The generitype of Delesseria J. V. Lamour. is D. sanguinea (Huds.) J. V. Lamour., a species occurring in the colder waters of the northeastern Atlantic Ocean, ranging from Arctic Norway and Iceland to Spain (South and Tittley 1986), as well as a reduced form extending into the Baltic Sea (Levring 1940, Nellen 1966, Luning 1990). Its handsome image depicted by Oeder in Flora danica (1766, Pl. 349, as Fucus sanguineus) and reproduced in Wynne (2006) was based on a specimen collected in Iceland by Johann Gerhard Konig, a student of Linnaeus, in 1764-1765 and initially illustrated in the field by Helt, a young illustrator who accompanied Konig to Iceland. Helt’s preliminary sketch was later refined in Copenhagen by Rosler (information from Peter Wagner, communicated by Ruth Nielsen, Copenhagen). Being a genus described so early (Lamouroux 1813), Delesseria has had numerous species assigned to it over the years. For example, J. Agardh (1872) assigned 48 species to his broadly defined Delesseria. Fifteen species are currently recognized in the genus (Guiry and Guiry 2012). This list of taxa now placed in Delesseria

includes two species occurring in the North Pacific, D. decipiens J. Agardh (1872) and D. serrulata Harvey (1857). Although these species share with D. sanguinea the same “Delesseria-Type” apical organization (Kylin 1923, 1924, Mikami 1972), that is, with a single transversely dividing apical cell terminating each indeterminate axis, the absence of intercalary divisions in cells of the primary row but with the presence of intercalary divisions in cells of the second-order rows, some differences have long been recognized that distinguish this pair of North Pacific species of Delesseria from the European generitype. The primary difference appears to be that of a life-history strategy, D. sanguinea being a perennial species that produces small reproductive bladelets from its perennating axes, whereas D. decipiens and D. serrulata are annual species, producing their reproductive structures on the surfaces of non-specialized blades.

We have extended our studies to include species assigned to the related genera Membranoptera and Pantoneura occurring on both sides of the North Atlantic and in the northeastern North Pacific. Membranoptera alata (Hudson) Stackhouse, the generitype, has been recognized on both sides of the North Atlantic (Rosenvinge 1923-1924, Taylor 1962, Bird and McLachlan 1992, Maggs and Hommersand 1993, Sears 1998, Loiseaux-de Goer and Noailles 2008). Although some treatments (e.g., Mathieson et al. 1969, South 1984, Sears 1998) recognized only M. alata as present in the northwestern North Atlantic, Taylor (1962) also recognized M. denticulata (Montagne) Kylin as present in this region. The basionym of M. denticulata is Delesseria alata var. denticulata Montagne, with a type locality of Labrador, Canada (Montagne 1849). The binomial M. denticulata (Montagne) Kylin (Kylin 1924), however, is a later homonym, predated by M. denticulata Kuntze (1891), a synonym of Heterodoxia denticulata (Kuntze) J. Agardh, an Australian species. Thus, the name M. denticulata (Montagne) Kylin is not available to apply to the taxon occurring in the northwestern North Atlantic.

The species Membranoptera spinulosa (Ruprecht) Kuntze, with syntype localities in the Sea of Okhotsk and St. Paul Island, Bering Sea (Ruprecht 1850) has recently been reported from the northeastern Atlantic (Mathieson et al. 2010). Earlier in an abstract, Hommersand and Lin (2009) stated that samples of Membranoptera from Atlantic North America were separated by only six base pairs in rbcL sequence from North Pacific samples of M. spinulosa. They also indicated that M. alata appeared to be restricted to Europe, a finding that our work will confirm.

