Survey of the vascular plants of Alert (Ellesmere Island, Canada), a polar desert at the northern tip of the Americas

Long-term monitoring is critical to guide conservation strategies and assess the impacts of climatic changes and anthropogenic activities. In High Arctic ecosystems, information on distribution and population trends of plants is dramatically lacking. During two field expeditions in 2018 and 2019, we conducted a systematic floristic survey together with opportunistic collecting in the polar desert surrounding Alert (Nunavut, Canada) to update past vascular plant inventories. We recorded 58 species, of which 54 species were recorded over the last seven decades, and four species that are additions to the local flora (Draba pauciflora R. Brown, Festuca edlundiae S.G. Aiken, Consaul, & Lefkovitch, Festuca hyperborea Holmen ex Frederiksen, and ×Pucciphippsia vacillans (T. Fries) Tzvelev). With the addition of 19 species that were previously reported but not found in our survey, we estimate the species richness in the study area at 77 species.


Introduction
Polar deserts are among the most extreme environments in the world. These seemingly barren landscapes are characterized by a short growing season, low annual and summer mean temperatures, low precipitations, severe limitation in nutrients, and soil disturbance due to freeze-thaw cycles (Billings 1987;Peterson 2014). Because of these important environmental constraints, High Arctic ecosystems support persistent but sparse and short vegetation Lévesque 1997;Peterson 2014;Zwolicki et al. 2020). These ecosystems are expected to be the most impacted by the ongoing climate change due to Arctic amplification, which refers to the greater climatic changes occurring near the North pole compared to the rest of the globe (Smith et al. 2019). With the steep climate warming occurring over the last decades (Meehl et al. 2007), Arctic regions have already experienced changes in vegetation composition, biomass, and diversity (Elmendorf et al. 2012a(Elmendorf et al. , 2012bRavolainen et al. 2020). Therefore, establishing ecological baselines and updating past inventories are critical scientific endeavors to monitor, predict, and manage current and future impacts of climate change (Mihoub et al. 2017). Despite the remoteness and considerable logistical constraints associated with working in the High Arctic, there is a long history of botanical collecting at High Arctic research stations, particularly in the mid-20th century (Panchen et al. 2019). However, current collections from some High Arctic locales are lacking and synthesized, published data regarding plant biodiversity in polar deserts remains scarce (CAFF 2013).
Located at the northern tip of Ellesmere Island in the Canadian Arctic Archipelago, only 817 km from the North Pole, Alert is the northernmost permanently inhabited settlement on Earth. Here, an Environment and Climate Change Canada weather station operates year-round, and a Canadian Forces Station (CFS Alert) was established in 1958 (Johnson 1990). The Alert region is characterized by continuous permafrost and is one of the most arid places in the Northern Hemisphere (Parks Canada 1994). Alert occurs in Circumpolar Arctic Bioclimate Subzone A, which is the coldest bioclimate subzone in the Canadian Arctic, with an average July temperature of 3°C (CAVM Team 2003). This subzone landscape is mostly barren, with some lichen and moss cover, and a <5% vascular plant cover (CAVM Team 2003;Christensen et al. 2013).
The history of vascular plant collecting in the Alert region began with the British Arctic Expedition in 1875-1876, when H.W. Feilden, the expedition's naturalist, made the first known vascular plant collections from the site (Bruggemann and Calder 1953). Later, during R. Peary's expeditions to the North Pole in 1905-1906 and 1908-1909, plants were collected by expedition members L.J. Wolf, R.A. Bartlett, and J.W. Goosdell (Bruggemann and Calder 1953). It was only during the establishment of the weather station at Alert in 1950 that the next plant collections occurred. From April to September 1950, an airstrip mechanic, J.P. Johnson was sent to Alert, where he explored the area and collected various samples (Johnson 1990), including vascular plants deposited at the National Herbarium of Canada at the Canadian Museum of Nature (CAN; herbarium acronyms follow Thiers 2020). In subsequent years, collections of the local flora greatly expanded. From April to September 1951, P.F. Bruggemann collected insects and carried out botanical investigations for the Northern Insect Survey at and around Alert, at Mt. Grant in the United States Range, and at the Wood River, located 40 km and 65 km west of Alert, respectively (Bruggemann and Calder 1953). During the botanical survey, Bruggemann was accompanied by S.D. MacDonald of the National Museum of Canada (the precursor to the Canadian Museum of Nature); MacDonald deposited vascular plants of his own collecting at the National Herbarium of Canada (CAN). Bruggemann deposited his collection at National Collection of Vascular Plants at Agriculture and Agri-Food Canada (DAO) and later summarized his own 1951 vascular plant inventory with the addition of the previous regional surveys from Feilden in 1875-1876, Bartlett in 1908, and Polunin in 1940(Bruggemann and Calder 1953. In this paper, he also estimated that 65 species of vascular plants were present in the eastern Canadian Arctic north of 82°N. In 1952, P. Gadbois and C. Laverdière, from Université de Montréal, were sent to Alert by the Canadian Department of Mines and Technical Surveys to perform geographical surveys and mapping of the region (Gadbois and Laverdière 1954). They made several vascular plant samplings deposited at the Marie-Victorin Herbarium (MT) (Gadbois and Laverdière 1954). During the International Geophysical Year from 1957Year from to 1959nologists, earth science researchers, and biologists were sent to Alert. Among them, C.R. Harington went back to Alert in 1959, where he also collected many plant specimens (GBIF 2020). Following this, K.V. Pilon in 1975, D. Drew in 1976, and R. Pomerleau in 1983 vascular plant specimens at Alert that they deposited at the University of British Columbia Herbarium (UBC), Canadian Museum of Nature (CAN), and Louis-Marie Herbarium (QFA), respectively (GBIF 2020).
Here, we report the results of a systematically conducted vascular plant survey supplemented with opportunistic collecting in the polar desert surrounding Alert in 2018-2019. This provides a current inventory of the vascular plants, and a temporal snapshot of current species diversity to inform continued botanical work at the station.

