Slightly mat-forming graminoid herb growing in open tussocks or small stands due to extravaginal, short, branched rhizomes, typically with rhizome branches of 0.5–2 cm between aerial shoots. Aerial shoots ascending from rhizome. Culms usually 10–15(20) cm, stout, erect, smooth and glabrous. Aerial shoots with a few withered leaves forming a loose sheath. Leaves and culms glabrous, lower parts of leaf sheaths, blades and culms usually deep purple (if not growing in shade).

Leaves with keel, flat to moderately folded (convolute), smooth and glabrous. Basal leaves 5–10 mm long, moderately narrow, 1–2.5 mm wide, evenly tapering towards apex. Culm leaves usually 2, similar to basal leaves, blade 2–4 cm long, the flag leaf blade attached near the middle of the culm or higher. Ligula 1–2 mm, obtuse, often fringed.

The units of the inflorescence of Poaceae are the spikelets, nearly always numerous in a panicle or spike-like inflorescence. Spikelets are composed of 2 glumes (bracts for the spikelet) and one or more flowers (the term used below) or rather floral units often named ‘florets’ because we do not know what is the exact flower. A flower or floret is composed of a lemma with 1 mid vein (probably the floral bract), a palea with 2 mid veins (either 2 fused bracteoles or perhaps 2 fused perianth leaves), 3 small organs called ‘lodiculae’ and essential in the opening of the flower at anthesis (possibly transformed perianth leaves or transformed stamens). Stamens 3. Gynoecium of 2 fused carpels with 2 feathery stigmas and one seed.

Inflorescence an open, pyramidal panicle 3.2–5.0(7.0) × 2.0–4.0(4.2) cm, occupying 1/3–1/5 of culm length, with spreading, fairly stiff branches. Panicle with 4–6 nodes with 1–2(3) branches at each of the lower nodes, smooth. Branches long (the longest ones ca. 5–10 mm), usually with 1 spikelet. Spikelets 4–7 × 1.8–2.5 mm, with 2–4 flowers. Bracts (glumes and lemmas) with keels, often fairly sharp. Glumes acute to acuminate, smooth and glabrous, purplish throughout or green at base, with bronze yellow or white, hyaline apex. Lower glume 3–4.5 × 0.6–0.8 mm, usually more than 1/2 the length of spikelet, narrowly lanceolate, with 1–3 veins; upper glume 3.5–5.5 × 0.85–1.0 mm, more than 1/2 the length of spikelet, lanceolate, with 3 veins and a hyaline margin usually less that 1/2 the width of glume. Lemmas 3.7–4.2 × 0.95–1.2 mm, acute, with 3–5 veins, with long, more or less straight hairs on the keel and veins and short hairs between the veins, purplish or green at base and with a broad hyaline margin, often variegated in bronze yellow and white. Paleas shorter than lemmas, with pubescent veins. Anthers shrivelled (non-functional).

Fruit an achene (with one seed).

Asexual reproduction by seeds; restricted and very local vegetative reproduction by rhizome growth. Seeds are probably produced more regularly in ssp. caespitans than in ssp. arctica, suggesting a difference in the balance between reproduction by seeds and by rhizome fragmentation. Subspecies caespitans is also the commoner of the two in disturbed situations along roads, tracks and in scree, i.e., in site types where regular recruitment by seed is needed to establish and uphold populations. Reproduction by bulbils in ssp. caespitans has been stated to occur but has not been observed by us anywhere and is not documented by any convincing specimen.

