of the tribes Orchideae ( Orchis , Dactylorhiza and their relatives) and Neottieae ( Epipactis , Neottia and their relatives), revolving around Kew’s comprehensive Genera Orchidacearum project, have emerged gradually over the last seven years (e.g. Bateman et al. 2003, 2004). The original sequence data derived from the ITS region of the nuclear genes, which is inherited from both parents of a plant, have since been supplemented with sequences from two regions of the plastids (mainly chloroplasts), which are inherited only from the mother. The two different clusters of DNA (termed genomes) in the nucleus and plastids are under different constraints and so can inform us about different aspects of the evolution and genealogy of the plants being analysed. The resulting phylogenies are considered more reliable when these contrasting sources of sequence data are broadly in accord. Comparison of the nuclear and plastid results confirms the original ITS-based interpre¬ tations (Fig. 2) in all cases. ‘Acer as' anthropophora is in fact a true ‘anthropomorphic’ Orchis. In contrast, ‘ Orchis ’ ustulata has no close relationship with Orchis purpurea , being part of the group epitomised by Neotinea macidata , and ‘ Orchis ’ morio has no close relationship with Orchis mascula , being nested between Anacamptis pyramidalis and A. (‘ Orchis ’) laxiflora. ‘ Coeloglossum ’ viridis is actually a near-basal diploid Dactylorhiza. More recent discoveries include the fact that Neottia nidus-avis is closely related to, and may have evolved from, ‘ Listera ’ ovata ; species formerly in the genus Listera should therefore be transferred to Neottia (Bateman et al. 2005). Within Tribe Malaxideae, a case could be made for sinking Hammarbya paludosa into Liparis, but current evidence suggests that it is 92 CURRENT TAXONOMIC RESEARCH O - o X X • X o Ophrys sphegodes Ophrys fuciflora Ophrys apifera Ophrys insectifera [Serapias lingua] Serapias parviflora Anacamptis (Orchis) morio Anacamptis pyramidalis [ Anacamptis (Orchis) laxiflora ] Himantoglossum hircinum Neotinea (Orchis) ustutata Neotinea macuiata Orchis (Aceras) anthropophora Orchis simia Orchis militaris Orchis purpurea Orchis mascuia Pseudorchis albida Piatanthera bifoiia Platanthera chiorantha Gymnadenia conopsea Gymnadenia borealis Gymnadenia densiflora Dactyiorhiza (Coeioglossum) viridis Dactyiorhiza fuchsii Dactyiorhiza macuiata Dactyiorhiza occidentalis Dactyiorhiza purpurella Dactyiorhiza ebudensis Dactyiorhiza traunsteinerioides Dactyiorhiza praetermissa Dactyiorhiza incarnata Herminium monorchis Spiranthes aestivalis Spiranthes spiralis Spiranthes romanzoffiana Goodyera repens Hammarbya paludosa Liparis loeselii Corallorhiza trifida Epipogium aphyllum Epipactis sancta Epipaciis dunensis Epipactis leptochila Epipactis helleborine Epipactis purpurata Epipactis phyllanthes Epipactis atrorubens Epipactis paiustris Neottia nidus-avis Neottia (Listera) ovata Neottia (Listera) cordata Cephalanthera damasonium Cephalanthera longifolia Cephalanthera rubra Cypripedium calceolus Figure 2. Grafted aggregate phylogeny of all British and Irish orchid species, based on several more focused analyses. Thick branches denote the more reliable relationships. Crosses indicate obligate myco- heterotrophs; spots indicate obligate autogams, circles indicate facultative autogams. Circled numbers indicate sources of component cladograms: 1 Soliva et a/. 2001, 2 Bateman & Hollingsworth 2004, 5 Bateman et al. 2003, 3 Bateman et al. in prep., 4 Pillon et al. in press., 6 Salazar et al. 2003, 7 Salazar et al. in prep., 8 Bateman et al. 2005, 9 Cox et al. 1997; 10 Dressier 1993, Cameron et al. 1999, Freudenstein & Rasmussen 1999. ORCHID SPECIES NATIVE TO THE BRITISH ISLES 93 sufficiently isolated and distinct to warrant maintenance as a monotypic genus (Salazar, Chase & Cribb in prep.; contra Bateman 2004). The almost complete coverage of species achieved now greatly facilitates further sys- tematics research, both within and outwith the British Isles. Firstly, it is easy to place phylogenetically the few remaining unplaced Eurasian species. For example, analysis of the contentious Balkan taxon ‘ Pseudorchis ’ frivaldii clearly demonstrates that this species is nested within Gymnadenia , close to (but apparently not sister to) the former genus Nigritella (Bateman, Rudall & James 2006). This evolutionary tree shows that floral reduction has occurred three times in Continental Europe within Gymnadenia (it has already been demon¬ strated to have occurred separately in ‘ Nigritella ’ and G. odoratissima). Secondly, the evolutionary tree provides a robust framework for re-examining records of natural hybridi¬ sation (cf. Stace 1975). For example, applying molecular phylogenetic techniques to a suspected hybrid from Mallorca new to science, between Anacamptis (formerly Orchis) fragrans and A. robusta , demonstrated not only that its parentage had been correctly identified but also that A. fragrans was its mother and had passed on to the adjacent hybrid more of its morphological characteristics than had its father, which was located at least 100 m distant (Bateman & Hollingsworth 2004). Such techniques are now being applied to hybrids in more challenging genera such as Dactylorhiza and Ophrys , as well as explaining degrees of historical introgression among the ‘anthropomorphic’ Orchis species O. anthro- pophora, O. simia , O. militaris and O. purpurea (Qamaruz-Zaman 2000; Bateman in press; Fay et al. in prep.), three of which are now awarded Schedule 8 conservation status. MORPHOMETRICS AND POPULATION GENETICS: SPECIES DELIMITATION Armed with both morphometric and population genetic tools, we can in theory re-examine the status of all of the supposed orchid species presently regarded as native to the British Isles. EPIPACTIS Several techniques have already been applied to one of the most troublesome genera, Epipactis (Squirrell et al 2001, 2002; Bateman et al. 2005). Sampling across Europe indicates at least a dozen independent and randomly-distributed origins of self-pollinated lineages from within the cross-pollinated E. helleborine complex. Each self-pollinated line is genetically distinct and, because of being self-pollinated, possesses exceptionally low genetic diversity. However, this supposed genetic weakness has not prevented species such as E. phyllanthes and E. leptochila from becoming widespread across western Europe. Epipactis dunensis is a British endemic that is genetically distinct from the more widespread E. leptochila ; moreover, the single small population of Epipactis on the dunes of Lindisfame in northeast England also appears sufficiently distinct genetically to provide some support for Delforge’s molecularly-inspired decision to describe it as a new species, E. sancta (Colour Plate If) (Delforge & Gevaudan 2002). However, another supposed British endemic Epipactis , E. ‘ youngiana\ does not pass muster. First described in 1982 and soon elevated to Schedule 8, it was subsequently explored using protein-based allozyme analysis (Harris & Abbott 1997). Results indicated that, in any one population, E. youngiana could not be separated effectively from the inevitably co-existing populations of E. helleborine\ in other words, individuals assigned to 94 CURRENT TAXONOMIC RESEARCH E. youngiana were merely plants of E. helleborine with unusually small rostella that consequently were unable to guard against self-pollination. This observation has since been confirmed using more sophisticated sequence-based population genetic techniques (Hollingsworth et al. 2006). Current evidence suggest two main causes of diversification of British Epipactis : mutationally-driven rostellum loss, and expansion into subtly different habitats and soils that may in part reflect switches of mycorrhizal partners (Bidartondo et al. 2004; Bidartondo, Reid & Bateman unpubl.). DACTYLORHIZA An even broader panoply of molecular techniques has been applied to the most taxonomical- ly troublesome of all of Britain’s orchid genera, Dactylorhiza. Many authors ascribe the morphological complexity in the genus to ‘hybridisation’. There is some truth in this statement, but the most crucial mode of hybridisation in Dactylorhiza is when two morpho¬ logically and genetically contrasting diploid species (most commonly D. incarnata s.l. and D. fuchsii s.l.) simultaneously hybridise and double their chromosome number in the progeny. This process, termed allopolyploidy (e.g. Stace 1975), immediately confers on the progeny fertility and at least partial genetic isolation from their parents. Key questions to address are therefore how frequently this process results in successful establishment of the resulting stabilised hybrid lineages, and which species is the mother and which the father in each case. By comparing results from a wide range of different analytical techniques (Table 1), it has become clear that members of the D. incarnata s.l. and D. fuchsii s.l. groups have combined repeatedly to generate large numbers of subtly distinct ‘species’ (Hedren et al. 2001; Pillon et al. subm.). In Greece and Asia Minor, D. euxina replaces D. incarnata and D. saccifera replaces D. fuchsii as parents of the allopolyploids (Hedren 2001, 2003), whereas in Ireland and northwest Scotland D. maculata sometimes replaces D. fuchsii. A similar situation characterises north-west Russia, where D. baltica replaces the morphologi¬ cally similar northern British/Nordic D. purpurella as the dominant allotetraploid (Shipunov & Bateman 2005). In this case, the lip shape (as summarised via landmark analysis) of an individual correlates well with species identity. The strong first axis separates D. fuchsii from the less deeply-lobed D. incarnata and D. maculata , and places D. baltica (the allotetraploid derivative of hybridisation between D. fuchsii and D. incarnata ) between its two parental diploids. Moreover, lip shape can successfully be related to the proportion of alleles of the Internal Transcribed Spacer (ITS) nuclear ribosomal DNA region estimated to have been inherited from each parental diploid by D. baltica (Fig. 3). This technique should now be applied to British and Irish populations. As with Epipactis, some peripheral populations of Dactylorhiza have proved to be both recently evolved and genetically unique. A good example is D. ebudensis (Colour Plate lb), found only on the Hebridean island of North Uist, which is unusual among British allotetra¬ ploid dactylorchids in having D. incarnata rather than D. fuchsii as its mother (it appears to be D. incarnata coccinea x D. fuchsii hebridensis ). Also, some allopolyploid species have proved to have multiple origins; for example, D. traunsteineri has separate origins in the Alps (where the type locality is situated), Scandinavia and the British Isles (Hedren et al. TABLE 1. DIPLOID (BOLDFACE) AND TETRAPLOID SPECIES OF DACTYLORHIZA FROM THE BRITISH ISLES GROUPED ACCORDING TO SEVEN DIFFERENT KINDS OF ANALYSIS. Asterisked studies were based largely on Continental material. ORCHID SPECIES NATIVE TO THE BRITISH ISLES 95 (/> 0. O' I QC O 0. ;o ■*-> ro * o o CM TO 0 Q 5 CB 5 0 O c CB CM CD =5 a * O "C 0 O E *2 V) ■c o >5 IS) 0 CB if) O o 1 ■V C/5 CB C/D 0 0 l— Q. "CB 0 C o CB X ro > o YZ 2 2 IS S c CB O c $2 co a _ 05 0 03 3 =5 C -o O 0 V) ;6 ,C s .52 cb a 3 § 3 03 £ o 2 o 5 o CO 0 03 0 C 0 to a 3 CD 0 0 0 .2 •2 E c ^ O 05 .0 Q. 0 g g-gS 3- Q.CN. 0 .CO 0 0 ==: C 2 £ 8-E 3 0 a o CM 0 0 0 05 :s ■4- 0 0 2-5 0 0 O c 0 3 CT 0 0 0 cb E o jz c 0 Q o a c 0 E 0 c o CO .a 3 • CL C 3 - 0 0 0 CO 0 § -C CO CM O IS 0 5 0 o c 0 is c 0 :3 0 0 •c 0 .3 O c 0 ■g o Q. i — i 0 "O J= ; 0 £ £ = 0 < 0 0 0 u c 0 3 cr 0 0 CO 0 _0 o 3 CO o o CM 0 C Q5 Bs c 2 TO 05 § 0 = CO 0L 0 x 0 ■+—> > 0 YY IS 0 0 P *2 C _ 0 0 03 3 3 £. -Q ,co •52 0 co co 2 2 2 -Q S-E 3 0 3 O 0 Q. O 0 $ ,c s .co 0 IS S § 3 -Q I ^ E o 0 0 0 ■2 CM 0 0 0 ^ 0 0 -C 0 -g o a-: 0 0 'VJ •*— < O C CM 0^0 o. o x: 0.0 u 0 O 3 3- cv c: 0 a. _i LL < O O CM 0 0 c -0 03 0 X 0 x ro -*— < > 0 YY IS ^ IS 5 c CB 0 o a r» W ■5 O 0 ■6 'c 0 •c o .3 0 .5 C 0 .°0 S'l I las S? 0-0 -3^0 .co co 3 E -Q Is Q.CV 0 0 E N O o o CM * CO 05 05 c 0 03 0 X "O 0 0& c JD n si 0 .s> G3 03 i- CO O 3 Q. c 3 0 O c c LU 0 X 0 -4— * > 0 ZL 2 0 o — ’ 3 c 0 c ■4— » c o O IS 0 0 + Ec^ 0 0 Z) 0 3 — E? •3 O C 0 c L— i C o O 0 ■6 'C s 0 ■c u .3 0 0 0 l -E .0 0 £ 2-2, Q.CO 0 0 ?3 O a ^ c 0 .52 E'S c i&l 3 Q 3 pa cb b 3= 0 .0 0 0 1 | E-E 3 0 Q. O 0 O 0 E o £ a t. o c 0 Q oQ * CO 05 ' 05 0 0 C _ 0 0 « c 0 0 CO ^ 00 T3 0 05 CD CO CL 0 X 0 4— » > 0 ■A 2 2 2 E c 0 0 o a ■E b V) ■6 'C ■*«> 0 X 0 > 0 0 0 ■c o 5 0 0 0 0 0 £ O a 0 0 03 .5 a 0 .a 0 0 .52 o co a a O <35 S 0.