Taxonomic synopsis of invasive and native Spartina (Poaceae, Chloridoideae) in the Pacific Northwest (British Columbia, Washington and Oregon), including the first report of Spartina ×townsendii for British Columbia, Canada

Abstract Five species of the grass genus Spartina are invading salt marshes along the Pacific coast of North America, of which three have been documented in British Columbia, Canada, in only the last decade. A taxonomic synopsis of the two native (Spartina gracilis, Spartina pectinata) and five introduced Spartina taxa (Spartina anglica, Spartina alterniflora, Spartina densiflora, Spartina patens, Spartina ×townsendii) in the Pacific Northwest is presented to facilitate their identification, including nomenclature, a new taxonomic key, new descriptions for a subset of taxa, and representative specimens. Spartina ×townsendii is newly reported for the flora of British Columbia. The non-coastal species Spartina pectinata is reported from an urban site in British Columbia, the first confirmed report of the taxon for the province. Lectotypes are newly designated for Spartina anglica C.E. Hubb., Spartina maritima subvar. fallax St.-Yves, and Spartina cynosuroides f. major St.-Yves.


Introduction
Spartina Schreb. (cordgrass) is a small grass genus of some fi fteen species native to North America, South America, and the Atlantic coasts of Europe and Africa occurring in such coastal habitats as intertidal mud fl ats, estuaries, salt marshes, and inland in marshes, sloughs and dry prairie. Spartina includes several globally invasive species (e.g., S. alternifl ora Loisel., S. anglica C.E. Hubb., S. densifl ora Brongn.) that are rapidly altering salt marsh and estuary ecosystems (e.g., Grosholz et al. 2009, Strong andAyres 2009) and is an important model system for studying allopolyploid speciation and the relationship between hybridization, polyploidy, and invasion (Ainouche et al. 2004a, Baumel et al. 2002a, Salmon et al. 2005, Fortune et al. 2007, Chelaifa et al. 2010. Consequently, a large and broad body of biological, genetic, and ecological knowledge on cordgrasses has been generated. Nevertheless, identifi cation of Spartina species can be challenging, as several are morphologically similar and rapidly expanding their ranges into new areas in which up-to-date taxonomic resources for the genus are not available. Th is is the situation in the Pacifi c Northwest of North America (Oregon, Washington, British Columbia), where fi ve Spartina species are invading coastal areas (S. alternifl ora, S. anglica, S. densifl ora, S. patens (Aiton) Muhl., S. ×townsendii H. Groves & J. Groves).
Th is taxonomic study was prompted by diffi culties encountered in determining recent herbarium collections of invasive Spartina from British Columbia. Existing regional taxonomic resources do not include all taxa known in the province (Hitchcock et al. 1969, Douglas et al. 2001, Kozloff 2005 since several species are new or recent invaders, and continent-wide treatments (Barkworth 2003) were found initially to be unsatisfactory to use in this taxonomically diffi cult grass genus. Since Spartina invasions are causing serious ecological damage on the Pacifi c coast, there has been considerable eff ort to develop information sources for fi eld-based identifi cation, probably in response to a lack of easy-to-use and easy-to-access locally relevant information in the scientifi c literature, in combination with an urgent need for information to facilitate Spartina identifi cation by as broad a user group as possible to minimize the impacts of the invasions. Th ese mostly emphasize vegetative characteristics easy to observe in the fi eld, such as plant height, blade width and height, leaf color, and plant habit (e.g., Anonymous no date, 2009, Pfauth and Sytsma 1998, O'Connell 2005, Eissinger 2008) but, unfortunately, none of these are among the best taxonomic characters to identify Spartina species with confi dence, particularly those that are closely related. Several of these resources do not clearly list authors, publication dates or places other than websites, and they do not identify the original sources (e.g., fl oras, the primary literature, herbarium specimens) from which the information was derived.
Th e purpose of this paper is to provide up-to-date taxonomic information for specimen-based identifi cation of Spartina species in the Pacifi c Northwest. Although fi eldbased identifi cation of invasive Spartina taxa is possible, reliable determinations should be made or confi rmed from specimens, as most of the diagnostic characters require magnifi cation and careful, accurate measurements. Specimens should be deposited in herbaria, where they become part of the scientifi c record, are available for study by other scientists, and document the distributions of species in time and space. Voucher specimens for invasive plants such as Spartina are particularly important, as they provide the raw materials from which reliable and repeatable identifi cations can be made, and they contribute to long-term understanding of the distribution and spread of these new invaders. Unfortunately, herbaria often have relatively few specimens of weedy species, a situation recently documented for noxious weeds in Washington (Zika 2006). Accordingly, there are relatively few herbarium specimens for invasive Spartina taxa in the Pacific Northwest, despite considerable eff orts to eradicate Spartina infestations (e.g., Hedge et al. 2003), eff ectively representing the collection of hundreds to thousands of plants.
Here, I present a taxonomic synopsis of the two native and fi ve introduced taxa of Spartina known from British Columbia, Washington, and Oregon (S. alternifl ora, S. anglica, S. densifl ora, S. gracilis Trin., S. patens, S. pectinata Link, and S. ×townsendii). I provide a new taxonomic key for the region, nomenclature for all taxa, including previously unpublished details on several type specimens and new lectotypifi cations, references to published illustrations that clearly show diagnostic characters, specimen citations, and notes on how to distinguish the taxa. New descriptions are given for the closely related and morphologically similar taxa S. alternifl ora, S. anglica, and S. ×townsendii, and the aggressively invading S. densifl ora, which has recently appeared in British Columbia. Spartina ×townsendii is newly reported from British Columbia. Th e descriptions and keys are based on study of specimens collected within and outside the region, in consultation with the global primary and secondary taxonomic literature; these data should be useful for distinguishing the taxa wherever they occur globally, including in Alaska where Spartina has not been reported but is anticipated to become a problem in the future (Morgan and Sytsma 2010). I also include detailed reviews of the evolutionary origins and introduction histories of the invasive Spartina taxa, with a focus on their introductions in British Columbia, Washington, and Oregon. To facilitate species identifi cation, images of exemplar herbarium specimens of each taxon are included in the main text, and images of fi fty herbarium specimens (a subset of the material examined) are included as supplementary information.