Study Area
Our study area surrounds CFS Alert (82°30′N, 062°20′W) and is located on the north-eastern tip of Ellesmere Island, Nunavut, Canada (Fig. 1). It is roughly delimited by the Lincoln Sea to the north and the boundaries of CFS Alert property in other directions. The study site encloses an area of 170 km 2 , including five lakes (White Pond [0.12 km 2 ], Self Pond [0.25 km 2 ], Kirk Lake [0.58 km 2 ], Lower Dumbell Lake [0.94 km 2 ], Upper Dum bell Lake [1.18 km 2 ]), three bays (Ravine Bay [3.01 km 2 ], Colan Bay [3.49 km 2 ], Dumbell Bay [4.65 km 2 ]), and one inlet (Alert Inlet [1.09 km 2 ]). The local topography consists of rugged and undulating terrain with mountains (to a maximum height of 525 m a.s.l.), hills, valleys, and creeks. The surficial deposit ranges from 2.4 to 4 m thick and is mostly composed of till (sandy clay loam) and shattered rock filled with ice (Taylor et al. 1982;Tam 2014). The underlying bedrock is highly calcareous, composed of argillite with greywacke in some places, and the permafrost is >600 m thick (Smith et al. 2012). Because of isostatic rebound, the post-glacial marine limit is estimated at 135 m a.s.l (England 1976) and many beds of marine silts and clays occur, containing masses of recent marine shells (Bruggemann and Calder 1953). No glacial striae, trough-shaped valleys, or cirques are observed at Alert, and the present topography has been shaped entirely by water erosion and frost action (Bruggemann and Calder 1953). Alert is free of the Innuitian Ice Sheet since ~9.6 ka calibrated BP (Dalton et al. 2020). The uplands are mostly mesic, xeric, or barren, and consist mainly of boulder, frost-shattered rock, gravel, and polygonal nets of till, with very low vegetation cover growing inside soil interstices. In the lowlands, where some soil moisture accumulates, a more continuous vegetation cover develops, consisting primarily of grasses and sedges (Bruggemann and Calder 1953).
At Alert, a 24-h direct sunlight period occurs from early April to early September, while the sun remains under the horizon from mid-October to late February. July temperatures average 3.4 °C and annual snowfall and rainfall average respectively 184.6 and 1.7 cm (corresponding to a combined water equivalent of 158 mm; Government of Canada 2010). The ground is generally covered with snow from early September to mid-June, leaving only about 2.5 months per year for plant growth (E Desjardins, S Lai, and D Berteaux personal observations).