On the Pacific coast of North America, a larger number of species of Membranoptera have been recognized, as many as six to eight (Gardner 1926, Wynne 1970, Abbott and Hollenberg 1976, Gabrielson et al. 2004, 2006, Lindeberg and Lindstrom 2010). In the North Atlantic two species of Pantoneura have usually been recognized, P. fabriciana (Lyngbye) M. J. Wynne [formerly P. baerii (Ruprecht) Kylin] and P. angustissima (Turner) Kylin (Wynne 1997). The former species is also reported to occur in Alaska (Lindstrom 1977), and P. juergensii (J. Agardh) Kylin is known from the Aleutian Islands and the Bering Sea (Wynne 1970, Klochkova et al. 2009).

The difficulty in separating taxa of Membranoptera and Pantoneura from the northwestern North Atlantic using morphology was discussed in detail by Lamb and Zimmermann (1964). According to them, the most common form of M. alata occurring in New England was a “very narrow form of the species.” In reference to Pacific North American species of Membranoptera, Hawkes et al. (1978) observed that it was “nearly impossible to distinguish Membranoptera multiramosa from M. platyphylla” because of inconsistences in the descriptions of the former species. Similarly, Gabrielson et al. (2006) referred to the need to re-examine the relationship between M. tenuis Kylin and M. weeksiae Setchell & N. L. Gardner, and they also questioned the taxonomic relationships among the three species M. multiramosa N. L. Gardner, M. platyphylla (Setchell & N. L. Gardner) Kylin, and M. spinulosa. Thus, a background exists of workers expressing frustration with the separation of species of Membranoptera and Pantoneura.

In the present paper analyses of rbcL, ribosomal large subunit (LSU), and COI-5P data have demonstrated that these life-strategy differences between D. sanguinea (Fig. 1), the generitype and the North Pacific pair, D. decipiens (Fig. 2A & B) and D. serrulata (Fig. 2C & D), are taxonomically significant with the last two joining Cumathamnion sympodophyllum M. J. Wynne & K. Daniels (Fig. 3), a species with a distribution restricted to northern California (Wynne 2009) and at present placed in a monotypic genus (Wynne and Daniels 1966). Adding internal transcribed spacer (ITS) to our previous list of genetic markers, our studies on the related species Membranoptera alata (Fig. 4A-C) show that specimens assigned to this species on the North American Atlantic coast are genetically distinct from this European-based species and conspecific with Pantoneura fabriciana. A taxonomic resolution for this result is proposed. Further, specimens from the North American Pacific coast assignable to four to six morphological species (Fig. 4D-G) (Gabrielson et al. 2012) were all resolved as a single genetic species, which again prompted taxonomic change.

Samples for molecular investigation (Table 1) were processed and DNA extracted following Saunders and McDevit (2012). Sequence data were generated for the mitochondrial cytochrome c oxidase 1 gene barcode region (COI-5P) (Saunders and McDevit 2012), the nuclear internal transcribed spacer (except the primer MEMR4 5′-AATTCAGCGCGTCACCTTATC-3′ replaced the standard reverse primer G4 in collections for which diatom contaminants were a problem) (Tai et al. 2001), the LSU of the ribosomal cistron (Harper and Saunders 2001), and the plastid RUBISCO large subunit (rbcL) (except two new internal sequencing primers were used-forward TLF5 TCWCARCCWTTYATGCGTTGG, and reverse TLR1 AAYTCWGCTCTTTCRTAYAT) (Vis et al. 2007) following established protocols. In total 103 COI-5P, 30 ITS, 20 LSU, and 21 rbcL sequences were generated for this study (Table 1).

Six alignments were constructed using MacClade 4 (version 4.06) for OSX (Maddison and Maddison 2003). Firstly, a COI-5P alignment (88 individuals and 664 bp) was generated for collections of Delesseria and Membranoptera from Canada and contiguous waters (Table 1) to assign collections to genetic species groups using neighbour-joining analyses (K2P corrected distances) as implemented in PAUP* (version 4.0b10) (Swofford 2003) in Geneious Pro version 5.6.2 (Drummond et al. 2012). To confirm the COI-5P results for a subsample of Membranoptera and Pantoneura collections (Table 1), ITS data were aligned (30 individuals and 1,041 characters) and similarly subjected to neighbour-joining analyses. Subsequent phylogenetic analyses placed these Canadian genetic species groups into an evolutionary context considering individual (COI-5P-21 taxa, 664 sites included; LSU-21 taxa, 2,957 sites, 2,631 included in analyses; and rbcL-23 taxa, 1,358 sites included) and a combined (COI-5P + LSU + rbcL-20 taxa, 4,653 sites included; P. fabriciana data were redundant and were removed prior to combined analyses) alignment (Table 1).