Higher-level classification of Spartina
Spartina is a member of the grass subfamily Chloridoideae Kunth ex Beilschm., one of six major lineages (subfamilies) in the grass PACMAD clade, which also includes the subfamilies Panicoideae Link, Arundinoideae Burmeist., Micrairoideae Pilg., Aristidoideae Caro and Danthonioideae H.P.Linder & N.P.Barker (reviewed in Saarela and Graham 2010). Chloridoideae includes some 1420 species in approximately 140 genera (Peterson et al. 2010). In the most recent classifi cation of Chloridoideae, Spartina is included in the tribe Zoyseae Benth., a lineage characterized by glabrous culm nodes, ligules that are a line of hairs or a fringed membrane, bisexual spikelets that are laterally compressed (or terete, rarely dorsiventrally compressed), spikelets with a single fl oret, lower glumes nerveless or 1-veined, lemmas 1-or 3-veined and awnless, ovaries glabrous, a base chromosome number of x = 7-10, and a preference for saline habitats (Peterson et al. 2007(Peterson et al. , 2010. Zoyseae includes the incertae sedis genus Urochondra C.E. Hubb. and the subtribes Zoysiinae Benth. (one genus, Zoysia Willd.) and Sporobolinae Benth. Sporobolinae is characterized by spikelets oriented abaxially along the axis with the lemma facing the rachis, lemmas that are similar in texture to the glumes, paleas that are relatively long and approximately equal in length to the lemma, and free (reluctantly free in Spartina) pericarps (Peterson et al. 2007). Sporobolinae includes the large polyphyletic genus Sporobolus R.Br. (some 200 species; Simon et al. 2011) in which the genera Calamovilfa (A. Gray) Hack. ex Scribn. & Southw., Crypsis Aiton, Pogoneura Napper, and Spartina are embedded -a phylogenetic structure that has been identifi ed in multiple molecular studies (e.g., Hilu and Alice 2001, Peterson et al. 2010. Peterson et al. (2010) recommended future expansion of Sporobolus to include all of these genera, including the monophyletic Spartina, refl ecting their origins from within the Sporobolus line. If (or more likely, when) these nomenclatural changes are proposed, species currently treated in Spartina would be treated as species of Sporobolus, more accurately refl ecting our understanding of the evolutionary history of these grasses and their closest relatives.

Spartina in North America
Taxonomic revisions of Spartina have been produced by Merrill (1902), Saint-Yves (1932) and Mobberley (1956), the latter being the most recent and comprehensive, though it does not cite the specimens examined during the study. Mobberley (1956) recognized some fi fteen taxa in the genus, including three native to the Old World (S. anglica, S. maritima (Curtis) Fernald, S. townsendii), and the remainder native to the New World. Th ere are thirteen species of Spartina in North America, of which nine are native (including one natural hybrid), and four are introduced (Mobberley 1956, Barkworth 2003. Th ey are distributed in low to high intertidal marsh (brackish), inland brackish and freshwater habitats and dry prairie, and have been classifi ed ecologically by habitat as low marsh, high marsh, and freshwater (Maricle et al. 2009). Marsh taxa include the native species S. alternifl ora, S. patens, S. bakeri Merr. (sand cordgrass), S. spartinae (Trin.) Merr. (gulf cordgrass) and S. foliosa Trin. (California cordgrass), and the introduced species S. anglica, S. ×townsendii and S. densifl ora. Freshwater/inland taxa include S. cynosuroides (L.) Roth (big cordgrass), S. gracilis, S. pectinata, and the hybrid taxon S. ×caespitosa A.A. Eaton (mixed cordgrass; S. patens × S. pectinata).
Spartina has a base chromosome number of x = 10, and all species are polyploids (e.g., Marchant 1963, 1986b, Fortune et al. 2008, Renny-Byfi eld et al. 2010. Phylogenetic analyses based on two nuclear (ITS and Waxy) and one plastid locus identifi ed two major lineages in the genus, one comprising the tetraploid (2n = 40) species S. patens, S. gracilis, S. ciliata Brongn., S. pectinata, S. cynosuroides, S. bakeri, and S. arundinacea (Th ouars) Carmich., and the other comprising the hexaploid (2n = 60, 62) species S. alternifl ora, S. foliosa, and S. maritima (Curtis) Fernald (Baumel et al. 2002b). In this analysis one tetraploid species, S. argentinensis Parodi, was sister to the hexaploid clade, and placement of the heptaploid S. densifl ora (2n = 70, , Fortune et al. 2008) diff ered among data partitions (Baumel et al. 2002b). A subsequent phylogenetic study, based on extensive cloning of the low copy nuclear gene Waxy, focused on the hexaploid taxa and revealed a more complicated phylogeny, and an allopolyploid origin for the hexaploid clade was proposed (Fortune et al. 2007). Further phylogenetic study of S. densifl ora revealed the taxon to be of hybrid origin involving lineages related to the tetraploid species S. arundinacea and the hexaploid clade (Fortune et al. 2008 St-Yves (1932) listed twelve specimens (syntypes) in the protologue, including the one designated here as lectotype.
Comments. Spartina alternifl ora is often misspelled in the literature as "Spartina alternifolia". Spartina alternifl ora was described from Bayonne in southwestern France (Loiseleur-Deslongchamps 1807), where it has been introduced for over two centuries. Th e taxon was collected in France as early as 1803, and was known from southern England (Southampton Water) by 1829; it is presumed to have been introduced from North America to these European coastal regions in shipping ballast (see Bromfi eld 1836, Marchant 1968). Spartina alternifl ora is one of the parents of the sterile F1 hybrid S. ×townsendii (see notes under this taxon), which was formed when this introduced species hybridized with the European Atlantic coast native species S. maritima. By 1969 S. alternifl ora was extremely rare in Southampton Water, known from only a single locality Goodman 1969a, 1969b), and it remains rare in Britain (Cope and Gray 2009).