In some grass genera, Poa among them, there is a fairly good correlation between ploidy levels and modes of reproduction. Diploids (mostly 2n = 14) and tetraploids (mostly 2n = 28) usually have sexual seed-set, whereas higher ploidy levels are often (but not always) asexual with either seed-set without fertilization (agamospermy) or vegetative propagation by bulbils replacing the flowers in the spikelet (vivipary). Poa arctica ssp. caespitans has some local growth by the rhizomes but is usually found as well separated, dense or loose tussocks indicating little or no fragmentation of plants. Seed reproduction is efficient but no pollen development has been observed, i.e., the seed-set is almost certainly fully asexual. However, there are some strange data in the study of Haugen (2000). She studied seed-set in a laboratory situation without and with removal of anthers (emasculation) in a pilot study of a few plants of Svalbard P. arctica subsp. caespitans. She found 24 % seed-set (113 fruits from 465 florets) in control plants (not manipulated) and much more fruits, 41 % (117 fruits from 286 florets), in emasculated plants. That emasculation should stimulate seed-set, even in an agamospermous plant, is unexpected and not explainable; however, the data set is restricted and the difference may be a random effect.

Fruits (inside florets) are probably spread mainly by wind.

Poa arctica s. lat. and the P. pratensis group differ from all other Poa species in Svalbard in their extensive, extravaginal rhizomes, evident even in apparently tussocky individuals when dug up. The rhizomes may be short or long but always run some distance horizontally before bending up and producing aerial shoots. These two species or species complexes also differ from the other Svalbard species of Poa (except for P. alpina) in usually larger and more open panicles with longer branches.

The differences between P. arctica s. lat. and the P. pratensis group are more subtle (see Haugen 2000). The key difference lies in the lemmas: in P. arctica the lemmas have veins with more or less straight hairs and there are short hairs or spinules between the veins; in the P. pratensis group the lemmas have veins with crisped hairs and the lemmas are glabrous and smooth between the veins. Another feature is that the hyaline margin of the lemmas in P. arctica is broad, ca. 1/2 the width of the lemma in the upper part and narrows to less than half the width towards base, in the P. pratensis group it is more narrow, less that 1/2 the width of the lemma the upper part but narrows less towards the base. The panicle branches are smooth in P. arctica but at least slightly scabrous in the P. pratensis group. Less reliable but more easily observable differences are found in the shape of the panicle (triangular in outline with spreading to even retrorse branches in P. arctica, more elongated ovate in outline with more ascending branches in the P. pratensis group), number of branches at the lower nodes in the panicle, 1–2(3) in P. arctica and usually more than 3 in the P. pratensis group, and the number and position of spikelets on branches (one of very few and only distally in P. arctica, several and along the branches in the P. pratensis group).

The differences between Poa arctica ssp. arctica and ssp. caespitans are less pronounced. "Typical" plants of the two are rather different: ssp. caespitans with the very short rhizome branches and dense mat or tussocky growth, broader leaves, stiffer culms, deep purplish culm leaf sheaths and a stouter panicle with stiffer branches and slightly larger spikelets; ssp. arctica with the long rhizome branches and loose mat growth, very narrow leaves, slender culms, mostly green or only purplish culm leaf sheaths and the graceful panicle with smaller spikelets. Haugen (2000) emphasized the following additional characters for separation: In ssp. caespitans panicle branches with usually 1 spikelet, lowest panicle branch 5–10 mm, lower glume 0.6–0.8 mm wide, upper glume 0.8–1.0 mm wide with hyaline margin less that 1/2 of width, lemmas 0.95–1.2 mm wide, and anthers always shrivelled; in ssp. arctica panicle branches with usually 2–more spikelets, lowest panicle branch 7–15 mm, lower glume 0.5–0.7 mm wide (keel to margin), upper glume 0.75–0.9 mm wide with hyaline margin 1/2 or more of width, lemmas 0.85–1.0 mm wide, and anthers usually well developed. These characters are quantitative and overlapping rather than qualitative, and some plants not conforming to this list of differential characters are found (especially in the herbaria where abnormal plants are over-represented). The genetic support for ssp. caespitans, as different from ssp. arctica, is perhaps better than the morphological one (see Haugen 2000).

Stout plants of P. arctica ssp. caespitans may perhaps be confused with P. alpina but are easily recognized by their extravaginal basal branching, their much narrower culm leaves, 1–2.5 mm broad, tapering more gradually towards the apex, and their more open panicle with fewer and more narrow spikelets, 4–7 × 1.8–2.5 mm. In P. alpina all branching takes place inside the sheath cylinder, the leaves are broader and tapering in the very apex, and the panicle is denser with much more numerous and broader spikelets 4–6 × 3–4 mm.

