Poa secunda J. Presl. s.l. is a morphologically highly variable bunchgrass that is a valuable forage species in western North America. There has been much controversy as to whether multiple taxa should be recognised and at what rank in this taxonomically challenging apomictic complex. Here we propose an infraspecific classification for Poa secunda of six varieties within two subspecies, juncifolia and secunda. New combinations are P. secunda vars. ampla, gracillima, juncifolia, nevadensis and scabrella. Conflicting plastid and nrDNA phylogenies show that P. sect. Secundae is of ancient hybrid origin. Based on this and its distinct morphology, the section is raised to the rank of subgenus. A key is presented for P. secunda infraspecies and closely related non-arctic species. Suppl. materials are provided of chromosome counts for Secundae taxa and D.D. Keck specimen annotations of taxa here included in P. secunda.

Apomixis, hybridisation, Poa secunda , Poaceae , polyploidy, reticulation, Secundae , taxonomy

Poa secunda J. Presl. s.l. is a morphologically highly variable species found primarily in western North America. It is common to dominant in grasslands extending from Alaska to Northern Mexico and eastwards in the northern Great Plains and scattered more eastern locations to the Gaspé Peninsula in Quebec (Marsh 1952, Kellogg 1985a, Soreng 2007). The species is disjunct but not common in South America, occurring in Patagonian Argentina and Chile (presumed origin of the type collection of P. secunda). This perennial bunchgrass is a valuable forage species that greens up and flowers in early spring and is abundant across western grasslands, coastal chaparral, Great Basin steppe, uplands of the Mojave Desert and dry forests surrounding these and reaches into alpine meadows. It exhibits diverse ecotypes across this wide geographic and climatic range. Poa secunda s.l. is abundantly represented in herbaria across the United States; for example, the United States National Herbarium (US) has more than three very full herbarium cases of specimens for the United States and Canada. It is distinguished from most Poa species by its obscurely keeled lemmas, calluses often with a minute crown of hairs surrounding the base of the lemma (in subsp. secunda) and elongated spikelets (mostly 3.5–5 × longer than wide).

Differing taxonomies of the Poa secunda J. Presl complex continually appear in the literature. There are some 45 formal names applied to P. secunda s.l. Some taxonomists follow A.S. Hitchcock’s (1935, 1951) taxonomic revision in the Manual of the Grasses of the United States, in recognising his two informal groups of species, “Scabrellae” and “Nevadenses”. In the former group, Hitchcock included P. scabrella (Thurb.) Benth. ex Vasey, P. gracillima Vasey, P. secunda and P. canbyi (Scribn.) Howell; in the latter group P. nevadensis Vasey ex Scribn., P. curtifolia Scribn., P. juncifolia Scribn. and P. ampla Merr. Most of these taxa are still often recognised as species, subspecies or varieties (e.g. Keck 1950, 1959, Hitchcock et al. 1969, Dorn 1977, 1988). Keck (1950, 1959) additionally recognised P. tenerrima Scribn., a serpentine endemic of the Sierra Nevada foot hills and P. incurva Scribn. & T.A. Williams, a western sub-alpine/alpine element, within the Scabrellae group. Keck (1950, 1959) furthermore treated North American P. secunda as P. sandbergii Vasey, restricting P. secunda to South America and Hitchcock et al. (1969) accepted this split. Arnow (1981) concluded P. secunda was the correct name for plants of the P. sandbergii form. Hitchcock (1935, 1951) included P. sandbergii and P. incurva in P. secunda. Marsh (1950, 1952) and Kellogg (1985a, b) quite independently lumped most of these taxa in P. secunda without any infraspecies (followed by Lesica 2012, for Montana). Kellogg’s work, based on morphometric analyses, separated P. curtifolia; Marsh maintained that and also P. tenerrima. Soreng (1991b) divided P. secunda s.l. into two subspecies, subsp. secunda and subsp. juncifolia (Scribn.) Soreng, corresponding to the taxa in the Scabrellae and Nevadenses groups, respectively and accepted P. curtifolia and P. tenerrima. Dorn (1988) accepted as full species the two subspecies that Soreng recognised and established a few varieties within those. Skinner (2010) followed the subspecies split, but provided separate pages for each of the common names that Hitchcock (1951) applied to species aligned within the two subspecies: Sandberg bluegrass (P. secunda var. secunda), Sandberg bluegrass [P. secunda var. incurva (Scribn. & T.A. Williams) Beetle], Pine bluegrass (P. scabrella), Canby bluegrass (P. canbyi), Nevada bluegrass (P. nevadensis), Big bluegrass (P. ampla) and Alkali bluegrass (P. juncifolia). Several cultivars correspond to these different forms (Alderson and Sharp 1995). Thus, there appears to be a consistent desire and effort to maintain some or all of the diversity of forms commonly recognised, at some taxonomic rank.