-0 a o 3 Cp 0 05 b A 0 ■52 0 0 a = 100000 m (windspeed of 5 m/s=18km/h, 5 h half-lifetime gives around this number); C is a normalisation constant, b is the power-law exponent, c is a constant describing the distribution ‘shoulder’ and x is the distance from the field (m). Combination of isolation profile data with the airborne pollen decay relationship allowed long-range wind-borne poljen delivery to B. rapa populations to be estimated. Use of this value to calculate hybrid seed set relies on the relative competitiveness of oilseed rape pollen. This was estimated empirically using hybrid frequencies observed over short distances where pollen densities were known. SPATIALLY EXPLICIT PREDICTIONS OF HYBRID ABUNDANCE The final part in the process of amassing a national scale model of hybrid abundance and distribution is to assemble the distribution profiles of both species, together with hybridization rates calculated for each isolation distance, to compile a spatially explicit prediction of hybrid abundance. Wilkinson et al (2003) combined data in this way to produce the first estimate of this kind for gene flow from Brassica napus to B. rapa in the United Kingdom. They calculated that approximately 48 000 hybrids form each year in the UK between cultivated oilseed rape and B. rapa. However, the significance of these F i hybrids rests largely on their ability to survive and produce introgressed B. rapa plants, and also on the capacity for subsequent spread of transgenes into other populations. It is in the latter of these two areas that we will now focus. THE SCOPE FOR TRANSGENE SPREAD AFTER HYBRIDIZATION AND INTROGRESSION There are some key features of the distribution of waterside B. rapa that provides important clues over the likely nature and speed of transgene movement between B. rapa populations following initial hybrid formation. The first of these is that waterside B. rapa is markedly more abundant along rivers than along canals, with a mean of 1 1 .4 populations over 1 km of river compared to a mean of 0.09 populations/km for canals (Wilkinson et al. 2003). A tentative explanation of this observation can be made by careful examination of the phenom¬ enon at sites where canals and rivers coincide. For example, this discrepancy is even apparent in instances where canals runs parallel to the course of the feeding river or when sites 118 CURRENT TAXONOMIC RESEARCH upstream and downstream of river feed points into canal systems are compared. One obvious difference between connected canals and rivers lies in the far greater propensity of the latter to flood, especially during the winter. Canals characteristically possess numerous overflow points to reduce the incidence of flooding; a feature generally lacking from natural rivers. However, navigable rivers do contain similar points of flood control immediately upstream of locks that are used for boat navigation. This is necessary to prevent flooding of the locks. However, the practice also means that comparisons can be made of B. rapa abundance in adjacent stretches of river where flood control is and is not evident. When this was done for 1 7 locks on the river Thames, all but three exhibited a much higher abundance of B. rapa in the 1 00 m downstream of the locks (where flooding is uncontrolled) compared to that seen in the 100 m upstream section (Fig. 1). Interestingly, in the exceptions, the flood control point was very close to the lock and B. rapa populations were upstream of this point. 120 -I 100 - □ Plants above lock ■ Plants below lock Figure 1. Relative abundance of Brassica rapa plants upstream (white) and downstream (black) of Locks along the river Thames. Further support for a role of flooding in determining the presence of waterside B. rapa can be taken from the location of numerous populations that are distant from the riverbank but are nevertheless located within the limit of the mean five year flood predictions provided by DEFRA. This finding has significance as it suggests that seed dispersal may be important in effecting gene exchange between populations as well as for population establishment. To test the capacity of B. rapa seed to withstand long periods of submersion under natural conditions, we dredged riverbed soil from three sites along the river Thames in early spring (i.e. before B. rapa flowering) and overlaid the resultant slurry onto a steam-steriled compost-sand mix. The dredged soil was noted to contain a small quantity of B. rapa seeds, of which 5 geminated and grew into plants. Taken collectively, it seems feasible from these data that B. rapa can be dispersed by flooding and can retain viability even when submerged for prolonged periods. This finding could have significance in terns of predicting the pattern of secondary spread of transgenes after initial hybrid formation. This is because, unlike pollen-mediated gene movement, the dispersal of genes HYBRIDISATION FROM GM OILSEED RAPE 119 by flooding would have a strong directional bias in the downstream direction. The relative importance of flood-mediated seed dispersal versus bee-mediated pollen dispersal between populations is therefore important in setting the pattern of secondary transgene spread after initial hybrid formation. This aspect is therefore currently under investigation. A second feature of the B. rapa distribution that warrants attention only becomes manifest when the same stretch of river is surveyed over several years. When this was done for the river Thames between Eton and Oxford using GPS-assisted positioning of the populations identified, it was noted that the location of individual populations was not fixed but was subject to radical year-to-year movement (e.g. Fig. 2). FIGURE 2. Location of Brassica rapa populations on the river Thames near Shiplake in 2001 (Grey circles) and 2002 (Light circles). Furthermore, even when the precise positions of individual plants was marked in a popula¬ tion that reappeared in the same site (identical grid reference), it was evident that the plants occupied a different physical space to the preceding year in a manner that was independent of the direction of water flow (Fig. 3). <9 0 2002 • Fence 0 2001 Figure 3. See text for details 120 CURRENT TAXONOMIC RESEARCH Perhaps the most reasonable explanation of these observations is that recruitment of B. rapa seedlings into the flowering plant population is highly dependent upon riverbank disturbance and that the most common cause of such disturbance is flooding. We have tested the capacity of extinct sites that no longer contain B. rapa to yield populations by simply removing surrounding vegetation and disturbing topsoil in spring following the annual winter floods. We noted that B. rapa seedlings germinated and became recruited only in the disturbed quadrats, supporting the suggestion that vegetation disturbance is a necessary prerequisite for B. rapa establishment. Overall then, we can infer that the presence of B. rapa populations is highly variable between years and that flooding is probably important in creating the opportunity for recruitment from the soil seed-bank but also in enabling interpopulational gene movement. Accordingly, we are now in the process of modelling the spread of genes within and between B. rapa populations based on a life history that is dominated by recruitment dynamics from the seed-bank. CONCLUSIONS There is currently only a limited number and complexity of transgene constructs in GM cultivars, although this situation seems set to change as advances in genomics and post-genomics research radically increase the availability of new constructs. This trend sets new challenges for risk assessment research and dictates that regulators assemble as much generic data as possible. For risks relating to gene flow from GM oilseed rape in the UK, several points arising from the work described here should be considered for future submissions for commercial release. 1 Hybrids will form with B. rapa in scattered locations, mostly across eastern England 2 The limited number of sites of hybridization means that the extent and speed of subsequent infraspecific gene movement, together with any fitness advantage, will detennine the ultimate transgene distribution 3 Infraspecific gene flow between populations is mediated by pollen and seed dispersal (flooding) 4 The directional bias of flood-mediated seed dispersal means that the site of transgene recruitment (hybridisation) will have a profound effect on its capacity to spread to other populations 5 The appearance of plant populations is partly a function of disturbance (flooding and river management). It follows that the rate of spread and fixation of any transgene will be partly influenced by the amount of disturbance. 6 These data are currently being assembled to develop a spatially explicit model of transgene spread REFERENCES Chevre, A.M., Eber, F., This, P., Barret, P., Tanguy, X., Brun, H., Delseny, M., Renard, M. (1997). Selection of stable Brassica napus, B. juncea recombinant lines resistant to blackleg (Leptosphaeria maculans). 1. Identification of molecular markers, chromosomal and genomic origin of the introgression. Theoretical and Applied Genetics 95:1104-1111. - Daniell, H. (1999). New tools for chloroplast genetic engineering. Nature Biotechnology / 17: 855-856. HYBRIDISATION FROM GM OILSEED RAPE 121 Davenport, I.J., Wilkinson, M.J., Mason, D.C., Jones, A.E., Allainguillaume, J., Butler, H.T.& Raybould, A.F. (2000). Quantifying gene movement from oilseed rape to its wild relatives using remote sensing. International Journal of Remote Sensing 21: 3567-3573. Elliott, L.J., Mason D.C, Wilkinson M.J., Allainguillaume J., Norris C., Alexander, M., Welters, R. (2004). Using satellite image processing for large-scale study of gene flow from genetically modified rapeseed. Journal of Applied Ecology 41: 1174-1184. Elliott, L.J., Mason, D.C., Wilkinson, M.J. & Allainguillaume, J. (2005). The value of remote sensing technology for GM risk assessment. In: van Emden, H.F. & Gray, A. J. (eds.) GMOs-Ecological dimensions. AAB, Warwick UK pp. 47-53.. Friesen, L.F., Nelson, A.G. & Van Acker. R.C. (2003). Evidence of contamination of pedigreed Canola ( Brassica napus) seedlots in Western Canada with genetically engineered herbicide resistance traits. Agronomy Journal 95: 1342-1347. Giddings, G. (2000). Modelling the spread of pollen from Lolium perenne. The implications for the release of wind-pollinated transgenics. Theoretical and Applied Genetics 100: 971-974. Jorgensen, R.B. & Andersen, B. (1994). Spontaneous hybridization between oilseed rape (Brassica napus) and weedy Brassica campestris (Brassicaceae): a risk of growing genetically modified oilseed rape. American Journal of Botany 81:1 620-1626. Lack, G. (2002). Clinical risk assessment of GM foods. Toxicology Letters 127: 337-340. Linder, C.R. (1998). Potential persistence of transgenes: Seed performance of transgenic canola and wild x canola hybrids. Ecological Applications 8: 1180-1 1 95. Loos, C., Seppelt, R., Meier-Bethke, S., Schiemann, J. & Richter, O. (2003). Spatially explicit modelling of transgenic maize pollen dispersal and cross-pollination. Journal of Theoretical Biology 225: 241-255. Metz, P.L.J., Jacobsen, E., Nap, J.P., Pereira, A. & Stiekema, W.J. (1997). The impact on biosafety of the phosphinothricin-tolerance transgene in inter-specific B. rapa x B. napus hybrids and their successive backcrosses. Theoretical and Applied Genetics 95: 442-450. Oliver, M.J., Quisenberry, J.E., Trolinder, N.L.G. & Keim, D.L. (1999). Control of Plant Gene Expression. US Patent no. 5,925,808. Ooms, G., Bains, A., Burrell, M., Karp, A., Twell, D. & Wilcox, E. (1985). Genetic manipulation in cultivars of oilseed rape (Brassica napus) using Agrobacterium. Theoretical and Applied Genetics 71: 325-329. Preston, C.D., Pearman, D.A. & Dines, T.D. (2002). New Atlas of the British & Irish Flora. Oxford University Press, Oxford. Rosellini, D., Pezzotti, M. & Veronesi, F. (2001). Characterization of transgenic male sterility in alfalfa. Euphytica 118: 313-319. Reiger, M.A., Lamond, M., Preston, C., Powles, S.B. & Roush, R.T. (2002). Pollen- mediated movement of herbicide resistance between commercial canola fields. Science 296: 2386-2388. Squire, G.R., Brooks, D.R., Bohan, D.A., Champion, G.T., Daniels, R.E., Haughton, A.J., Hawes, C., Heard, M.S., Hill, M.O., May, M.J., Osborne, J.L., Perry, J.N., Roy, D.B., WOIWOD, I.P. & Firbank, L.G. (2003). On the rationale and interpretation of the Farm Scale Evaluations of genetically modified herbicide-tolerant crops. Philosophical Transactions of the Royal Society, London series B 358: 1779-1799. 122 CURRENT TAXONOMIC RESEARCH SCHEFFLER, J.A. & Dale, P.J. (1994). Opportunities for gene transfer from transgenic oilseed rape ( Brassica napus) to related species. Transgenic Research 3: 263-278. Scott, S.E. & Wilkinson, M.J. (1998). Transgene risk is low. Nature (London), 393: 320. Song, K. & Osborn, T.C. (1992). Polyphyletic origins of Brassica-napus - new evidence based on organelle and nuclear RFLP analyses. Genome 35: 992-1001. STACE, C.A. (1975). Hybridization and the flora of the British Isles. London: Academic Press. STACE, C.A. (1997). New Flora of the British Isles , edn 2. Cambridge: Cambridge University Press. Timmons, A.M., Charters, Y.M., Crawford, J.W., Burn, D., Scott, S.E., Dubbels, S.J., Wilson, N.J., Robertson, A., O’Brien, E.T., Squire, G.R., & Wilkinson, M.J. (1996) Risks from transgenic crops. Nature (London) 380: 487. Waines, J.G. & Hedge, S.G. (2003). Intraspecific gene flow in bread wheat as affected by reproductive biology and pollination ecology of wheat flowers. Crop Science 43: 451- 463. Wilkinson, M.J., Davenport, I.J., Charters, Y.M., Jones, A.E., Allainguillaume, J., Butler, H.T., Mason, D.C. & Raybould, A.F. (2000). A direct regional scale estimate of transgene movement from genetically modified oilseed rape to its wild progenitors. Molecular Ecology 9: 983-991. Wilkinson, M.J., Elliott, L.J., Allainguillaume, J., Shaw, M.W., Norris, C., Welters, R., Alexander, M., Sweet, J. & Mason, D.C. (2003). Hybridization between Brassica napus and B. rapa on a National Scale in the United Kingdom. Science 302: 457^159. Current taxonomic research on the British & European flora Bailey, J. & Ellis, R.G. (eds) 123-133 (2006), BSBI. London. 