Several introductions of S. alternifl ora have occurred along the west coast of North America where it is not native. Th e species was reported in 1945 from a single estuarial location in Willapa Bay, Washington, where occasional plants had been observed since around 1911, and thought to have been introduced in the early twentieth century with oyster culture (Scheff er 1945). Th e earliest known record documenting S. alternifl ora at Willapa Bay is a photograph taken in the early 1940s by T. Scheff er, housed at the California Academy of Sciences Herbarium (Civille et al. 2005). Based on a review of historical records of the oyster trade, Civille et al. (2005) concluded that S. alternifl ora was probably introduced to Willapa Bay between 1893 and the late 1910s via seeds attached to oysters during their railway transport from the Atlantic coast to the Pacifi c coast. Since the mid-twentieth century S. alternifl ora has expanded dramatically in Willapa Bay and northwards to Grays Harbour and the Copalis River estuary in Washington (Stiller and Denton 1995), radically transforming these tidal ecosystems from mudfl ats to dense cordgrass swards (e.g., Feist and Simenstad 2000, Davis et al. 2004a. Civille et al. (2005) used historical records and remote sensing to reconstruct the dramatic expansion of S. alternifl ora that occurred in Willapa Bay between 1945 and, and concluded that its rapid colonization likely originated from multiple introductions over several decades, as opposed to the long-standing hypothesis of postestablishment colonization following a single introduction to the region (e.g., Scheff er 1945, Stiller andDenton 1995, Feist andSimenstad 2000). Considerable research has been conducted on understanding the invasion dynamics of S. alternifl ora in Willapa Bay (e.g., Davis et al. 2004a, Taylor and Hastings 2004, Grosholz et al. 2009), and substantial eff orts and resources have been directed at its management, control and eradication (e.g., Grevstad et al. 2003, Taylor andHastings 2004). Spartina alternifl ora is also known from the southern end of Padilla Bay in northwestern Washington, where it was apparently introduced independently between 1941 and 1945 to aid in beach stabilization (see Wiggins andBinney 1987, Riggs 1992). Stiller and Denton (1995) confi rmed the genetic distinctiveness of this stand compared with the Willapa Bay population, supporting its putative independent origin.
In Oregon, S. alternifl ora has been reported from the Siuslaw River estuary and Coos Bay (Howard 2005;Wilson et al. 2012). In California S. alternifl ora occurs in southeastern San Francisco Bay (Alameda Creek Flood Control Channel and along the shoreline some three km south) where it was planted for restoration in the late 1970s (Spicher andJosselyn 1985, Ayres et al. 2003). By some 20 years later it had spread considerably, and was known from seven diff erent areas in San Francisco Bay (Callaway and Josselyn 1992). Spartina alternifl ora has also been reported from Bolina's Lagoon and Point Reye's National Seashore north of San Francisco Bay; these occurrences are thought to have originated from fl oating seed originating in San Francisco Bay, but this hypothesis has not been confi rmed .
Th e only native Spartina taxon in California is S. foliosa, and by the 1990s it was known that S. alternifl ora was in the process of competitively excluding S. foliosa (Callaway and Josselyn 1992). Substantial evidence has accumulated indicating that the two taxa have successfully interbred producing a highly invasive fertile hybrid capable of backcrossing with its parental taxa, resulting in a genetically heterogeneous hybrid cordgrass swarm (i.e., S. alternifl ora × S. foliosa) that is much more invasive than either parental taxon (Daehler and Strong 1997, Antilla et al. 2000. Morphological variation of these hybrids is not well characterized in the literature. Spartina alternifl ora is not known from British Columbia, Canada. Daehler and Strong (1996) identifi ed estuaries along the Pacifi c coast extending to the Canada / United States border, which they predicted to be potentially vulnerable to invasion by Spartina species, including S. alternifl ora. Workers should be aware of the potential for its invasion and keep watch for the species in British Columbia. In a recent study of saltmarsh diatoms in central mainland British Columbia some 40 km northeast of Vancouver Island, Roe et al. (2009) reported S. alternifl ora as a dominant species in low and mid marsh components of Waump (51°11'15N, 126°55'15W) and Wawwat'l saltmarshes (51°11'36N, 126°40'5W) in Seymour Inlet, based on fi eldwork conducted in 2002. If the species is correctly identifi ed as S. alternifl ora in this study, this would be the fi rst known site of the taxon in British Columbia. Alternatively, the taxon may be S. anglica, but it is not possible to further confi rm its identifi cation based on the study (voucher specimens are not mentioned). Spartina has not otherwise been reported from Seymour Inlet. Th is region should be re-visited for further fi eld study and collection of voucher specimens. Whatever the identity of the species is, this report suggests that by 2002 invasive Spartina was present in coastal British Columbia at sites further north than those fi rst documented in 2003 near the international border south of Vancouver.
Morphology. Spartina alternifl ora and the European species S. maritima are the parents of the sterile F1 hybrid S. ×townsendii; unsurprisingly, S. alternifl ora is morphologically similar to S. ×townsendii and the amphidiploid S. anglica. It can be distinguished from these taxa by its shorter spikelets [8-14(-16.5) mm vs. 14-25 mm], narrower branch rachises [0.4-1 mm wide between spikelets vs. 1-2.2 mm wide], glumes glabrous or weakly pubescent [vs. glumes weakly to densely pubescent], and leaf blades erect, forming an angle of 15-18° with the culm [vs. leaf blades spreading, forming an angle of 30-60° with the culm]. Spikelets of S. alternifl ora are shown in Fig. 1, and an exemplar specimen is shown in Fig. 2. Glumes in S. alternifl ora vary from glabrous to pubescent (details on this variation are given in Mobberley 1956). In  individuals with glabrous glumes, this is a good character for distinguishing S. alternifl ora from S. ×townsendii and S. anglica, which consistently have pubescent glumes, but the character is more diffi cult to interpret and less reliable in S. alternifl ora individuals with pubescent glumes. Spartina alternifl ora also tends to have narrower infl orescences than S. ×townsendii and S. anglica [(5-)6-10(-15) mm wide vs. 7-25 mm wide], but there is considerable overlap in this character. Recent taxonomic keys (Barkworth 2003, Kozloff 2005) distinguishing these three taxa emphasize variation in number of panicle branches (3-25 in S. alternifl ora vs. 1-12 in the other taxa). I found this character unreliable in material examined here, and do not include it in the key. Spartina alternifl ora can further be distinguished from S. anglica by its shorter anthers [3-6 mm long vs. 7-10 mm long], and from S. ×townsendii by its fully exserted, dehiscent anthers at anthesis and fertile pollen [vs. anthers not or incompletely exserted, indehiscent and sterile pollen]. Spartina alternifl ora is readily distinguished from S. densifl ora, S. gracilis, S. patens and S. pectinata by it glabrous leaf blade margins [vs. scabrous leaf blade margins].
