Checklist of vascular plants for Wind River Indian Reservation (USA) high-elevation basins: ecological drivers of community assemblages

Background and aims – Native American reservations in the United States provide biodiversity critical for conservation and ecosystem functions. Unfortunately, botanical inventories are less common for reservations than other land jurisdictions. Such ecological importance and needs are apparent for the Wind River Indian Reservation (WRIR), the 7th largest reservation in the US (>890,000 ha) that is shared by the Eastern Shoshone and Northern Arapaho. Material and methods – A botanical study for two WRIR high-elevation basins (Saint Lawrence Basin (SLB) and Paradise Basin (PB)) to (1) reconcile a 1960 plant list, and (2) quantify plant communities ecologically was conducted. In 2017, 106 monitoring sites were established to quantify species presence. Across basins, 231 total vascular plant taxa (221 to species and 10 to genus) were identified, or > 3× more plant species than noted in the 1960 list. In SLB, 222 plant taxa (213 to species and 9 to genus) were identified and in PB 98 plant taxa (90 to species and 8 to genus) were identified. In 2018, sites were re-sampled to quantify species abundance, soil pH, organic matter, soil nutrients, CEC, salts, and texture. Key results – Slope and elevation explained species distributions in the topography ordination and soil organic matter, pH, texture, P, and K explained species distributions in the soil ordination. Eleven exotic species, and one rare endemic species were documented with implications for empowering tribal management. Using a classification approach followed by an indicator species analysis and fidelity (Phi) assessment, we identified 14 unique plant communities and related these to 6 alliances and 7 associations across 6 macrogroups from the US National Vegetation Classification database. These indicator species of communities included sedges (Carex aquatilus), grasses (Pseudoroegneria spicata, Elymus elymoides, Achnatherum lettermanii, Elymus trachycaulus subsp. trachycaulus, Poa glauca subsp. rupicola), forbs (Polygonum bistortoides, Balsamorhiza incana, Castilleja flava), shrubs (Artemisia tridentata, Betula glandulosa, Dasiphora fruticosa subsp. floribunda) and trees (Pinus contorta). Conclusion – The plant taxa, plant communities, and ecological drivers documented in this study will enhance tribal and federal monitoring of these high-elevation WRIR basins.


INTRODUCTION
Quantifying botanical diversity patterns is fundamental to conserving global biodiversity, particularly at ecotones and extreme margins within the context of a changing climate (Pauchard et al. 2009;Lesica 2015). Recognizing plant communities as valuable resources, the United States (US) federal government regularly works with researchers to document and study the natural resources in national parks (Carter et al. 2006). Native American Indian reservations also function as pools of biodiversity for a broad spectrum of ecosystems and sizes of conservation units (Luna & Bahls 2017). However, even though the US federal government manages trust land on behalf of tribes, plant inventory projects, or projects quantifying biodiversity of any kind, are less common for reservations than other land jurisdictions such as national parks or forests (Crumpacker et al. 1988;Lesica 1993).
Friday & Scasta, Wind River Reservation high basin plant checklist There are currently 573 federally recognized American Indian and Alaskan Native tribes and villages in the US (Bureau of Indian Affairs 2019). The federal recognition is formalized, particularly in the relationship with the Bureau of Indian Affairs via the following definition: "A federally recognized tribe is an American Indian or Alaska Native tribal entity that is recognized as having a government-togovernment relationship with the United States, with the responsibilities, powers, limitations, and obligations attached to that designation, and is eligible for funding and services from the Bureau  These reservations are diverse in terms of size, management, organization, and natural resources. Some tribes have no designated trust land. Many reservations have a complex checker-board arrangement that includes a patchwork of trust, allotted, and fee land (Bureau of Indian Affairs 2019). Trust lands are accessible by tribal members year-round and during the summer months non-tribal members can purchase recreation use permits with varying levels of accessibility across the many tribes and their associated tribal laws. There is a complex history of policy decision-making regarding tribal land involving the US Congress, the Department of the Interior and individual tribes themselves, although tribes currently function with more autonomy regarding natural resource management than in past eras of policy (McCarthy 2004;Wilkins & Stark 2017; Bureau of Indian Affairs 2019). Such autonomy can be reflected in a tribal natural resource, environmental, or wildlife office and current practice demonstrates how tribes have the authority to make decisions regarding the depth and types of land use, which has the potential to reduce or increase human and livestock interactions with ecosystems. It is also common for most tribes to work with BIA staff to further address natural resource issues and such co-management of tribal resources is mandated because the BIA is "charged with carrying out the United States' trust responsibility to American Indian and Alaska Native people, maintaining the federal government-to-government relationship with federally recognized Indian tribes, and promoting and supporting tribal self-determination" (Bureau of Indian Affairs 2019). The federal Indian trust responsibility is a legally enforceable fiduciary obligation by the United States to protect tribal treaty rights, lands, assets, resources and executes the mandates of federal law (McCarthy 2004; Bureau of Indian Affairs 2019).