Molecular (Patterson et al. 2005) and cytogenetic (Hiesey and Nobs 1982) studies show that P. secunda s.l. is a highly complex, apomictic species. Apomixis, although it is facultative to varying degrees (Clausen 1961, Kellogg 1987), is apparently the main mode of reproduction in P. secunda (Kellogg 1987). This has led to the production of various “strains”, ecotypes and races that are fairly monomorphic, but also to much intermediacy. Taxonomy in apomictic complexes is notoriously difficult, as is evident in the case of P. secunda s.l. However, we concur with Marsh (1952) and Kellogg (1985a, b) that the rank of species is not justifiable for most of the Scabrellae and Nevadenses taxa.

Poa secunda s.l. belongs to P. sect. Secundae, a primarily North American section of about eight species (Soreng 1991a, b, 1998, 2007, Soreng et al. 2003a, b, Gillespie et al. 2007). Soreng (1991a) established sect. Secundae with two subsections to accommodate species that share similar, apparently plesiomorphic Poeae traits, along with a derived chloroplast type (Soreng 1990) and suggested the section was of hybrid origin (Soreng 1990, 1991a, b). Our phylogenetic studies, based on plastid and nuclear ribosomal DNA, demonstrate that sect. Secundae is reticulate in origin (Gillespie et al. 2008, Cabi et al. 2017, Soreng et al. 2017). Secundae species exhibit several atavistic traits (Soreng 1991a, Soreng et al. 2015) that are otherwise odd in Poa, including; a crown of hairs around the base of the callus, upper culm leaf sheaths margins often free more than 80–90% of their length and lemmas that are often somewhat rounded on the back. These traits are common in genera outside of Poa and have led to many names for Secundae taxa being published in Atropis, Glyceria and Puccinellia. Secundae species are tufted (rhizomes occur in some putative hybrids, see Species hybrids involving Poa secunda in the Taxonomy section) and perfect flowered, with anthers ranging from 1.2–3 mm long. Soreng (1991a, 2007) recognised two subsections: Secundae and Halophytae V.L. Marsh ex Soreng. In addition to P. curtifolia, P. secunda s.l. and P. tenerrima, subsection Secundae also includes two Arctic species, P. ammophila A.E. Porsild and P. hartzii Gand. (Soreng 1991b) and is characterised by elongate, weakly compressed spikelets and lemmas rounded on the back. Species of subsect. Halophytae (including P. napensis Beetle, P. stenantha Trin., P. unilateralis Scribn. ex Vasey) commonly have distinctly keeled lemmas and sometimes a papilliate epidermis on pedicels and leaf blades (Soreng 1991a, b, 2007).

Our goals in this paper are: 1) to provide a current overview of the taxonomy of P. secunda s.l. and present an up-to-date infraspecific classification including new combinations for the forms often recognised as species and 2) to document what we know of the relationships and hybrid origin of P. secunda s.l. and sect. Secundae. We also provide a review of the cytology of P. secunda s.l. and other species in sect. Secundae.

In the interest of facilitating land-managers, ecologists, plant breeders, seed-storage facilities and collections managers in maintaining the understanding of variations in morphological forms in P. secunda s.l. that are often recognised as species, we here provide varietal names within P. secunda s.l. subspecies. This would also maintain herbarium collections that are organised or understood along A. S. Hitchcock’s taxonomic concepts, in which the taxa can be viewed to correspond to ecotypes or ecologically adapted apomictic lineages within P. secunda. There is extensive intermediacy between these taxa and, thus, the rank of species is viewed as untenable (Marsh 1950, 1952, Kellogg 1985a, b, Soreng 1991b, 1993, 2007, Soreng et al. 2003a, b). RJS’s revised classification of P. secunda s.l. is presented in the Taxonomy section below.

As a whole, P. secunda s.l. is relatively easily split into two subspecies (juncifolia and secunda, Fig. 1), but finer distinctions are often inconstant and overlapping. Within subspecies secunda, it is a futile exercise to attempt to consistently separate P. canbyi, incurva, sandbergii and secunda forms and so these are all included in var. secunda. The epithet gracillima is often misapplied to plants RJS would classify as var. secunda. Both gracillima and scabrella are extreme forms that many botanists seem to think are worthy of taxonomic recognition at some level. Kellogg (1985a, b) concluded that the P. ampla form was the most distinct element within P. secunda s.l., but the difficulty noted in the literature (see Hitchcock et al. 1969) of separating that from subspecies juncifolia s.s., leads RJS to treat it as a variety in the subspecies. Range maps of the two P. secunda subspecies are provided in Fig. 2.