123 Endemic vascular plants in the Nordic flora BENGT JONSELL Ber gins Foundation, Royal Swedish Academy of Sciences, Box 5001 7, SE-104 05 Sweden. THOMAS KARLSSON Swedish Museum of Natural History, Box 50007, SE-104 05 Sweden. ABSTRACT The endemic taxa of vascular plants of the Nordic countries amount to about 130, of which 46 are currently regarded as species ( Hieracium , Taraxacum , and the Ranunculus auricomus complex excluded). A postglacial origin, i.e. a time span of maximally 10,000 years of evolution, is now postulated for all Nordic endemics. There are five allopolyploid sexual endemic species; Arabidopsis suecica, Corydalis gotlandica, Draba cacuminum, Primula scandinavica and Saxifra- ga osloensis. Outside the large apomictic groups mentioned above, only about 40 apomictic endemics are recognized. A large proportion of the non-hybrid and non-apomictic endemics have probably originated from ecotypes. Many Nordic endemics are members of intricate species complexes with subtle and often disputed taxonomic limits. The Nordic endemics are geographical¬ ly concentrated, but far from exclusively, to a number of ‘hot’ areas with particular ecological and historical conditions, the most important of which are the Scandes, the Baltic land-lift shores and the Baltic limestone islands of Oland and Gotland. A smaller number of endemic taxa are found at the coasts of NW Jylland (Denmark), along the Arctic Ocean coast, and in Iceland. The isolated Nordic islands and the island groups to the north and west (Spitzbergen, Jan Mayen, and the Faroes) have few or no endemic taxa. Keywords: endemism, Scandinavia, North Atlantic, Scotland, glaciation. INTRODUCTION There are few endemic taxa in the Nordic vascular flora (as long as apomictic groups are disregarded), and their taxonomic rank is usually low. The latest glaciation in Norden terminated about 10,000 years ago. With present evolutionary models that time span is considered long enough to allow for the amount of differentiation by ecological and geographical isolation, though many endemics appear to be much younger. There are no indications for the previously much-favoured view that Nordic endemics should emanate from glacial survivors. For more comprehensive discussions of endemism in Norden see Borgen (1987), Brochmann et al. (2003), Dahl (1989), Jonsell (1988, 1990a, 1990b, 1997), Jonsell & Karlsson (2004). The area considered here is that covered by Flora Nordica (Jonsell 2000, 2001). Some consideration is also given to the area between the Finnish-Norwegian border and the White Sea, i.e. the Kola Peninsula and Russian Karelia. This area is geologically a part of the Fennoscandian shield and viewed from the point of natural history it is connected to Fennoscandia. With microspecies of the three major groups of apomicts excluded the 124 CURRENT TAXONOMIC RESEARCH number of endemics at species, subspecies and varietal level in the Flora Nordica area is c. 130. By including northwestern Russia the number is increased to about 160, and if the whole Baltic area (i.e. the southeastern and southern coasts) is included, the number reaches 180. Inclusion of the whole North Atlantic area, i.e. Greenland and the British Isles as well, would add between 25 and 30 taxa. In comparisons with the endemism of other regions it is important that they are made on similar taxonomic levels, and take into account the breeding systems. The origins of ende¬ mism are multifarious and often poorly understood, even within each region quite different evolutionary situations may have led to endemism. According to the mode of origin many, but far from all, Nordic endemics can be grouped into the three categories discussed in detail below. The taxa considered here are comparatively readily distinguishable, but their taxonomic status is not always quite clear. Some of them belong to aggregates not sufficiently analysed even within Norden and it is perhaps not possible to maintain them all as separate taxa. Some may have a wider distribution than presently assumed and occur in Russia, between the White Sea and northern Urals and even further east. Another question is how to define ‘endemic’ taxonomically. The number of endemics found in Norden will depend very much on whether or not we count the endemic apomictic taxa. Some apomictic groups, e.g. Rubus , Sorbus and Alchemilla , have a moderate number of distinct agamospecies with local, regional or even wider distribution; they are mentioned here as equivalents to non-apomictic taxa. There are also large apomictic groups, with hundreds of described ‘ microspecies ’ and still many undescribed ones, viz. the Ranunculus auricomus group and the genera Taraxacum and Hieracium. They are excluded here mainly because the distribution of their species is poorly known outside Norden. The apomictic groups are enumerated below. Some of the more distinct endemics probably originated from ecotypes. By definition (Turesson 1925) these may have polytopic origins; most cases are poorly analysed. Few ecotypes have been formally described. ENDEMIC ALLOPOLYPLOIDS Only a few of the endemic sexual taxa can undoubtedly be ranked as species. Among those that can, five are the results of allopolyploidy, an abrupt mode of speciation on an evolution¬ ary time scale: Corydalis gotlandica (2n = c.30) is known from a few localities in the western part of the Baltic island of Gotland. It grows, and is expanding, in a small scale agricultural landscape and on an open limestone island some five kilometres off the coast. Its distribution and behaviour suggest a comparatively recent origin. The parental species are C. intermedia (2n = 16) and C. solida (2n = 16) (Liden 1991), which are both present on Gotland. Arabidopsis suecica (2n = 26) (Colour Plate 2a). This species is widely distributed on disturbed ground in Sweden and Finland, particularly on gravelly roadsides and railway areas, and not in anything like natural habitats. Its possible origin is in southeastern Finland with its vast glaciofluvial gravelly moraines, which in early postglacial times offered natural open gravelly areas free from competition. The parental species are A. arenosa (2n = 16, 32) and A. thaliana (2n = TO) (O’Kane et al. 1996). They are both common within much of the area of A. suecica. ENDEMICS IN THE NORDIC FLORA 125 Draba cacuminum (2n = 64) is a rare high mountain species with different races in two separate areas of the Scandes: the south Norwegian mountains from N Buskerud to Sor Trondelag (subsp. cacuminum ), and further north in Nordland and adjacent parts of Sweden from Lycksele to Lule Lappmark (subsp. angusticarpa). The species has a multiple origin with the common D. norvegica (2n = 48) as one parental species and probably the much rarer D. fladnizens is (2n = 16) (Brochmann & Elven 1992) as the other. Saxifraga osloensis (2n = 44) (Colour Plate 2b). This species is confined to base-rich outcrops in a zone across middle Scandinavia from the Oslo region in Norway to Uppland in E Sweden. It probably originated in early post-glacial time from the montane, northern S. adscendens (2n = 22) and the southern, calcicole S. tridactylites (2n = 22) (Knaben 1954, Brochmann et al. 1996). The parental species are still widespread to the north and the south, respectively, of the zone of S. osloensis , but the three species are not known to grow on the same spot anywhere. Primula scandinavica (2n = 72). This species is widespread in more oceanic parts of the Scandes from Norway (Rogaland) and W. Jamtland in Sweden north to western Lapponia tomensis in Sweden and Troms in Norway. According to Knaben (1982) it probably originated from P. farinosa (2n =18, 36), nearest in C Sweden, and P. scotica (2n = 54), now only in Scotland. Today the parental species are far apart but Knaben speculated that they might have been in contact in glacial and early late-glacial times on the North Sea Continent, which was apparently of great importance for glacial survival and as a source of immigrants in late glacial time. Other interpretations are, however, possible (Richards 1993). HYBRIDS There are various asexual ways for hybrids to reproduce and spread more or less independ¬ ently of their parents. Four well defined taxa of this kind are among the Nordic endemics. Salix xarctogena is basically a triple hybrid between S. herbacea, S. polaris and S. phylicifolia , but in some localities there is evidence that S. glauca and S. lapponum also participate. S. xarctogena is pollen and seed fertile; it has arisen polytopically in at least seven areas in the Scandes (Elven in Jonsell 2000). Saxifraga xopdalensis is a hybrid between S. cernua and S. rivularis with one local occurrence in south central Norway (the Dovre mountains). It does not set seed but reproduces by means of bulbils. Further populations, probably of the same parentage, are known from some places in northern Norway and in Swedish Lapland. A similar plant, described as ‘S. xSvalbardensis\ has arisen in Svalbard from a cross between Svalbard forms of the same two species (Steen et al. 2000). All these plants are here included in S. xopdalensis. Poa xherjedalica is the viviparous hybrid between P. alpina and P. pratensis subsp. alpigena. It has a wide distribution throughout the Scandes and has almost certainly arisen polytopically. Carex halophila is a member of the C. recta complex, an amphi-Atlantic aggregate of hybrids between, on the one hand, C. paleacea and C. salina , and on the other hand, C. acuta , C. aquatilis and C. nigra. C. halophila occurs along the coasts of northern Norway and the Kola Peninsula, and, in deviating forms, in the Gulf of Bothnia. 126 CURRENT TAXONOMIC RESEARCH APOMICTS Agamospermy is another process that leads to rapid establishment of new taxa. There are 1 3 genera with agamospermous endemics in Norden. This is by far the largest element of the Nordic endemics. The three largest agamospermous groups, viz. the Ranunculus auricomus group and the genera Hieracium and Taraxacum, comprise the great majority of taxa. The Ranunculus auricomus group. 605 Nordic species have been described, and most of them are only known from Norden. The treatment in Flora Nordica includes a number of examples and a checklist to all known Nordic taxa (Ericsson in Jonsell 2001). In all likelihood, very many taxa remain to be described. Alchemilla is represented in Norden by 22 indigenous or archaeophytic species, most of them with a rather wide distribution. Only three Nordic endemics, A. faeroensis (in the Faroes and E. Iceland), A. semidivisa (both of the Splendentes group and intermediate to A. alpina) and A. taernaensis, have been described; a further two, A. borealis and A. oxyodonta , have but little of their range outside Norden. There is also a moderate number of undescribed (and partly insufficiently studied), mostly very local taxa. Cotoneaster. There are four indigenous Nordic species of this largely agamospermous genus. Three of them occur also in the Baltic states; the fourth, C. kullensis, is a local taxon from southernmost Sweden. Potentilla. Numerous agamospermous species have been described within this genus, but only two Nordic endemics are known, viz. P. insularis from Svalbard and P. sterneri from coastal SE Sweden. Rubus subgen. Rubus. Of the 98 accepted spontaneous taxa of brambles known from Norden only 1 1 are Nordic endemics. All except one belong to sect. Corylifolii, which comprises hybrid derivatives probably of fairly recent origin. Some further local taxa are known, and some of them have been validly described, but bramble biotypes with a very limited distribution are not taxonomically accepted nowadays (Weber 1972). Sorbus. In addition to the three diploid sexual species (2n = 34) there are nine currently accepted, described apomictic triploids (2n = 51) and tetraploids (2n = 68). One of these (. S . aria) is fairly widespread in Europe. One (X. intermedia) is present in Norden and Estonia only, the occurrence in Scotland being a recent introduction. The other seven are endemic to Norden. S. hybrida and S. meinichii have comparatively wide distributions (southern Nor¬ way, coastal southern Sweden), while the other five are restricted to fairly small areas in the west. The genus is under revision and the number of described endemics may possibly increase. Hieracium. Within Hieracium more than 5000 presumably apomictic species have been described from Norden. The native species are arranged into 14 sections with tens to hundreds of species, some mainly occurring in the man-made landscape, others montane or alpine. In all likelihood most of the species are endemic to Norden. The Hieracium flora is well investigated in large parts of Norden, but in and near the Scandes numerous species probably remain to be described. Pilosella. This genus is notoriously difficult due to the presence of facultative agamos¬ permy, leading to the formation of numerous microspecies whose distinctness is partly concealed by more continuous variation among the sexual forms. A new taxonomic system, ENDEMICS IN THE NORDIC FLORA 127 which seems to work well at least within Norden, was recently proposed by Tyler (2001). He distinguished 16 indigenous taxa, of which three are endemic. Taraxacum. The genus comprises about 1000 described species in Norden distributed among 14 sections, including between two and hundreds of species. The rate of endemism varies greatly among sections and between habitats, being relatively high in alpine and montane areas, fairly low in the man-made landscape. The smallest section includes T. dovrense, endemic in the S. Norwegian mountains and closely related to the circumarctic T. arcticum. A limited number of yet undescribed species have been identified. Gymnadenia. This genus comprises four species, of which two are agamospermous and endemic to Norden. G. ( Nigritella ) nigra subsp. nigra is a triploid (2n = 60) with related sexual taxa in Central Europe, whereas G. (Gymnigritella) runei is a tetraploid (2n = 80) which has arisen locally with G. nigra subsp. nigra as one parent and G. conopsea (2n = 40) as the other (Teppner & Klein 1989, Hedren 1999). Calamagrostis. This grass genus comprises seven Nordic species, three of which are agamospermous. One of these, G. chalybaea , is endemic. It is widespread in northern Sweden and known also from Nord-Trondelag and Nordland in Norway as well as from one locality in the Kola Peninsula. Its closest relative appears to be the sexual C. obtusata of eastern Russia. Hierochloe. As currently understood (Weimarck 1971, 1986) the genus comprises seven taxa in Norden, all except one ( H . australis) at least facultatively apomictic. Most taxa have fairly wide to very wide distributions, but H. odorata subsp. baltica appears to be endemic to Norden and Estonia. It probably developed from the amphi-Atlantic subsp. odorata. Poa. Apomixis is widespread in this genus, but within Norden, agamospermous endemics have been described only within P. arctica (Nannfeldt 1940). In the southern Scandes there are three fairly well-delimited races, one of which (subsp. stricta) is viviparous. In the northern Scandes the variation is more continuous and only two very local taxa have been possible to discern (apart from subsp. caespitans, which has a wide amphi-Atlantic distribution). AREAS OF ENDEMISM AND EXAMPLES OF TAXA There are three major areas of endemism in Norden. Apparently, these areas have, or have had, ecological conditions promoting evolution. The areas are presented below with a few examples of their endemics. In addition the Arctic coast of Norway extending into the Russian Kola Peninsula, coastal dunes and sea-facing hillsides in northwestern Jylland in Denmark, and Iceland each house a few endemic taxa (see further Jonsell & Karlsson 2004). The island groups of Svalbard and the Faroes have only one or two endemics each, and the very isolated, volcanic island of Jan Mayen none. On the other hand some 60 endemics at all taxonomic levels have distributions not coinciding with any others, among those the allopoly¬ ploid Saxifraga osloensis. The Scandes and adjacent areas The majority of the endemics occur in the low-alpine belt, and some extend also to the birch belt below and to the mid-alpine above (Jonsell 1990 b). Few taxa are exclusive to any of the latter belts. Out of the total of 1 80 endemics in Norden and adjacent areas, 47 taxa belong to this group. Of these, nearly all are known from Norway, which has the core of the Scandinavian mountain ridge. The number of taxa declines eastwards (27 in Sweden, 15 in Finland, 1 1 in northwestern Russia). 