The BIA Branch of Agriculture and Rangeland Development administers the trust responsibility by improving the management of land and natural resource assets on trust land. BIA staff provide oversight and technical assistance in eight major categories: (1) inventory; (2) farm and range planning; (3) rangeland improvements; (4) rangeland protection; (5) leasing and permitting services; (6) contract monitoring; (7) agriculture extension; and (8) noxious weed eradication (Office of Trust Services 2019). BIA also facilitates cooperative efforts between tribes and other federal agencies for soil survey and rangeland vegetation classification to guide management and rangeland improvement projects (Hodgkinson 1984;Pease et al. 1991; Office of Trust Services 2019). Integrated management and conservation plans have been created for tribes to meet the technical assistance goals expected of the BIA (Fred Phillips Consulting and Bureau of Indian Affairs 2013). While BIA is a federal agency, it is not always the case that national level resource inventories are implemented on trust lands. For example, Ecological Site Descriptions (ESDs) are the world's largest land management framework created by scientists in collaboration with land managers. ESDs can provide detailed information on climate, soil, geomorphology, hydrology and vegetation to assist in land management decisions (USDA 2013;Twidwell et al. 2013), yet on trust lands ESDs either do not exist or often contain gaps of data yet to be determined by ground verification, making these ESDs useless for writing range management plans. Thus, the currently available vegetation inventory data is not always operational.
Such issues of biodiversity, bureaucracy, and opportunities intersect on the Wind River Indian Reservation (WRIR). WRIR is the 7 th largest Indian reservation by land area in the US, consisting of more than 890,000 hectares. The WRIR has approximately 26,000 residents (U.S. Census Bureau 2010). The WRIR is located within Fremont and Hot Springs Counties of Wyoming USA and the federally-recognized Eastern Shoshone and Northern Arapaho tribal nations share the reservation ( fig. 1). The WRIR is situated in the Wind River Basin and is surrounded by the Wind River, Owl Creek, and Absaroka mountain ranges. The Wind River Mountain Range is the largest mountain range in Wyoming and has the highest peak in the state -Gannett Peak at 4,209 m. The aesthetic and cultural uniqueness of these high-elevation mountains illustrate the societal values that the WRIR and society at large place upon the Wind River Mountain Range. To date, detailed botanical data has primarily been limited to species presence and absence lists with no ecological site information (supplementary file 1).
Due to the expansive rangeland resources of the WRIR and its ecological, agriculture, and cultural importance, a more intricate study of botanical relationships is needed for both conservation and rangeland management. In 2015 a two-year, comprehensive plant inventory study was initiated to (1) reconcile a dated (55-year-old) plant presence/absence study, and then (2) to quantify the plant communities of high-elevation basins with a greater level of ecological detail relative to environment features. Specifically, quantifying plant communities and dominant functional groups in the basins, coupled with topographic, soils and disturbance data is critical to understanding where plant assemblages occur and provides resources to assist in making informed management decisions. The impact of such studies is that more comprehensive data allows tribal land managers to better detect how environmental change such as variation in land use, temperature, precipitation, and snowpack levels impacts sensitive plant communities in these high-elevation basins -information that has been lacking to date.

Justification
In 1960, Field & Tidd conducted an inventory of the plants in Paradise Basin (or "Paradise Park") and Saint Lawrence Basin in the Wind River Mountain Range with voucher specimens reconciled by C.L. Porter who was the curator of the University of Wyoming's Rocky Mountain Herbarium at that time (see supplementary file 1). A second set of voucher specimens are housed at the Bureau of Indian Affairs (BIA) Wind River Agency in the Range Department Herbarium. This initial inventory, while useful as a basic plant species list, is severely lacking in applicable value for several reasons: (1) it only includes the presence or absence of 74 species with some repeating, (2) makes no estimate of species abundance, (3) makes no differentiation between the two basins relative to the presence of a species, and (4) does not measure any other ecological explanatory information to explain species occurrence. Thus, while this survey has some base-level value, it is not useful in explaining the rangeland plant communities in these high-elevation basins, has no abundance data that would be useful for determining rangeland health and condition, and is unable to develop predictive models for where certain plant species and plant assemblages occur relative to other environmental features. These lacking features of the 1960 inventory preclude the tribal use of the information to understand the current rangeland conditions and development of management plans.