Methods
Data collection. The floristic study was carried out during two field expeditions between 26 June and 24 August 2018, and 25 June and 4 September 2019. We conducted a systematic survey, which involved a random stratified design, as described below. The survey was supplemented with opportunistic sampling, collecting vascular plants of interest along marked trails and while traveling within the study area.
Selection of survey locations. We selected systematic survey locations based on a habitat map (Appendix 1) constructed from a multispectral satellite picture of the study area (Fig. 1). The habitat map consisted of six classes, including snow/water and five broad vegetation gradients classified into habitat types, namely barren ground, xeric, xeric-mesic, mesic, and wetland (Fig. 2). A 2 × 2 km grid was overlaid on the study area and, for each grid cell, we performed vegetation surveys at one or more random locations for each of the five habitat types present in the grid (Fig. 3A). ArcGIS 10.6.1 (ESRI 2018) was used to complete these steps.
Field surveys. Each vegetation survey corresponded to a vegetation plot consisting of five 1 m × 1 m quadrats, each located 5 m from a central point and at equal distance from one another (Fig. 3B;Bay 1998). All species inside the quadrats (Fig. 3C) were identified and recorded (Fig. 3D). If a species was opportunistically encountered at a surveyed site but out of a quadrat, specimens were collected to contribute to our floristic survey.
Specimen collection. The first flowering (or fruiting) specimens encountered of each species (in or outside the vegetation plots) were collected and dried in a standard plant press. Specimens were identified subsequently by the authors (ED, SL, SP, MD, PS, AS, JL, and LS) using keys from the Flora of the Canadian Arctic Archipelago ), Flora of North America North of Mexico (Flora of North America Editorial Committee 1993, 2005, 2006, 2009, Flore nordique du Québec et du Labrador (Payette 2013(Payette , 2015(Payette , 2018, Vascular Plants of Continental Northwest Territories, Canada (Porsild and Cody 1980), The Flora of Canada (Scoggan 1978a(Scoggan , 1978b(Scoggan , 1979, and other taxonomic literature (Aiken et al. 1995;Aares et al. 2000;Brysting and Elven 2000;Consaul and Gillespie 2001;Harris 2006;Consaul et al. 2008aConsaul et al. , 2008bElven and Al-Shehbaz 2008;Al-Shehbaz and Mulligan 2013;Saarela et al. 2020;Solstad and Elven, unpublished). Collected specimens were also compared to those conserved at the Louis-Marie Herbarium (Université Laval, Quebec City, Quebec, Canada). Species names were updated based on the Database of Vascular Plants of Canada (VASCAN) (Brouillet et al. 2010+;Desmet and Brouillet 2013) and synonyms used in Saarela et al. 2020 (and other current Arctic floristic treatments; Elven et al. 2011) are also indicated. These specimens are deposited at the Louis-Marie Herbarium (QFA), with duplicates deposited at the Canadian Museum of Nature Herbarium (CAN).

Vascular plant descriptions and comparisons.
Descriptions of species include their plant habit and distinctive characteristics and are roughly organized according to the template provided by Aiken et al. (2007). All size measurements and character states (e.g., pubescence, shape, color) represent specimens collected at Alert. Species descriptions are based on at least 10 specimens of various sizes, with the exception of Draba pauciflora R. Brown and Pedicularis hirsuta Linnaeus, for which only three specimens were available. In the species comparisons, we indicate the characteristics which, based on literature sources, are important to distinguish a given species from congeneric species found at Alert or elsewhere on Ellesmere Island. We also indicate when our collected specimens present characteristics that do not fully correspond to those reported in published floras. Botanical terms used in the descriptions and comparisons follow Beentje (2016) and Payette (2013Payette ( , 2015Payette ( , 2018.

Vascular plant habitats.
The main habitat of each species was determined as follows. For species recorded in vegetation plots, we considered the species' main habitat as the habitat with the highest frequency of occurrence (highest number of vegetation plots). For species recorded only opportunistically (i.e., outside vegetation plots), we copied the GPS location of the collection on our habitat map to identify the species habitat.