Model parameters were estimated (Akaike’s information criterion [AIC]) for each of the four phylogenetic alignments in Modeltest version 3.06 (Posada and Crandall 1998) as implemented in PAUP* through Geneious Pro on a Mac Pro (OS X version 10.6.8). The selected model was used to complete maximum-likelihood analyses in PHYML version 3.0 (Guindon and Gascuel 2003) with BIONJ used to designate the starting tree and nearest neighbour interchanges (NNIs) branch-swapping in effect. Branch support was estimated for the single gene analyses using the Shimodaira-Hasegawa-like (SH) approximate likelihood ratio test (aLRT) and for the multigene alignment with nonparametric bootstrap resampling (500 replicates). In addition to maximum likelihood for the combined alignment, Mr. Bayes (version 3.1.2) (Huelsenbeck and Ronquist 2001) was used to complete two independent trials (each with parallel runs) of Bayesian inference under a GTR + I + G model. Parallel runs

of four Markov chains were completed with two million generations and sampling each 100 generations. The data were partitioned (by gene, and then by codon for COI-5P and rbcL), and the parameters were unlinked across partitions with the overall rate allowed to vary across partitions. The burn-in for each run was determined by plotting overall likelihood scores against generation, which established the stationary phase of each run for estimating the posterior probability distribution; the final estimate was based on pooled samples from two independent runs.

Despite a range of morphologies for our collections of D. decipiens, all of the collections resolved in a single genetic species group with COI-5P data (0-0.92% divergence) consistent with current taxonomic thought (Fig. 5). M. alata from a diversity of sites in the northeastern waters of North America (0% divergence, Table 1), however, failed to join European collections (type locality) of this species (Fig. 5). More interesting, the former had COI-5P sequences identical to collections of P. fabriciana

from Newfoundland (Table 1, Fig. 5). Equally dramatic were our results for collections from British Columbia (one from California, Table 1) that were variously field identified to all of the four-six species recognized in those waters (Table 2), but which nonetheless formed a single genetic group (0-0.46% divergence) in COI-5P analyses (Fig. 5, M. platyphylla). This would suggest that the sub-stantial morphological variation that has previously been used at the species level for taxonomic discrimination is more likely phenotypic plasticity in response to the environment (Table 2).

Internal transcribed spacer data for six diverse Pacific morphs of Membranoptera (Tables 1 & 2) were virtually identical (four sites with ambiguities and / or substitu-

tions indicating low level within individual and population variation) consistent with the COI-5P results (Fig. 5) that all of these collections should be assigned to a single species (Fig. 6). For the Atlantic populations ITS were generated for six individuals assignable to P. fabriciana and 10 individuals assignable to M. alata from North America (Fig. 6). All individuals were virtually identical (two specimens with an ambiguity [C and T] at the same site) indicating that the ITS data are consistent with the COI-5P data in assigning all of these collections to a single genetic species group. European collections of M. alata (n = 8) similarly clustered together (Fig. 6); two collections with C and T ambiguity at the same site, one with G and T ambiguity, and one with variable number of T’s (6 and 7) and were distinct from the North American collections, again consistent with the COI-5P data (Fig. 5).