Poa arctica s. lat. occurs in both stable and unstable vegetation covers and on open ground, sometimes as the sole or one of very few vascular plants (e.g., in boulder screes). It is also largely indifferent as to soil moisture, ranging from the tussock level in mires and moderately soaked sediment plains to dry ridges and screes, but with its main occurrences in moist to moderately dry mossy sites. The species is probably not a strong competitor but may sustain (sometimes vegetatively) in very adverse conditions as on coastal cliffs, in over-manured patches in bird colonies, or in deep shade. It is largely indifferent as to soil reaction (pH), present on both acidic and basic substrates.

Of the two subspecies, ssp. arctica is the plant of the moister site types and is the one found in meadows and moss mats, ssp. caespitans is the plant of drier and more often unstable sites types and predominates in screes and on dry gravel slopes. The subspecies do not seem to differ much in soil reaction preferences; both are found on both acidic and basic soils.

Poa arctica s. lat. is largely independent of climate and found in a broad range of site types and probably throughout the islands, i.e., along the entire gradients from the middle arctic to the polar desert zones and from the weakly oceanic to the clearly continental sections. Both subspecies are found more or less throughout the investigated parts of Svalbard (including the outlying islands of Kong Karls Land and Hopen). There are no records from Edgeøya and only one from Barentsøya, however, we suspect the lack of records to be due to under-collecting.

Outside Svalbard Poa arctica s. lat. is one of the most widespread of all arctic plants, present in nearly all areas, and also with extensive ranges in temperate mountains south to S Scandinavia, C Asia, and the southern Rockies. Whereas ssp. arctica may be circumpolar or nearly so (doubts about its range in Beringia), ssp. caespitans is amphi-Atlantic. Confirmed records of ssp. caespitans are from the eastern parts of the Canadian Arctic, Greenland, Svalbard, N Scandinavia, and Novaya Zemlya. No reports from other regions have been confirmed.

Poa arctica s. lat. is a large and much disputed group within northern Poa. The variation in N Canada, Greenland, and arctic Europe has usually been treated as two subspecies (Böcher et al. 1978; Elven et al. 2011, 2022): ssp. arctica and ssp. caespitans. Edmondson (1980) did not accept any races for Europe. Soreng et al. (2003) provisionally accepted ssp. caespitans, with their weaker "acceptance level 2", and Soreng (2007) accepted ssp. caespitans as one of five North American subspecies. Our view (Elven, Haugen) concerning the plants of Greenland and N Europe is that there are two disjunctly different groups (‘arctica’ and ‘caespitans’) without documented signs of transition (even if some incomplete plants collected may be difficult to assign). Additional races and species of a P. arctica group are reported from the Beringian areas, both on the American and on the Asian side.

Poa arctica ssp. caespitans is still disputed and treated differently in Russia, NW Europe, and North America. Tzvelev (1976) synonymized ssp. caespitans with P. tolmatchewii Roshev. and stated it to be "undoubtedly hybridogeneous (or even a modern hybrid: P. arctica × glauca)". This raises two questions: whether the two names are synonymous and whether the hybrid hypothesis can find support. The Svalbard material of this group was studied by Haugen (2000). Morphological investigations demonstrated that ssp. caespitans as described and with type from NE Canada (Ellesmere Island) differs in the majority of the diagnostic characters reported by Tzvelev for P. tolmatchewii. The synonymization was not accepted. Neither did Tzvelev's hybrid hypothesis receive any molecular support. Subspecies caespitans did not show any additive pattern in isoenzymes compared with the parents proposed by Tzvelev (Haugen 2000). Tzvelev's hypothesis may be relevant for the Siberian P. tolmatchewii but not for P. arctica ssp. caespitans (Elven et al. 2011).