128 CURRENT TAXONOMIC RESEARCH Most of these endemics are currently evaluated as subspecies, whilst six are regarded as varieties. Three taxa are hybrids ( Salix ^arctogena , Saxifraga xopdalensis and Poa xherjedalica). Of the nine species two are allopolyploids ( Draba cacuminum and Primula scandinavica), whereas four are agamospermous ( Alchemilla semidivisa , A. taernaensis , Taraxacum dovrense and Gymnadenia runei). Among the few sexual endemics currently evaluated as species are Silene wahlbergel- la, a pronounced selfer within the intricate arctic S. uralensis group, as well as Euphrasia hyperborea and Antennaria nordhageniana, both members of complexes, in which the differences between species are disputed. Isolation in various mountain massifs leading to vicariant endemics in a species or species group is of little significance. Papaver radicatum with its many local or regional subspecies in the Scandes is much cited in Norden as an example, even held to indicate a history of glacial survival. It is now rather thought that the degree of differentiation between most of the P. radicatum subspecies hardly surpasses that found within more or less coherent Arctic populations (Solstad et al. 1999, 2003). This is also true of subsp. laestadianum , which is an octoploid (2n = 56) in contrast to all other Nordic P. radicatum , which are decaploid (2n = 70). Oland and Gotland In Oland and Gotland open habitats on limestone pavements (alvar plains) play a major role and in Gotland semi-open calcareous pine forest is still common. These and other habitats on the two islands harbour a rich flora which comprises 20 taxa endemic to Norden or nearly so. Five of these are known only from Oland ( Helianthemum oelandicum var. oelandicum and var. canescens, Galium oelandicum , Artemisia oelandica and Crepis tectorum subsp. pumila), four only from Gotland {Pulsatilla vulgaris subsp. gotlandica , Coiydalis gotlandica , Euphrasia salisburgensis var. schoenicola and Crepis tectorum var. glabrescens). Arenaria gothica var. gothica has one occurrence outside Gotland, see below. The remaining ten taxa are known from both Oland and Gotland, two of them {Cotoneaster canescens and Artemisia maritima subsp. humifusa) also from Estonia. The majority of the Baltic island endemics are classified as varieties or subspecies. Among the five species the allopolyploid Corydalis gotlandica was presented above. Coton-easter canescens and Pilosella dichotoma are agamospermous apomicts. Galium oelandicum and Artemisia oelandica are sexual and not evidently hybridogenous; the former is a member of the G. pumilum complex where the limits between the often rather narrowly endemic species are subtle. A. oelandica is only doubtfully distinct from the polymorphic A. laciniata of eastern Europe and western Asia. The spectacular Helianthe¬ mum oelandicum with its two endemic races on Oland is conspecific with taxa in western and central Europe, the Caucasus and Turkey. Not all endemics occur in the various alvar habitats. Euphrasia salisburgensis var. schoenicola grows in spring fens, Artemisia maritima subsp. humifusa on seashores, and Coiydalis gotlandica and Euphrasia stricta var. suecica belong to the man-made land¬ scape. Several of the endemics appear to be relics from a much wider range in Scandinavia and other parts of Europe in late-glacial or early post-glacial times. Among those are Arenaria gothica, which is widespread on calcareous gravel in Gotland and endemic to that island but for a local disjunct population in a similar habitat in Mt Kinnekulle, Vastergot- ENDEMICS IN THE NORDIC FLORA 129 land, Sweden. It seems to be a relic from the early post-glacial, differentiated within the A. ciliata complex, which shows many disjunctions in NW Europe. Material referred to as A. gothica var .fugax from Swiss Jura remains to be critically reassessed. Possibly the isolation between Oland and Gotland has promoted differentiation in some taxa. There are clear morphological differences between the islands within both Euphrasia stricta var. gotlandica (Karlsson 1986) and Pilosella cymosa var. gottlandica (Tyler 2001), and the local dwarf variants of Crepis tectorum are currently regarded as different taxa. However, parallel evolution of variants within these taxa on the two islands is also possible; in Crepis tectorum the latter hypothesis is the most likely according to Andersson (1990). The Baltic shores Numerous endemics are bound to the shores of the Baltic. In its northern parts (the Gulf of - Bothnia, the Gulf of Finland and the northern part of the Baltic proper) there is still land uplift of up to ca 80 cm in a century on very flat shores, which means that large areas of land arise (Fig. 1). Therefore a permanent pioneer situation prevails on these shores, which is supposed to promote evolution and establishment of new taxa (Jonsell 1988, 1990a). The isolation of populations of wide-spread species on the coasts of the brackish, non-tidal Baltic has also been important. Flooding during storms contributes to make the habitats extremely unstable. The salinity gradient along the Bothnian coast also influences the distribution of the constituent species. Fourteen endemic taxa are restricted to the northern part of the Baltic as defined above, three of which are generally treated at species level: Alisma wahlenbergii , Deschampsia bottnica (Colour Plate 2c) and Euphrasia bottnica. Alisma wahlenbergii occurs now only in the innermost part of the Bothnian Gulf and the former bay of the Baltic which is now Lake Malaren (W of Stockholm). It appears to be a brackish- water derivative of the widespread freshwater species A. gramineum (Jacobsson 2003). Deschampsia bottnica (although mor¬ phologically quite distinct) belongs to the D. cespitosa complex and has close affinities to taxa on the Arctic coasts of N Russia, Siberia and in Beringia. Euphrasia bottnica on the contrary is a very distinct taxon for its genus, endemic to the shores of the Bothnian Gulf in Sweden and Finland, with its closest relatives in N America. Among the hemiparasites there are two more endemics, but less pronounced, in Euphrasia and in Odontites. Three more taxa, Primula nutans subsp. finmarchica, Sonchus arvensis var. maritimus and Carex halophila , occur also along the Arctic coast of Norway and northwestern Russia. The local variant of C. halophila in the northern Baltic appears, however, to be distinct from the variants on the Arctic coast. P. nutans is represented by var. jokelae in the northern Baltic (and the White Sea area), whereas var. finmarchica grows on the Arctic coast. Eight taxa have a more southern distribution in the Baltic and thus have part of their distribution area on shores which are nowadays affected by the land uplift to a somewhat lower degree (about 40 cm in a century on less flat shores). The most clear-cut of these are the almost glabrous Mentha aquatica var. litoralis and the narrow-leaved Veronica longifolia var. maritima. Finally, five taxa are found in the southern Baltic and on the western coasts of southern Scandinavia, viz. Polygonum oxyspermum, Cakile maritima subsp. baltica , Lotus cornicula- tus var. alandicus and var. carnosus and Cuscuta europaea subsp. halophyta. Despite their deviating distribution their origin is probably in some way connected to the history of the Baltic and they are best treated in this context. Polygonum oxyspermum is closely related to 130 CURRENT TAXONOMIC RESEARCH FIGURE 1 . Isopleths of the present-day land uplift in the Baltic area im mm/year. The values are relative, i.e. not corrected for the eustatic rise in sea-level (c.l mm/year). From Ericson & Wallentinus (1979). P. ran of the Atlantic coasts. It has probably developed as a consequence of isolation on brackish, non-tidal Baltic shores. All in all, there are 30 endemic seashore taxa associated with the Baltic and present in Norden. All of these occur in Finland; in Sweden all are known except three Gulf of Finland endemics ( Silene vulgaris var. littoralis , Lotus comiculatus var. maritimus and Odontites littoralis subsp .fennicus). About half the number (14 taxa) are found in the Baltic outside Norden. ENDEMICS IN THE NORDIC FLORA 131 CONNECTIONS TO SCOTLAND The immigration of taxa which have evolved into endemics in Norden is multifarious. In this context only connections to the British Isles and its glacial surroundings, the North Sea Continent will be considered. The somewhat enigmatic origin of Primula scandinavica was mentioned earlier. Other examples relate to the famous taxa restricted to special habitats in Yorkshire, including the Teesdale limestone. Among these are Viola rupestris , the British populations of which have obvious isoenzymatic similarities with V rupestris subsp. relicta , an endemic of the Scandes on limestone from W Central Norway northwards (Jonsell et al. 2000). Other examples probably recruited from Britain are to be found in Cochlearia officinalis , which has two endemic subspecies in Norway, and Salix caprea subsp. sphacela- - ta in most of the Scandes at low levels. If the geographical scope is extended to include Scotland a few more endemics can be added: Arenaria nomegica subsp. nowegica, common in Iceland, scattered along the Scandes and extremely local in north-western Scotland and western Ireland. In Yorkshire it is represented by a different subspecies, subsp. anglica, which links this species complex, often called the A. ciliata complex, to that important area of relict distributions. The species is accordingly a North Atlantic endemic. Arenaria pseudofiigida, is an Arctic - North Atlantic endemic of the same complex, whrch reaches W Greenland in the west and Novaya Zemlya in the East. Car ex saxatilis subsp. saxatilis. This subspecies is widespread in the Scandes, N Scandinavia, Iceland and the Faroes, and local in northern and western Scotland. The species is circumpolar and over most of its area represented by subsp. laxa. Cerastium nigrescens var. laxum. This taxon is the non-Arctic component of the previ¬ ously widely circumscribed C. arcticum and was defined by Brysting & Elven (2000). It occurs in the Scandes, Iceland, the Faroes, Scotland and even Mt Snowdon in Wales. The type variety, var. nigrescens, is a local endemic on serpentine on the island of Unst in the Shetlands (Brysting & Borgen 2000, Jonsell 2001). Accordingly this species is a North Atlantic endemic, not reaching the Arctic. Poa xjemtlandica , the viviparous asexual hybrid between Poa alpina and P. flexuosa , which contrary to earlier views (Nannfeldt 1937) certainly originated separately in each of its areas, the Scandes. Iceland and central Scotland, where it is very local (Brysting et al. 1997,2000). Poa flexuosa. Also one of the parental taxa of P. xjemtlandica, P. flexuosa, is a North Atlantic endemic with a distribution very similar to that of P. x jemtlandica although wider (south and central Scandes, Iceland and locally in central Scotland; Nannfeldt 1935). In Scotland the degree of endemism appears to be dramatically lower than in Scandina¬ via, despite the fact that the regions in some respects are similar. Preston (2003) listed 1 1 Scottish endemic taxa, 8 of which are taxonomically doubtful, or apomictic microspecies. There are only three with a seemingly undoubted specific status, Calamagrostis scotica, Cochlearia micacea and the previously mentioned Primula scotica, apparently a parent of the Scandinavian endemic P. scandinavica. On the other hand only taxa at some time given species rank are considered by Preston (2003). Even so endemism in Scotland appears low compared with that of Scandinavia. 