Sagebrush-Associated Bunchgrasses Drive Invasion Resistance in a Greenhouse Experiment

ABSTRACT Invasion of non-native annual grasses is a significant threat to the sustainability of sagebrush steppe ecosystems. Ecological resilience, the ability to bounce back after a disturbance, and resistance, the ability to withstand invasion, are influenced by both abiotic factors, such as soil temperature, moisture, elevation, and aspect, and biotic factors, such as plant community composition. We quantified the effects of moss biocrusts, native shrubs, and native perennial grasses on invasion resistance in a greenhouse experiment containing dominant sagebrush ecosystem plants and invasive grasses. We saw greatest suppression of invasive annual grass biomass in treatment replicates containing native bunchgrass species (P < 0.01). Final invasive grass biomass was 4.79 g on average when perennial grasses were not present and was reduced to 1.59 g with perennial grass competition (P < 0.01). Presence of shrubs and moss biocrusts did not decrease annual grass biomass (P= 0.38 and P= 0.25, respectively). We saw complex interactions between native plants grown in these ideal greenhouse conditions such that native perennial grass seedlings grown with sagebrush seedlings had a mean of 4.50 g more biomass (P < 0.001) relative to pots grown with bitterbrush or without shrubs, but shrubs were an average of 7.9 cm (P < 0.001) shorter and had biomass 4.75 g lower (P < 0.001) in pots grown with perennial grasses compared with shrubs grown without perennial grasses. Our results demonstrate that with increased treatment complexity, we see greater invasion resistance, but that nuanced relationships between plant community members should also be considered in managing and restoring these imperiled ecosystems.