Interestingly, rbcL and LSU data generated for North American M. alata and P. fabriciana were also identical, as had been detected previously in our COI-5P and ITS analyses, consistent with these two distinct morphologies belonging to a single species. In phylogenetic analyses all of the single gene alignments essentially resolved the same topology as the combined alignment, but the rbcL phylogeny is presented because it included two sequences from Genbank, which were not ultimately included in our combined analyses (Fig. 7). First, Genbank data for Membranoptera weeksiae (AF257384) were identical to those generated here for M. platyphylla consistent with our suggestion that only a single species should be recognized for the variety of morphologies currently con-

sidered multiple species in the Northeast Pacific (Fig. 7). Second, a sequence in Genbank attributed to Membranoptera tenuis (AF257383) for a collection from Alaska is very similar to our data for North American (Atlantic) M. alata / P. fabriciana-differing at only seven (likely six, one difference is at a highly conserved codon position and may represent an error in the Genbank entry) nucleotides out of the 1,354 available for comparison or 0.5%. These two taxa are thus very closely related and need further study to assess their status as distinct species.

Only maximum likelihood results for the combined analyses are presented as once again all of the trees were similar (Fig. 8). The tree for the most part was solidly resolved and showed a close affiliation between the genera Cumathamnion, Delesseria and Membranoptera (Fig. 8), which are currently assigned to three different tribes on the basis of morphological features (discussed below). Further, D. decipiens and D. serrulata joined the type species of Cumathamnion rather than D. sanguinea, the generitype of Delesseria (Fig. 8). Our molecular results thus necessitate substantive taxonomic changes at the species, genus and tribe level for species assigned to the

genera Delesseria, Membranoptera and Pantoneura in the Canadian flora. Although beyond the scope of the current manuscript, we have also uncovered phylogenetic evidence that representatives of the diverse genera Branchioglossum, Chauviniella, Claudea, Grinnellia, Hemineura, Patulophycus and Phitymophora, which have been previously assigned to other tribes, resolved closely to our Cumathamnion, Delesseria and Membranoptera lineage and are all in need of future taxonomic revision (Figs 7 & 8).

Much of the currently accepted classification of the Delesseriaceae is based on the investigations on the comparative vegetative organization and reproductive patterns made by Kylin (1923, 1924). Limiting the scope of our discussion to the genera included in our study, we can refer to Delesseria being assigned by Kylin (1924) to the “Delesseria-Gruppe” and Membranoptera and Pantoneura being assigned to his “Membranoptera-Gruppe.” When describing their new genus Cumathamnion, Wynne and Daniels (1966) recognized a new category, the “Cumathamnion Group” for that genus. Subsequently, Wynne (2001) replaced these informal “Gruppe” names with corresponding tribal names.

The critical characteristic that was used to separate Membranoptera and Pantoneura in their tribe, or Gruppe, from Delesseria in its tribe was that in the former genera intercalary cell divisions are absent in second-order cell rows, but they do occur in species of Delesseria. Such intercalary cell divisions also occur in the second-order cells rows of Cumathamnion, but it was separated into its own tribe because of its sympodially organized axes.

Although several entries of gene-sequence data are now listed in Genbank for D. sanguinea, D. decipiens, D. serrulata, and D. lancifolia, actual published information has been limited. On the basis of rbcL sequence analysis, Lin et al. (2001) showed a closer relationship between D. decipiens and two species of Membranoptera than with D. sanguinea, which was the deepest lineage in that clade. Small subunit rDNA data for a Korean collection of D. serrulata has been published by Choi et al. (2002). Sequence data on C. sympodophyllum has not been previously published.

Previous molecular phylogenetic research on the family Delesseriaceae has indicated in a preliminary way that Delesseria was a polyphyletic genus. Lin et al. (in press) demonstrated that several southern hemisphere species that had been assigned to Delesseria were more closely related to Apoglossum ruscifolium (Turner) J. Agardh and fell out in their newly described tribe Apoglosseae. According to Lin et al., the name Paraglossum J. Agardh (1898), originally based on the two species P. lancifolium (J. Agardh) J. Agardh and P. epiglossum (J. Agardh) J. Agardh and later lectotypified with the former species (Lin et al. 2001), could be reinstated to accommodate those species related to A. ruscifolium. According to Lin et al. (in press), the tribe Apoglosseae accommodates Apoglossum and the reinstated Paraglossum. On the other hand, D. decipiens fell out in a clade close to two species of Membranoptera with D. sanguinea as sister to that group.