132 CURRENT TAXONOMIC RESEARCH REFERENCES Andersson, S. (1990). A phenetic study of Crepis tectorum in Fennnoscandia and Estonia. Nordic J. Botany 9: 589-600. Borgen, L. (1987). Postglasial evolusjon i Nordens flora - en oppsummering. Blyttia 45: 147-169. Brochmann, C. & Elven, R. (1992). Ecological and genetic consequences of polyploidy in arctic Dr aba (Brassicaceae). Evol. Trends in Plants 6: 111-124. Brochmann, C., Nilsson, T. & Gabrielsen, T. (1996). A classic example of postglacial allopolyploid speciation re-examined using RAPD markers and nucleotide sequences: Saxifraga osloensis (Saxifragaceae). Symb. bot. upsal. 31(3): 75-89. Brochmann, C., Gabrielsen, T.M., Elven, R. & Nordal, I. (2003). Glacial survival or tabula rasa. The history of North Atlantic biota revisited. Taxon 53: 4 1 7^4-50. Brysting, A.K. & Borgen, L. (2000). Isozyme analysis of the Cerastium alpinum - C. arcticum complex (Caryophyllaceae) supports a splitting of C. arcticum Lange. PI. Syst. Evol. 220: 199-221. Brysting, A.K. & Elven, R. (2000). The Cerastium alpinum - C. arcticum complex (Caryophyllaceae): numerical analysis of morphological variation and a taxonomical revision of C. arcticum Lange s. lat. Taxon 49: 189-216. Brysting, A.K., Elven, R. & Nordal, I. (1997). The hypothesis of hybrid origin of Poa jemtlandica supported by morphometric and isoenzyme data. Nord. J. Bot. 17: 199-214. Brysting, A.K., Holst-Jensen, A. & Leitch, I. (2000). Genomic origin and organization of the hybrid Poa jemtlandica (Poaceae) verified by genomic in situ hybridization and chloroplast DNA sequences. Ann. Bot. 85: 439^145. Dahl, E. (1989). Nunatakk-teorien II. Endemismeproblemet. Blyttia 47: 163-172. Ericson, L. & Wallentinus, H.G. (1979). Sea-shore vegetation around the Gulf of Bothnia: Guide for the International Society of Vegetation Science, July-August 1977. Wahlenbergia 5: 1-142. Hedren, M. (1999). Kommentarer om brudkullan och dess ursprung. SvenskBot. Tidsh\ 93: 145-151. JACOBSSON, A. (2003). Diversity and phylogeography in Alisma (Alismataceae), with emphasis on Northern European taxa. Thesis. Univ. of Lund. JONSELL, B. (1988). Mikroendemism i det baltiska landhojningsomradet. Blyttia 46: 65-73. JONSELL, B. (1990a). Evolutionary trends among plants in the Baltic land uplift area. Sommerfeltia 11: 137-146. JONSELL, B. (1990b). Fjallendemism och annan endemism i Skandinaviens flora. Blyttia 48: 79-81. JONSELL, B. (1997). Endemism and its backgrounds among vascular plants in Scandinavia. Sprawd. z pozied. komisji nauk. (Polska Akad. Nauk, Krakow) 40(2): 75-80. JONSELL, B. ed. (2000). Flora Nordica 1. Stockholm. JONSELL, B. ed. (2001). Flora Nordica 2. Stockholm. JONSELL, B. & KARLSSON, T. (2004). Endemic vascular plants in Norden. In B. JONSELL (ed.) Flora Nordica. General Volume: 139-159. Jonsell, B., Nordal, I. & Roberts, F.J. (2000). Viola rupestris and its hybrids in Britain. Watsonia 23: 269-278. ENDEMICS IN THE NORDIC FLORA 133 KARLSSON, T. (1986). The evolutionary situation of Euphrasia in Sweden. Symb. Bot. Upsal. 27(2): 61-71. KNABEN, G. (1954). Saxifraga osloensis n. sp., a tetraploid species of the Tridactylites section. Nytt Mag. Bot. 3: 1 17-138. Knaben, G. (1982). Om arts- og rasedannelse i Europa under kvartaertiden 1. Endemiske arter i Nord-Atlanteren. Blyttia 40: 229-235. Liden, M. (1991). Notes on Corydalis sect. Corydalis in the Baltic area. Nord. J. Bot. 11: 129-133. Nannfeldt, J.A. (1935). Taxonomical and plant-geographical studies in the Poa laxa group. A contribution to the history of the North European mountain flora. Symb. Bot. Upsal. 1(5): 1-113. Nannfeldt, J.A. (1937). On Poa jemtlandica (Almqu.) Richt., its distribution and possible origin. A criticism of the theory of hybridization as the cause of vivipary. Bot. Notiser 1937: 1-27. Nannfeldt, J.A. (1940). On the polymorphy of the Poa arctica R. Br., with special reference to its Scandinavian forms. Symb. Bot. Upsal. 4(3). O’Kane, S.L. Jr., Schaal, B.A. & Al-Shehbaz, I.A. (1996). The origin of Arabidopsis suecica (Brassicaceae) as indicated by nuclear rDNA sequences. Syst. Bot. 21: 559-566. PRESTON, C.D. (2003). Scottish vascular plants in a global context. Botanical Journal of Scotland 55: 7-15. Richards, J. (1993). Primula. Batsford, London. Solstad, H., Elven, R. & Nordal. I. (1999). Are there too many species and subspecies in the Papaver radicatum complex? Skr. Norske Vidensk.-Akad., Oslo, I. Mat.- Naturvitensk. Kl. 38: 281-294. Solstad, EE, Elven, R. & Nordal, I. (2003). Isozyme variation among and within North Atlantic species of Papaver sect. Meconella (Papaveraceae) and taxonomic implications. Bot. J. Linn. Soc. 143: 225-269. Steen, S.W., Gielly, L.. Taberlet, P. & Brochmann, C. (2000). Same parental species, but different taxa: molecular evidence for hybrid origins of the rare endemics Saxifraga opdalensis and S. svalbardensis (Saxifragaceae). Bot. J. Linn. Soc. 132: 153-164. Teppner, H. & Klein, E. (1989). Gymnigritella runei spec, nova (Orchidaceae-Orchideae) aus Schweden. Phyton 29: 161-173. TURESSON, G. (1925). The plant species in relation to habitat and climate. Contributions to the knowledge of genecological units. Hereditas 6: 147-236. Tyler, T. (2001). Forslag till ny taxonomisk indelning av stangfibbloma (Pilosella) i Norden. Svensk Bot. Tid$h\ 95: 39-67. Weber, H. (1972). Die Gattung Rubus im nordwestlichen Europa. Phanerogamum Monographiae 7. Cramer, Lehre. Weimarck, G. (1971). Variation and taxonomy of Hierochloe (Gramineae) in the northern hemisphere. Bot. Notiser 124: 129-175. Weimarck, G. (1986). Hierochloe hirta subsp. praetermissa subsp. nova (Poaceae), an Asiatic - E European taxon extending to N and C Europe. Symb. Bot. Upsal. 27(2): 175-181. Current taxonomic research on the British & European flora Bailey, J. & Ellis, R.G. (eds) 135-148 (2006), BSBI. London. 135 Evolution in thyme enough? Rapid physiological evolution in re¬ sponse to pollution in two common British plants. A. John Richards, Helena Winsnes School of Biology, Ridley Building, University of Newcastle NE1 7RU, UK. Lynn Whitfield Department of Biological Sciences, Royal Holloway, University of London, Egham, Surrey TW20 OEX, UK ABSTRACT Speciation is often confused with evolution, but taxonomic recognition of a species depends on morphological distinctiveness. We argue that when plants are challenged by rapid and drastic environ¬ mental changes such as those resulting from pollution, climate change, or new predators and patho¬ gens, they commonly show dramatic and often sympatric physiological evolution, but because this is normally cryptic (‘silent’) it is not recognised by taxonomists and so tends to be ignored by conserva¬ tionists and policy makers. The principle is illustrated by wild thyme Thymus polytrichus from metal-rich calaminarian grasslands, and perennial rye-grass Loliwn perenne from salted road- verges, both of which appear to have evolved tolerance within the last 100 years. Zinc tolerance in T. polytrichus is biparentally dominant. We were unable to detect a metabolic cost to this tolerance, or for salt tolerance found in roadside L. perenne from several sites. We were unable to induce tolerance by early exposure to salt, and strains that appeared to be innately tolerant were not later affected by this early exposure. In contrast, susceptible strains showed a ‘hang-over’ effect from early exposure, even when grown in salt-free conditions. Keywords: Thymus polytrichus, Lolium perenne, salt tolerance, zinc tolerance, tolerance induction, tolerance costs. GENERAL INTRODUCTION While considering the massive contribution that Clive Stace has made to plant taxonomy, not only in the British Isles, AJR fell to contemplating, not for the first time, the very purpose of taxonomy itself. Undeniably, human communication requires living things to be known by reliable, internationally valid names. Unfortunately the multidimensional matrix of morpho¬ logical variation within which we identify noda for the purpose of this taxonomic recognition does not always fall into neat compartments. It is not surprising that evolution has resulted in patterns of variation which are not readily packaged within a simple system of nomencla¬ ture, the philosophical basis of which predates Darwin by more than a century. Consequent¬ ly, the relationship between a name (taxon) and the limits of the morphological variation which corresponds to this name can be problematical. These problems provide the rationale for the science which was once called ‘experimental taxonomy’. 136 CURRENT TAXONOMIC RESEARCH However, in recent years, taxonomy has been called upon to undertake other, quite new tasks whose philosophical bases are very different from the original need to ‘give a dog a name’. For instance, detailed studies of DNA sequence variation in well-understood genes are now routinely applied, so that we can trace unambiguously the pathways by which related organisms have evolved. As a consequence, phylogeneticists expect taxonomic frameworks to accurately reflect our understanding of these evolutionary pathways. But, as is discussed in more detail elsewhere in this volume, evolutionary radiation as revealed by the DNA may not always correspond well with either morphological radiation (which may not always be adaptive), or adaptive radiation (including adaptive characters which are not expressed morphologically). Our antiquated taxonomic system is particularly unsuited to a situation where a distinctive derived (‘apomorphic’) clade which demands generic recogni¬ tion (for example) is nested within a wider clade for which more than a single generic attribution would be considered inappropriate. A second ‘new’ task imposed upon our creaking taxonomic systems has resulted from the sudden rise in prominence of ‘biodiversity’ in our value judgements, resulting from the extraordinarily influential ‘Earth Summit’ held at Rio in 1992. The resulting ‘Convention on Biological Diversity’ has since been ratified by 186 nations, and has led to political systems as diverse as UK ‘Biological Action Plans’ (‘BAPs’), implemented in part at Local Council level (DOE 1994), and international commitments to stabilise loss of Biodiversity made at the World Summit on Sustainable Development in Johannesburg in 2002. Driven by the urgency of the ‘mass extinction catastrophe’ through which we are currently living, the rate at which we implemented policies designed to conserve biodiversity far outstripped our ability to validate our assessments of that biodiversity, although some attempts are currently being made to redress this (Crane 2003). Nevertheless, many concerns have been expressed about measures of biodiversity: for instance whether some groups of organisms provide more predictive estimates of biodiversi¬ ty than others (May et al. 1995); and at what taxonomic level (family, genus, species) estimates should be calibrated in various groups (Williams et al. 1998). However, one basic tenet, that morphologically-based philosophies of conventional taxonomy should provide the basic unit by which biodiversity is assessed (for instance the species), has rarely been questioned. This dogma involves a number of assumptions, usually tacit, and all of which are essentially untrue. For instance, it is assumed that the species represents a discrete evolutionary end-point; that all species possess roughly the same ‘value’ as units of evolutionary amplitude; and that species, which are largely based on morphological discon¬ tinuities, adequately represent the total significant evolutionary divergence and adaptation that has occurred. It is this last point which forms the subject of the present paper. It is a particular concern that many characters by which species are differentiated in Floras appear to be trivial and non-adaptive, and may have arisen by gene-drift and casual fixation. As a corollary, it seems that many very significant evolutionary characteristics are cryptic or ‘silent’, in that they are not expressed morphologically and so cannot be described taxonom- ically (Magurran 1998). In the last 200 years, we have enormously accelerated the rate at which species have become extinct, so that the current era has been compared with mass extinction episodes resulting from environmental catastrophes, such as that which massacred the dinosaurs 65 million years ago (Myers & Knoll 2001). What is much less often stated is that by creating new environments, and by bringing together in open habitats species which evolved in isolation, we have simultaneously greatly increased the rate at which new biological entities have evolved. We have no ‘base-line’ from earlier times against which EVOLUTION IN THYME ENOUGH? 