Introduction
The sagebrush steppe is one of the most threatened ecosystems in western North America ( Noss et al. 1995 ;Brooks et al. 2004 ;Davies et al. 2011 ) and is underrepresented within protected areas ( Caicco et al. 1995 ;Stoms et al. 1998 ;Aycrigg et al. 2013 ).This biome covers a broad geographic area and contains a suite of perennial bunchgrasses, flowering plants, shrubs, and biological soil crusts (lichens, mosses, and cyanobacteria on the soil surface; hereafter, biocrusts) that provide wildlife habitat, myriad ecosystem services ( Yahdjian et al. 2015 ), and economic opportunities to humans ( Maczko et al. 2011 ).Key biophysical processes in these semiarid rangelands include episodic wildfire, herbivory, seasonal drought, and periodic precipitation falling mostly as snow.
Invasive annual grasses such as cheatgrass ( Bromus tectorum L.) and medusahead ( Taeniatherum caput-medusae L.) pose a critical threat to the sagebrush steppe.These rapidly growing grasses can reduce light at the soil surface, which reduces native competitors' photosynthetic ability ( D'Antonio and Vitousek 1992 ; DiTomaso 20 0 0 ; Davies and Johnson 2017) .Cheatgrass germinates in most seasons, and soil seedbanks can remain viable for up to 5 yr ( Rice and Dyer 2001 ).As an early germinant, cheatgrass often successfully competes with native plants for water and nutrients ( Boyte et al. 2016 ;Kerns and Day 2017 ).Medusahead has sharp awns and a high silica content that limits herbivory and forms a thick, persistent thatch layer that reduces the establishment of native plants ( Davies and Svejcar 2008 ;Esposito et al. 2019 ).Medusahead also aggressively competes for resources, decreasing biodiversity and abundance of native vegetation ( Nafus and Davies 2014 ;Uselman et al. 2014 ).Both invasive grass species alter fire regimes in the sagebrush steppe, creating a feedback cycle between increased invasion and increased disturbance ( D'Antonio and Vitousek 1992 ), ultimately decreasing the success of management and restoration effort s ( Pilliod et al. 2021 ).While invasive grasses are found across the sagebrush biome, they are most persistent and destructive at the driest, least resilient end of the biome ( Chambers et al. 2014a( Chambers et al. , 2014b ) ).
Ecological resilience refers to the ability of plant communities to regain their structure and function after a disturbance.Resistance is the ability of a community to withstand invasion by nonnative species (Chambers et al. 2014a(Chambers et al. , 2014b(Chambers et al. , 2014c) ) .Resilience to disturbance and resistance of plant communities to invasion depend on local conditions such as topography, soil water, climate, available nitrogen, and native vegetation cover ( Brooks and Chambers 2011 ;Chambers et al. 2014b ).Annual grass invasions can affect multiple levels of ecological organization from population to ecosystem and alter competitive interactions with native plants and biocrusts ( D'Antonio and Vitousek 1992 ;Bowker et al. 2006 ;Havrilla et al. 2019 ), altering the landscape's inherent resistance and resilience (Gunderson 20 0 0 ; Reisner et al. 2013 ;Seipel et al. 2018 ;Williamson et al. 2020) .
Previous studies have linked the presence of shrubs, bunchgrasses, and biocrusts to site resistance against invasive plants.There is evidence that established native shrubs, mosses, lichens, and perennial bunchgrasses with small gaps between native plants greatly reduces the magnitude of invasion ( Condon et al. 2011 ;Reisner et al. 2013 ;Ellsworth and Kauffman 2017 ).Moreover, postfire recovery of sagebrush is slower when annual grasses are present ( Beck et al. 2009 ;Chambers et al. 2014c ;Ellsworth et al. 2016 ;Esposito et al. 2019 ).Deep-rooted bunchgrasses, in particular, may be key providers of resilience from and resistance to invasive grasses due to being locally abundant, stabilizing soils, controlling water flow, and resprouting after fire ( Chambers et al. 2007( Chambers et al. , 2014 a; a;Ellsworth and Kauffman 2010 ;Hulet et al. 2015 ;Germino et al. 2019 ;Ellsworth et al. 2020 ).
Biocrusts form complex networks of mosses, lichens, bacteria, algae, fungi, and liverworts, which bind the top millimeters of soil together and colonize the interstitial space between vascular plants ( Belnap 2003 ;Zhang et al. 2016 ;Bowker et al. 2018 ;Slate et al. 2019 ;Eldridge et al. 2020 ).Biocrusts can reduce soil erosion, increase water infiltration, reduce evaporation, and contribute to nutrient cycling ( Belnap 20 03 ;Warren 20 03 ;Eldridge and Leys 20 03 ;Bowker et al. 20 08;Chamizo et al. 2012) .Empirical evidence and theoretical models have both suggested that essential resources such as water, organic matter, seeds, and nutrients are distributed in patches ( Ryel et al. 1996 ;Stringham et al. 2003 ;Gasch et al. 2015 ;Liu et al. 2020 ).This uneven distribution of resources has been partly attributed to the presence of biocrusts because they tend to retain more water than crust-free soil ( Eldridge and Leys 2003 ;Chamizo et al. 2012Chamizo et al. , 2016;;Eldridge et al. 2020) .Biocrusts have variable relationships with vascular plants and can either positively or negatively alter germination and establishment ( Zhang et al. 2016 ;Havrilla and Barger 2018 ) of both invasive and native plants ( McIntyre et al. 2021 ).Vascular plants can benefit from the seed protection and increased soil-water availability provided by biocrusts.Conversely, biocrusts can also inhibit germination, establishment, and/or dispersal of vascular plants ( Su et al. 2007 ;Li et al. 2021 ).Invasive grass may be less effective at colonizing areas with an intact biocrust ( Peterson 2013 ;Chambers et al. 2016 ;Slate et al. 2019 ) because evidence strongly suggests that biocrust inhibits germination and root penetration of annual grasses ( Deines et al. 2007 ).The direction of this interaction seems to be dependent on microtopography: smooth biocrusts are likely to inhibit seed retention, increasing granivore predation, while rugose, pinnacled and rolling biocrusts are more likely to retain seeds and offer protection ( Havrilla and Barger 2018 ).
Disturbance of the biological soil crust has directly facilitated increased plant invasion in many ecosystems ( Deines et al. 2007 ;Hernandez and Sandquist 2011 ).Additionally, the loss or change in the mutualistic relationship that native plants (shrubs and/or bunchgrasses) have with biocrusts ( Chaudhary et al. 2020 ;Havrilla et al. 2020 ), which is often associated with disturbance ( Steven et al. 2018 ), can further limit native plant establishment and growth.
Moss-dominated biocrusts are commonly found in the Great Basin, particularly in association with sagebrush ( Condon and Pyke 2020 ).Moss-dominated biocrusts perform a key suite of ecosystem functions that are critical in dryland ecosystems ( Delgado-Baquerizo et al. 2016 ) and can be early, mid, or late-successional species depending on the disturbance type and severity, soil type, and relative aridity ( Weber et al. 2016 ).Much research has focused on their utility in restoration as they grow quickly, can be dominant, and have a suite of desirable ecosystem functions ( Antoninka et al. 2016 ;Condon and Pyke 2016 ;Doherty et al. 2018 ;Grover et al. 2022 ).
As dryland ecosystems around the world are at great risk of type conversion to invasive grasses, an understanding of the relative role of woody plants, perennial herbaceous plants, and biocrusts on ecosystem stabilization and invasion resistance is urgently needed.Our objective was to quantify the effects of moss biocrusts, native shrubs, and native perennial grasses on invasion resistance in a greenhouse experiment containing dominant sagebrush ecosystem plants.We hypothesized that 1) increased treatment complexity (shrub, grass, moss biocrust) would increase invasion resistance because there would be greater interspecific competition for resources by native plants, reducing biomass of invasive grasses; 2) the presence of moss biocrust would inhibit germination and growth of invasive grass, and 3) productivity of native shrubs and grasses would not be impacted by the presence of moss biocrusts.

