Table Of Content

American Fern Journal 91(3):150-165 (2001) and Diversification Relationships of Extant Homosporous Lycopods WlKSTROM NiKLAS Department The Museum, of Palaeontology, Natural History Cromwell Road, London Kingdom SVV7 5BD, United — A Abstract. series of phylogenetic analyses using nucleotide sequence data have resolved many aspects of the relationships in a group of land plants that until recently had received comparatively homosporous little attention, the lycopsids (Lycopodiaceae). The family includes no more than 400-500 living species but the group has evolved as an isolated lineage since the Early Devonian (390 Mya). Despite this ancient history, patterns emerging through the phylogenetic analyses imply most that diversification in this group is comparatively recent. The Lower Jurassic stem section minimum Lycoxylon indicuin indicates a age for the split between Lycopodium and Lycopodiella at 208 Mya, and reticulate fossil spores from the Early Jurassic indicate that early cladogenesis in Lycopodium is of equivalent age. In the diverse predominantly epiphytic Huperzia group, biogeo- % graphic data indicates that 85-90 of all living species result from cladogenic events postdating the final rifting of S. America and Africa in Mid to Late Cretaceous. The timing of these events coincides with the radiation of Angiosperms, and the diversification of epiphytic Huperzia was mediated by development likely the of broad leaved Angiosperm rain forests. Results indicate a single origin of epiphytism in Huperzia, hut there have been two at least reversals to the terrestrial The habit in the neotropics. diversification of a large secondarily terrestrial clade, including about montane 80 high altitude species, was likely triggered by the Andean orogeny in the Mid Miocene, no more than 15 Mya. an are herbaceous among species most are the easily recognizable and conspicuous elements group was soon recognized as the product of an early cladogenic event land in plant evolution (Banks, 1968, 1975; Bower, Recent 1908). phylogenetic analyses support view and this indicate that lycopsids are remaining sister to all ants (Doyle, 1998; Duff and Nickrent Kranz provides strong evidence that the three extant lineages (Lycopodiaceae, Sela- and ginellaceae Isoetaceae) also originated in the Palaeozoic. The occurrence of Protolepidodendrales Devonian in the Early (Kenrick and Crane, 1997} sug- minimum Myr gests a age of 390 for the split between Lycopodiaceae and the ligulate clade including and Selaginellaceae Isoetaceae. Furthermore, the occurrence of Isoetales in the Late Devonian (Kenrick and Crane, 1997] suggests Myr known these events, little is crown diversification of the groups. Although the fossil record includes taxa ^ V A K ^ & imilarities 1992), Selaginellaceae (Thomas, 1992, 1997) and Isoetaceae (Pigg, 1992; Re- tallack, 1997), their specific relationships to living species remain unclear. One of the major problems has been the lack of well developed phylogenetic WIKSTROM: EXTANT HOMOSPOROUS LYCOPODS 151 hypotheses. Without such hypotheses, no way whether there is of telling the crown fossils are nested within the groups, or are part of the stem lineages. Isoetes bestonii Retallack example has been back for identified as far as the shown Triassic [Retallack, 1997). If to be part of the crown group of living would Isoetes, this certainly contribute to our understanding of the patterns of diversification of living species. Separated from the ligulate lycopsids since the Early Devonian, the homosporous lycopods (Lycopodiaceae) have evolved an independent as lineage for show at least 390 Myr. Morphologically, living species striking similarities to some of the oldest vascular plants (Heuber, 1992; Kenrick and Crane, 1997], but from a palaeobotanical perspective Lycopodiaceae have been particularly problematic (Skog and Hill, 1992; Thomas, 1992]. The macrofossil record is meager and work hampered by relatively palaeobotanical recognition prob- is lems. Absence of clear morphological synapomorphies for the family makes