Introduced annuals mediate climate‐driven community change in Mediterranean prairies of the Pacific Northwest, USA

How climate change will alter plant functional group composition is a critical question given the well‐recognized effects of plant functional groups on ecosystem services. While climate can have direct effects on different functional groups, indirect effects mediated through changes in biotic interactions have the potential to amplify or counteract direct climatic effects. As a result, identifying the underlying causes for climate effects on plant communities is important to conservation and restoration initiatives.


| INTRODUC TI ON
Climate change is altering plant communities by promoting widespread invasions (Thuiller et al., 2008), reshuffling compositional assemblages (Pecl et al., 2017) and contributing to the loss of species diversity (Harrison et al., 2015). These changes play out through direct effects of temperature and moisture on individual species, as well as indirect effects of temperature and moisture on interspecific interactions such as competition. A major challenge to effectively managing plant communities under climate change is identifying the direct and/or indirect pathways by which these changes occur (Avolio et al., 2015;Gornish & Tylianakis, 2013). For example, moisture, rather than temperature, could be the primary influence on plant community responses (Franklin et al., 2016), or vice versa. Biotic interactions can act as important intermediaries (Brooker, 2006). Nutrient availability may be the driver of change but be dependent on climate (Brooks, 2003). In many cases, the underlying mechanisms by which temperature, moisture, nutrients and biotic interactions influence change in plant communities remain a black box.
There is debate whether direct or indirect effects dominate climate-driven plant community change. Theoretical evidence suggests that direct climatic effects are more important in some grassland plant communities (Chu et al., 2016), which has been corroborated in some empirical studies. For example, in a Kansas prairie, the direct effects of temperature had greater influence than species interactions on the population dynamics of ten forb species (Adler & HilleRisLambers, 2008). However, alternative evidence highlights the important roles of indirect effects, given their potential to amplify or counteract direct effects (Adler et al., 2009).
Studies in California grasslands with Mediterranean-climate systems support both direct and indirect effects of climate on plant communities. For example, Levine et al. (2010) found that the precipitation responses of native annual forbs were minimally affected by the responses of their competitors. However, LaForgia et al. (2020) found that drought effects on native annual forbs (especially resource-acquisitive ones) were exacerbated when invasive winter-annual grasses were present, suggesting a strong mediation of climate through competition. Suttle et al. (2007) found that while extending the wet season initially boosted plant species richness, these effects were reversed once increased nutrient availability prompted a shift from forbs to winter-annual grasses, which suppressed other species through their litter accumulation. In prairie and grassland systems such as this, winter-annual invasives may amplify or counteract direct climatic effects on other functional groups by altering soil resources such as moisture or nitrogen (Prevéy & Seastedt, 2014), potentially shifting competitive hierarchies in their favour (Everard et al., 2010).

The Pacific Northwest (PNW) of the United States is a
Mediterranean-climate region defined by its mild, wet winters and warm-to-hot, dry summers (Kottek et al., 2006). Like many other Mediterranean-climate ecosystems across the globe, PNW prairies are critically endangered (Klausmeyer & Shaw, 2009;Noss et al., 1995;Sala et al., 2000). Following Euro-American settlement, most of these prairies have been lost to land use change or considerably altered by non-native plant invasions (Dunwiddie & Bakker, 2011). While introduced winter-annual species achieved dominance in the hotter, drier California grasslands (Clary, 2012), introduced perennial grasses were the primary invaders in the cooler, moister PNW (Sinclair et al., 2006). Looking towards the future, how climate change will alter these (and other Mediterranean climate) plant communities is an important conservation question.
Models for the PNW project a temperature increase of at least 2.5°C by the 2080s (compared to 1970-1999), with changes in precipitation being less certain (Dalton & Fleishman, 2021;Mote & Salathé, 2010). In general, the seasonality of the Mediterraneanclimate system is expected to become more extreme, with winters becoming wetter and summers becoming longer and drier. Increased precipitation variability and extreme events are likely (Pendergrass et al., 2017). Drought potential may rise due to increasing evaporative demands, particularly in the valley lowlands (Jung & Chang, 2012), where many of these prairies occur.
Previous research by our group suggested that future climate change will alter plant community composition and decrease diversity in PNW prairies. Population-level studies of native annual and perennial species revealed declining demographic performance under experimental warming (Pfeifer-Meister et al., 2013;. At the community level, warming increased the cover of introduced annual species, causing a "Californication" phenomenon in which communities became more similar to invaded grasslands further south in California (Pfeifer-Meister et al., 2016). Although this was hypothesized to be due to a reduction in soil moisture, temperature and moisture were confounded: adding +20% precipitation to warming did little to alleviate the warming-induced drying effects. Additionally, the role of introduced annuals in mediating climatic effects, or whether nutrient availability played any role, was unexplored. Thus, it is difficult to ascertain whether these climate-driven community changes can be primarily attributed to temperature, moisture, nutrients or other factors.
Here, we built upon this previous research with 3 years of new data, decoupling the direct warming effects from their drying effect. We used structural equation modelling to examine the extent to which abiotic drivers (i.e. temperature, moisture and nitrogen) controlled functional group cover, and how these groups in turn determined diversity, richness and evenness. Utilizing a climate We hypothesized that warming would favour introduced annual species at the expense of other functional groups and would thus negatively affect diversity. In particular, we expected a direct positive effect of temperature on introduced annual cover, as many of these species in this system are winter annuals (Dennehy et al., 2011) that may benefit from more favourable early growing season conditions (Dunwiddie et al., 2014;Pfeifer-Meister et al., 2016). Concurrently, we expected increasing temperatures to reduce soil moisture, augmented by feedback from introduced annuals, whose rapid growth draws down available soil water (Prevéy & Seastedt, 2014). We also expected a decline in nitrogen (N) availability due to reduced soil moisture and greater uptake by introduced annuals, shifting the competitive hierarchy in their favour (Everard et al., 2010;Prevéy & Seastedt, 2014). As a result, we expected the cover of later-growing functional groups (i.e. introduced perennials, native perennials and native annuals) to decline with warming and drying, and for this to be primarily mediated through increasing competitive pressure by introduced annuals. Finally, we expected these changes to result in a decline in diversity, induced by losses in both richness and evenness.