When the new genus Cumathamnion was established by Wynne and Daniels (1966), they stressed its sympodial development as a primary distinction from related genera of Delesseriaceae. The monotypic genus was later placed in its own tribe, the Cumathamnieae (Wynne 2001). The pattern of growth with not all tertiary initials reaching the thallus margin, the presence of intercalary divisions in cells rows of the second order, and the production of branches along the midrib are features shared by Cumathamnion and Delesseria. So the question arises: Is the sympodial development present in C. sympodophyllum in contrast to the monopodial development present in D. decipiens and D. serrulata too significant a difference to override their obvious affinity as expressed in the genesequence data?

Two types of sympodial growth in the order Ceramiales were recognized by Norris et al. (1984). In “cellulosympodial” growth, the apical cell initiates a determinate lateral branch, and the lateral cell or branch formed by the subapical cell carries on development of the thallus axis, and that process is continually repeated. Cellulosympodial development occurs in the family Dasyaceae (Parsons 1975). Ramisympodial branching, on the other hand, occurs when development in a given axis ceases, and further growth of the plant is from new branches initiated in an intercalary position. Ramisympodial organization is known in some genera of Ceramiales. Some instances are known where genera include species with monopodial growth and species with ramisympodial growth. Examples include Hypoglossum in the Delesseriaceae, in which most of the species have a monopodial organization, but thalli of H. revolutum (Harv.) J. Agardh are ramisympodially organized. Likewise, in Crouania in the Ceramiaceae some species have monopodial organization, whereas other species have a ramisympodial organization (Norris 1986, Schneider 2004). Thus, it is not unprecedented where we now interpret Cumathamnion to have species with both ramisympodial branching (in the type species, C. sympodophylum) and monopodial branching, in these two species formerly assigned to Delesseria, D. decipiens and D. serrulata.

Delesseria decipiens was first recognized by Harvey (1862) under the name “Delesseria hypoglossum var. arborescens” collected from the Strait of Juan de Fuca by Dr. Lyall, who was a surgeon with the British Boundary Commission. When J. Agardh (1872) later described Delesseria decipiens, he cited Harvey’s nom. ined. in taxonomic synonymy. J. Agardh indicated Vancouver Island (British Columbia, Canada) as the only location for this species. Subsequently, J. Agardh (1898) transferred this species to Apoglossum, A. decipiens (J. Agardh) J. Agardh, but its placement in Delesseria has been followed by most workers (Kylin 1924, Smith 1944, Abbott and Hollenberg 1976, Scagel et al. 1989, Gabrielson et al. 2000). It is now known to have a distribution in the eastern North Pacific from the Kodiak Archipelago, Alaska, to Baja California, Mexico (Abbott and Hollenberg 1976, Hawkes et al. 1978, Scagel et al. 1989, Lindeberg and Lindstrom 2010, Riosmena- Rodriguez et al. 2011). The known distribution for D. serrulata is from northern Japan (Okamura 1908), Korea (Lee and Kang 1986), and eastern Russia (Perestenko 1996, Kozhenkova 2009). The Delesseria serrulata depicted from Australia by Harvey (1858, Pl. 59) has been shown by Kurogi (1979) and Womersley (2003) to be an incorrect application of the name, and the Australian alga is now known as Hypoglossum harveyanum (J. Agardh) Womersley & Shepley (1982).