assignments of fossil taxa difficult, and this is often based on absence of the more groups such Another characteristics of distinctive as Selaginellaceae. problem involves the superficial similarities between Lycopodiaceae and organ systems of other plant groups. These problems notwithstanding, Skog and Hill hypothesized major groups within Lycopodiaceae were (1992] that the living established during the Late Jurassic-Cretaceous. In other words, despite its modern apparent species diversity in Lycopodiaceae evolved relict status, more and with comparatively recently in parallel that in diverse, ecologically prominent groups such as flowering plants. my Over the couple of years, and coauthors have been developing a last I Lycopodiaceae (Wikstrom and phylogenetic hypothesis living species of for Wikstrom Using molecular sequence Kenrick, 1997, 2000a, 2000b; et ah, 1999]. and data from the plastid rbcL gene, and from the trnl^-truF intron spacer we number regions, have conducted a of phylogenetic analyses at different These analyses provide empirical support levels of the hierarchy in the family. for the idea that most cladogenesis in this ancient family is comparatively made recent. Here review some of the progress that has been in our under- I standing Lycopodiaceae relationships and the implications for interpreting of the evolution of the family. Relationships monophyly and PnyLLOGLOSSL/M.—Recent analyses Lycopodiaceae cladistic weakly supported by have shown that monophyly of Lycopodiaceae is at best Kenrick and Crane, Other stud- morphological characters [Crane, 1990; 1997). synapomorphies (Wagner and have unequivocal family ies failed to identify may Lycopodiaceae be paraphyletic to a hetero- Beitel, 1992), or indicate that and sporous lycopsid clade comprising Selaginellaceae, Isoetaceae extinct rel- Molecular characters, on the other hand, furnish very (Bateman, atives 1992). (Wikstrom and Kenrick, 1997, monophyly family strong support of the for mono- Lycopodiaceae unequivocally support 2000a). Their rbcL sequence data Lycopodiaceae phvlv. Furthermore, molecular data resolve a basal split in sep- AMERICAN FERN JOURNAL: VOLUME NUMBER 152 91 3 (2001) Marchantia polymorphs Andraea rupestris Sphagnum palustre nudum Psilotum 51 Equisetum arvense 57 Angiopteris evecta Ephedra tweediana Zamia 64 inermis 13 99 Pin us edulis 20 100 Psedotsuga menzi Isoetaceae melanopoda Isoetes 63 11 84 apoda Selaginella 50 Selaginellaceae 100 Sefaginella selaginoides Phyllogfossum drummondii 94 Huperzia sefago 22 Huperzia wilsonii (Epiphyte) 100 Huperzia hippuridea 53 100 Huperzia campiana (Epiphyte) Lycopodiaceae Huperzia cumin 64 gii Huperzia HnifoHa (Epiphyte) 18 100 Lycopodiella inundata 41 100 Lycopodiella alopecuroides "Strobilate clade" 88 Lycopodium clavatum 1 Lycopodium obscurum 2 70 Lycopodium digitatum A Fig. 1. single most parsimonious tree indicating relationships of Lycopodiaceae. The tree is reproduced from the analyses of Wikstrom & Kenrick (1997) of rbcL sequence The data. tree is printed as a phylogram with branch lengths proportional to the amount of change along each branch and branch support is indicated by decay indices (above) and bootstrap values (below) for each branch. arating a "strobilate clade", including Lycopodium and L. Lycopodiella Holub, Hup Phylloglossum Kunze The clade. single most parsi- Wikstrom shows branch support measured by decay as indices (Bremer, 1988; Donoghue and et al, 1992) bootstrap values (Felsenstein, 1985). The support mono- for phyly Lycopodiaceae of not entirely unexpected. Perhaps more is surprising Huperzi m WIKSTROM: EXTANT HOMOSPOROUS LYGOPODS 153 and morphology made highly divergent has its relationships to other spe- its cies in Lycopodiaceae understand. Previous taxonomic treatments difficult to have with either implicitly or explicitly classified incertae sedis respect to it other groups in the family (Holub, 1985; 011gaard, 1987; Wagner and Beitel, A between Phylloglossum and Huperzia however 1992). close relationship is number consistent with a of morphological features