| Experimental design
We conducted this study at three prairie sites spanning a 520-km latitudinal gradient across the western PNW ( Figure 1a). Sites were selected to reflect the Mediterranean-climate gradient of increasing temperature and summer drought severity from north to south. All plots were initially established for native plant demography experiments (Pfeifer-Meister et al., 2013;. In 2009, the southern and central plots were treated with herbicide (2% Glyphosate (RoundUp) and 0.75% grass-specific Fusilade) and then sown with identical mixtures of 31 native prairie species to establish the first demography experiment   . Finally, all 60 plots had additional sowings of 14 focal native species in fall 2015, 2016 and 2017.
By September 2016, we initiated four climate treatments: control, drought, warming and warming + irrigation (five plots per treatment per site; Figure 1b). Controls experienced ambient temperature and precipitation, the drought treatment reduced annual precipitation by ~40% using fixed rainout shelters, and the warming treatments increased canopy temperature by approximately +2.5°C with infra-

| Climate and nitrogen data
Between fall 2016 and summer 2019, we continuously logged plot canopy and soil temperatures and soil volumetric water content aggregated to daily averages. Missing values due to equipment malfunction were interpolated (see Appendix S1). To compare moisture across sites with different soil characteristics, we calculated soil matric potential from volumetric water content, soil texture and soil carbon (Saxton & Rawls, 2006; see Appendix S1). Matric potentials are negative values where 0 kPa indicates saturation and −1500 kPa indicates permanent wilting point. In the Mediterranean climate of the PNW, active vegetation growth occurs mainly in the wet season (October-June), followed by several months of dormancy during the summer drought (July-September). Therefore, we aggregated soil temperature and moisture data to the growing season from October 1 to June 30 (Reed et al., 2019). We also investigated annual, winter and spring seasonal values, but only report growing season here because trends were consistent.
We collected data on total inorganic nitrogen (N) availability (µg/10 cm 2 strip/burial period) using Plant Root Simulator and 2018 (August 2017-July 2018) but lacked these data for 2019.

| Plant community data
In each spring 2017-2019, we measured plant cover at peak standing biomass (approximately mid-May, late-May and mid-June at the southern, central and northern sites, respectively) using the point intercept method (Elzinga et al., 1998). Using a 1-m 2 quadrat in a fixed location within each plot, we dropped 25 equally spaced pins (4.8 mm diameter) through the plant canopy. We counted each plant contact with a pin (hit) to the species level and multiplied hits by four (to scale to 100). Due to canopy layering, >100% cover occurred.
The same four researchers counted pin hits during all 3 years of the study and calibrated pin-hit counting techniques together to ensure consistency across researchers. We assigned a cover of 0.4% to species, which were present in the quadrat but not hit by a pin. Because pin diameter affects counts, the method is more precisely described as a cover index than an absolute cover measure, and pin size can be adjusted for a given study without introducing bias (Jonasson, 1988).
Point-intercept methods have proved effective and accurate across a wide range of grassland productivity and diversity (Barkaoui et al., 2013).
To determine the functional group covers, we assigned each species to an origin (native/introduced), duration (annual/perennial, with the few biennials assigned as annuals), and growth form (grass/ where p i is the relative cover of species i and s is the richness, or total number of species. Lastly, we calculated Simpson's evenness (Morris et al., 2014;Simpson, 1949) as:

| Analyses
We conducted analyses using R version 4.0.2 (R Core Team, 2020).
Due to repeated sampling of the plots, we ran all analyses using mixed models (lme4 package; Bates et al., 2015) with plot as a random effect. Marginal and conditional R 2 values were calculated based on Nakagawa et al. (2017), with R 2 m providing the variance explained by the fixed effect(s) alone and R 2 c providing the variance explained by both fixed and random effects. Because heaters and irrigation were turned off after July 2018, we excluded data from the warming and warming + irrigation plots in 2019 to prevent any legacy treatment effects from obscuring patterns.
To improve the normality of distributions for several variables, we used the following transformations. We transformed soil s moisture using −log(1 − matric potential) so that the transformed values stayed consistent with raw values (more negative = drier, less negative = moister). We used square root transformations for N availability and the cover of both introduced annuals and introduced perennials. For native perennial and native annual cover, we used a log transformation after adding a constant of 0.4% as these functional groups had 0% cover in some plots. Lastly, because diversity and evenness were proportional data, we used logit transformations (Warton & Hui, 2011).
Differences in the abiotic variables (temperature, moisture and N availability) were examined with ANOVA with site as a fixed effect followed by Tukey's post hoc comparisons if site was significant (p < .05). We explicitly left site out as a predictor from all subsequent analyses because sites were chosen to represent a climate gradient and would obscure the continuous-data climate signals.
Using the piecewiseSEM package (Lefcheck, 2016), we examined our a priori hypothesized causal pathways with structural equation models (SEM), which linked together a series of multiple regressions fitted using lme4 with plot as a random effect. In SEMs, overall model fit is determined by testing the significance of unspecified causal pathways (i.e. a well-supported model has a high goodnessof-fit p-value). We investigated Pearson's correlation coefficients and checked for multicollinearity using variance inflation factors (VIFs). Concerns over multicollinearity were minimal as the greatest VIF was 3.4 (Table S1) and, in general, VIFs >5 may indicate multicollinearity (James et al., 2013). We ran separate SEMs for diversity, richness and evenness. Although we hypothesized that biotic effects would primarily occur through introduced annuals, we also accounted for the possibility of introduced perennials affecting the two native functional groups, as introduced perennials are already widely prevalent throughout this system and have strong competitive effects on native prairie species (Sinclair et al., 2006;Stanley et al., 2011). We

| Abiotic variables
Across all treatments, mean growing season soil temperature, moisture and N availability differed by site (p < .001). From north to south, average plot conditions became warmer (northern: 9.8 ± 1.1°C (mean ±s.d.), central: 11.8 ± 1.4°C, southern: 11.9 ± 0.9°C) and drier (northern: −11.4 ± 10.8 kPa, central: −66.3 ± 44.7 kPa, southern: −239.0 ± 172.6 kPa), although the central and southern plots were not significantly different for temperature (Figure 1). Treatmentspecific means in temperature and moisture differed by site and year ( Figures S1 and S2). Mean N availability was lowest in the central plots, while the northern and southern plots did not significantly differ ( Figure S3). There was a significant positive correlation between temperature and introduced annual cover, and significant negative correlations between temperature and soil moisture, introduced annual cover and soil moisture, and soil moisture and N availability ( Figure S4).

| Functional group cover
Overall, introduced annuals had the greatest average cover (505.4%, range: 0%-1596.4%), followed by introduced perennials (228.3%, range: 0%-900.8%), native perennials (121.0%, range: 0%-564.8%) and native annuals (30.0%, range: 0%-328.4%). Native annual, native perennial and introduced perennial cover all had significant, positive correlations with one another but significant, negative correlations with introduced annual cover ( Figure S5). Introduced annuals increased with temperature and had a negative relationship with soil moisture (Figure 3a,e; Table S4), while the other three functional groups all decreased with increasing temperature and increased with soil moisture (Figure 3b-d,f-g; Table S4). Functional group cover was not related to N availability (Figure S7a-d; Table S4), nor did N availability ever act as a significant mediator in the SEMs ( Figure S6; Table S2). Therefore, we focus on the SEMs, which include 2019 data (and exclude N availability).
We found the effects of temperature on introduced perennials and native perennials to be primarily mediated through its effects on introduced annuals and soil moisture (Figure 2). In particular, soil

| Diversity, richness and evenness
Diversity declined with increasing temperature and increased with soil moisture (Figure 4a,d; Table S4). It also declined with increasing introduced annual cover but increased with cover for the three other functional groups (Figure S8a-d; Table S4). Using SEM, we found that introduced annual cover had a negative effect (−0.37 std. est.) and native annual cover a positive effect (0.29 std. est.)