Results
We conducted a total of 264 vegetation surveys, hence systematically inventorying 1,320 quadrats of 1 m 2 each. The xeric and xeric-mesic habitats were too similar to be distinguished in the field. We, therefore, merged them in the xeric class for further analyses, yielding four main vegetated habitat types (Table 1).
In Table 1, we present the species found during our two field seasons and indicate if each taxon was recorded by previous collecting expeditions. Occurrences from Table 1. List of species found in the study area (Alert, Ellesmere Island, Nunavut, Canada) based on 2018-2019 inventory, including specimens accession numbers at the Louis-Marie Herbarium (QFA) (Québec City, Québec, Canada) and Canadian Museum of Nature Herbarium (CAN) (Ottawa, Ontario, Canada). Previous literature records for each taxon from Bruggemann and Calder (1953) and occurrences recorded by other collectors found in GBIF (2020) are also indicated, along with the main habitat, and the proposed status by IUCN Red List of Threatened Species (2020) and NatureServe (2020). Names are based on VASCAN (Brouillet et al. 2010+;Desmet and Brouillet 2013), while synonyms used in Saarela et al. 2020 andElven et al. 2011 are also indicated. A star symbol (*) following a name indicates an Arctic endemic species (Daniëls et al. 2020). Habitat: B = barren ground, X = xeric, M = mesic, W = wetland. IUCN status: NA = not available, LC = least concern. NatureServe status: NA = not available, NSR = no status rank, S = secure, AS = apparently secure, H = hybrid, I = imperiled. The NatureServe status is indicated separately for Canada (Can) and Nunavut (Nu) when they differ.
Characteristics allowing to distinguish B. thorild-wulffii subsp. thorild-wulffii from the other Braya species on Ellesmere Island can be found under B. purpurascens.
Draba micropetala and D. pauciflora both have narrow petals (≤2 mm wide) compared to all the other yellow-flowered Draba species on Ellesmere Island (D. alpina, D. arctogena, D. oblongata, and D. simmonsii; Aiken et al. 2007;Al-Shehbaz et al. 2010b;GBIF 2020). In addition, pedicels are usually shorter (1-4 mm long) in D. micropetala than in D. alpina and D. simmonsii, which have 2.5-30.0 mm long pedicels (Al-Shehbaz et al. 2010b). D. micropetala and D. pauciflora can usually be distinguished on other characters: D. micropetala has leaves with an obtuse to rounded apex, whereas D. pauciflora has leaves with an acute apex; the leaves of D. micropetala have a predominance of cruciform hairs on the leaves, whereas those of D. pauciflora have a simple and forked hairs; and D. micropetala has ovateelliptic, 2.0-3.2 mm wide fruits, whereas D. pauciflora has obovate, 3-5 mm wide fruits (Elven and Al-Shehbaz 2008;Saarela et al. 2020). The difference in the fruits did not fit exactly with our specimens from Alert. Instead, we found mainly obovate or oblanceolate, 2.7-3.5 mm wide fruits in D. micropetala, and lanceolate or elliptic, 1.8-4.0 mm wide fruits in D. pauciflora. However, based on the specimens from Alert, we found that using the dimensions of the styles and stigmas persisting on the fruits were good additional criteria to differentiate the two species; D. micropetala had shorter styles (0.05-0.20 mm long) with stigmas as wide as the styles (or very slightly wider), whereas D. pauciflora had longer styles (0.2-0.4 mm long) with stigmas always wider than the styles. As for the other yellow-flowered Draba found at Alert, D. corymbosa, it had even longer styles (0.4-0.7 mm long) and the stigmas were even more distinctly wider than the styles (the two structures appearing like a T).

Cerastium arcticum Lange
Cerastium Linnaeus species can be differentiated from Arenaria Linnaeus and Stellaria Linnaeus species by the petals: Cerastium species have petals cleft at most to 25 %, whereas Arenaria species have unlobed petals and Stellaria species have petals cleft nearly to the base . C. arcticum is very similar to C. alpinum Linnaeus subsp. alpinum and subsp. lanatum (Lamarck) Cesati, C. beeringianum Chamisso & Schlechtendal, and C. bialynickii Tolmatchew that are present on Ellesmere Island (GBIF 2020), but a few characters help differentiate them. C. arcticum has leaves with hairs that are always >0.3 mm long, whereas both C. alpinum subspecies have long but also short hairs (<0.3 mm long); bracts have more distinct hyaline margins in both C. alpinum subspecies (0.3-0.8 mm wide) than in C. arcticum (≤0.3 mm wide; Blondeau 2015a). As for the differences between C. beeringianum and C. arcticum, hairs on leaves are longer in C. arcticum (≥0.9 mm long) than in C. beeringianum (≤0.9 mm long); bract leaves of C. arcticum (4-9 mm long) are longer than those of C. beeringianum (2.5-3.0 mm long); and fruits of C. arcticum are also larger (3-5 mm wide) than those of C. beeringianum (2-3 mm wide; Aiken et al. 2007;Blondeau 2015a). C. bialynickii has a pulvinate growth form with short stems (1-10 cm long), whereas C. arcticum, both C. alpinum subspecies, and C. beeringianum have rhizomatous or mat-forming, not pulvinate growth form (cushion-like) with longer stems (5-50 cm long; Morton 2005).