Greenhouse
We used a full factorial greenhouse experiment (all possible combinations of moss/no moss, bitterbrush/sagebrush/no shrub, native bunchgrass/no bunchgrass, and cheatgrass/medusahead/no annual grass-for a total of 36 treatments with 5 replicates per treatment combination) to assess the effect of treatment group complexity, interspecific competition, and biological soil crusts on the growth of shrub seedlings, native perennial grass seedlings, and invasive annual grass establishment ( Fig. 1 ).Sagebrush ( Artemisia tridentata Nutt.ssp.tridentata ) seedlings were sourced from WinterCreek Nursery in Bend, Oregon, United States as 16.4 cm 3 -plugs and were transferred to 4.5-L (20.3 cm diameter by 21 cm height) greenhouse pots on June 12, 2018.Sagebrush was not inoculated with mycorrhizal fungi at the nursery and was grown in a sterile growth medium.Replicates containing sagebrush were started 3 wk before no shrub and bitterbrush replicates due to earlier availability of plant materials.Bitterbrush ( Purshia tridentata [Pursh] DC.) and native bunchgrasses (bluebunch wheatgrass [ Pseudoroegneria spicata (Pursh) A. Löve] and Idaho fescue [ Festuca idahoensis Elmer]) were grown from seed sourced from Great Basin Seed in Ephraim, Utah, United States and started in 128-cell greenhouse starter trays.Bitterbrush seedlings were transferred to 4.5-L pots on July 2, 2018.Actively photosynthesizing moss (mostly Syntrichia ruralis [Hedw.]F. Weber & D. Mohr) and invasive grass seeds (cheatgrass and medusahead) were collected from field sites near Dayville, Oregon, United States June 12, 2018.Moss biocrusts were propagated from field-collected dried fragments.Fifteen g of dried field-collected moss biocrust fragments were sieved through a 2-mm soil sieve, well mixed, and then applied across the soil surface, with a target initial cover of 20−25% moss fragments ( Serpe et al. 2006 ;Condon and Pyke 2016 ).Moss biocrust and native grasses were added to pots on June 20, 2018 for sagebrush pots and on July 11−13, 2018 for bitterbrush and no shrub plots.Each sagebrush and bitterbrush replicate had a single shrub seedling.Pots with native grasses contained one Idaho fescue seedling and one bluebunch wheatgrass seedling (see Fig. 1 ).We maintained pots at optimal conditions for plant germination and growth (12−29 °C, 51−106% humidity, no supplemental lighting) in greenhouses at Oregon State University in Corvallis, Oregon, United States.Seedlings were watered gently every 2−3 d, depending on ambient conditions, and were allowed to completely dry out between watering.Sunshine LA4-P growing mix was used for all treatments, which includes peat, perlite, and pumice media.
To simulate invasion by non-native annual grasses into treatment replicates, we spread 15 g of either cheatgrass or medusahead seeds on the surface of each pot on July 30, 2018 for replicates containing sagebrush and on September 5, 2018 for replicates containing bitterbrush or no shrub.Additional replicates were left as uninvaded controls.Heights of all shrubs were measured immediately before invasion to account for variability in starting size.Pots were watered when dry but not otherwise disturbed for 37−40 d.On September 5, 2018 (sagebrush) and October 15, 2018 (bitterbrush and no shrub), shrub heights were measured; aboveground plant biomass was clipped, sorted by functional group (shrub, native grass, or invasive grass), dried, and weighed to compare productivity; and cover of moss biocrust was calculated using a gridded 117-point frame set over the soil surface and relativized to get percent cover.