Nordic authors (e.g., Nannfeldt 1940; Hylander 1953; Elven et al. 2022) have considered ssp. caespitans a major race of P. arctica. Poa arctica ssp. caespitans seems to be simpler genetically than the main body of P. arctica ssp. arctica, with fewer enzyme multilocus phenotypes (Haugen 2000) and generally at a constant and lower ploidy level (8×, 2n = 56) than the majority of ssp. arctica (9× or higher). Haugen (2000) found little variation in ssp. caespitans, both in morphology and in molecular markers, and a much higher level of variation in both aspects in ssp. arctica. However, the variation found in ssp. caespitans was outside or at the end of the range of that in ssp. arctica in most aspects, i.e., ssp. caespitans seemed to be a rather monomorphic segregate besides the much more polymorphic ssp. arctica. Haugen (2000) concluded that "Poa arctica ssp. caespitans appears uniform and is separated from P. arctica ssp. arctica to such a degree that it deserves to be recognized as a distinct taxon, at least at the rank of subspecies." She found a higher number of morphological features and enzymatic markers to characterize ssp. caespitans than she found separating between P. arctica ssp. arctica and P. alpigena in Svalbard material.

The Svalbard plants of ssp. caespitans are pollen-sterile and agamospermous as are the Canadian plants we have investigated, including the pollen-sterile type specimens. If ssp. caespitans is result of hybridization, it might be a fairly well stabilized and widespread agamospermous offspring from a parental taxon in the P. arctica aggregate and a more low-ploid species. Note that P. glauca, as a proposed parent, is highly polyploid in the High Arctic (mostly 8–10×) and an improbable parent. In Nordic opinion, ssp. caespitans is an acceptable taxon and more distinctly different from ssp. arctica than is the more often recognized Beringian ssp. lanata. If ssp. caespitans is accepted as species, the name P. filipes Lange 1890, based on P. trichopoda Lange 1877 (a homonym), is the relevant one. Lange's illustration, which is original material and could function as a lectotype, clearly shows a plant corresponding to our concept of ssp. caespitans. The genetic data available at present support rank as species, the morphological data rank as subspecies.

Böcher, T.W., Fredskild, B., Holmen, K. & Jakobsen, K. 1978. Grønlands flora. Ed. 3. – P. Haase & Søns Forlag, København.

Edmondson, J.R. 1980. Poa L. – In: Tutin, T.G. et al. (eds.), Flora Europaea. 5. Alismataceae to Orchidaceae (Monocotyledones): 159–167.

Elven, R., Bjorå, C.S., Fremstad, E., Hegre, H. & Solstad, H. 2022. Norsk flora. Ed. 8. Det Norske Samlaget, Oslo.

Elven, R., Murray, D.F., Razzhivin, V. & Yurtsev, B.A. (eds.) 2011. Annotated Checklist of the Panarctic Flora (PAF). http://panarcticflora.org/

Haugen, J. 2000. Possible hybrid origins of Poa hartzii and Poa arctica ssp. caespitans (Poaceae) investigated by morphology and isoenzymes. – Cand. scient. Thesis, Univ. Oslo, Oslo.

Hylander, N. 1953. Nordisk Kärlväxtflora, I. – Almqvist & Wiksell, Stockholm.

Nannfeldt, J.A. 1940. On the polymorphy of Poa arctica R. Br., with special reference to its Scandinavian forms. – Symbolae Botanicae Upsaliensis 4(4). 86 pp.

Soreng, R.J. 2007. Poa L. – In: Flora of North America Editorial Committee (eds.), Flora of North America north of Mexico 24. Magnoliophyta: Commelinidae (in part): Poaceae, part 1: 486–601.

Soreng, R.J., Peterson, P.M., Davidse, G., Judziewicz, E.J., Zuloaga, F.O., Filgueiras, T.S. & Morrone, O. 2003. Catalogue of New World grasses (Poaceae): IV. Subfamily Pooideae. – Contributions of the U.S. National Herbarium 48. 730 pp.

Tzvelev, N.N. 1976 [1977]. Zlaki SSSR. – Nauka, Leningrad.