Sabulina rubella (Wahlenberg) Dillenberger & Kadereit
Characteristics allowing to distinguish S. rubella from the other Sabulina species present on Ellesmere Island can be found under S. rossii.
Four Eriophorum Linnaeus species are present on Ellesmere Island (GBIF 2020). Only E. triste and E. angustifolium Honckeny subsp. angustifolium have more than two spikelets, whereas E. scheuchzeri and E. callitrix Chamisso have a solitary spikelet (Saarela et al. 2020). E. scheuchzeri differs by having rhizomes and ≤7 empty proximal scales, whereas E. callitrix is caespitose (i.e., without rhizomes) and has usually ≥10 empty proximal scales (Saarela et al. 2020). In addition, two subspecies of E. scheuchzeri are found on Ellesmere Island (GBIF 2020) and can be differentiated according to the following characters: subsp. arcticum has spherical spikelets, whereas subsp. scheuchzeri has hemispherical spikelets; proximal fertile scales of subsp. arcticum are darker and gradually passing to paler grey tones with conspicuous hyaline margins, whereas proximal fertile scales of subsp. scheuchzeri are completely dark with dark margins or reduced, sharply differentiated hyaline margins (Saarela et al. 2020). Lower bracts shorter than the inflorescence; 13-24 mm long. Peduncles 5-18 mm long; scabrous all around the surface. Lateral spikelets and terminal spikelet bisexual, with both sexes in each floret; pedunculate; erect or pendent. Proximal scales 3.8-5.3 mm long, 2.0-3.1 mm wide; ovate or lanceolate; greyish black sometimes purple or red-tinged, with or without narrow, hyaline margins at the apex; apices acute or acuminate. Perianth represented by bristles; white or yellowish-brown. Androecium with 3 stamens and 1.9-3.8 mm long anthers. Gynoecium with 1 style and 3 stigmas. Fruit an achene; 2.0-2.4 mm long, 0.9-1.9 mm wide; obovoid; keeled; glabrous.
Eriophorum triste is similar to E. angustifolium subsp. angustifolium by having more than two spikelets, but can be differentiated by culms, peduncles, scale, and anther characters: E. triste has shorter culms (≤30 cm) than E. angustifolium (20-100 cm; Ball and Wujek 2002); E. triste has scabrous, arcuate, ≤2 cm peduncles, whereas E. angustifolium has glabrous (or scabrous on angles), drooping, ≤10 cm long peduncles (Ball and Wujek 2002;Saarela et al. 2020); E. triste has grey to black scales with or without hyaline margins, whereas E. angustifolium has brownish grey, greyish, reddish or ferruginous scales with broad hyaline margins (Ball and Wujek 2002;Saarela et al. 2020); anthers of E. triste are usually shorter (1.8-3.0 mm long) than those of E. angustifolium (2.5-5.0 mm long; Saarela et al. 2020); and fruits are obovoid, 2.0-2.5 mm long in E. triste, whereas they are oblong-obovoid or oblong-elliptical, 2.5-3.5 mm long in E. angustifolium (Saarela et al. 2020). The culm height from the specimens of Alert (9-16 cm), as well as the peduncle length and scabrousity, and the fruit dimensions and shape correspond with those reported in the literature, but not the scale color and the anther length. The scales were not only grey to black, but some scales also had a reddish-purple tinge in the center. The anthers were also longer than the maximum anther length known in E triste. Only E. triste has previously been found at Alert and the surrounding areas (GBIF 2020). Identification. Plants 2-7 cm high; herbaceous; not caespitose. Fibrous roots and rhizomes present. Rhizomes black or reddish-brown. Vegetative stems 2-7 cm long; annual; ascending or decumbent; green or yellowish-green; simple or branched from the whorls (branches ascending). Leaves reduced to fused sheaths in whorls. Cauline sheaths 0.5-2.0 mm long; yellowish-green proximally and purplish-brown apically, and sometimes with hyaline margins; ending with 4-8 teeth. Cauline teeth lanceolate; apices acute. Rameal sheaths 0.3-0.8 mm long; attenuate; green and sometimes brown apically; ending with 3 or 4 teeth. Teeth triangular or lanceolate; apices acute. Sporiferous stems annual; erect; green or yellowish-green; simple. Strobilus 4.5 mm long, 2 mm wide.

Equisetaceae -Horsetail family
Three Equisetum Linnaeus species are present on Ellesmere Island, although E. pratense Ehrhart has only been found once near Eureka (GBIF 2020). Equisetum arvense and E. pratense both have vegetative stems bearing whorls of branches, whereas the vegetative stems are unbranched in E. variegatum (Hauke 1993). Equisetum arvense has ascending branches with attenuate, rameal sheath teeth, whereas E. pratense has spreading branches with deltate, rameal sheath teeth (Hauke 1993).  8 Aug. 2019; habitat: wetland on the margins of a small river, with peat and till as substrates, dominated by Alopecurus magellanicus and moss; QFA0634993. Identification. Plants 2-10 cm high; herbaceous; not caespitose. Fibrous roots and rhizomes present. Rhizomes black, purplish brown, or orange. Vegetative stems 2.5-9.0 cm long; evergreen; ascending or decumbent; green or yellowish-green; simple. Leaves reduced to fused sheaths in whorls. Cauline sheaths 1-3 mm long; green or yellowish-green proximally, with an apical broad black band, and white margins; ending with 4-8 teeth. Teeth triangular or lanceolate; apices acute or awned. Sporiferous stems not seen at Alert.
Deschampsia brevifolia and D. sukatschewii (Poplav skaja) Roshevitz are the two Deschampsia Palisot de Beauvois species on Ellesmere Island GBIF 2020). They can be differentiated by the following: D. brevifolia has dense panicles with strongly imbricate spikelets, whereas D.