137 hypothetical changes in rates of evolutionary diversification after the industrial and agricul¬ tural eras can be assessed, so that such assertions of recent increases in the rate of evolution¬ ary change are essentially untested, but this does not make them untrue. Nevertheless, there are numerous ‘headline’ examples where morphologically distinc¬ tive new species have evolved recently, usually as a result of long-distance migration, hybridity and polyploidisation, Spartina anglica C.E.Hubb. and Senecio cambrensis Rosser in the UK alone for instance. An even greater volume of rapid contemporary evolution may have passed largely unnoticed by students of biodiversity, because it does not in the main involve morphological changes (is ‘silent’) and so has not been recognised taxonomically. A vital function of measures of biodiversity is predictiveness, so that we can construct models which show how potentially ameliorative shifts in environmental policy might alter forecast rates of extinction (Pimm & Askins 1995, Pitman et al. 2002). Such models usually assume zero evolutionary change, so that the possible role played by ‘Red Queen’ in allowing threatened organisms to escape recent environmental damage by adaptation to new niches is generally ignored. Yet the literature of ecological genetics contains many thousands of examples where such evolution has occurred, unrecognised by taxonomy or measures of biodiversity. This paper reports two previously unreported but rather typical examples of such recent physiological ‘silent’ evolution of plants in response to polluting activities by man. ZINC TOLERANCE IN THYME INTRODUCTION Historic mining has produced widespread lead, cadmium and zinc contamination of the fluvial deposits of the South Tyne and Tyne river basins in north-east England (Hudson- Edwards et al. 1996). Although some mining for lead has occurred over 2000 years, mining for zinc started in about 1880 and peaked between 1900 and 1910, virtually ceasing by 1920 (Macklin & Smith 1990). Most contaminated deposits now lie above normal flood level, but remain phytotoxic so that plant communities are very open and carry a specialised metallophyte flora. This habitat, from which more than 40 species have been recorded in this region, seems to be equivalent to OV37 ‘ Festuca ovina-Minuartia verna community' in the National Vegetation Survey (Rodwell 2000), although the latter states that only ‘a few’ such river gravel sites are known (in fact more than 50 have been recorded) and admits that no such sites had been included in their survey. Locally, such sites have become known as ‘calaminarian grasslands’. We assume that most of the metallophyte species that occur here, including Thlaspi caerulescens J.S. & C. Presl, Minuartia verna (L.) Hiem., Viola lutea Huds., Silene vulgaris Garcke and Armeria maritima (Miller) Willd. (Richards et al. 1989), show unusually high levels of tolerance to metals such as lead and zinc. In the present context it is of interest to discover whether such tolerance is constitutive (typical of the species as a whole, as may be the case for the species above); inducible; or has evolved recently from relatively susceptible races on non-toxic habitats (Baker 1987). In order to demonstrate metal tolerance, we would expect that after transplantation to non-polluted soil, propagants of individuals originating from metal-rich habitats would grow more readily in conditions with high levels of the metal 138 CURRENT TAXONOMIC RESEARCH under investigation than did individuals from low metal (control) habitats, and that this tendency could be inherited between generations. We suspect that metal tolerance may have evolved within a number of species on Tyne calaminarian grasslands within the last 100 years, although it is possible that these habitats have been colonised by metal-tolerant strains of longer standing. In no case has a morpho¬ logically distinctive ecotype arisen. In this case, our investigations centred on zinc tolerance in wild thyme Thymus polytrichus A. Kemer ex Borbas subsp. britannicus (Ronn.) Ker¬ guelen which is a characteristic component of the most apparently toxic parts of this habitat, but elsewhere is typical of non-polluted soils. ORIGIN OF MATERIAL AND MEASUREMENT OF ZINC TOLERANCE To test for the occurrence of zinc tolerance in thyme, propagants of adult plants from the most apparently toxic parts of four calaminarian grassland sites (M1-M4) and two control sites on limestone grassland (Cl, C2) (Table 1) were brought into cultivation and established under glass where they were kept in active growth throughout the year. Total and extractable average soil concentrations of Zn, Pb and Cd, average soil pH, and extractable P are also given for each site, for soils taken from amongst the roots of the plants propagated (methods in Whitfield 2002). To measure tolerance to a toxic metal such as zinc, clones were propagated from semi-hardened cuttings, and root growth of these ramets studied when suspended in a liquid medium (solution of 0.5 g f1 calcium nitrate (0.72 g T1 Ca(N03)2.4H20 in distilled water)) in transparent containers (root extension technique, Wilkins 1957). Growth rate in this medium was studied for 2-3 weeks and when the rate became constant was recorded between days 3 and 6 in the medium (a) and between days 9 and 1 2 in the medium with 0.1 mg l'1 Zn 2+ (as ZnS04.7H20) added on day 6 (b). Tolerance index (TI) for a particular clone was the average of the b/a ratios. Absolute growth rates for days 9-12 (TIab) were also compared to avoid the possibility of scalar bias. To study inheritance of tolerance, controlled crosses were made onto pollen sterile (female) individuals in this gynodioecious species, using parents of known TI, and the TI of their offspring investigated using the same technique. RESULTS The average tolerance index (TI) for clones did not differ significantly between the toxic sites, but in every case was significantly greater than for the control sites, indicating that all the clones tested in all the toxic sites were tolerant to zinc (Table 2). For plants from toxic and control sites, there was a loose but significant relationship between the TI displayed by the individual and the extractable concentration of zinc taken from amongst the roots of that plant (r2 = 0.15, Fi,58= 9.82, P < 0.005), suggesting that some selection for individuals of a given level of tolerance occurs between microsites within the toxic areas. Offspring do not differ significantly in TI from the more tolerant parent, whether the male or female parent to the cross (Fig. 1), but when they result from crosses between a more tolerant parent and a less tolerant parent, their tolerance is significantly higher than that of the latter. Consequently, a pattern of dominance for zinc tolerance is clearly demonstrated, although offspring tend to have a higher TI than the average of their parents TI. Root growth rate in calcium nitrate (reading (a)) in individual clones shows no relation¬ ship to TI (Fig. 2), so that there is no indication from this technique that zinc tolerance causes the plant a metabolic cost. TABLE 1. LOCATIONS FROM WHICH EXPERIMENTAL PLANTS OF THYMUS POLYTRICHUS WERE TAKEN, WITH AVERAGE LEVELS OF TOTAL AND EXTRACTABLE METALS, PHOSPHORUS (P), CARBON (C) AND PH OF SOILS TAKEN FROM THEIR ROOTS. SITE AND SOIL CHARACTERISTICS (MEAN±SE). EVOLUTION IN THYME ENOUGH? 139 nd=not determined. Within each row, values followed by the same letter are not significantly different (P<0.05, /-test). * Gunnerton Crag: short grassland with thyme growing mainly in thin soil overlying rocky outcrops. Sherburn Hill: sparsely vegetated hilltop with discrete patches of thyme growing on thin, stony, limestone soils, f Mean values for soils sampled at a depth of 20-30 cm (total/extractable): Cd 7.26/2.3 1, Pb 141 7/540.4, Zn 2523/406.6 mg kg1. 140 CURRENT TAXONOMIC RESEARCH TABLE 2. AVERAGES OF ZINC TOLERANCE INDEX (TI%) AND TIAB (ABSOLUTE GROWTH RATE IN ZINC SOLUTION) FOR THYMUS POLYTRICHUS FROM CONTROL (Cl, C2) AND METALLIFEROUS (M1-M4) LOCATIONS TOGETHER WITH SIGNIFICANCE OF DIFFERENCES AGAINST Cl AND C2 RESPECTIVELY. Site No. of clones TI (%) P vs Cl, C2 (Mest) TIab (mm) Pv sCl,C2 (ANOVA*) Cl 3 6.67±1.20 a 0.55i0.14 a C2 10 10.44±2.92 a 1.49i0.34 a Ml 14 22.02i3.69 b <0.01, <0.05 4.26i0.82 b <0.001, <0.01 M2 8 24.95±4.16 b <0.05, <0.01 3.24i0.61 b <0.001, <0.01 M3 17 29.50±4.07 b <0.001, <0.001 4.1 li0.55 b <0.0001, <0.001 M4 11 25.84i2.20 b <0.01, <0.001 3.22i0.39 b <0.001, <0.01 Within columns, values followed by the same letter are not significantly different (P<0.05). * Values log-transformed for ANOVA: F5,57=6.17, FO.OOOl. Figure 1 . Results of four crosses between Thymus polytrichus individuals with a high (PI), medium TI (P3) and low TI (P2, P4) average zinc tolerance index (TI%). Male parent to cross on left. Average TI of offspring (01-04) shown with standard error. Inheritance of TI appears to be fully dominant and reciprocal (* P<0.05, ** P< 0.005). EVOLUTION IN THYME ENOUGH? 141 Tl (%) FIGURE 2. Average zinc tolerance index (TI%) (x axis) of each clone of Thymus polytrichus and root growth rate of each clone in calcium nitrate solution in the absence of zinc (y axis) are not correlated (r: = 0.003, F i,62= 0.017), suggesting that there is no innate cost to tolerance. SALT TOLERANCE IN RYE-GRASS INTRODUCTION De-icing salt (mostly sodium chloride) has been used on British roads since the 1940s, although heavy and widespread use with mechanical application did not occur until the 1960s (HMSO 1968). Deposition levels of 34g/m2 per individual application have been suggested (Davison 1971). De-icing salt is spread into the roadside environment by run-off, infiltration, use of snow-ploughs, and splash/spray from vehicles. However, salt levels decrease exponentially with distance from road and more than 90% of salt deposition occurs within 20 m of the road, so that in Northumberland, UK, Scott (1985) found the highest soil conductivities at 0.5 m from the verge; at 1.5 m conductivities were scarcely raised beyond those typical away from roads. Scott also recorded a striking annual fluctuation of soil conductivity in roadside soils, the greatest inferred salinities occurring in February. High salinity affects plants in many ways, by decreasing osmotic potential within the plant, through the toxicity of sodium and chlorine ions (Westing 1969, Norrstrom & Bergstedt 2001) and by altering soil pH, nutrient availability, and structure (Davison 1971). By replacing other cations, sodium causes soil particles to lose their capability for aggrega¬ tion and flocculation, so that the ability of the soil to hold water and oxygen decreases. Consequently, salt can be both the primary and secondary source of plant decline (Dobson 1991). 142 CURRENT TAXONOMIC RESEARCH As a result, it is frequently found that ‘grassland’ immediately adjacent to the road is essentially devoid of vegetation, particularly where trunk roads on which vehicle speeds are relatively high are not provided with a protective ‘hard shoulder’ lane. In these bare areas it is possible to find isolated individuals of several species of grass. Puccinellia distans (Jacq.)Parl. naturally occurs on salt-marshes and is well-known to have colonised salted verges throughout much of Britain (Scott & Davison 1982) so it is expected to be constitu- tively salt-tolerant. Most other individuals occurring in these circumstances are either Poa pratensis L. or Lolium perenne L. As reproduction in Poa pratensis is primarily apomictic, we decided to concentrate on perennial rye-grass Lolium perenne. Lolium perenne is a variable outcrossing tetraploid. When many thousands of seedlings were screened, a very few proved to be salt tolerant as assessed by their capability for root growth in a saline solution (Ashraf et al. 1986a) and this capability was associated with greater dry matter productivity in sand culture with added saline when compared with low tolerance individuals (Ashraf et al. 1986b). Further selection of ‘salt-tolerant’ lines led to individuals which performed even better in saline conditions (Ashraf 1990). Nevertheless, although it has been shown that Lolium perenne can have the potential to vary genetically for tolerance to salt, there has been no indication that this has actually led to the evolution of wild strains which have been able to colonise salt-polluted soils. ORIGIN OF MATERIAL AND MEASUREMENT OF SALT TOLERANCE To test the hypothesis that the isolated individuals of Lolium perenne which occur in salt-polluted bare patches were salt tolerant, we took four such individuals (‘verge’) into cultivation and tested them in comparison with four individuals occurring about 1 5 m from the carriageway in a grass sward of healthy appearance (‘control’). In April 2002, samples were dug up from the verges of the A 167, a dual carriageway in the suburbs of Newcastle upon Tyne at NZ2 15.672. Soil was sampled from amongst the roots of each clone and the salinity estimated by conductivity using the technique of Davison (1971) which is adapted from that recommended by the Soil Survey of England and Wales. Plants were then washed clean with tap water and each divided into 40 single-shoot tillers each with a fragment of root. These tillers were individually grown under glass in standard volume plastic pots in a single carefully mixed batch of compost approximating to John Innes 2 watered from below for two weeks; those few tillers which failed to establish and grow were abandoned. The experimental period lasted nine weeks, during which time batches of 80 tillers were each evenly watered from above in a fine spray every three days using 1 0 1 of either distilled water or 0.1 5M sodium chloride (8.76g/l) (‘saline’). Soil conductivities were checked at three week intervals, and on average increased from 8 to 17 mS in pots watered with saline for nine weeks, but remained below 1 mS at weeks three and six in pots watered with distilled water. In a second experiment, plants which had been treated with saline for the first three weeks only, and the first six weeks only, and were watered with distilled water for the remainder of the nine weeks were compared with those watered only with distilled water. Also, plants watered with saline for only the final three weeks, and the final six weeks were compared with those which had been treated with saline for all nine weeks. At harvest, the performance of each initial tiller was judged in two ways: final tiller number; and above-ground dry weight. EVOLUTION IN THYME ENOUGH? 143 RESULTS The average conductivities of soils from amongst roots of plants collected in mid April varied from 0.9 to 2. 