Study area
The study area consists of two adjacent high-elevation basins in the Wind River Mountain Range on the WRIR: (1) Saint Lawrence and (2) Paradise basins ( fig. 2). The Saint Lawrence basin is located approximately 26 kilometres northwest of Fort Washakie, Wyoming and has an area of ~94.52 km 2 . The Saint Lawrence basin receives, annually, 64.3 cm of precipitation, and temperature ranges from -4 to 8. 7°C (1988-2018) with an elevation ranging from 2,560 to 3,352 m a.s.l. (PRISM Climate Group 2019). Dominant land uses in Saint Lawrence basin are livestock grazing, firewood harvesting, outfitter operations (June-September annually) and outdoor recreation. Paradise basin is located approximately 35 kilometres northwest of Fort Washakie, Wyoming and encompassing an area approximately 41.73 km 2 ( fig. 2). Paradise basin receives, annually, 78.8 cm of precipitation and temperature ranges from -4.7 to 6. 7°C (1988-2018) with an elevation ranging from 2926-3429 m asl (PRISM Climate Group 2019). Within Paradise basin, outdoor recreation and outfitter client services are the dominant uses.
The geology of the Wind River mountains is very complex as it spans a long time period dating back over 3 billion years with multiple episodes of major developments particularly sea inundations, intense magmatism and tectonic plate activity, and ultimately dramatic uplift and thrusting of Paleozoic and Mesozoic sediments (Wells et al. 2015). For the specific basins in this study, the parent material is primarily Precambrian rocks consisting of mostly granites and gneisses of sedimentary origin (Donohue & Essene 2005). The older strata near the mountains rise to elevations almost as tall as the main peaks themselves. The combination of geologic parent material, alpine glaciation, and geologic forces have formed high-elevation basins as these foothills transition to high peaks and they are dominated by herbaceous vegetation important for livestock and wildlife. For both basins, forested areas are dominated by Lakehelen-Hazton complex soils and steeper slopes and ridges are dominated by Alpine Ridges Rubbleland-Tundra complex soils. For Paradise Basin, sedge-rush areas are dominated by Venapass-Silas loam soils and graminoid-shrub dominated areas are dominated by Nathale-Pishkun-Rock outcrop complex soils. For Saint Lawrence Basin graminoid-shrub dominated areas are dominated by Barbarela-Hapjack-Sawcreek complex soils and sedge-rush areas are dominated by Vensora clay loam soils (USDA NCRS 2020).
Most of the study area is designated as a roadless area and named the Wind River Reserve (Cornell Law School 2019). The roadless area was created in response to the WRIR tribal councils' concerns about the future construction of mountain passes for highways for tourists travelling to visit Yellowstone or the Grand Teton National Parks (Aragon 2007). The Wind River Reserve sets aside ~73,000 hectares of land as roadless area and restricts the construction or establishment of roads, highways, truck trails, work roads, and all other types of motor transportation passage ways (Aragon 2007;Cornell Law School 2019). More than half of Saint Lawrence basin is designated roadless area. Paradise basin is designated entirely as roadless area with no livestock grazing permitted and the use of motorized vehicles and tools is restricted unless approved by the BIA Wind River Agency Superintendent (Cornell Law School 2019).

Study design
The BIA Wind River Agency Range Program's assessment and monitoring protocol (originally adapted from Wyoming Range Service Team 2008) was used as the foundation for data collection. The protocol consists of a 30.5 m transect that is established in a north to south direction, photos are taken in four cardinal directions at the north end, line point intercept data is recorded every 0.3 m (i.e., a foot), and quadrats for cover data are placed every 7.62 m starting from zero (Wyoming Range Service Team 2008). The protocol is used for long-term vegetation monitoring by range staff and interested livestock producers are trained with the protocol throughout the WRIR. Prior to conducting this study, the Field & Tidd plant list from 1960 (see supplementary file 1) was reconciled and updated with current taxonomic designations and a determination of plant species status (native/exotic, conservation status, rare/endangered, toxic/poisonous, etc.). Random points were generated from Geographic Information System (GIS) analysis of digitized soil maps (Soil Survey Staff 2019) and study area maps using ArcGIS Software (Esri 2019). One hundred and six (106) total monitoring sites for sampling were selected from the random points maps to establish data collection sites during the study and for long-term monitoring by the BIA Wind River Agency Range staff. Transects were randomly established in both basins across habitat types (riparian, meadow, upland, or forested). For each transect, we noted aspect, slope, and elevation. Additionally, in Saint Lawrence basin we established 4 transects inside and outside of a pre-existing BIA exclosure. In 2012 a wildfire named the Alpine Lake fire burned ~16,732 hectares and in September 2017 another wildfire burned 66 hectares in Saint Lawrence basin. We thus also established transects within each burn area.