Festuca hyperborea Holmen ex Frederiksen
Characteristics allowing to distinguish F. hyperborea from the other Festuca species present on Ellesmere Island can be found under F. baffinensis and F. brachyphylla subsp. brachyphylla. Figures 11E, 12D Materials examined. CANADA -Nunavut • Ellesmere Leaf cross section in F. hyperborea (C) usually shows even less welldefined ribs than F. brachyphylla subsp. brachyphylla (which is not the case here) but retains the same looking midrib. Compared to these two species, F. edlundiae (B) presents midrib grooves that are deeper and go beyond the vein level while the other lateral ribs are also consistently well-defined. In F. viviparoidea subsp. viviparoidea (D), the grooves are shallow with well-defined but still small ribs, giving that species unique anatomical facies among Festuca species in Canada (M Dubé personal observation). Ribs, identified by R letters, are the longitudinal bumps containing veins and are delimited by the contiguous grooves (however, some veins may not be included within the ribs). erect; glabrous and scabrous on angles below the inflorescence. Leaves basal and cauline. Sheath margins glabrous or hairy, with short simple hairs. Ligules 0.2-0.5 mm long. Basal leaf blades 12-25 mm long, 0.3-0.7 mm wide; linear; folded or rolled in bud; abaxial surface glabrous or scabrous; adaxial surface scabrous; margins glabrous or scabrous. Flag leaf blades 10.9-11.2 mm long, 0.2-0.3 mm wide; tip linear. Inflorescence a dense panicle, with bulbils allowing vegetative reproduction; 26-30 mm long. Branch at lowest inflorescence node 1; 1.5-2.2 mm long. Pedicels scabrous. Spikelets 10-14 mm long, 1.1-2.0 mm wide. Florets per spikelet 2 or 3. First glumes 2.8-3.1 mm long; lanceolate; surface glabrous or hairy, with hairs at the apices only; margins ciliate; apices acuminate. Second glumes 3.9-4.0 mm long (shorter than the lowest lemma); lanceolate; veins 3; surface glabrous and scabrous apically; margins ciliate; apices acute. Lemmas 4.1-6.7 mm long, 0.6-0.8 mm wide; lanceolate; keeled; veins 5; surface dull, sparsely scabrous, and hairy apically (on the projection), with short hairs; apices acuminate and glabrous; awnless. Palea vestigial or absent. Rachilla absent. Bulbils 3.3-9.6 mm long. Androecium and gynoecium absent.

Phippsia algida (Solander) R. Brown
Most authors recognize two Phippsia (Trinius) R. Brown species, P. algida and P. concinna (Th. Fries) Lindeberg (Steen et al. 2004;Aiken et al. 2007;Consaul and Aiken 2007), although Soreng and contributors (2003) treated the latter taxon as Phippsia algida subsp. concinna (Th. Fr.) Á. Löve & D. Löve; both taxa are known from Ellesmere Island. The most reliable characters to differentiate the species are the number of stamens (1 or 2 stamens in P. algida and 1 stamen in P. concinna); caryopsis shape (ellipsoid in P. algida and ovoid in P. concinna); pedicel angle (5-8° in P. algida and 20-108° in P. concinna); lemma hairiness (hairs on lower 1/3 or entirely glabrous in P. algida and hairs on the 1/2-2/3 in P. concinna); and lemma hair length (0.02-0.15 mm long in P. algida, and 0.19-0.24 mm long in P. concinna; Aares et al. 2000). In addition, P. algida has spikelets that are less than twice as long as wide, whereas P. concinna has spikelets 2-3 times longer than wide . We found that the lemmas of the P. algida specimens from Alert had longer hairs (up to 0.17 mm long) on a larger surface (½ of the surface) than reported in the literature.
Phippsia algida can also be mistaken for the hybrid ×Pucciphippsia vacillans, which is also known from Alert and is somewhat intermediate between Phippsia algida and Puccinellia vahliana (Hedberg 1962;Steen et al. 2004). They both have very short glumes (≤0.8 mm long; Aiken et al. 2007), but ×P. vacillans differs from P. algida by having longer leaf blades (≥30 mm long), more than one floret per spikelet, and panicles with predominantly ascending branches, not erect as in P. algida ).
Characteristics allowing to distinguish ×P. vacillans from Phippsia can be found under P. algida.
At Alert, P. pulchella is highly polymorphic in terms of hairiness. Most individuals have silvery-white appearance due to dense hairs on leaves, whereas few individuals are sparsely hairy (Fig. 15D, E). A molecular study of this species on Svalbard has shown that there is nearly no genetic (RAPD) variation among subpubescent and pubescent plants (Hansen et al. 2000). Instead, pubescence variation is phenotypic and is associated with different abiotic conditions: pubescent plants growing on cliffs, ridges, scree slopes, and silt shore terraces, and subpubescent plants on gravel shore terraces (Hansen et al. 2000). Figure 16A  long anthers. Gynoecium with hairy (white hairs) pyriform ovaries (2.2 mm long, 1 mm wide), 1.1-1.3 mm long styles, and 0.3-0.6 mm long stigma lobes.