1 mS for verge samples (mean 1 .28 mS), but from only 0.3 to 0.5 mS in control samples (mean 0.39 mS). At harvest, verge plants watered with saline had grown on average twice as many tillers as control plants treated with saline, and the number of tillers formed by verge plants in saline did not differ significantly from those watered with distilled water. After treatment with distilled water we found no significant difference between the number of tillers formed by verge and control plants, so that this result revealed no appreciable metabolic cost to verge plants compared to control plants in the absence of saline (Fig. 3). Analysis of dry weight by GLM showed a significant interaction between source (verge or control) and treatment (saline or distilled water) (Fi,273= 20.53, P < 0.001), thereby demonstrating that verge plants showed more tolerance to saline than control plants. There was no significant difference in tolerance noted between clones from the same source (Table 3). 14 1 12- 10- Weeks DS = Salt exposed plant grown in distilled water DC = Control plants grown in distilled water SS = Salt exposed plants grown in saline solution SC = Control plants grown in saline solution Figure 3. Mean and standard error of tiller number developed from single tillers of Lolium perenne over nine weeks, indicating verge plants (DS, SS) and control plants (DC, SC) grown in saline (SS. SC) and distilled water (DS, DC). Verge plants grow much better than control plants in saline and do not differ from those grown in distilled water. 144 CURRENT TAXONOMIC RESEARCH TABLE 3. ANOVA, EFFECT OF SOURCE AND TREATMENT ON DRY WEIGHT IN LOLIUM PERENNE. Source DF Seq SS Adj SS Adj MS F P Treatment 1 6.82 6.27 6.27 80.63 <0.001 Source 1 2.83 2.78 2.78 35.79 <0.001 Clone 6 0.32 0.25 0.04 0.55 0.772 Interaction, treatment and source 1 0.50 0.59 0.59 7.57 0.007 Interaction, treatment and clone 6 0.74 0.74 0.12 1.59 0.158 Error 103 8.00 8.00 0.08 Total 118 19.2 TABLE 4. ANOVA, EFFECT OF SOURCE AND TIME OF TREATMENT ON DRY WEIGHT IN LOLIUM PERENNE WATERED WITH SALINE. Source DF Seq SS Adj SS Adj MS F P Source 1 1.48 1.37 1.37 18.9 <0.001 Clone 6 0.87 0.80 0.13 1.85 0.097 Time 2 1.40 1.43 0.71 9.87 <0.001 Interaction, time and source 2 0.02 0.03 0.01 0.18 0.832 Interaction, time and clone 12 0.38 0.38 0.32 0.44 0.942 Error 112 8.10 0.07 0.07 Total 135 TABLE 5. ANOVA, EFFECT OF SOURCE AND TIME OF PRETREATMENT WITH SALINE ON DRY WEIGHT IN LOLIUM PERENNE WATERED WITH DISTILLED WATER. Source DF Seq SS Adj SS Adj MS F P Source 1 8.62 8.78 8.78 143.2 <0.001 Clone 6 1.76 1.67 0.28 4.54 <0.001 Time 2 2.43 2.41 1.21 19.7 <0.001 Interaction, time and source 2 0.12 0.13 0.06 1.06 0.349 Interaction, time and clone 12 0.66 0.66 0.05 0.89 0.555 Error 113 6.93 6.93 0.06 Total 136 20.5 EVOLUTION IN THYME ENOUGH? 145 Control plants which had been pretreated with saline for the first three and six weeks of the experiment performed less well over the final three weeks of the experiment in saline than plants which had been pretreated with distilled water and then tested with saline (Table 4). In contrast, verge plants were not affected by this early exposure to saline. We concluded that tolerance to salt is not induced by exposure to salt in this case, but that salt tolerance expressed by verge plants is probably under genetic control. When control plants are subjected to saline in early weeks, their performance in the final three weeks of the experiment in distilled water is affected compared with those watered with distilled throughout, but verge plants show no such ‘hang-over' effect from this early exposure to saline (Table 5). DISCUSSION We have shown for the first time that wild thyme Thymus polytrichus growing on metal- polluted river gravels are tolerant to zinc, and that perennial rye-grass Lolium perenne on heavily salted road-verges is salt tolerant. The pollution of both these habitats has resulted directly or indirectly from the activities of man. We believe that the following protocols must be satisfied if the occurrence of adaptive tolerance is to be established: • in the concentrations in which it typically occurs in polluted habitats, the pollutant should be sufficiently phytotoxic that vegetational cover is open and bare; • concentrations of the pollutant should normally and consistently be at a much higher level in polluted habitats than in those where susceptible ‘wild type’ plants (which are used as controls and which are deemed to be typical of stock from which tolerant forms evolved) are found; • experimental plants should be grown in pollutant-free conditions prior to the experimen¬ tation. to lessen the possibility of 'hang-over' or inducible' effects from prior exposure to the pollutant; • replicated propagants taken from polluted habitats should for every clone grow signifi¬ cantly better in the presence of the polluting substance in otherwise uniform controlled conditions than those taken from most or all clones from an unpolluted (‘control’) source(s); • in the absence of the polluting substance (control conditions), propagants from polluted habitats would not however be expected to grow better than those from control habitats; • EITHER the heritability of the tolerance should be established by comparing the toler¬ ance of the offspring of two susceptible parents with those of two tolerant parents, OR the potential for tolerance induction is investigated by comparing the tolerance of clones of control origin which have, and which have not, been pretreated with the pollutant. We believe that these conditions have been met in each of the two cases we have investigated. These two examples differ in many attributes. One species is a woody dicotyledon, the other a herbaceous monocotyledon. In one case it is a metal that is the toxin, in the other the toxicity is more indirect in the sense that soil and soil water properties are affected by salt in complex ways. Metal pollution on river gravels is decreasing, but this is probably untrue of roadside salt deposition. Nevertheless, we are more struck by a number of features shared by both cases. 146 CURRENT TAXONOMIC RESEARCH • The polluting activities are relatively recent, in that most zinc pollution of the River South Tyne has occurred in the last 100 years, and most roadside salting in the last 40 years. • Thus, we suppose that in both cases, the occurrence of tolerant individuals in these habitats is also very recent. However, we need to explore the likelihood that these species could have been exposed to the pollutant at an earlier date, and that tolerant plants could have dispersed from there to the present sites. In the case of Thymus polytrichus, ‘natural’ metalliferous habitats occur in the River South Tyne catchment, but are excep¬ tionally localised (Hudson-Evans et al. 1996). Lolium perenne is not usually a constitu¬ ent of naturally saline conditions (but T. Rich (in litt.) provides one anecdotal example of L. perenne in a salt-marsh from, appropriately enough, Rye in Sussex). • Tolerant genotypes have spread into other suitable localities. Zinc tolerant individuals of Thymus polytrichus are known to occur at a number of sites, possibly as the result of colonisation of polluted habitats by water-borne seed from an upstream founder where evolution of tolerance originated. Although salt tolerance of Lolium perenne has only been tested at one site, plants occur within similar heavily polluted areas throughout greater Newcastle upon Tyne and we expect these also to be salt tolerant and to have been spread between sites by the movement of vehicles. • We can establish no apparent cost expressed as vigour associated with the tolerance. • Evolution has been for physiological features, and as far as we are aware there are no morphological features that differentiate between tolerant and susceptible individuals in either case. It is likely that such examples of very recent ‘silent’ physiological evolution are very common. Although many already exist in the literature, many more probably lie undetected in the absence of experimental work. Physiological evolution may not only occur in response to chemical pollution, but also to climate change, or to the introduction or evolution of ‘new’ predators, pathogens and competitors, and such evolution is likely to be much more signifi¬ cant in future considerations concerning the survival of biodiversity, than are the few ‘headline’ examples of new distinctive hybridogenous species such as Senecio camhrensis. We need to consider whether ‘alpha taxonomy’ alone will be adequate to answer the needs of future conservationists. By attempting to quantify ‘silent’ physiological evolution more accurately, perhaps the time has come to challenge the popular conception that biodiversity and taxonomic diversity are synonymous. REFERENCES Ashraf, M. (1990). Selection for salt tolerance and its genetic basis in perennial rye-grass (Lolium perenne L.). Hereditas 113: 81-85. Ashraf, M., McNeilly, T. & Bradshaw, A.D. (1986a). Tolerance of sodium chloride and its genetic basis in natural populations of four grass species. New Phytologist 103: 725-734. Ashraf, M., McNeilly, T. & Bradshaw, A.D. (1986b). The response of selected salt- tolerant and normal lines of four grass species to NaCl in sand culture. New Phytologist 104:453-461. Baker, A.J.M. (1987). Metal tolerance. New Phytologist 106 (suppl.) 93-1 1 1. Crane, P. (2003). Measuring Biodiversity for Conservation. Royal Society of London, Policy Document 11/03. EVOLUTION IN THYME ENOUGH? 147 Davison, A.W. (1971). The effect of de-icing salt on roadside verges. I. Soils and plant analysis. Journal of Applied Ecology 8: 555-561. DOBSON, M.C. (1991). De-icing salt damage to trees and shrubs. Forestry Commission Bulletin 101. HMSO, London. DOE (Department of the Environment) (1994). Biodiversity: the UK Action Plan. HMSO, London, UK. HMSO (1968). Salt treatment of snow and ice on roads. Road Note 18, edn 2. Road Research Laboratory, Crowthome, Berkshire, UK. Hudson-Evans, K.A, MacKlin, M.G., Curtis, C.D. & Vaughan, D.J. (1996). Processes of formation and distribution of Pb-, Zn-, Cd-, and Cu-bearing minerals in the Tyne basin, north-east England: implications for metal-contaminated river systems. Environmental Science and Technology 30: 72-80. Macklin, M. & Smith, R.S. (1990). Historic riparian vegetation development and alluvial metallophyte plant communities in the Tyne basin, north-east England. In: J.B. Thornes, (ed.), Vegetation and erosion: processes and environment, pp. 239-256. John Wiley & Sons, Chichester. MACURRAN, A.E. (1998). Population differentiation without speciation. Philosophical Transactions of the Royal Society of London B 353: 275-286. May, R.M., Lawton, J.H. & Stork, N.E. (1995). Assessing extinction rates. In: J.H. Lawton & R.M. May, (eds), Extinction Rates, pp. 1-24. Oxford University Press, Oxford. Myers, N. & Knoll, A.H. (2001). The biotic crisis and the future of evolution. Proceedings of the Natural Academy of Sciences , Washington DC 98: 5389-5392. Norrstrom, A.C. & Bergstedt, E. (2001). The impact of road de-icing salt (NaCl) on colloid dispersion and base cation pools in roadside soils. Water, Air and Soil Pollution 127:281-299. Pimm, S.L. & Askins, R.A. (1995). Forest losses predict bird extinctions in eastern North America. Proceedings of the Natural Academy of Sciences, Washington DC 92: 9343- 9347. Pitman, N.C.A, Jorgensen, P.M., Williams, R.S.R., Leon-Yanez, S. & Valencia, R. (2002). Extinction-rate estimates for a modem neotropical flora. Conservation Biology ; 16: 1427-1431. Richards, A.J., Lefebvre, C., Macklin, M. Nicholson, A. & Vekemans, X. (1989). The population genetics of Armenia maritima (Mill.) Willd. on the River South Tyne. New Phytologist. 112: 281-293. RODWELL, J.S. (2000). British Plant Communities Vol. 5. Cambridge University Press, Cambridge, UK. SCOTT, N.E. (1985). The updated distribution of maritime species on British roadsides. Ph.D. thesis, University of Newcastle upon Tyne. Scott, N.E. & Davison, A.W. (1982). De-icing salt and the invasion of road verges by maritime plants. Watsonia 14: 41-52. WESTING, A.H. (1969). Plants and salt in the roadside environment. Phytopathology 59: 1174-1180. Whitfield, L. (2002). Heavy metal tolerance and mycorrhizal colonisation in Thymus polytrichus A. Kerner ex Borbas ssp. britannicus (Ronn.) Kerguelen (Lamiaceae). Ph.D. thesis, University of Newcastle upon Tyne. Wilkins, D.A. (1957). A technique for the measurement of lead tolerance in plants. Nature 180: 37-38. 148 CURRENT TAXONOMIC RESEARCH Williams, P.H., Gaston, K.J. & Humphries, C.J. (1998). Mapping biodiversity value world-wide: combining higher-taxon richness from different groups. Proceedings of the Royal Society of London B 264: 141-148. Current taxonomic research on the British & European flora Bailey, J. & Ellis. R.G. (eds) 149-152 (2006), BSBI. London. 