Field methods and data
In 2017, plant species presence/absence inventory was recorded by identifying plant species present within a 180 m 2 plot centred on a 30.5 m transect. Of the 106 monitoring sites, 86 sites are located in Saint Lawrence basin and 20 sites are located in Paradise basin ( fig. 2). We then compiled a species list for both basins from the presence inventory (supplementary file 2) and taxonomical nomenclature follows USDA NRCS (2019). In some instances, a specimen could only be identified at the genus level, including for Arnica, Artemisia, Astragalus, Carex, Cirsium, Erigeron, Eriogonum, Lupinus, Potentilla, and Salix. These 10 genera were included in all analyses with the exception of functional group summaries.
Plant voucher specimens were prepared for storage at the BIA Wind River herbarium for a majority of the plant species that were identified. Each specimen included at a minimum the following information: (1) collection number; (2) date collected; (3) identification of the plant; (4) location, including township, range, section, county, elevation and vegetation type; (5) environmental aspects, including aspect, slope and elevation; (6) notes on flower colour, plant size, variability; and (7) collector name (Elzinga et al. 1998;Martin 2010).
Soil samples were taken at every 3 m along a 30.5 m transect line for a total of 10 samples per transect. Soil samples of 10 cm in depth were extracted using a standard cylindrical soil core sampler with a 2.54 cm diameter. The soil samples were air dried on paper bags in full sunlight and stored in 1-quart plastic bags. At the end of the field season the soil samples were shipped to Ward Laboratories, Inc. in Kearney, Nebraska, USA for chemical and texture analysis. In addition, soil compaction data was recorded in the field using a pocket soil penetrometer at 0 m, 8 m, 15 m, 23 m, and 30 m along a 30.5 metre transect line. The measurements were then averaged and recorded in kg/cm 2 . Soil characteristics included in analyses included soil pH, organic matter (OM), total nitrogen (N), total phosphorus (P), potassium (K), cation exchange capacity (CEC), soluble salts, and soil texture percentages (clay, sand, and silt).

Statistical analysis
We first calculated the number of species by functional group (forb, grass, sedge/rush, shrub, tree) and origin (native or exotic) for each basin. Varieties and subspecies were included as distinct taxa in all analyses. Then to understand ecological relationships and plant species distributions relative to topographic and soil features within the two high-elevation basins (combined for all subsequent analyses), we first used multivariate statistical techniques and CANOCO version 5 statistical software to perform analyses (Šmilauer & Lepš 2014). This approach allows for the assessment of complex community data relative to variation and similarity and the identification of primary drivers in multi-dimensional space (Frye 2009). We performed a Detrended Correspondence Analysis (DCA), an unconstrained ordination technique, to quantify plant species in multidimensional space data. DCA has been widely used in ecology because of its non-linear model, its ability to address arch effects of correspondence analysis, and its application for predicting vegetation patterns (Zhang et al. 2008;Shetie et al. 2017). We then performed two Canonical Correspondence Analyses (CCA) to quantify plant species distributions relative to topographic features and soil data. In the topographic CCA, we used the topographic features of percentage slope, aspect northness, aspect eastness, and elevation as explanatory variables. In the soil CCA, we used soil pH, CEC, OM, soluble salts, N, K, P, texture percentages (silt, clay, and sand), and SC as explanatory variables (Dingaan et al. 2017). The species response data was binary (i.e., presence/absence or 1/0), data were not transformed, there was no downweighting of rare species, sample diversity was expressed as number of species, and all the first axis and then all constrained axes com- bined were tested with a permutation test (1,000 iterations) for significance in each of the CCAs.
In order to move towards identifying plant communities and classification, we then applied a hierarchical clustering approach to initially identify groups of sites using a classical Ward's method algorithm (PAST version 3.0; Hammer et al. 2001). In this step we used a distance cut-off criterium of 12.5 to identify fourteen groups (number of plots within groups ranged from 3 to 14). We then conducted an indicator species analysis using the "indicspecies" package in R (De Cáceres 2020). We then calculated adjusted phi (φ) coefficients (adjusted for some groups having more sites than others; Tichý & Chytrý 2006) as indicators of fidelity to each group and an indication of a positive or negative preference of a species for a group (Chytrý et al. 2002) using CANOCO version 5 (Šmilauer & Lepš 2014). Finally, we evaluated the presence, significance, and fidelity of species within groups and compared to classifications in the United States National Vegetation Classification (USNVC), relative to indicator species and invasive species, to identify similar USNVC alliances (A) or associations (CEGL) and macrogroups when possible (USNVC 2020).  Explained fitted variation for axis 1 was 38.36% and axis 2 explained fitted variation was 29.38%. The first axis is explained by slope and northness/ eastness and the second axis is explained by elevation. The pseudo-F value for the test on the first constrained axis was 3.0 and p = 0.001 and for all four axes was 2.4 and p = 0.001, indicating that the topographic covariates are significant explanatory variables for the plant community data. Plant species are represented by a four to five letter code assigned by the USDA Plants database. Other letter codes are associated with Daubenmire cover classes (Daubenmire 1959).