Th ere is considerable morphological variation in S. alternifl ora throughout its native range with northern plants from Canada and Maine tending to have looser infl orescences, weakly overlapping spikelets, and less glume pubescence, and southern plants tighter infl orescences, more strongly overlapping spikelets, and more pubescent glumes. Th is variation has been recognized taxonomically in the past at the species and infraspecifi c levels; however, Mobberley (1956) demonstrated the variation to be clinal and therefore taxonomically inconsistent, and only a single taxon is now generally accepted (e.g., Barkworth 2003). Specimens examined from the Pacifi c coast in Washington have moderately overlapping spikelets.   (1978), Hubbard S.17868A, is mounted on two sheets at K: K000710269 and K000710270. Th e latter is clearly labeled 'sheet II', but the former is not reciprocally cross-labeled as 'sheet I' therefore the two sheets are duplicates according to the Code. Th e specimen indicated as 'sheet II" is more robust, and is here designated as the lectotype.

Comments.
Spartina anglica is an amphidiploid taxon that arose in Britain in the 18 th century from chromosome doubling of the sterile F1 hybrid taxon S. ×townsendii (see the discussion under that species for details, and reviews in Marchant 1968 andGray et al. 1991). Th e origin of S. anglica is a remarkable and well-studied example of allopolyploid evolution.
Spartina anglica is a problematic invasive species in coastal areas of western North America, and has been present on the continent for over fi fty years. In the United States it is known from Washington and California. It was planted in Puget Sound, Washington in 1961 (Spicher andJosselyn 1985, Frenkel 1987) to provide forage for cattle and to stabilize a dike system (Hacker et al. 2001). By 1997, it had expanded substantially in Puget Sound, occurring in 73 sites and covering 397 solid hectares of intertidal habitat (Hacker et al. 2001, see Hedge et al. 2003. Eradication eff orts in this region began in 1997 (Hacker et al. 2001). In 1977 S. anglica was deliberately transplanted from Puget Sound to California where it was introduced at Creekside Park Marsh in San Francisco Bay, Marin County (Spicher and Josselyn 1985), and it persists at this single location in the state (e.g., Grijalva et al. 2006, San Francisco Estuary Invasive Spartina Project 2012. Spartina anglica is mentioned, but not treated in Baird and Th ieret (2012).
Spartina anglica was discovered in British Columbia in 2003 on Roberts Banks in the Fraser River estuary and in Boundary Bay along the British Columbia and Washington border (Williams 2004). Williams (2004) noted the infestation to be in its early stages in 2003, and estimated the species to have arrived in the area some three to fi ve years earlier. Additional collections were made in 2004 in Boundary Bay (Lim s.n., V-two specimens) and on Roberts Banks (Williams 2004-1, 2004-2 [Fig. 3], 2004. Following this initial documentation of S. anglica in British Columbia, extensive fi eld surveys and mapping exercises were undertaken to determine the extent of the species at Roberts Bank and in Boundary Bay, and an eradication plan was implemented (Dresen et al. 2010;Williams et al. 2010). Despite removal and eradication eff orts, by 2009 S. anglica had increased dramatically in these regions (Ducks Unlimited Canada 2010). It is currently reproducing by seed production and vegetatively by outward clonal growth forming meadows and by drifting plant parts (R. Knight, personal communication 2010). Th ere are only a few herbarium collections documenting S. anglica in the province, collected by G. Williams and P. Lim when the taxon was fi rst found. Additional collections should be made at these and other sites where the species is found, to properly document its continued existence and spread.
Morphology. Th e description here is based on collections from Washington, Oregon, and British Columbia, and Old World material housed at CAN and UBC (see Specimens Examined). Spartina anglica is morphologically similar to S. ×townsendii, and the two can be challenging to distinguish. Diff erences between these species were characterized in detail by Marchant (1968a), and my observations here are in general accordance with his fi ndings. Mobberley (1956) commented only briefl y on their differences in his treatment of S. ×townsendii, the name under which both ploidal forms of the taxon were recognized at the time. Only a few taxonomic keys in North America include both taxa (Barkworth 2003, Kozloff 2005, emphasizing variation in ligule length, anther length, and anther dehiscence for identifi cation. In addition to these characters, the key presented here includes spikelet length, upper glume venation, and upper glume length. Overall, plants of S. anglica tend to be larger than those of S. ×townsendii, including the lengths of reproductive structures useful in distinguishing the taxa. Th e species can be distinguished with careful measurements on herbarium specimens, though there is some overlap in the diagnostic morphological characteristics. When making a determination, multiple characters should be examined and multiple measurements should be made on a single plant when possible. Spartina anglica is distinguished from S. ×townsendii by its longer spikelets [(15-)16.5-25 mm long vs. 14-17.5 mm long]; longer anthers [7-10 mm long vs. 5-7(-8.5) mm long]; anthers that are fully exserted and dehiscent [vs. anthers that are not or incompletely exserted and indehiscent;  Marchant (1968a) observed that leaf blade angle with respect to the culm can be used to diff erentiate the taxa, as the blades tend to be more erect in S. ×townsendii [blades forming an angle 30-40° with respect to the culm] and more spreading in S. anglica [blades forming an angle of 30-60°with respect to the culm], and this character is noted in many of the recent fi eld identifi cation resources for the Pacifi c Northwest. Leaf angle is sometimes diffi cult to determine on herbarium specimens. Marchant (1968a) also noted swards of S. anglica in the fi eld (in England) diff ered from swards of S. ×townsendii as being less dense in tillers (ca. 52/100 cm 2 vs. ca. 96/100 cm 2 ) and having more robust but fewer rhizomes.