149 Some thoughts about the future of electronic floras RUUD VAN DER MEIJDEN National Herbarium of the Netherlands, P.O.Box 9514, 2300 RA Leiden. In Europe a few countries already have electronic floras: Swiss (Lauber & Wagner 1997), Sweden (Anderberg 1998), the Netherlands (Van der Meijden 1999), Germany (Seybold 2001), Bulgaria (Petrova 2002), British Isles (Stace 2004). The advantages of electronic floras are that they contain more illustrations than any printed book may do (for the British one: 7000 colour slides and 2000 line drawings), as well as distribution maps. In future editions illustrations can be added or replaced easily, and maps can be updated easily as well. An extra feature of the British and the Dutch flora is the fact that they contain all basal distribution records, which enables the user to withdraw those data for a certain area; one can easily get a list of all species known in a certain square or selection of squares. A disadvantage is that a computer is (still) heavier than a book. I expect that printed floras will remain to be used in the future, especially in the field, and that electronic floras will be used at home or in the base camp. Nearly all electronic floras use a traditional dichotomous key, usually literally the same as in the printed book. The Interactive Flora of the British Isles (Stace 2004) has made the standard dichotomous keys much more user-friendly by a series of devices. Firstly, the user is confronted with only the relevant couplet, and once the choice is made he is taken straight to the next relevant couplet (Fig. 1). Secondly, at any point, the user is told how many taxa he has already eliminated and how many taxa remain as possibly correct (Fig. 1 , 3); the latter can be listed and by selecting any of them a picture of that taxon can be displayed (Fig. 5). Thirdly, at any point the user can obtain a list of the decisions that he has already made, and after studying them he can back-track to one that he thinks might have been made in error, and change it accordingly to take him on a different pathway (Fig. 3). Finally, by clicking on any term unknown to the user he is taken straight to its definition in the Glossary (Fig. 1, 2); this facility applies equally to the plant descriptions as well. By these means it is hoped that the two major problems encountered by beginners, the complex terminology and the mysteries of dichotomous keys, can be alleviated. Next to this, the electronic floras sometimes include multi-access keys, which are more easily used on a computer than in a book. At present, most multi-access keys are available for difficult groups only. The challenge, however, is to make really user-friendly general keys, like those that Richard Pankhurst is making for the British flora. Another solution may be the construction of a kind of sorting program, like the one in the Dutch flora. The characters used there are simple: leaves (compound, simple; length- width ratio, arrangement), flower colour and actual date of finding the plant in flower and its height, habit and habitat. With the help of these easy characters the user gets a quick selection of species; he or she can click on their names, attempting to find the proper species. 150 CURRENT TAXONOMIC RESEARCH igtrlS X File Edit Go Window Help Page 610: Festuloliurn(X) and Festulpia(X) Choices made: 32 Possible species/taxa: 2 610a, Pointed auricles present at junction of leaf and sheath; lemmas rather abruptly narrowed to awn; upper glume with 5-9 veins. Genus Festuhlium(X) (= Genus Festuca x Genus Lofium) Genus Festulolium(X) 610b. Auricles absent; lemmas very gradually narrowed to awn; upper glume with 1-3 veins, Genus Vutpia) (X) (= Genus Festuca x Genus Genus Festulpia(X) A, jf?- Find Decision Path EES y - jnTxji File Edit Go Window Help ABCDEFGHI KLMNOP RSTUVW z f Definition Synonyms Multimedia lid basal exten sion of a leaf-blade, especially in a. Family Poaceae — ; ▼ J fl cj> - Contents Find Previous Next Back * , M w “"I /* ** First Previous Back Figure 1. Page from Stace (2004) showing a text key couplet (with glossary screen inset), decisions made and number of taxa remaining. Figure 2. Pages from Stace (2004) showing glossary illustrations linked to choices in fig. 1. FUTURE OF ELECTRONIC FLORAS 151 Figure 3. Page from Stace (2004) showing list of decisions already made and taxa remaining. -□jxl ,■ Lyy ,,.4. . . File Edit Go Window Help i Description i f Key to ta a [ Synonyms Classification Genus Festulpia (X) Scientific name: X Festulpia Melderis ex Stace & R. Cotton (= Genus Festuca X Genus Vulpia). Diagnostic features Plants perennial and vegetatively close to the Festuca parent, but with fewer and shorter (or absent) rhizomes and some overlapping sheaths, Panicles narrower and less branched than in Festuca, with markedly longer awns, Lower glume c.1/2 as long as upper, Anthers indehiscent with ± empty pollen grains, but a very small degree of fertility may exist, Festuca arenaria and Festuca rubra hybridise with Vulpia fasciculata, Vulpia bromoides and Vulpia myuros, and 4 of the 6 possible combinations have been found: - See Festulpia(X) melderisii Stace & R, Cotton (= Festuca arenaria x Vulpia fasciculata) - See Festulpia(X) hubbardii Stace & R. Cotton (= Festuca rubra x Vulpia fasciculata) - See Festuca rubra-X-Vulpia-bromoides. - See Festuca rubra-X-Vulpia-myuros. This species is keyed out on Page 610: Festulolium(X) and Festulpia(X') in the Text Key. Contents Find Save Print Previous Next Back Figure 4. Page from Stace (2004) showing a Genus page with diagnostic features and links to various other pages. 152 CURRENT TAXONOMIC RESEARCH Species File Edit Go Window Help _ n x : Y Y Genus Family Synonyms Classification Literature Multimedia Festulpia(X) hubbardii Scientific name: X Festulpia hubbardii Stace & R, Cotton (= Festuca rubra x Vulpia fasciculata) Diagnostic features Lower glume 2.4-4, 4mm; upper glume (including awn) 3. 5-7. 2mm. Lemmas 6-9,5rnrn plus awn 2-5.5rnrn. Anthers 3, l,5-2mrn. Chromosome number: 2n = 35, Habitat Native; on open sand-dunes with parents. Distribution Probably frequent in Channel Islands and Britain North to Westmorland. This intergeneric hybrid is only keyed out in the Text Key as Genus Festulpia(X) on Page 610: Festulolium(X) and Festulpia(X) , \ Hide Picture Contents Find Save Print Previous Nett Back Figure 5. Page from Stace (2004) showing a Species page with illustration and links to various other pages, including more illustrations. Although this sorting program must still be improved considerably, its advantage is that the user does not need much botanical knowledge. This is an important factor, as botany now plays a very minor part in education programs at schools and at home. If we succeed in making our floras as easy to handle as possible, we will get again the attention of the younger generations. Then the electronic flora will become a tool to widen the potential pool of plant observers, and to rejuvenate this group. This is necessary for the public awareness of the ongoing changes in the wild flora and the effects of this for the whole environment. REFERENCES ANDERBERG, A. & A.-L. (1998). Den virtuella floram see http://linnaeus.nrm.se/flora/ LAUBER, K. & WAGNER, G. (1997). Flora der Schweiz. Haupt, Berne. Meijden, R. VAN DER (1999). Heukels ’ Interactieve Flora van Nederland. ETI, Amsterdam. Petrova, A., Asov, B., Vasilev, R., Dimitrov, D., & Jotov, G., (2002). Flora Bulgarica : see http://bsbcp.inet.bg/flora Seybold, S. (2001). Schmeil-Fitschen Interaktiv. Quelle&Meyer, Wiebelsheim. STACE, C.A. (2004). The Interactive Flora of the British Isles. ETI, Amsterdam. Cwrent taxonomic research on the British & European flora Bailey, J. & Ellis, R.G. (eds) 153-154 (2006), BSBI. London. 153 New field botanists for the future FRANKLYN PERRINGt Clive Stace was a precocious field botanist who contributed to the first Maps Scheme as a teen-aged boy in the mid-1950s, significantly encouraged by his biology master and the local natural history society in Tunbridge Wells and by the nearby Francis Rose. And he was not alone - other youngsters, notably Raymond Harley at Harrow and David Allen at Rugby, were similarly encouraged to high quality work. But where are such young people today? Today very few biology teachers have the taxonomic skills, or the realisation of the need for field botanists, for them to be able or willing to promote it as a hobby, let alone as a career. Yet it is clear that there never was a time when there was so much demand for trained field botanists and never a time when there were so few available. To meet this gap the BSBI is working with many others to produce materials, literature and courses to make opportuni¬ ties for careers in field botany widely known. These projects include: SPOTLIGHT ON PLANTS 32 free places offered annually to second-year sixth formers in collaboration with the Field Studies Council (FSC). TREES AND SHRUBS WEBSITE Developed by BSBFSAPS (Science and Plants for Schools), this is now being expanded to cover winter trees and twigs (www-saps.plantsci. cam. ac.uk/trees/home.html) (see fig. 2, p. 9). BOTANIC GARDENS LEAFLET Information on how to get to know British Plants, available via the Botanic Gardens Education Network (BGEN) in all Botanic Gardens. CERTIFIED COURSES IN BIOLOGICAL RECORDING AND PLANT IDENTIFICATION Birmingham University has a range of courses, starting with the part time field based certificate in Biological recording and species identification. An identification masterclass for experienced botanists to study with one or two expert tutors for five days to get their ID skills to a high enough level to sit a mock IDQ administered by the Natural History Museum. Finally, there is the part-time MSc on Biological Recording. Leicester University is also active in this field and offers a certificate in Applied plant studies: wild and garden plants. The BSBI will be working with the NHM to develop a range of IDQ qualifications for vascular plants. 154 CURRENT TAXONOMIC RESEARCH CAREERS IN FIELD BOTANY A two-page leaflet, Why choose Plant Science as a Careerl , has been prepared and is being distributed to schools and careers agencies. LEARNING ABOUT PLANTS A very productive partnership with the Wildlife Trusts which has resulted in many local one-day courses and has generated a 4-page leaflet to give to all actual or potential students. Editors’ note - The above notes were prepared by Franklyn Perring when he was still intending to attend the meeting shortly before his untimely death. They were delivered by Philip Oswald. It should be emphasised that Franklyn was deeply involved with some of the above schemes; he was running field days at FSC centres and for the Leicester University course up to his final year, and the excellent Trees and Shrubs Website for school pupils is largely due to him. 155 List of participants (including all authors* (those not present**)) Joel Allainguillaume** John Bailey Richard Bateman* Paddy Braithwaite Michael Braithwaite Mary Briggs Ailsa Bums Julia Carey Pilar Catalan* Arthur Chater Clare Coleman Ann Conolly Alessandra Contento Tom Cope Arthur Copping Robyn Cowan Mick Crawley Jane Croft Ingrid de Kroot Nicholas De Sausmarez Ian Denholm Bob Ellis Gwynn Ellis Graham French Miss M Fryer Mr P.G. Gamer Mary Ghullam Margot Godfrey Mike Grant Eric Greenwood Dr H.J. Harvey Trevor Harwood Rob Heijman Pat Heslop-Harrison Yolande Heslop-Harrison Pete Hollingsworth* Michelle Hollingsworth* Dr J.D. Hopton George Hutchinson Martin Ingrouille* Clive Jermy Andy Jones University of Reading Leicester (Ed.) NHM, London Roxburghshire Roxburghshire West Sussex Derbyshire/Staffordshire Buckinghamshire University of Zaragoza, Spain Ceredigion Surrey Leicester Leicester RBG Kew Norfolk RBG Kew Imperial College, London Cambridgeshire The Netherlands Cornwall Rothamsted Norfolk Cardiff (Ed.) Edinburgh North Yorkshire Worcestershire Norfolk Kent RHS Wisley West Lancashire Gloucestershire Lancashire The Netherlands Leicester Leicester RBG Edinburgh RBG Edinburgh North Yorkshire NMW, Cardiff Birkbeck College, London Herefordshire CCW, Ceredigion 156 CURRENT TAXONOMIC RESEARCH Bengt Jonsell* Uppsala, Sweden Lena Jonsell Uppsala, Sweden Stephen Jury Reading Thomas Karlsson** Swedish Museum of Natural History, Sweden Dr R.J. Kemp Buckinghamshire Anna Krahulcova** Czech Republic Franta Krahulec* Czech Republic Mr N.J. Law Derbyshire Bob Leaney Norfolk Mr R. Maskew Worcestershire Victoria Matthews RBG Kew Douglas McKean Edinburgh Andy McVeigh Buckinghamshire Jochen Muller* * University of Jena, Germany Rose Murphy Cornwall Philip Oswald Cambridge Richard Pankhurst* RBG Edinburgh Peter Payne Suffolk Franklyn Perringf ** Sarah Priest Berkshire Richard Pryce Carmarthenshire, BSBI President Martin Rand Hampshire Tim Rich NMW, Cardiff John Richards* University of Newcastle Jose Angel Lopez Rodriguez** Zaragoza, Spain Mary Clare Sheahan RBG Kew Roy Smith Derbyshire (Excursion Leader) Jane Squirrell RBG Edinburgh Margaret Stace Leicester Clive Stace* Leicester Trude Schwarzacher Leicester Peter Thompson Leicester Peter Thomson Hereford Pedro Torrecilla** Universidad Central de Venezuela, Venezuela Ruud Van der Meijden* Leiden, Netherlands Mike Walpole Loughborough Lynn Whitfield* * Royal Holloway, University of London, Mike Wilkinson* University of Reading Helena Winsnes** University of Newcastle Ray Woods CCW, Powys Goronwy Wynne Flint PLATE 1. (a) Platanthera chlorantha (left) versus P. bifolia (right) from Skye; (b) Dactylorhiza ebudensis, a recently evolved species confined to North Uist; (c-e) Marshland (c), grassland (d) and heathland (e) species of Gymnadenia ; (f) Epipactis sancta , a recently evolved species confined to Lindisfame; (g-h) Early-flowering (g) versus late-flowering (h) Neotinea ustulata in southern England (p. 93-102). PLATE 2a. Arabidopsis suecica on a railway in PLATE 2b. Saxifraga osloensis, Osthammar, PLATE 2c. Deschampsia bottnica on the Baltic coast Funbo, Uppland, Sweden. Uppland, Sweden. at Hallnas, Uppland, Sweden. Photo © B. Jonsell (p. 124). Photo © B. Jonsell (p. 125). Photo © B. Jonsell (p. 129). PLATE 3. Clive relishes the sight of Sesleria caerulea at it’s southernmost outpost in Monk’s Dale, Derbyshire (p. 2). PLATE 4. John Richards identifying a yellow composite at Monk's Dale, Derbyshire (p. 2). 5. The party at Houndkirk Moor, Derbyshire, site of one of the most southerly stations of TrientaHiL emanea in the British Isles (p. 2). This volume contains papers from the ‘Current Taxonomic Research Work on the British & European Flora’ meeting organised by the B SB I to commemorate Clive Stace’s retirement from the University of Leicester. The papers closely mirror his particular interests, but still form a cohesive whole in the area of Floristics and Phylogeny. Papers range from regional Floras, through new approaches to Keys and electronic Floras to more detailed studies of particular plant groups, often involving DNA based techniques. The book takes an optimistic look at the future directions of plant taxonomy and will be of value to anyone interested in the European flora. 780901 158376 Upsai.. ERIAL. ISBN 0-901 1 58-37-2 r A S * A IS hfsi w ^ iJ fa g JLj