Plant species distributions -unconstrained ordinations
In Saint Lawrence Basin (SLB) 222 plant taxa were identified (213 to species and 9 to genus), with 96% native, and the dominant functional group was forbs followed by native grasses (fig. 3). In Paradise Basin (PB) 98 plant taxa were identified (90 to species and 8 to genus), with 94% native, and similarly forbs are the dominant functional group followed by native grasses. 89 of the plant taxa occurred in both basins, so there were 9 unique species to PB. The total number of exotic taxa was less in PB than in SLB (6 and 8 respectively) ( fig. 3).

Topography and soil influences on plant speciesconstrained ordinations
The topographic CCA explained fitted variation for axis 1 was 38.36% and axis 2 explained fitted variation was 29.38% (fig. 5), a substantial improvement over the explained variation in the DCA alone ( fig. 4). The first axis is strongly explained by slope and northness/eastness, the  (Salts), nitrogen (N), potassium (K), phosphorus (P), and texture percentages (silt, clay, and sand). Explained fitted variation for axis 1 was 32.72% and axis 2 explained fitted variation was 16.14%. The first axis is explained by organic matter (OM), pH, and potassium (K), the second axis seems to be explained by phosphorous (P) and salts, and a soil texture gradient is also apparent from sand (left) to clay (right). The pseudo-F value for the test on the first constrained axis was 5.2 and p = 0.001 and for all four axes was 2.0 and p = 0.001, indicating that the soil covariates are significant explanatory variables for the plant community data. Species are represented by four to a four to five letter code assigned by the USDA Plants database. Other letter codes are associated with Daubenmire cover classes. second axis is strongly explained by elevation. Regarding slope, species such as Artemisia (ARTEM), spike fescue grass (Leucopoa kingie; LEKI2), prairie sagewort (Artemisia frigida; ARFR4), yellow Indian paintbrush (Castilleja flava; CAFL7), and slender wheatgrass (Elymus trachycaulus subsp. subsecundus; ELTRS) seem to be explained by increasing slope, while sedges and rushes such as water sedge (Carex aquatilis; CAAQ), Sierra hare sedge (Carex leporinella; CALE9), slenderbeak sedge (Carex athrostachya; CAATS), Baltic rush (Juncus balticus var. montanus; JUBAM), and certain shrubs such as willows (Salix species; SALIX), resin bog birch resin bog birch (Betula glandulosa; BEGL), and shrubby cinquefoil (Dasiphora fruticosa subsp. floribunda; DAFRF) are associated with gentler slopes and more north and/or east facing aspects. Several species seem to be explained by increasing elevation such as cushion buckwheat (Eriogonum ovalifolium; EROV), fireweed (Chamerion angustifolium; CHAN9), spearleaf stonecrop (Sedum lanceolatum; SELA), Drummond's rockcress (Arabis drummondii; ARDR), slender wheatgrass (Elymus trachycaulus subsp. subsecundus; ELTRS), and lupines (Lupinus species; LU-PIN). The pseudo-F value for the test on the first constrained axis was 3.0 and p = 0.001 and for all four axes was 2.4 and p = 0.001, indicating that the topographic covariates are significant explanatory variables for the plant community data ( fig. 5).
The soil CCA explained fitted variation for axis 1 was 32.72% and axis 2 explained fitted variation was 16.14% ( fig.  6), a substantial improvement over the explained variation in the DCA alone ( fig. 4). The first axis is strongly explained by organic matter (OM), pH, and potassium (K), while the second axis seems to be strongly explained by phosphorous (P) and salts. A soil texture gradient is also apparent from sand (left) to clay (right). Regarding this first axis, species of willow (Salix species; SALIX), shrubby cinquefoil (Dasiphora fruticosa subsp. floribunda; DAFRF), water sedge (Carex aquatilis; CAAQ), and resin bog birch (Betula glandulosa; BEGL) are associated with greater OM, greater percentage silt and clay, and lower pH. In contrast, species such as Artemisia species generally (ARTEM), big sagebrush (Artemisia tridentata; ARTR2), prairie sagewort (Artemisia frigida; ARFR4), Astragalus species generally (ASTRA), and spike fescue grass (Leucopoa kingii; LEKI2) are associated with decreasing OM, more K, greater percentage sand, and higher pH ( fig. 6). Regarding the second axis, lodgepole pine (Pinus contorta; PICO), fireweed (Chamerion angustifolium; CHAN9), and spreading wheatgrass (Elymus scribneri; ELSC4) are associated with more P and sand, and lower salt content and CEC. The pseudo-F value for the test on the first constrained axis was 5.2 and p = 0.001 and for all four axes was 2.0 and p = 0.001, indicating that the soil covariates are significant explanatory variables for the plant community data ( fig. 6).