Pollen Fertility. Determining pollen fertility by staining anthers with lactophenol cotton blue is a useful way to distinguish male sterile hybrid plants from those that are fertile, as the cytoplasm of fertile pollen grains readily takes up the stain whereas sterile (i.e., aborted) pollen grains do not. Pollen staining is thus an eff ective, though more technically involved method, to defi nitively distinguish the fertile S. anglica from the sterile F1 hybrid S. ×townsendii, as demonstrated by Marchant (1968a). To confi rm identifi cations of these taxa here, and to demonstrate the utility of this technique, I determined pollen fertility for multiple herbarium specimens of S. anglica and S. ×townsendii using lactophenol cotton blue; collections examined in this manner are identifi ed with the symbol † in the specimen list below and under S. ×townsendii. One to three anthers were sampled from a spikelet on each specimen, and these were macerated with forceps in a drop of water on a glass slide to release the pollen. One or a few drops of lactophenol cotton blue were added to the slide and after sitting for a minimum of ten minutes, the slides were examined at 20× to 60× magnifi cation with a compound microscope. In general, the number of pollen grains recovered per specimen varied with the stage of pollen development. For example, in specimens of S. anglica that had been collected during or after dehiscence, fewer pollen grains were available for study, whereas in S. ×townsendii dehiscence does not occur and abundant pollen grains were recovered from the anthers. In fertile pollen grains the cytoplasm was evidently visible and stained, whereas in sterile pollen there was no or little evidence of cytoplasmic staining (Fig. 5). All plants that were determined morphologically to be S. anglica had fertile pollen, and all plants determined as S. ×townsendii had sterile pollen. Description. Culms to 96 cm tall, cespitose from hard knotty bases, rarely with short rhizomes, forming dense tufts. Sheaths glabrous, often purple-tinged; ligules 1-2 mm long; blades to 32 cm long × 1-2 mm wide, involute for most or all of their length, wider proximally when fl at, adaxial surfaces scabrous, abaxial surfaces glabrous, margins scabrous. Infl orescences 10.5-17 cm long × 6-8(-10) mm wide at midpoint, with (2-)6-9(-15) branches; branches 3-6(-7.5) cm long × 2-3 mm wide, appressed, conspicuously decreasing in length towards infl orescence apex, rachises 0.8-1 mm wide between spikelets, not extending beyond terminal spikelet, glabrous, margins glabrous or scabrous. Spikelets 9-13 mm long × 1.5-2 mm wide, tightly appressed, strongly overlapping; calluses 1-1.5 mm long. Glumes glabrous or scabrous, when present hairs < 0.1 mm long, keels scabrous, teeth 0.1-0.2 mm long, margins usually purple-tinged; lower glumes 4-7 mm long × 0.5-0.7 mm wide, 1-veined; upper glumes 7.5-11.5 mm long × 1-1.5 mm wide, 1-veined; lemmas 6-9 mm long, glabrous or minutely scabrous, keels scabrous distally, glabrous proximally; paleas exceeding lemmas by 0.5 mm, glabrous; anthers 3-4 mm long, yellowish, exserted at maturity, pollen fertile. 2n = 70 , Fortune et al. 2008) Common Name. Austral cordgrass. Etymology. Th e epithet densifl ora refers to the densely-fl owered infl orescences of the species.
Illustration. Barkworth 2003:251. Distribution. Native to South America in temperate coastal regions of southern Brazil, Uruguay, Argentina, and on the coast of Chile (Bortolus 2006). Introduced in Spain, Portugal, the Mediterranean (see Castillo et al. 2000, Nieva et al. 2005, United States (California, Washington), and Canada (British Columbia).
Comments. Th e taxonomy, biogeography, and natural history of S. densifl ora is reviewed by Bortolus (2006). Th e synonymy presented here is not complete; several taxa described from South America are considered to be synonyms by Mobberley (1956), including S. montevidensis Arech. (see Carrasco et al. 2002 for the location of a probable type specimen), S. patagonica Speg., and several taxa described by St-Yves (1932). A modern taxonomic revision is needed for S. densifl ora to clarify its taxonomy and morphological variation throughout its range (Bortolus 2006), particularly given its rapidly increasing global footprint.
In North America, S. densifl ora is present in California, Washington, and British Columbia. It was fi rst established in California. It occurs in Humboldt Bay, where it is thought to have been introduced by shipping in the late 1800s (Spicher andJosselyn 1985, Kittleson andBoyd 1997). Prior to the early 1980s, this population was erroneously thought to be a native form of S. foliosa (Bortolus 2008), and seed collected from plants growing in Humboldt Bay was introduced to San Francisco Bay (see Faber 2000), and independently to Richardson Bay. Hybrids between S. densifl ora and S. foliosa have been confi rmed recently from San Francisco Bay in a region where both parental species are present . Th ese are not well characterized morphologically in the literature.
Spartina densifl ora was documented in Washington a decade ago. Th e fi rst collection was made in 2001 on Whidbey Island at the northern boundary of Puget Sound (Heimer 01-1 WTU, UBC). Caplow (2002) reported a 1/10 acre infestation of S. densifl ora at the west end of Grays Harbor (46.9408°N, 124.04583°W) in the southeastern part of the state. By 2006 this infestation had expanded substantially, covering some 10 acres (Murphy et al. 2007). Kozloff (2005) included S. densifl ora in his Spartina treatment, noting only 'coastal salt marshes' as its distribution. Spartina densifl ora is now invading British Columbia, where it was fi rst found in 2005 in Bayne's Sound, a channel between Vancouver Island and Denman Island (Morgan and Sytsma 2010). It is not reported for the province in several recent fl oras that consider the region (Douglas et al. 2001, Barkworth 2003, Kozloff 2005, but is included in the online E-Flora BC (Anonymous 2011). In the Bayne's Sound region, only a few herbarium collections documenting the species have been made; these are from Fanny Bay (Lomer 5723, CAN, Fig. 6) and the Buckley Bay Ferry Terminal (Lomer 7377,CAN). Some data on the extent of the invasion are given by Morgan and Sytsma (2010) and Dresen et al. (2010), who mapped the species in Bayne's Sound where the extent of its local distribution ranges from individual plants to multiple fi ve meter patches. Detailed characterization of this invasion has not otherwise appeared in the literature.