Analysis
To account for differences in shrub height at the start of the experiment, we ran Pearson correlations (r) to determine the extent to which shrub height at the time of invasion predicted end shrub biomass and end shrub height.Significant predictors were used as covariates in shrub biomass and height models.Linear mixed models describing shrub height and biomass also included fixed factors indicating shrub species (sagebrush or bitterbrush), presence or absence of moss biocrust, presence or absence of perennial grasses, and annual grass (cheatgrass, medusahead, or no annual grass addition).Linear mixed models for perennial grass biomass included presence of shrub (sagebrush, bitterbrush, or no shrub); annual grass (cheatgrass, medusahead, or no annual grass); and presence or absence of moss biocrust.Models for annual grass biomass included presence of shrub (sagebrush, bitterbrush, or no shrub), perennial grass (presence/absence); and moss biocrust (presence/absence).Analyses were performed with IBM SPSS 24 ( IBM Corp 2016 ).

Shrubs
At the experiment's end, the average height of sagebrush was 31.4 cm (SE = 1.7) and the average height of bitterbrush was 14.8 cm (SE = 0.8) ( Fig. 2 ).Final average biomass was 12.1 g (SE = 1.1) for sagebrush and 2.0 (SE = 0.3) for bitterbrush.Day 0 shrub height was a strong predictor of both end shrub height ( r = 0.90) and end shrub biomass ( r = 0.86) and was included in shrub models to control for this variability.Final height did not vary by shrub species after controlling for initial height (i.e., shrub growth rate; P = 0.86).Similarly, biomass did not vary by shrub species after controlling for initial height ( P = 0.48).
We saw evidence of competition between shrubs and native perennial bunchgrasses such that all planting treatments containing perennial grass reduced shrub height and shrub biomass ( Figs. 2 −3 ; Table 1 ).In the presence of perennial grass, shrub final height was reduced by an average of 7.9 cm ( P < 0.001) and final shrub biomass reduced by 4.75 g ( P < 0.001), with no significant

Table 1
Linear mixed models predicting the shrub height and shrub biomass as a function of initial shrub height, shrub species (sagebrush or bitterbrush), addition of moss biocrust, presence of perennial grasses, and presence of annual grasses (medusahead or cheatgrass) in a greenhouse experiment.difference by shrub species ( P > 0.05).Shrub height was not impacted by the presence of either invasive grass species ( P = 0.87; see Fig. 2 ).Shrub biomass in uninvaded plots was no different from that in cheatgrass and medusahead-invaded plots ( P = 0.08; see Fig. 3 ).After controlling for initial shrub height, shrub species, and presence of annual and perennial grasses, there was no evidence that the addition of moss biocrust to the soil surface modified shrub height ( P = 0.18; see Fig. 2 ) or shrub biomass ( P = 0.91; see Fig. 3 ).

Bunchgrasses
Final average biomass for perennial bunchgrasses across all treatments was 16.6 g (SE = 0.5; Fig. 4 ).Perennial bunchgrasses grown with sagebrush had an average of 4.5 g more biomass ( Table 2 ; P < 0.001) relative to the pots with bitterbrush or the pots with no shrubs.The presence of bitterbrush did not alter bunchgrass biomass ( P = 0.40) relative to the pots without shrubs.Perennial grass biomass was not significantly impacted by the presence of medusahead or cheatgrass ( P = 0.21), nor was there evidence that the addition of moss biocrust to the soil surface modified perennial grass biomass ( P = 0.48).

Invasive grasses
At the end of the experiment, average cheatgrass biomass across all pots was 2.5 g (SE = 0.4) and average medusahead biomass was 0.7 g (SE = 0.2) ( Fig. 6 ; see Table 2 ).The presence of perennial grasses suppressed invasive grass germination and establishment ( P < 0.01), but shrubs and moss biocrusts did not have an effect on biomass of invasive grasses ( P = 0.38 and P = 0.25, respectively).Final invasive grass biomass was 4.79 g on average when perennial grasses were not present and was reduced to 1.59 g on average with perennial grass competition .Uppercase letters denote statistically significant differences between planting groups, where pots with sagebrush had higher perennial grass biomass than planting groups without sagebrush.There were no significant differences between annual grass treatments.