Figure 1. 

Poa secunda habit and panicles: AP. s. subsp. secunda var. secunda (Soreng 9359) BP. s. subsp. juncifolia var. ampla (Soreng 9358) C Panicles of subsp. secunda var. secunda (left) and juncifolia var. ampla (right) (Photos. RJS, Deschutes River near Madras, Jefferson Co., Oregon).

Figure 2. 

Distribution of Poa secunda subspecies in North America: subsp. secunda (yellow) ; subsp. juncifolia (blue).

An exhaustive summary of the nomenclature, protologues and types for the more than 45 names included in P. secunda s.l. is beyond the scope of the present paper, but names of synonyms are listed. Hitchcock’s and some other historical synonymies for the nominal taxa can be found in his widely available "Manual of the Grasses of the United States" (Hitchcock 1951) and on-line: http://tropicos.org. See also the "Catalogue of New World Grasses" (Soreng et al. 2003a, and on-line Soreng et al. 2003b: http://tropicos.org/NameSearch.aspx?projectid=10) for synonymy for the two subspecies accepted.

A compilation of chromosome numbers for Poa secunda s.l. and other species in sect. Secundae (P. hartzii, P. napensis, P. stenantha, P. tenerrima and P. unilateralis) is presented in Suppl. material 1: Table S1. This shows that Poa secunda s.l. and all other taxa in sect. Secundae (for which chromosome numbers are known) are polyploid with a hexaploid base chromosome number of 2n = 42. There are many higher numbers, particularly in P. secunda (2n = 42 to 105 or 106; some 140 counts). Subsp. secunda has a mode at 2n = 84, while subsp. juncifolia has a mode at 2n = 63 (2n = 42 to ca. 100) (Soreng 1991b). Maintenance of a dysploid - nanaploid mode of ca. 2n = 63 in subsp. juncifolia suggests this subspecies is highly apomictic.

Suppl. material 2 provides Keck’s annotations for each of the nominal taxa he accepted that are here included in P. secunda. David Daniels Keck (1903–1995) had a deep interest in Poa secunda s.l. and other western North American species of Poa and, in 1986, he kindly gave RJS his manuscript on those, as well as his extensive specimen annotation lists for Poa of North America. He worked at the Carnegie Institute of Washington, at Stanford University, with J. Clausen and W. Hiesey for more than two decades (up to 1950), on the nature of species. He ceased his work on the lists by 1958 when he retired from the New York Botanical Garden. His lists focused on western Continental United States species, but included some mainly non-arctic Alaskan and Canadian and Mexican (Baja California) records, along with representative records of eastern United States species. Keck’s annotations represent hundreds of historical collections widely distributed in herbaria as vouchers for P. secunda infraspecies. His annotations are considered to be sound by RJS, although here we recognise the taxa at the rank of variety. Copies of the full lists are stored in the reprint files in the Grass Lab in the Department of Botany, Smithsonian Institution. The lists are reproduced here in a semi-searchable form.

The whole section Secundae is shown to be of reticulate origin (Ns plastid / nrDNA genotype combination) consistent with our previous results based on fewer samples (Gillespie et al. 2008, Cabi et al. 2017, Soreng et al. 2017). All sampled taxa of the section have the N plastid genotype marker that is otherwise known only from Europe and SW Asia (sect. Nanopoa J. R. Edm. and unclassified species) and the snrDNA genotype of subg. Stenopoa (Dumort.) Soreng & L.J. Gillespie. While it remains a mystery how the N plastid type came to North America, it is evident that the snrDNA type came from P. subg. Stenopoa (Ss genotype combination), either from sect. Stenopoa or, as we predict, from the primarily western North American sect. Abbreviatae (where P. hartzii was historically placed; see notes in Soreng 1991b) or possibly from the morphologically related Russian P. sect. Kolymenses Prob. (still lacking DNA data). Since N genotypes are otherwise absent from South America and S genotypes are otherwise rare there, we believe Secundae (P. secunda subsp. secunda, subsp. juncifolia and P. stenantha) arrived there secondarily by long distance dispersal from North America.

This ancient hybrid origin, together with its unusual morphology, make P. sect Secundae rather difficult to place within Poa. To better accommodate P. sect. Secundae within the infrageneric classification of Poa, we here raise the section to the rank of subgenus (see Taxonomy section below).