such as spore morphology (Breckon and Falk, 1974; Tryon and Lugardon, 1991), sporangial epidermis morphology phytochemistry (Markham and (011gaard, 1975), et ah, 1983) chro- mosome number [Blackwood, Other and 1953). characters, in particular the morphology of the perenniating tuber in Phylloglossum, led workers such as Bower (1935), Bruce (1976a) and Hackney (1950) to consider a close relationship between Phylloglossum and Lycopodiella. The rbcL sequence how- data, ever, lend unequivocal support for the Huperzia-Phylloglossum relationship. The Phylloglossum from Huperzia Lycopodium- evolution of either a or a Lycopodiella like ancestor implies a remarkable architectural transformation. new This include the development of a organ (perenniating tuber), a substan- stem and development reduction of the microphyll-bearing system, the of tial a unique stem anatomy. noteworthy though that the aspects of morphology is It likely to be the least affected by this overall architectural change, such as spore and sporangial epidermis morphology and phytochemistry, are also the most consistent with the rbcL data in supporting a Huperzia relationship. — The "strobilate clade". The "strobilate clade" includes the two genera Ly- copodium and Lycopodiella (sensu 011gaard, 1987), and although comprising than My One and subgeneric names follows that of 011gaard (1987). factor contributing subgeneric groups the segregation of several divergent to the inflation of is m mor imilar Works morpho- between such groups. looking at various to find similarities such stem anatomy (Bierhorst, 1971; Jones, 1905; Ogura, logical features as morphology (Tryon and Lugardon, 1991; Wilce, 1972), distribu- spore 1972), and trophophylls sporophylls (Bruce, 1976b), tion of mucilage canals in (W, •phology and 011gaard ogy 1976c) have tended to reinforce this pattern, (1990) (Bruce, groups represent ancient evolutionary line- hypothesized the subgeneric that ages. work taxonomic recognizing subgeneric in Despite the success of recent morphological knowledge about the variation in groups and our increased have been few explicit attempts to investigate the phylo- these there plants, Wagner and Lycopodium and Lycopodiella. Beitel of genetic relationships American and Wikstrom and of N. species analyzed the relationships (1992) representatives of both genera in their analyses. Both Kenrick included (1997) AMERICAN FERN VOLUME NUMBER JOURNAL: 154 91 3 (2001) selago H. 1 52 Phyllogfossum hippuridea H. 21 100 campiana H. L magellanicum 13. Section Magellanica 100 L ^ fastigiatum 96 L Pseudodiphasium volubile Section L alpinum A 15 L wightianum Section Complanata 1 100 L digitatum ^^ 1 L annotinum Section Annotina L clavatum 82 IS Lycopodium Section 1 100 L vestitum 69 L Obscura obscurum Section L 79 jussiaei 17 Diphasium 8 Section 100 L 93 scariosum Pseudolycopodium i Section L casuarinoides Section Lycopodiastrum 58 100 La. alopecuroides 32 Section Lycopodiella 100 La. inundata cernua La. 41 42 glaucescens 100 100 La. Section Campyiostacinys 5 99 penduHna La. 77 4 La. lateralis Section Lateristachys 64 La. caroliniana Section Caroliniana Fig. Strict consensus of two most parsimonious 2. trees indicating relationships within the "stro- The bilate clade". tree is reproduced from the analyses of Wikstrom and Kenrick (2000b) of combined rbcL gene and trnL intron sequence data. Branch support indicated by decay indices is and (above) bootstrap values (below) for each branch. Subgeneric groups (sections sensti 011gaard, shown 1987] are the to right. Studies however, only included number a limited and of species the majority of the subgeneric groups recognized were not included. To address the relationships of Lycopodium and and Lycopodiella, specifically the relationships among we subgeneric groups, recently undertook an analysis using combined taxa (Wikstrom and The Kenrick, 2000b]. consensus from strict tree this analy. reproduced in figure 2 showing branch support (decay and indices bootstrap and values), subgeneric groups sensu (sections 011gaard, 1987) to the As