Indicator species analysis -plant communities and indicator species
The classification and indicator species analysis revealed 14 unique plant communities. These plant communities generally had similar classifications to the United States National Vegetation Classification (USNVC) with two exceptions. Here, we provide details about each plant community relative to indicator species, invasive species, and USNVC similar Alliances (A) or Associations (CEGL) and/or gaps in the USNVC system relative to our findings. We note that the classifications were derived from six USNVC macrogroups, including: M020 (Rocky Mountain Subalpine-High Montane Resin bog birch (Betula glandulosa) plant community (table 1) -This was the preferred group for shrubs and trees that prefer moist environments such as resin bog birch (Be-tula glandulosa), Wolf's willow (Salix wolfii), Engelmann spruce (Picea engelmannii). Similarly, smallflowered woodrush (Luzula parviflora) and alpine timothy (Phleum alpinum) preferred this group. All these species had a significant association (p < 0.01) and a high level of fidelity (φ > 0.7) (table 1). This group was also preferred by three exotic species (Kentucky bluegrass (Poa pratensis), timothy (Phleum pratense), and prostrate knotweed (Polygonum aviculare)). This group is similar to USNVC CEGL000357 Picea engelmannii / Caltha leptosepala Swamp Forest (M034) with additional similarities to A4096 Dasiphora fruticosa / Festuca campestris -Festuca idahoensis Shrub-steppe Alliance because of the invasion by Phleum pratense and Poa pratensis (M048) and to A3770 Salix wolfii -Salix brachycarpa -Betula glandulosa Wet Shrubland Alliance (M893). Water sedge (Carex aquatilus) -Willow (Salix species) plant community (table 2) -This was the preferred group for two sedges including water sedge (Carex aquatilis) and Sierra hare sedge (Carex leporinella). In addition, this was the preferred group for the forb redpod stonecrop (Rhodiola rhodantha). All these species had a significant association (p < 0.02) and a moderate to high level of fidelity (φ = 0.8903, 0.6820, and 0.4171 respectively) (table 2). This group included similar shrubs found in the resin bog birch plant community (resin bog birch (Betula glandulosa), willows (Salix species), and shrubby cinquefoil (Dasiphora fruticosa subsp. floribunda), as well as other sedges and rushes. This group is similar to USNVC A3770 Salix wolfii -Salix brachycarpa -Betula glandulosa Wet Shrubland Alliance (M893); Salix wolfii was present but non-significant in the data. Shrubby cinquefoil (Dasiphora fruticosa subsp. floribunda) -Idaho fescue (Festuca idahoensis) plant community (table 3) -This group included shrubby cinquefoil (Dasiphora fruticosa subsp. floribunda) and Idaho fescue (Festuca idahoensis) both of which were significantly associated with the group (p < 0.001) and displayed high fidelity (φ >  all of which were significantly associated with the group (p < 0.001) and displayed high fidelity (φ > 0.6). This was also the preferred group for tufted hairgrass (Deschampsia cespitosa) (p < 0.001; φ > 0.6) (table 4). Wolf's willow (Salix wolfii), shrubby cinquefoil (Dasiphora fruticosa subsp. floribunda), and other sedges and rushes were significantly associated with the group (p < 0.001) and displayed high fidelity (φ > 0.6). In addition, two exotic clover species (white clover (Trifolium repens) and red clover (Trifolium pratense)) were noted to prefer this group although significance and fidelity levels were low for both (p > 0.05; φ < 0.3) (  [Group 10 plot means (n = 6): Elevation 2,832 m, % Sand-Silt-Clay 63-21-17, Organic Matter 6.0%, pH 5.9]. Phi (φ) is an indicator of fidelity and the asterisk (*) suggests this is the preferred group for this species. [Group 11 plot means (n = 3): Elevation 3,008 m, % Sand-Silt-Clay 53-29-18, Organic Matter 5.0%, pH 6.4]. Phi (φ) is an indicator of fidelity and the asterisk (*) suggests this is the preferred group for this species.