Salix arctica Pallas
Salix arctica and S. arctophila Cockerell ex A. Heller are the only Salix Linnaeus species present on Ellesmere Island (GBIF 2020) and can be distinguished based on the following: the largest medial leaf blade abaxial surface is glabrous in S. arctophila, whereas it is usually pilose in S. arctica (sometimes the midrib is sparsely short-silky and the apex is long-silky bearded; Saarela et al. 2020); leaf blade margins are serrulate or crenulate (sometimes entire) in S. arctophila, whereas the margins are entire in S. arctica (Saarela et al. 2020); and the ovary hairs are white and rust-couloured, appressed, crinkled, and ribbon-like in S. arctophila, whereas the hairs are only white, flattened, and wavy (not crinkled) in S. arctica ).

Saxifraga tricuspidata Rottbøll
Characteristics allowing to distinguish S. tricuspidata from the other Saxifraga species present on Ellesmere Island can be found under S. cernua.

Discussion
The total of 58 species recorded during this survey (including 15 Arctic endemics) is a representative number of taxa for a polar desert ecosystem (Aleksandrova 1980;Bay 1992). This is also within the range of vascular plant species richness found in other areas on or near Ellesmere Island. For example, from south to north, 85 species were found at Alexandra Fiord (Muc et al. 1989), 40 in the central part of Ellesmere Island (Lévesque 1997), 75 at Sverdrup Pass (Bergeron andSvoboda 1989), 151 at Quttinirpaaq National Park (Parks Canada 1994), and 36 at Ward Hunt Island (Vincent et al. 2011). Part of the variation among surveys likely reflects the size of the surveyed area, as well as environmental conditions and landscape heterogeneity (e.g., topography, hydrology, substrate, and microclimate; Ravolainen et al. 2020;Taylor et al. 2020).
Among the 14 families recorded, the Poaceae (18 species) accounted for 32% of the inventoried vascular plant species. The predominance of this family is higher than in other High Arctic ecosystems, notably Ward Hunt Island (11% with 4 Poaceae species; Vincent et al. 2011), North of Greenland (20% with 19 Poaceae species; Bay 1997), Svalbard (14% with 25 Poaceae species; Svalbard Flora 2020), and Cape Chelyuskin in Russia (22% with 10 Poaceae species; Matveyeva and Chernov 1976). Poaceae are early successional species in some Arctic ecosystems and they are most abundant within the first 20 years of colonization of a new environment, including bare soils and enriched or disturbed areas (Forbes and Jefferies 1999;Forbes et al. 2001;Cray and Pollard 2015). This relatively high diversity in Poaceae for a polar desert ecosystem may reflect perturbations related to land use, soil contamination, and operation of heavy vehicles on the soft ground during snowmelt in early summer near CFS Alert.
We recorded four species that were new for the Alert region, but that had been already reported on Ellesmere Island by other collectors (distance and direction from nearest previous records are indicated in parentheses): Draba pauciflora (Grant Land Mountain, ca. 40 km W; Kershaw s.n. (DAO6935)), Festuca edlundiae (Tanquary Fiord, ca. 300 km WSW; Aiken 94-016E (CAN10012792)), Festuca hyperborea (Lake Hazen, ca. 140 km WSW; So ko loff & McMullin 951 (CAN10064714)), and ×Pucciphippsia vacillans (Craig Harbour, ca. 790 km SSW; Malte s.n. (ALTA-VP33562)). For Festuca edlundiae, this record at Alert corresponds to an extension of the known northern distribution limit in the Canadian Arctic from 82°25′N to 82°31′N (GBIF 2020). Similarly, the record of the hybrid ×Pucciphippsia vacillans extends its known northern distribution limit from 76°12′N to 82°31′N in the Canadian Arctic (GBIF 2020). Draba pauciflora and Festuca hyperborea have been found previously at the same latitude or higher on Ellesmere Island (GBIF 2020). Draba pauciflora and Festuca hyperborea are rare and were found only in a few plots (one and five plots, respectively), which may explain why they had not been found previously. ×Pucciphippsia vacillans was more abundant (found in 20 plots), but they may have been mistaken for Phippsia sp. in the past. Festuca edlundiae was relatively recently separated from other known Festuca species (Aiken et al. 1995), and the re-identification of existing specimens from Alert may uncover more records of this species.
Despite our intensive systematic search, we did not find 21 species that had been previously recorded within the study area (GBIF.org 2020). It is possible they were located in microhabitats that we missed while focusing more on systematic than opportunistic sampling.