Foremost among the morphological differences between D. sanguinea and these two North Pacific species of the genus is the fact that D. sanguinea is a perennial plant that produces a conspicuous foliose vegetative stage in the summer growing season, but it does not bear reproductive organs on the surfaces of those large blades (Fig. 1A). In the winter, when the vegetative blade is largely eroded, the persistent midrib produces very small special fertile proliferations on which the reproductive structures (carpogonial branches and tetrasporangia) are formed. The male proliferations are also formed on the midribs, but this occurs while the blade lamina is still present (Maggs and Hommersand 1993). There is a short mid-winter fertile season, with peak production of spores (carpospores and tetraspores) from December to March (Kain 1982), although male organs appeared 3 months earlier. This production of specialized reproductive proliferations has long been recognized in D. sanguinea (Smith 1804-1805, Pl. 1041; Turner 1807-1808, Pl. 36; Cuvier 1816-1829; Harvey 1848, Pl. 151; Phillips 1898, Pl. 15). The persistent midribs also serve to sprout out the next generation of leafy shoots in the next growing season (Fig. 1B). In contrast to the pattern of reproduction in D. sanguinea, D. decipiens (Fig. 2A & B), and D. serrulata (Fig. 2C & D) produce their reproductive organs directly on the vegetative blades at the end of the growing season, sometimes on the final order of vegetative branches [Kylin 1924, for D. decipiens; Okamura 1908 (as Apoglossum violaceum), Mikami 1972 (as Delesseria violacea) for D. serrulata]. In their annual life-history pattern and the production of reproductive organs on non-special blades, these two species of North Pacific Delesseria parallel the life history pattern of C. sympodophyllum (Wynne and Daniels 1966).

The results of a comparison of the sequence data (Figs 7 & 8) lead to the conclusion of the closer relationship of Delesseria decipiens and D. serrulata with C. sympodophyllum than with D. sanguinea, the European generitype. The following transfers are proposed:

Basionym. Delesseria decipiens J. Agardh 1872, Lunds Univ. Arsskr. 8(6), p. 58.

Homotypic synonym. Apoglossum decipiens (J. Agardh) J. Agardh 1898, p. 194.

Also. Delesseria hypoglossum var. arborescens Harvey 1862, p.170, nom. ined.

Basionym. Delesseria serrulata Harvey 1857, Narrative Perry Exped. 2, p. 331.

Homotypic synonyms. Membranoptera serrulata (Harvey) Kuntze 1891, p. 904; Hydrolapatha serrulata (Harvey) Kuntze 1898, p. 410.

Heterotypic synonym. Delesseria violacea J. Agardh 1872, p. 52, nom. illeg. [type locality: Japan]; this name is illegitimate because the valid and legitimate name D. serrulata Harvey (1857) was cited in synonymy by Agardh (1872); Apoglossum violaceum J. Agardh (1898) nom. illeg.; cf. Kurogi (1979).

Key to species of Cumathamnion:

1. Thallus monopodially organized, that is, with percurrent indeterminate axes……...........................................2

1. Thallus ramosympodially organized, that is, with indeterminate axes continually being replaced with lateral branches assuming the focus of growth and they then being replaced………......................C. sympodophyllum

2. Blade margins entire, smooth……................C. decipiens

2. Blade margins serrate, toothed.....………….C. serrulata

The next question is: What name should be applied to the taxon of so-called Membranoptera and Pantoneura fabriciana occurring in the northwestern North Atlantic? The molecular results (Figs 7 & 8) demonstrated that Membranoptera and so-called Pantoneura of the northwestern North Atlantic were assignable to a single genetic group distinct from the European-based populations of M. alata. Wynne (1997) reviewed the complicated history of the name Gigartina fabriciana Lyngbye (1819, Pl. 11D), based on a collection from Greenland. He examined type material in C and confirmed that it was a member of the Delesseriaceae and that it was an older name for Pantoneura baerii (Ruprecht) Kylin. Thus, the name Pantoneura fabriciana (Lyngbye) M. J. Wynne was proposed. The three currently recognized species of cold-water Northern Hemisphere Pantoneura have been placed in that genus because of their morphological similarity to type species of Pantoneura, P. plocamioides Kylin (in Kylin and Skottsberg 1919), with a type locality of South Georgia. According to Hommersand et al. (2009), however, unpublished rbcL sequence data by S. -M. Lin show P. plocamioides to be unrelated to the Arctic species P. fabricana, but the latter taxon is related to M. alata, in agreement with our results.