Morphology. Th e description presented here is based on the few specimens that have been collected in Washington and British Columbia and deposited in herbaria (see Specimens Examined). Variation in some characters, particularly vegetative characteristics such as leaf length and culm height, is probably greater than recorded here. For example, Mobberley (1956) reported that culms in the species range to 1.5 m tall. For a description based on material from the native range of S. densifl ora, see Mobberley (1956).
Spartina densifl ora can be distinguished from S. alternifl ora, S. anglica, and S. ×townsendii by the following combination of characters: plants cespitose [vs. strongly rhizomatous]; blades involute for all or most of their length [vs. blades fl at proximally, involute distally]; blade margins scabrous [vs. blade margins smooth]; branch rachises not prolonged beyond the terminal spikelet [vs. branch rachises prolonged beyond the terminal spikelet as a bristle, rarely not prolonged]; glume margins often purple-tinged [vs. glume margins whitish, not purple-tinged] (Fig. 7); and spikelets tightly appressed and strongly overlapping, concealing the rachis between spikelets [vs. spikelets weakly appressed and weakly to moderately overlapping, with portions of the rachis usually visible between spikelets]. In the fi eld, the leaf margins of S. densifl ora may feel smooth to the touch, as the involute blades often conceal the leaf margins where the scabrous teeth are located; these scabrous teeth are best observed with a microscope (F. Lomer, personal communication, 2012    Description: See Mobberley (1956), Barkworth (2003), and Douglas et al. (2001). 2n = 40 (Marchant 1968b, Reeder 1977. Common Name. Alkali cordgrass, big cordgrass. Illustrations. Hitchcock 1935:493, Hitchcock 1951: 513, Hitchcock et al. 1969:706, Barkworth 2003:251, Douglas et al. 2001 Etymology. Th e Latin epithet gracilis means "thin, slender" (Stearn 1992), probably given in reference to the general habit of the species.
Etymology. Th e Latin epithet patens means "spreading, outspread" (Stearn 1992), given in reference to the general habit of the species, which forms dense, monotypic stands.
Illustrations. Hitchcock 1935: 494;Hitchcock 1951:514;Holmgren et al. 1998:750, Douglas et al. 2001:271, Barkworth 2003 Distribution. Native to the east coast of North America and Central America, distributed along the Atlantic coast of Newfoundland and Labrador to Texas, the Atlantic coast of Mexico, and throughout the Caribbean Islands (e.g., Barkworth 2003, Beetle 1977, Howard et al. 1989, Liogier and Martorell 2000, Peterson 2001). It has been introduced to coastal regions in Spain (Campos et al. 2004, SanLeón et al. 1999, Prieto et al. 2011, China ), North Africa, and the west coast of the United States (Oregon, Washington), and Canada (British Columbia).
Comments. Spartina patens grows in coastal salt marshes and brackish waters, where it usually forms dense stands above the intertidal zone and into higher and drier areas of the salt marsh (Mobberley 1956). In Europe this taxon is sometimes treated as a separate species, S. versicolor Fabre, a traditional circumscription that some contemporary authors maintain (e.g., Sánchez Gullón 2001; Clayton et al. 2002 onwards;see SanLeón et al. 1999 andPrieto et al. 2011 for summaries of this taxonomic problem). In North America S. patens hybridizes with S. pectinata in areas where they grow together, forming the hybrid taxon S. ×cespitosa. Th is hybrid is known from scattered locations along the Atlantic coast from Prince Edward Island to Maryland (Mobberley 1956, Dore and Marchant 1968, Barkworth 2003), but has not been recorded along the Pacifi c coast. Th e biology of S. patens is reviewed by Lonard et al. (2010).
In Oregon, Spartina patens grows in the Siuslaw estuary on Cox Island (Lane Co.), where it has been introduced since at least 1939 and has expanded considerably since that time (Frenkel and Boss 1988). It is not reported from elsewhere in Oregon (Wilson et al. 2012). In Washington it has been reported from Dosewallips State Park on the west shore of Hood Canal (Frenkel 1987). In California it is present in Southampton Marsh in San Francisco Bay (Spicher andJosselyn 1985, Baird andTh ieret 2012).
Spartina patens was the fi rst of the invasive cordgrasses to be collected in British Columbia. It was discovered in 1979 in the Comox Estuary on Vancouver Island (Brayshaw 79-1143, V); nearly a decade later, in 1988, it was collected on the adjacent mainland coast in Burrard Inlet, North Vancouver (Lomer 88-140, UBC, Fig. 9). Spartina patens was recognized as part of the provincial fl ora by Douglas et al. (1994) based on collections from these two localities, and is reported for the region in recent fl oras (Barkworth 2003, Douglas et al. 2001. Large stands of the species were reported in 2000 at the Comox locality in the Courtenay River estuary at the delta of Brooklyn Creek to the embayed area of Goose Spit (Asp and Adams 2000). Spartina patens is also reported from Port Moody Arm of Burrard Inlet in British Columbia (Dresen et al. 2010). Voucher specimens from these latter two sites should be collected and deposited in herbaria.
Morphology. Spartina patens exhibits considerable morphological variation and several authors have recognized two infraspecifi c taxa (see Mobberley 1956). Mobberley (1956) studied this variation across the native range of the taxon and found plants at the northern and southern extremes of the taxon's range to be distinct, but considerable overlap and clinal variation from north to south in areas between the extremes. Based on these data he argued against the recognition of infraspecifi c taxa in S. patens, a circumscription followed by subsequent authors.