Draba alpina, D. cinerea, D. fladnizensis, and D. ni va lis were all previously recorded for Alert by Bruggemann and Calder (1953). However, we were not able to access these records for confirmation in any online database or at the National Collection of Vascular Plants (DAO). Draba lactea and D. oblongata were not reported in the initial paper of Bruggemann and Calder (1953); they were since re-identified at DAO (D. lactea: Bruggemann s.n. (DAO08985) and (DAO08986)). However, we were not able to verify sheets of D. oblongata R. Brown ex de Candolle (Bruggemann s.n. (DAO10096) and (DAO10097); Harington s.n. (ALTA-VP9754)), reported in Aiken et al. (2007). Draba simmonsii was reported from the station by Elven and Al-Shehbaz (2008) as a paratype of this newly described High Arctic taxon. This specimen was previously determined as D. alpina (Mac-Donald 37 (CAN10056945)). We did not encounter D. simmonsii in 2018-2019.
Neither Cerastium alpinum Linnaeus (Bruggemann and Calder 1953) nor C. beeringianum (Harington 81 (CAN10046052)) were encountered during our 2019-2019 field seasons. Cerastium alpinum and C. arcticum were previously differentiated within a C. alpinum-C. arcticum complex or lumped into a widely defined C. alpinum (Hultén 1956;Böcher 1977;Brysting and Elven 2000). The identification of all three species is particularly problematic in the Arctic because they overlap in most characters and a combination of characters is required to differentiate them unambiguously (Brysting and Elven 2000). Silene sorensenis was previously recorded at Alert by Bruggemann and Calder (1953), MacDonald (MacDonald 23 (CAN10047218)), and Harington (Harington 112 (CAN10047213)). We did not encounter this taxon ourselves but accept the two species records at CAN based on examination of online specimen images.
Poa hartzii Gandoger subsp. vrangelica var. vivipara Polunin was recorded by Bruggemann and Calder (1953) but not recollected by us. Bruggemann and Calder (1953) also recorded Puccinellia pumila (Vasey) Hitchcock at the station, which we suspect had been misidentified, since this species is usually found at much lower latitudes and is not currently recorded on Ellesmere Island (GBIF 2020). However, we were not able to verify the current status of this collection as most of the collections deposited by P.F. Bruggemann do not appear on the database of the National Collection of Vascular Plants (CBIF 2020), and we were unable to visit this herbarium.
Based on this analysis of the floristic inventories from the last 70 years and by the exclusion of Papaver radicatum and Puccinellia pumila, we conclude that 77 vascular plant species are currently present in our study area at Alert.
An interesting observation is that we found Poa pratensis subsp. colpodea on a perching site used by birds of prey and on an arctic fox den. These faunal sites are productive vegetation patches due to the regular deposits of urine, faeces, casts, and prey remains, which release nutrients Gharajehdaghipour et al. 2016).
Our survey provides new information that contributes to biodiversity monitoring efforts and conservation management of vascular plants at Alert. In addition to being exposed to climatic warming (CAFF 2013), the vegetation at Alert may have already experienced direct anthropogenic disturbance, including habitat degradation from off-road use of military vehicles and soil contamination, due to past waste management issues. Introduction of new plant species through the transportation of personnel and heavy equipment at Alert may also provide opportunity for plant dissemination, although our survey revealed no obvious case of establishment of a non-native species. A better understanding of the above threats may be important regarding the imperiled Festuca viviparoidea subsp. viviparoidea, which has a scattered distribution and low occurrence in Nunavut (NatureServe 2020; GBIF 2020), and other arctic endemic species with restricted distribution (e.g., Draba pauciflora, Festuca edlundiae, and ×Pucciphippsia vacillans). In this context, identifying areas of highest conservation concerns is essential. The status of several Arctic species is currently not assessed and precise information on distribution and population trends is often lacking (CAFF 2013). Long-term monitoring of plant communities and regular surveillance for the establishment of non-native species are thus critical to assess the effects of climatic changes and anthropogenic impacts on High Arctic ecosystems. The relative accessibility of Alert to scientists offers invaluable opportunities to better understand High Arctic biodiversity and its ongoing changes.