The exotic species rock dandelion (Taraxacum laevigatum) preferred this group (

Comparison with the 1960 inventory list
Field & Tidd listed 74 plant species on their 1960 collection list from Saint Lawrence basin and Paradise park areas.
Through the reconciling process, we eliminated repeat entries and those specimens for which identification was questionable producing a reconciled list of 56 identifiable plant species. Our 2017 plant inventory significantly expanded on the number of plant species in the high-elevation basins to a total of 221 plant taxa and 10 genera for which species identification was not possible (for a total of 231

DISCUSSION
Our study identified 231 plant taxa (221 to species and 10 to genera) across various vegetation communities in the highelevation Saint Lawrence and Paradise basins of the Wind River Reservation. Previously, for these two basins, Wind River BIA range staff were working with an outdated and insufficient plant list with only 56 reconcilable plants. The plant inventory list and ecological explanatory data collected from our study will serve as comprehensive baseline data for the BIA range staff to effectively monitor plant communities and basin resources in the future. Given Paradise basin's roadless area designation and Saint Lawrence basin's partial roadless designation with limited livestock grazing and logging activities, the data can be used to gauge anthropogenic changes over time which will become increasingly important in these high-elevation basins particularly in the context of climate change forecasts (Gildar et al. 2004). There is also the potential for BIA to consider prescribed burns within the basins in order to reach management objectives by using the baseline data to build GIS-based vegetation mapping and spatial modeling to strategically plan burns (Young et al. 2017). Topographic, soil, and disturbance data proved to be informative for understanding vegetation in the two high-elevation basins. Slope, aspect, and elevation influence where and how certain plant communities occur. Our vegetation inventory associates topographic data with vegetation data, which we have extrapolated to plant communities. Understanding how plant communities interrelate across varied topography has important management implications (Banta et al. 2005). Soil nutrient content was also an important factor determining where specific plant species grow and their level of abundance (Menzel et al. 2016) and operationalizes soil and vegetation data for BIA range staff. Disturbances identified in the basins were burned areas, livestock grazing, outdoor recreation, two-track roads, and firewood harvesting. In 2012, the Alpine Lake Fire burned a northwest portion of Paradise Basin and in 2017 the Saint Lawrence Fire burned 66 hectares. Establishing long-term transects in burned areas allows land managers to monitor plant species presence and abundance over time (Kost & De Steven 2000) which is important because some plant species are early colonizers of burn areas and monitoring is critical to identify problems or changes.
It is important to thoroughly understand sensitive plant communities that are marginally or extremely distributed to detect changes in plant species abundances caused by environmental changes (Lesica 2015). Kelly & Goulden (2008) found that a dominant plant species identified in 1977 had shifted up ~65 m in elevation in 2007. The researchers attributed the increase in elevation to changes in regional climate that consisted of warming regional temperatures, increasing precipitation variability, and decreased snowpack (Kelly & Goulden 2008). High-elevation environments are also at risk for loss of habitat as plant communities shift higher in eleva-tion. Species that live and thrive in native plant communities at current elevations can become threatened, endangered, or extinct as dominant plant communities shift upwards in elevation (Dirnböck et al. 2011).
As regional climates warm and native plant species move higher in elevation, habitat loss and inadvertent introductions of invasive plant species to high elevation environments can threaten native plant communities (Pauchard et al. 2009;Expósito et al. 2018). Our study documented eleven introduced species, two of which are designated as noxious weeds by county and state authorities (Wyoming Weed and Pest 2019). Plant species are designated as noxious weeds, a legal designation, when they are harmful to the health and welfare of the environment and/or animals (Wyoming Weed and Pest 2019). We identified Canada thistle (Cirsium arvense) in both basins and nodding plumeless thistle or musk thistle (Carduus nutans) in Paradise basin -both noxious species in Wyoming. Canada thistle invades plant communities by primarily reproducing by asexual and sexual mechanisms (Wyoming Weed and Pest 2019). Musk thistle generates an excess of 20,000 seeds per plant that are viable in the soil up to 10 years making it difficult to manage the species (Wyoming Weed and Pest 2019). Downy brome (Bromus tectorum), more commonly known as cheatgrass, is a declared a noxious weed for Fremont County and it was found in Saint Lawrence basin along a road leading to a popular fishing spot (Wyoming Weed and Pest 2019). Cheatgrass is an introduced winter annual grass that matures earlier than native plant species and has been implicated in dramatic changes in composition and function of western US plant communities (Mealor et al. 2012). The early detection of these noxious species in the high-elevation basins provides the BIA range staff the opportunity to implement management decisions to reduce the abundance and potentially eradicate them from the basins.