Discussion
In this greenhouse study, the competitive and facultative effects of native perennial grasses, moss biocrust, and shrubs, as well as native community resistance to annual grass invasion, were quantified.We hypothesized that with greater treatment group complexity, there would be greater invasion resistance, but instead it was the presence of native bunchgrasses (alone and in combination) that drove invasion resistance.This result is consistent with our understanding of perennial bunchgrasses as a foundation species for ecosystem resistance and resilience and for postdisturbance stabilization of ecological processes ( Rodhouse et al. 2014 ;Davies and Johnson 2017 ).Bunchgrasses are critical to the structure and function of sagebrush steppe ecosystems for capture of soil nutrients ( James et al. 20 06 ,20 08 ), limiting gap space between perennial plants ( Reisner et al. 2013 ;Rayburn et al. 2014 ;Pyke et al. 2022 ), and providing a seed and plant material source for ecosystem stabilization and postdisturbance recovery.However, not all bunchgrasses may provide equivalent suppression of invasive grasses: In one study, Blank et al. (2020) found greater suppression of cheatgrass by non-native crested wheatgrass than by native Snake River or bluebunch wheatgrass, likely due to greater utilization of nutritional resources by the forage grass.Similarly, invasive annual grasses vary in the conditions in which they are most likely to invade.We found much less medusahead invasion in this study compared with cheatgrass, which is corroborated with recent work by Applestein and Germino (2022) , who found that medusahead invasion was more likely to be associated with existing cheatgrass invasion than with native dominance.Both invasive species occurrences, however, were negatively associated with deep-rooted perennial bunchgrasses ( Applestein and Germino 2022 ).
Contrary to our expectations, we saw no evidence that the addition of moss biocrust to the soil surface inhibited invasive germination and growth in this greenhouse experiment.Our experimental design tested the hypotheses under warm, moist growth conditions inside of a greenhouse with potting substrate starting with moss fragments.It is possible that the interactions seen in other studies ( Su et al. 2007 ;Peterson 2013 ;Chambers et al. 2016 ;Slate et al. 2019 ) were not present in our study because they are contingent on environmental factors, life stage, or growth form of plants involved in the interaction ( Pando-Moreno et al. 2014 ;  ( Slate et al. 2019 ;Root et al. 2020 ) or facilitation ( Ferrenberg et al. 2018 ) of invasive grass germination by biocrusts in field settings, but these often uncontrolled real-world settings also have complicated interactions with altered soil texture (Chambers et al. 2016) or fertility ( Ferrenberg et al. 2018 ), livestock grazing ( Root et al. 2020 ) or grazing exclusion ( Condon et al. 2020 ), wildfire ( O'Connor and Germino 2021 ), and inherent ecosystem resistance and resilience.However, in these more complex systems, important nuances become evident: Slate et al (2019) found, for example, that in a field experiment there was greater inhibition of exotic than of native grasses by biocrusts but equivalent suppression of exotics and natives in a parallel greenhouse setting.Our study and others have used greenhouse methods that result in rapid establishment and growth of early successional biocrusts that are predominantly Syntrichia ruralis moss, with limited presence of cyanobacteria or lichens, which may limit our inference.Other studies have found that intact, more complex cyano-lichen-moss biocrust communities do limit exotic plant germination or establishment ( Hernandez and Sandquist 2011 ;Eldridge et al. 2020 ).Relationships among perennial grasses, sagebrush and other shrubs, invasive grasses, and moss biocrusts are also affected by local environmental factors such as moisture regimes, soil types, soil temperatures, elevation, climate, and annual precipitation ( Chambers et al. 2007 ;Reisner et al. 2013 ).
Finally, we hypothesized that productivity of native shrubs and grasses would not be impacted by the presence of moss biocrusts, which was consistent with our results.This study also supports perennial bunchgrasses as key providers of resistance and resilience in sagebrush steppe.As we might expect in small pots with dense plantings, both sagebrush and bitterbrush plants were larger when grown without native bunchgrasses.Unexpectedly, however, we saw that perennial bunchgrasses had higher biomass when they were grown with sagebrush than when they were grown with bitterbrush or without a shrub, indicating that the sagebrush may be providing a nurse plant effect or other increased access to resources for the bunchgrasses ( Huber-Sannwald and Pyke 2005 ; Padilla and Pugnaire 2006 ).Because the sagebrush plants were the only ones in our experiment sourced externally, we are cautious in interpreting this result, though we ruled out fertilization or inoculation by personnel at the source greenhouse.Reisner and colleagues (2015) found evidence of facilitation between sagebrush and grasses at low levels of water and heat stress, but competition when stress was high, and it is possible that our greenhouse conditions are mirroring that low-stress environment and interspecific interaction.