right. seen monophyly in figure the analyses support 2, of both Lycopodium and m low number relatively a groups and some for of of the questions concerning WIKSTROM: EXTANT HOMOSPOROUS LYCOPODS 155 among relationships subgeneric groups remain unresolved. There are however some interesting patterns that not only provide an opportunity calibrate the to tree against the fossil record (see belov^] but that are supported by morphological data. Lycopodium casuarinofdes Spring, [section Lycopodiastrum) grouped as is sister to all remaining Lycopodium. This basal split in Lycopodium sup- is ported by at least two morphological features: the lack of spore muri (Tryon and Lugardon, 1991; Wilce, 1972] and the presence of thick, lignified and sinuate sporangium cell walls in L. casuarinoides (011gaard, 1975). Both features are reminiscent of Huperzia and Pbylloglossum and are best explained as retained plesiomorphic features. Another well supported grouping the is — Pseudodiphasium [Lycopodium Magellanica Even volubile G. Forster) clade. though there are few similarities in habit between L. volubile and section Ma- gellanica, features of spore morphology (Wilce, 1972] and the absence of basal mucilage canals in their sporophylls (Bruce, 1976b; 011gaard, 1987) support this grouping. In Lycopodium, two more subgeneric group relationships are supported. Section Obscura groups with section Diphasium and section An- notina groups with section Lycopodium. more The Relationships within Lycopodiella are difficult to resolve. analyses of Wikstrom and Kenrick (2000b) included datasets from both the plastid rbcL gene and the plastid trnL intron regions. Results from the separate analyses and however with respect subgeneric group relationships the sup- differed to combined was moderate. The port obtained in the analyses results are also somewhat morphological Section Caroliniana considering features. surprising, example has including isovalvate sporangia (011gaard, several features, for 1987) and absence of leaf veinal mucilage canals (Bruce, 1976b], that are best would explained retained plesiomorphic conditions. This suggest a basal as from remaining Lycopodiella separating section Caroliniana species, split in supported by molecular data (Wikstrom and but such relationship not a is Kenrick, 2000b]. clade.—The Huperzia-Phylloglossum clade by Phylloglossum far Huperzia- is and Huperzia alone includes more species diverse than the "strobilate clade", The occur epiphytes an 300-400 majority of these either as estimated species. montane South America, Africa and low mid rain forests of in to altitude open upper montane ground dwellers in habitats of the South-East Asia, or as Andean and high altitude alpine vegetation of the neotropical rain forests mountains. and had been no phylogenetic treatments of the group, Until recently, there morpho- document identifying discrete most taxonomic treatments difficulties 011gaard relationships. (1987) for ex- infrageneric indicating logical features be premature and considered ample considered formal taxonomic decisions to K and Schrank group to be the only reasonably Martins his selogo C. (L.) He however recognize 22 informal species did distinct infrageneric entity. are vicanance AMERICAN FERN VOLUME 156 lOURNAL: possibly linked to the rifting of Pangea, or more recent transoceanic dispersal. He recognized and also several different epiphytic groups, but has terrestrial it how been unclear these are related each other and whether epiphytism to evolved once or iteratively. we To address these questions, recently developed a phylogenetic hypothesis oi Huperzia based on plastid trnL-trnF intron and spacer sequences (Wiks- trom The et al, 1999]. analyses included 46 Huperzia species, representing 18 of 011gaard's 22 informal species groups and most of the geographical regions where Huperzia The species are found. resulting consensus from strict tree these analyses reproduced showing branch is in figure 3 support {decay in- and dices bootstrap values), species groups (sensu 011gaard, growth hab- 1987), it (terrestrial/epiphytic) as well as the occurrence of neotropical and paleo- The many tropical clades. results indicate that of the species groups recog- H. H. and taxifolia (Sw.) Trevisan H. H ^fl the existence of other Epiphytism interesting patterns. appears have to originated once within Huperzia with two independent at least reversals to a terrestrial habit [H. liippuridea (Christ) Holub and the clade including H. sau- rurus, //. brevifolia (Grev. & Hook.) Holub and H. reflexa groups), both in the With two )pics. exceptions [H, ophiodossoides (Lam.) Rothm. and H. funifi Within and brevifolia H. refl groups constitute a large, secondarily These terrestrial clade. patterns are all well supported by and the trnL-trnF data the existence of neotropical and an (Wikstrom Kenrick, 2000a). Comparing these with results respect morphological to data difficuh. is m and groups, characteristics that have been used endpoints are morphological in continua, resulting in arbitrary decisions on character state delimitations. Not- withstanding some these problems, of the patterns obtained in the molecular analyses are unexpected. For example, and neotropical members paleotropical K of the H. phlegmaria and of the groups show verticillata conspicuous morphological similarities but, based on the molecular analyses, morpho- these logical similarities are the result of convergent evolution. Furthermore, as homosporous plants they have usually been considered to disperse but easily, biogeog Huperzia ametophyt ametophytes are (Wikstrom knowledge on aspects of the establishment and dvnamics ' WIKSTROM: EXTANT HOMOSPOROUS LYCOPODS 157 H. proVitera H. hippuris group , H. fordii H. hamiltonii group I B8 H. sieboJdii H. cancellata group 1 78 H. tauri H. phlegmaria group 99 H. hippuris H. hippuris group H carinafa group H. fockyen phlegmaria group H. bafansae H. H squarrosa group _4 H. squarrosa I 97 H. nummulariifolia Paieotropical H. phlegmarfoides dade 67 1 phlegmaria group 62 H. cf phlegmaria H. 75 8 H. horizontalis 99 4 H. phlegmaria H. funiformls group H. taxifolia 1 H. vertfciltata H. group verticillata 1 H. dacrydioides 62 H. gnidioides group hofsW 96 H. H. billardieri H. myrtffoHa group 1 mandiocana group H. wilsonii H. I hippmdea group 94 H. H. brongniartii I H 95 H. dichotoma dichotoma group 54 I H. tenuis H. vertidllata group I H. linifolia H. linifotia group \ 99 H. sarmentosa H. verb'dHata group I 97 H. capeffae H. cowpacta H. crasaa 1 H. hystiix saurunjs and H. 64 1 H 99 H. polydactyfa groups 64 brevifolia H. rufescens 76 Neotropical H. cwrJingU clade H. attenuata 96 H. tetmgona 87 _ H. eversa 62 _ H. reflexa group H. reflexa 64 98 _ H. unguiculata t 99 76 H. reflexa var, reflexa la H. rosenstockiana H. taxifolia group 100 campiana phlegmaria group H. H. group H. lindenii H. taxifolia n H. ericifolia 100 H. heteroclita H. phlegmaria group 100 94 H. subulata 7 H. ophiogfossoides L H. tacidula selago group H. i 77 H. sefago _ Phyflogfossum L magetlanicum L fasb'giatum L vofvbSe L deiAerodensum _ obscufum L. L afpintm L wightianum _ L annotifKtm L _ vestitum L casuaiinddes La. akjpecuroides inundata La. giaucescens La. p&KhiHna La. cemua La. La. fateraiis species Terrestrial parsimonious indicating relationships within the most trees consensus of 16,924 Fig. Strict 3. Wikstrom The reproduced from the analyses of et al. (1999J tree Hupe'rziaPhylloglossnw clade. is Branch support indicated by decay indices (above) and spacer sequence data. is of trnh-truF intron each branch. Informal species groups (sensu 011gaard. 