Regional climate and environmental changes may stress plant species as ideal growing conditions are altered causing migrations across landscapes or species extirpations (Dirnböck et al. 2011;Thuiller et al. 2005). In Wyoming, there are approximately 485 vascular plant species listed as either 'Species of Concern' or 'Species of Potential Concern' (Heidel 2018) and our study documented two such species. Weber's saw-wort (Saussurea weberi) is listed as a 'Species of Concern' and was identified in Saint Lawrence basin. Weber's saw-wort is considered a rare plant species because of its disjunct and infrequent occurrences in alpine habitats. In Wyoming, Weber's saw-wort is considered sensitive and imperilled at the state level and the US Forest Service lists it as sensitive in Region 4 consisting of National Forests in Wyoming: Bridger-Teton, Caribou, Targhee, Wasatch-Cache, and Ashley (including Flaming Gorge National Recreation Area) National Forests (Heidel 2018). At a global level, Weber's saw-wort is considered to be vulnerable and imperilled with a probability of species extinction (Heidel 2018). Limber pine (Pinus flexilis) is listed as a 'Species of Potential Concern' and was identified in both Saint Lawrence and Paradise basins. Warmer/drier climate conditions and associated mountain pine beetle (Dendroctonus ponderosae) outbreaks are potentially influencing its distribution (Cleaver et al. 2015). In Wyoming, limber pine is considered sensitive and imperiled at the State level but at a global level it is apparently secure (Heidel 2018).
Surprisingly, we did not find Northern sweetgrass (Hierochloe hirta subsp. hirta; henceforth 'sweetgrass') in either basin. Sweetgrass is commonly used by Indigenous people as a ceremonial smudge or medicine, and as an incense by non-Indigenous people (Cantrell et al. 2016;Shebitz 2005;Shebitz & Kimmerer 2005). The demand for sweetgrass also leads to gathering, braiding, and commercially selling sweetgrass to local businesses or individuals (Dhar et al. 2000;Shebitz 2005). The commercial demand for sweetgrass may have led to overharvesting and its depletion from its natural habitat in at least Saint Lawrence basin (Gaoue & Ticktin 2007;Shebitz 2005;Vihotogbé et al. 2014). Thus, the potential restoration of this culturally important plant species should consider human use, harvesting, economic value, and land use change (Droissart et al. 2019).

CONCLUSIONS
Federally recognized tribes hold roughly 22.7 million hectares (five percent) of trust land designated as Indian reservations in the United States (Bureau of Indian Affairs 2019; Stumpff 2000). The WRIR is an important example as the 7 th largest reservation in the US, encompassing more than 890,000 hectares, serving as a pool of biodiversity for conservation with a diversity of ecosystems (Cozzo 2004;Luna & Bahls 2017). This biodiversity is important at a global scale but is also critical at a local scale for local tribal communities. Ultimately, intergenerational transfer and revitalization of Indigenous ethnobotanical knowledge can enhance cultural identity (i.e., people's sense of place) and propel Indigenous perspectives for meaningful management decisions regarding their lands (Serenari et al. 2017).
The WRIR has expansive rangeland resources with ecological, agricultural, and cultural importance that requires understanding of intricate botanical relationships. Our study expanded on outdated plant inventories by identifying an additional 168 vascular plants, collecting data to quantify plant communities, and identifying the dominant functional groups in the Saint Lawrence and Paradise basins. This vegetation data was coupled with topographic, soil, and disturbance data to understand where plant assemblages occur and provides data driven information to guide management decisions of rangelands in the high-elevation basins. This comprehensive study will allow BIA range staff to detect the occurrence of environmental changes such as variations in land use and regional climate. These changes coupled with disturbances such as outdoor recreation, livestock grazing, and wildfire can introduce non-native species to plant communities. The early detection of noxious weeds and invasive species in the high-elevation basins is critical because they can outcompete native species and this knowledge will empower BIA range staff to act early and make management decisions that will either maintain, reduce, or eradicate these species from the basins. Another consideration associated with regional climate change is the negative stress on plants when ideal growing season shift up in elevations of montane environments. Our study documented two species of concern (Weber's saw-wort (Saussurea weberi) and limber pine (Pinus flexilis)) enabling BIA range staff to consider these species when developing management plans or making management decisions regarding the basins.

ACKNOWLEDGMENTS
Funding was provided by the Bureau of Indian Affairs -Grant/Cooperative Agreement Number A17AC00019 "Enhanced Ecological Inventory of High Elevation Basins on the Wind River Indian Reservation". Appreciation is also extended to Mr. Preston Smith, BIA Range Management Specialist for his support with all aspects of the project and to the field technicians through the years.