Management and research implications
Research has suggested that monitoring community complexity can provide early warning signs of ecosystem state change after disturbances ( Reisner et al. 2013 ).However, there is evidence that appropriately planning for the needed sequencing to build that complexity when restoring native ecosystem structure and function is important and may require a two-step process-first converting the annual grass−dominated site to a resilient perennial grass−dominated site, followed by planting less resilient native species ( Cox and Anderson 2004 ).Our results are congruent with those findings as treatments with perennial bunchgrasses were the only ones to show a decrease in annual grass productivity.Further understanding of the role of species-specific interactions, successional stages, and differential invasion niches (Applestein and Germino 2022) may be key to facilitating more successful restoration effort s by maximizing complexity and reducing competition with invasive grasses in the sagebrush biome and in dryland ecosystems around the world.

Figure 2 .
Figure 2. Sagebrush (top panel) and bitterbrush (bottom panel) seedling heights in a greenhouse experiment when shrubs (Sh) are grown in competition with invasive grasses (cheatgrass or medusahead) and perennial bunchgrasses ( G, Idaho fescue and bluebunch wheatgrass) and/or moss biocrust (C).Boxes represent the 25th−75th percentiles, with horizontal lines depicting the median.Error bars show 10th and 90th percentiles.Uppercase letters denote statistically significant differences between planting groups ( Sh, G, and/or C ), with lower shrub biomass in pots shared with perennial grass.

Figure 3 .
Figure 3. Sagebrush (top panel) and bitterbrush (bottom panel) seedling biomass in a greenhouse experiment when shrubs (Sh) are grown in competition with invasive grasses (cheatgrass or medusahead) and perennial bunchgrasses ( G, Idaho fescue and bluebunch wheatgrass) and/or moss biocrust (C).Boxes represent the 25th−75th percentiles, with horizontal lines depicting the median.Error bars show 10th and 90th percentiles.Uppercase letters denote statistically significant differences between planting groups ( Sh, G, and/or C ), with lower shrub biomass in pots shared with perennial grass.Medusahead had a minor suppressive effect on shrub biomass relative to cheatgrass, but neither invasive grass differed from uninvaded pots.

Figure 4 .
Figure 4. Perennial grass (G) seedling biomass in a greenhouse experiment when grown in competition with sagebrush (S), moss biocrust (C), and bitterbrush (B).Treatment plots were then invaded with medusahead (striped bars), cheatgrass (white bars), or no invasive grass (black bars).Boxes represent the 25th−75th percentiles, with horizontal lines depicting the median.Error bars show 10th and 90th percentiles.Uppercase letters denote statistically significant differences between planting groups, where pots with sagebrush had higher perennial grass biomass than planting groups without sagebrush.There were no significant differences between annual grass treatments.

Figure 6 .
Figure 6.Annual grass seedling biomass in a greenhouse experiment when grown in competition with sagebrush (S), moss biocrust (C), and bitterbrush (B).Treatment plots were then invaded with medusahead (striped bars) or cheatgrass (white bars).Boxes represent the 25th−75th percentiles, with horizontal lines depicting the mean.Error bars show 10th and 90th percentiles.

Figure 7 .
Figure 7. Summary of shrub, perennial and annual grass biomass, and moss biocrust cover (rows) in response to competition with shrubs (sagebrush or bitterbrush), perennial grass (Idaho fescue and bluebunch wheatgrass), moss biocrust, and annual grasses (cheatgrass or medusahead) additions (columns) in a greenhouse experiment.

Table 2
Linear mixed models predicting perennial bunchgrass biomass, moss biocrust cover, and invasive grass biomass as a function of shrub species (sagebrush or bitterbrush), addition of moss biocrust, presence of perennial grasses, and presence of annual grasses (medusahead or cheatgrass) in a greenhouse experiment.