1987) are and values [below) for bootstrap and epiphytic Huperzio into a neotropical a paieotropical The of partition indicated to the right. and H. ophioglossoides (marked with *) are exceptions to H. funiformis clade also indicated. is and remaining Taxa within grayed boxes are terrestrial species biogeographic pattern. this clear are epiphytic. AMERICAN FERN VOLUME NUMBER 158 JOURNAL: 91 3 (2001) of lycopod populations however limited, and before such knowledge ac- is is quired, one can only speculate. Patterns of Diversification Calibration of the phylogenetic tree critical to reconstructing the historical is patterns and two approaches have been adopted. Where possible, information from the fossil record has been used. This includes the use of fragmentary plant remains, isolated organs and spores. The calibrations discussed here are however and based on As tentative primarily a literature survey. seen in figure some further quired. Ultimately, the fossils should of course not only be used for calibrations, but should also be included in the analyses. Attempts to correlate biogeographic data on living species with other geological data such major as tectonic events have also been done. The macrofossil record of Lycopodium and Lycopodiella meager (Skog and is The middle Hill, 1992). Jurassic (Bajocian-Bathonian; 174 Mya) Lycopodites falcatus Lindley & Hutton is one of the better known species and mor- its phology was documented by He in detail Harris suggested an (1961), affinity with section Complanata, based on flattened branch system with its large lat- eral leaves and smaller dorsal and would ventral leaves. correct, this suggest If minimum Mya a age of 174 the between Complanata and for split section the clade including sections Lycopodium and Annotina Although the (Fig. 4). ar- affinity stem position along the stem lineage of section Complanata indicated as in figure highly uncertain. 4, is The mature stem anatomy Lycopodium of highly with distinctive the xy- is lem arranged bands in parallel This (plectostele). feature characteristic of is all living species. Srivastava (1946) documented what he considered be to a lycopodiaceous plectostele [Lycoxylon indicum from Srivastava) the Jurassic (Lias) of India Although completely (Fig. decorticated, the arrangement 5). of xylem the strands is highly reminiscent of living Lycopodium. This ev- fossil minimum idence suesest a ase for the snlit betwRRn J\^rn-nndhim ^nrl l^rrr^. Mya same Lycopodites falcatus two names if the represent different organ names the same would for species certainly contribute to our understanding of lycopod evolution. Using evidence fossil for calibrating the tree associated with number is a of problems and potential sources of error (Doyle and Donoghue, 1993). First of synapomorphies shared by must all, living taxa be observable and recognizable in fossil taxa. Secondly, fossils only provide estimates minimum of < maximum meager WIKSTROM: EXTANT HOMOSPOROUS LYCOPODS 159 H. selago Phyltoglossum H. hippuridea campiana H. L magellanicum Section Magellanica L fastigiatum L Pseudodiphasium volubile Section L alpinum L Complanata wightianum Section I L digitatum t My Lycopodites falcatus t (174 a) L annotinum Section Annolina L clavatum Reticulate spores Lycopodium Section L rl vestitum L Obscura Section obscurum L jussiaei Section Diphasium L scahosum - Mya) Reticulatispohtes t (208 L deuterodensum 1 Pseudolycopodium Section Plectosteie L t casuarinoides Section Lycopodiastrum Mya) Lycoxylon indicum t (208 alopecuroides La. Section Lycopodiella La. inundata cemua La. Campylostachys glaucescens Section La. pendulina La. Section Lateristachys La. lateralis La. caroliniana Section Caroliniana from Wikstrom and Kenrick (2000b) two most parsimonious trees consensus of Fig. 4. Strict and The were not part of the analyses used calibrating the tree. fossils including the fossils for basalmost position implied by their characteristics. Ultimately have simply been inserted the at knowledge current makes such an inclusion unfeasible at this stage. underestimate the true age of origin. can greatly its observation of feature this minimum based on information there- age estimate fossil Our confidence in a of finding the feature in earlier depends on the probability that failure fore meaning had absence" not yet origi- "true represents a it records geological information calibrating the ages of most useful for the nated. In this respect, comes the spore record. clade fi^om fossil groups within the strobilate rni J „r r „-i ^T^^T^oc TAHt>i rjntntivR T.vcoDodium affinities is auite ex-