Reinsch (1875) described Hypoglossum grayanum with three syntype localities: West Gloucester, Massachusetts, USA; Labrador, Canada; and Anticosti Island, Quebec, Canada. Farlow (1881) regarded Hypoglossum grayanum to be the same as M. alata of the New England coast. De Toni (1900) treated H. grayanum as a “Species mihi planae ignotae aut ulterius inquirendae,” saying that it perhaps had affinities with Caloglossa leprieurii. But the fact that Reinsch stated that his new species was epiphytic on Ptilota plumosa is ample evidence that it is not in the genus Caloglossa, which typically occurs in estuarine habitats. Reinsch’s (1875, Pl. 47) clearly depicts Membranoptera epiphytic on its common host Ptilota. But this name is predated by Lyngbye’s G. fabriciana. Therefore, we conclude that G. fabriciana is the name with priority to use for the Membranoptera present in the northwestern North Atlantic, and we propose the binomial:

Basionym. Gigartina fabriciana Lyngbye, 1819, Tent. Hydrophytol., p. 48, PI. 11D.

Lectotype in C. Type locality. Frederikshaab (61°59′ N, 49°42′ W), Greenland, collected by Bishop O. Fabricius in the period 1768-1773, fide Ruprecht (1851). See Wynne (1997, Figs 1 & 3) for depictions of the apical portions of the Type.

Homotypic synonym. Pantoneura fabriciana (Lyngbye) M. J. Wynne 1997, p. 325.

Heterotypic synonyms. Delesseria alata var. denticulata Montagne 1849, p. 62; Membranoptera denticulata (Montagne) Kylin 1924, p. 16, nom. illeg.; Hypoglossum grayanum Reinsch 1875, p. 55, Pl. 42.

In regard to the populations of Membranoptera from the northeastern North Pacific, the results of the molecular analyses (Figs 5 & 6) establish that only a single species is represented in this region, despite the fact that as many as six species have been recognized (Gabrielson et al. 2006). The names of two species have equal priority, dating from Kylin (1924), namely, M. platyphylla (Setchell & N. L. Gardner) Kylin and M. tenuis Kylin. We opt to apply the name M. platyphylla for the representatives of Membranoptera occurring in the northeastern North Pacific. The type locality of M. platyphyllya is Pleasant Beach, Kitsap County, Washington, USA (Setchell and Gardner 1903).

Membranoptera spinulosa (Ruprecht) Kuntze was reported from Alaska by Wynne (1970) and by Lindeberg and Lindstrom (2010) and also included in the flora of the northeastern North Pacific by Gabrielson et al. (2006). This species was based on Delesseria alata var. spinulosa Ruprecht (1850), with syntype localities of the Okhotsk Sea and St. Paul Island, Bering Sea. Because of the remote nature of these syntype localities to our collections, we have decided to apply the name M. platyphylla rather than M. spinulosa.

The earlier assignment of Delesseria, Membranoptera, and Cumathamnion to three different tribes within subfamily Delesserioideae (Wynne 2001) is clearly incorrect in view of their genetic relatedness seen in Figs 7 & 8. The separation of Apoglossum along with the reinstated Paraglossum J. Agardh from the Delesserieae to the new tribe Apoglosseae has been proposed by Lin et al. (in press). The newly circumscribed tribe Delesserieae includes Delesseria, Cumathamnion, and Membranoptera, and likely other genera such as Phitymophora, Grinnellia, and Hemineura, that have been previously assigned to other tribes. These latter genera, included in Figs 7 & 8, need to be re-evaluated for possible inclusion in the tribe Delesserieae.