Based on specimens of S. patens examined here collected in Oregon, Washington, and British Columbia, introduced plants in these areas are similar morphologically to those in the eastern Canada and the northeastern United States, which tend to be smaller than plants distributed further south (Mobberley 1956). Spartina patens can be distinguished from S. densifl ora by the following combination of characters: branches distinctly one-sided, appressed, ascending or spreading from main axis, distant or weakly overlapping, and approximately the same length within an infl orescence [vs. branches not distinctly one-sided, appressed, strongly overlapping, and conspicuously decreasing in length towards the infl orescence apex], rhizomes wiry, plants forming dense mats [vs. rhizomes absent, rarely short, plants cespitose, forming  Description. See Mobberley (1956), Barkworth (2003), and Douglas et al. (2001). 2n = 40 (Marchant 1963, 1968b, Reeder 1977. Common Name. Prairie cordgrass. Etymology. Th e epithet pectinata means comb-like or tooth-like, and was given in reference to the distinctly pectinate teeth on the glume keels of S. pectinata, one of its diagnostic characteristics. Illustrations. Hitchcock 1935:490, Hitchcock 1951:510, Holmgren et al. 1998:749, Barkworth 2003:251, Cope and Gray 2009 Distribution. Spartina pectinata is widespread across much of North America north of Mexico, distributed in southern Alberta, eastern Washington and Oregon, south to Texas, and east to Newfoundland and Labrador (Barkworth 2003). It grows in marshes, slough and fl oodplains, in dry prairies and along roads (Barkworth 2003). It is known from a few scattered sites in Britain and Ireland where it has been introduced and escaped from cultivation (Cope and Gray 2009).
Comments. Spartina pectinata is not considered to be a problematic invasive species, though it has been introduced sporadically to other regions, such as the United Kingdom. It occurs natively in eastern Washington and Oregon (Hitchcock et al. 1969, Barkworth 2003. Spartina pectinata is grown as an ornamental. British Columbia is the only Canadian province in which S. pectinata is not native, but the taxon has been variously recognized as part of its fl ora. In the fi rst major treatment of the British Columbia fl ora, Joseph K. Henry (1915) reported the nonnative S. cynosuroides, an Atlantic coast taxon confused at the time with S. pectinata according to Eastham (1947), which Henry suggested might be present in the southern interior based on its presence in adjacent Washington. No collections of S. pectinata from the southern interior of British Columbia have been made. Spartina pectinata was added to the fl ora of British Columbia by Eastham (1947), based on a single collection made by V. Brink on the shore of Sea Island at the mouth of the Fraser River (where the Vancouver International Airport is now located). Eastham suggested that it might have been transported to this location down the river from the interior, but this is unlikely since there are no reports of the taxon from interior British Columbia. Based on this single record, S. pectinata was recognized as part of the British Columbia fl ora in subsequent fl oristic works (Hubbard 1969, Taylor and MacBryde 1977, Scoggan 1978, but was later excluded by Douglas et al. (1994) on the basis of being an introduced spec ies not collected in over 40 years. Th is exclusion was followed in recent treatments (Qian andKlinka 1998, Douglas et al. 2001). Attempts to locate the Brink collection in British Columbia herbaria (V, UBC) were unsuccessful, and it is therefore not possible to confi rm this report.. Spartina pectinata was recently collected in a vacant lot in the city of Burnaby (Lomer 6778, Fig. 10, 6805, Suppl. Fig. 33), where it likely escaped or persisted from cultivation (F. Lomer, personal communication, 2011). Th e clump from which these specimens were obtained no longer exists, as the site has been built over (F. Lomer, personal communication, 2012). Th e specimens reported here are the fi rst confi rmed records of S. pectinata for British Columbia. Th e species is not considered to be invasive.
Comments. In the mid to late 1800s an unknown cordgrass of restricted distribution appeared and spread rapidly along the shores of Southampton Water, England (Stapf 1914), which diff ered morphologically (particularly by its sterile pollen) from S. maritima, the single cordgrass species native to the Atlantic coast of Europe and north Africa (Marchant and Goodman 1969c), and the introduced S. alternifl ora, which had been present in the region since the early part of the 19 th century. Th e brothers Henry and James Grove (1881) described this taxon as S. townsendii from plants collected near Hythe. In the 1890s a second form of S. townsendii, which was recognized and considered distinct by having fertile stamens, was collected at multiple localities in the region, and by the mid twentieth century it had expanded substantially on tidal fl ats across the British Isles (see Goodman et al. 1959, Hubbard 1957, 1965. For decades these two forms (one sterile, the other fertile) of S. townsendii were referred to collectively as the S. townsendii aggregate or S. townsendii sensu lato. Because of its vigorous growth and ability to rapidly colonize and stabilize mud fl ats, S. townsendii s.l. was considered to be a "useful" species and was distributed and planted widely for land reclamation, coastal protection, and animal feed across the British Isles, Europe, and in New Zealand (e.g., Oliver 1925, Harboard 1949, Goodman et al. 1959, Ranwell 1967. Soon after its formal description S. townsendii was considered to be a species of hybrid origin. Foucaud (1894) suggested that S. townsendii was probably a hybrid of the native S. maritima and the introduced S. alternifl ora, a hypothesis later supported by Stapf (1914) and Huskins (1930), who examined cytological evidence and hypothesized that fertile plants of S. townsendii s.l. originated from chromosome doubling following hybridization between its parent species. Marchant (1963) confi rmed this work, and reported chromosome numbers as 2n = 62 for S. alternifl ora, 2n = 60 for S. maritima, 2n = 62 for sterile plants of S. townsendii, and 2n = 120, 122, 124 for fertile plants of S. townsendii. Th e hybrid and chromosome doubling origins of the forms of S. townsendii have been confi rmed by multiple lines of molecular evidence (Guénégou et al. 1988, Ferris et al. 1997, Ayres and Strong 2001, Baumel et al. 2003). Hubbard (1957) observed the type specimen of S. townsendii to be the sterile F1 hybrid, and the taxon was subsequently referred to as S. ×townsendii. Th e fertile plants remained without a name until Hubbard (1978) later described them as S. anglica, and the two forms have since been recognized as distinct taxa. Molecular data have identifi ed S. alternifl ora as the female parent and S. maritima as the male parent in the origin of S. ×townsendii (e.g., Ferris et al. 1997).