Curiously, Poa cenisia All. and P. psychrophila Boiss. & Heldr. (P. sect. Cenisiae Asch. & Graebn.) also have an Ns genotype combination. This species group is confined to alpine habitats in central and southern Europe and Anatolian Turkey. The taxa are rhizomatous, with extravaginal branching, have a strongly keeled lemma and a dorsal tuft of cobwebby hairs (web) on the callus lemma. Morphological characteristics lead us to hypothesise that this group of species is derived from a cross between Nn taxa of Europe and Ss taxa from P. subg. Stenopoa sect. Stenopoa, whereas we predict Secundae originated from a cross with P. subg. Stenopoa sect. AbbreviataeSs taxa. The N and s genotypes are slightly different between Secundae and Cenisiae, but if they prove to have a common origin, sect. Cenisiae may be better placed within subgenus Secundae than with species they are usually associated with in P. sect. Malacanthae (Roshev.) Olonova (type: P. malacantha Kom.), which exhibit Px genotypes (e.g. P. arctica, P. macrocalyx Trautv. & C.A. Mey., P. smirnowii Roshev.).

Poa hartzii, an Arctic species distributed from Wrangel Island across the North American Arctic to Svalbard, represents an example in Secundae of more recent secondary reticulation. hartzii was shown previously (and again here) to include individuals with two different plastid types (N and S), interpreted as a case of recent and localised chloroplast introgression from P. glauca (Gillespie et al. 1997, 2007, Gillespie and Boles 2001). The sample with S plastid type, clearly has a P. hartziinrDNA type.

Poa arida Vasey, a rhizomatous species of the Great Plains and eastern slopes of the Rocky Mts., was suggested by Soreng (2007) to have arisen from hybridisation of P. secunda and some element of P. sect. Poa. The species appears to have the typical Secundae Ns genotype; it is shown here to have an N plastid genotype, whereas its nrDNA type is likely to be the s type as cloned nuclear DNA genes of P. arida track with P. secunda (Patterson et al. 2005). So, the postulated origin of rhizomes in P. arida deriving from Poa sect. Poa may be wrong, but deeper genetic analyses would be more convincing. Two collections, Soreng & Soreng 6055 and 6107, from Alaska, are viviparous plants identified by RJS as P. arctica × P. stenantha. Both form a clade with P. arctica (sect. Malacanthae) in the plastid tree. In the nuclear tree, while the single sample included (6107, from the Alaska Range) resolves amongst Secundae taxa, it forms a clade with P. unilateralis and these with P. secunda, rather than P. stenantha. Based on geography, the paternal parent could not be P. unilateralis (coastal Washington to California) and is unlikely to be P. secunda (occurs in SE Alaska, but not in the Alaska Range); the only co-occurring Secundae is P. stenantha. Further research is needed, including deeper sampling in P. stenantha to determine the precise origin of this common viviparous Alaskan form.

Given that P. subg. Secundae is an apomictic, polyploid (often with high and odd sets of seven and further dysploid chromosome numbers), hybridising complex of reticulate origin, it is not surprising that the detected phylogenetic structure within the group is mostly not well resolved nor supported. Neither the plastid nor nrDNA trees support the current division into two subsections, Secundae and Halophytae. Several clades that are weakly to moderately supported may be informative of possible changes to the taxonomy. One example is the status of P. macroclada. This taxon was considered as a southern variant of P. stenantha, subsequently tentatively treated as a separate species in P. sect. Stenopoa (Soreng et al. 2003a) and then most recently as a synonym of P. stenantha (Soreng et al. 2003b, Soreng 2007). Our data suggest that it may be a distinct species in subg. Secundae. Such taxonomic changes may be considered in the future, but we feel making changes within Secundae, based on the current phylogenetic trees, to be premature, especially given the small sample set within P. secunda s.l. and the nature of the beast.

RJS thanks Konstantyn Romaschenko for spurring him to write this paper as part of the US National Herbarium Poaceae reorganisation project. We thank Roger Bull for assistance with the molecular research, Alice Tangerini for assistance with artwork on maps and plant illustrations and Carly Casey for assistance with the phylogenetic figure and OCR efforts on Keck’s annotations. Marina Olonova provided some Russian material, Turkish material was collected with the assistance of Evren Cabi and Diane & Dave Bilderback and Kathleen Sayce provided leaf material of P. unilateralis subspecies. LJG acknowledges financial support received from the Canadian Museum of Nature and RJS acknowledges Smithsonian Institution support. We thank Maria Vorontsova and Steven Wagstaff for their helpful reviews.