An independent origin of S. ×townsendii is documented in France. In 1894 Jules Foucaud described S. neyrautii Fouc. from southwestern France and northern Spain. Spartina neyrautii was initially considered to be a variant of S. maritima (e.g., Chevalier 1923, Saint-Yves 1932, but was later recognized as a morphologically and cytologically distinct hybrid, S. ×neyrautii, with the same parentage as S. ×townsendii (e.g., Jovet 1941, Chevalier 1933, Marchant 1977. Baumel et al. (2003) confi rmed this with molecular data, demonstrating that S. ×neyrautii and S. ×townsendii originated independently by hybridization between the same maternal (S. alternifl ora) and paternal (S. maritima) taxa. Because both taxa apply to the hybrid S. alternifl ora × S. maritima, the later name S. ×neyrautii is a synonym of S. ×townsendii, in accordance with article H.2 of the Vienna Code (McNeill et al. 2006), and as noted earlier by Raybould et al. (1990). In recent decades it has been documented that these hybrid plants are highly restricted in distribution in France (Marchant 1977, Raybould et al. 1990, Baumel et al. 2003. Minor morphological diff erences between S. ×neyrautii and S. ×townsendii were noted by Marchant (1977). Measurements of spikelet characters in S. ×neyrautii type material examined at US fall within the range of variation reported here for S. ×townsendii.
Spartina ×townsendii has been introduced into North America, where apparently only a single occurrence has been reported in the literature. Hitchcock et al. (1969) noted a single known population of S. ×townsendii in Washington at Stanwood, Snohomish Co. At the time of that publication, the fertile (=S. anglica) and non-fertile forms of Spartina ×townsendii were not distinguished taxonomically, and it is not explicitly clear from the fl ora which form of the taxon was known from the site. A specimen collected in 1965 from this population [Austenson s.n. (WTU, Suppl. Fig. 35)] is here confi rmed to be the F1 sterile hybrid S. ×townsendii. Th e determination of a more recent collection from the Stanwood area identifi ed as S. ×townsendii requires confi rmation (Snohomish Co., Davis Slough west of Stanwood, 25 Aug 1990, M.Arnot 254, WTU-317391, not seen). A 2005 collection from Washington originally identifi ed as S. ×townsendii (Giblin & Legler 270 WTU) is here re-determined to be S. anglica. Barkworth (2003) included S. ×townsendii in her treatment of Spartina for North America, but did not include a distribution map or otherwise indicate a range for the species, suggesting some confusion in the literature on its status in North America. Kozloff (2005) also included S. ×townsendii (as well as S. anglica) in his Pacifi c Northwest fl ora, indicating only 'coastal salt marshes' for its distribution. Th ere apparently are no published data on the current status of the Stanwood population. If the population at Stanwood persists, new collections should be made to document its continued existence at the site, and if other populations are known or discovered, herbarium collections should be made to document their existences.
Spartina ×townsendii has not previously been reported from British Columbia. It is here reported as new for the province on the basis of two collections made in 2006 in in Boundary Bay at sites separated by some 4.4 kilometers (by air) [Taylor 80 (UBC,Fig. 11) and Saarela & Percy 791 (CAN, Fig. 12, UBC, Suppl. Fig. 34)]. Th ese appear to be the most recent confi rmed reports of the taxon in North America since it was collected at Stanwood, Washington. Herbarium specimens of these collections were initially determined (incorrectly) as S. anglica and S. alternifl ora, since S. ×townsendii was not expected in British Columbia. Subsequent study of this material, in combination with the Spartina taxonomic literature and comparisons with specimens of S. anglica and Old World specimens of S. ×townsendii at CAN and UBC, confi rmed the specimens to be S. ×townsendii, prompting the current taxonomic study. Pollen in these specimens is sterile, as determined by pollen staining (see discussion under S. anglica, Fig. 5), further confi rming their identities as S. ×townsendii. Specimens from which pollen was extracted and stained with lactophenol cotton blue to assess fertility are identifi ed with the symbol † in the Specimens Examined below.
Th e origin of S. ×townsendii in British Columbia is not known, and there are no data on the extent of the Boundary Bay sub-populations in 2006 aside from notes on the Saarela and Percy collection label indicating a single clump of the grass approximately one meter in diameter. It is not known if S. ×townsendii has persisted in British Columbia since collected some fi ve years ago. Since major eff orts are ongoing to remove Spartina Figure 11. Photograph of a specimen of Spartina ×townsendii collected in Boundary Bay, British Columbia (Taylor 80,UBC). Th is is the fi rst record of the taxon for British Columbia. Image published with the permission of the University of British Columbia Herbarium, Beaty Biodiversity Museum. plants from Boundary Bay where S. ×townsendii was found, it is possible the original stands from which the specimens were collected have been removed. Th e region should be studied to determine if the taxon is present. Since the taxon is sterile and does not set seed, it must have been introduced into Boundary Bay by vegetative reproduction, probably from rhizome fragments transported in tidal currents. It is possible that the British Columbia plants originated from the stand at Stanwood, Washington, if it persists, or there may be other extant occurrences of S. ×townsendii somewhere to the south of Boundary Bay. Workers searching Puget Sound for invasive Spartina (e.g., Benbrook 2011) should be aware that S. ×townsendii may also be present.
Morphology. Th e description here is based on the fi rst known collection from Washington, the two collections from British Columbia, and Old World material housed at CAN and UBC (see Specimens Examined), including collections made by H. & J. Groves who fi rst described the taxon over a century ago. Th e North American specimens of S. ×townsendii are morphologically similar to the Old World specimens examined. Spartina ×townsendii is distinguished from Spartina anglica by its shorter spikelets [(14-17.5  ×townsendii, compared to 30-60° in S. anglica (Marchant 1968a). Th is character can be diffi cult to evaluate on herbarium specimens depending on how they were pressed, but should be more reliable in the fi eld for distinguishing the taxa, particularly if they occur together. If stands of S. ×townsendii are relocated in British Columbia or elsewhere in the region, the taxonomic utility of this character should be carefully evaluated. Marchant (1968a) noted swards of S. ×townsendii to be distinct in appearance from swards of S. anglica in England, having high culm density and high tiller density (ca.