F I G U R E 2
Structural equation model using 2017-2019 data to determine how climate controls functional groups and how functional groups control diversity. Values within/next to arrows (direct effects) are standardized estimates of coefficients, and arrow thickness is scaled to their magnitude. Red = negative effects, black = positive effects, semi-transparent = non-significant effects (p > .05; Table S3). Within each response variable box (those at the receiving end of arrows), R 2 m provides the variance explained by the fixed effect(s) alone (boxes at the starting end of the arrows) and R 2 c provides the variance explained by both fixed and random effects (plot = random).       Table S3). There was a 45% reduction in predicted diversity from minimum to maximum introduced annual cover and a 31% increase from minimum to maximum native annual cover. The SEM for diversity had good model fit (goodnessof-fit: p = .973).
Like diversity, richness also declined with increasing temperature and increased with moisture (Figure 4b,e; Table S4), whereas evenness was not affected by temperature and weakly declined with moisture (Figure 4c,f; Table S4). Additionally, richness declined with increasing introduced annual cover but increased with cover for the three other functional groups (Figure S8e Figure S9b).

| D ISCUSS I ON
A major challenge in community ecology is to tease apart the complex relationships behind climate-driven community change. This challenge may be best-achieved using large-scale, multisite experiments analysed with path-analytic approaches (Brooker, 2006;Gornish & Tylianakis, 2013). Using a climate manipulation experiment embedded in three Mediterranean-prairie sites across a latitudinal gradient in the PNW, we used SEM and found that warmer conditions resulted in increasing cover of introduced annuals, along with subsequent declines in other functional groups and diversity.
Our data not only confirm results from our previous study (Pfeifer-Meister et al., 2016), but identify the causal pathways underlying the observed climate-community relationships.
Our first hypothesis, that warmer conditions favour introduced annuals, was strongly supported. Indeed, we found a direct positive effect of temperature on introduced annual cover, which we suspect was due to the early winter-growth habits common to this group. The most abundant introduced annuals in our study, Vulpia spp. (V. myuros and V. bromoides), Bromus spp. (B. hordeaceus and B. tectorum) and Trifolium subterraneum, were orders of magnitude more common than other annuals ( Figure S10) and are characterized as winter-growing, early-maturing species (Friddle, 2018;USDA-Natural Resources Conservation Service, 2012;USDA-NRCS, 2014). Warming could be advantageous to such species by promoting growth during the early growing season (Blumenthal et al., 2016;Cleland et al., 2006), at a time when temperature is most limiting and the rest of the community is still largely dormant.
Whether this favouring of introduced annuals came at the expense of other functional groups (i.e. through competition for limiting resources) was partially supported. Invasive annual grasses such as Bromus can rapidly exploit and draw down soil water resources (Dyer & Rice, 1999;Melgoza et al., 1990). These grasses generally reached peak flowering in our study by mid-April, whereas most other functional groups peaked around mid-May to mid-June (P.
Reed, pers. obs.), when soil moisture became a critically limiting resource ( Figure S2 completing their entire life cycle prior the onset of extreme summer drought. However, it is important to note that the direct negative effect of temperature on soil moisture was greater than that from introduced annuals, indicating that competition from this functional group amplifies, but does not entirely control, moisture limitation for the rest of the community. We found minimal support for N limitation acting as an important intermediary in these community dynamics. Soil moisture is the transport medium for plant-available N (Everard et al., 2010), but we actually found a slightly negative relationship between N availability and soil moisture ( Figure S4), driven by relatively high N availability at our driest (southern) site ( Figure S3). Legumes (e.g. Trifolium and Vicia spp.) were among the most common introduced annuals in this study ( Figure S10), and Trifolium subterraneum was especially common at the southern site in 2017. Thus, increases in N-fixation may have offset decreases in N availability due to drier conditions.
Alternatively, N availability could be controlled by other site-specific factors. Regardless, N availability did not affect the cover of introduced perennials, native perennials or native annuals, and was poorly predicted by temperature, soil moisture and introduced annual cover Figure S6). The direct effects of introduced annuals on the two perennial functional groups suggest that an additional competition mechanism is at play. In our parallel demography experiment, we speculated that poor recruitment among native perennials was driven by competition from introduced annual grasses .
Poor recruitment ultimately led to fewer adult-stage individuals, hence the lower cover observed here. In a similar California grassland system, litter accumulation positively reinforces the biomass of introduced annual grasses to suppress native recruitment (Mariotte et al., 2017). Likewise, seedling establishment for both native and introduced perennials declined with increasing litter depth in our plots (Brambila et al., unpublished data). Given that existing vegetation was initially removed from our plots, introduced perennials had to re-establish through the seedbank, while native perennials came in via our seed additions. As warmer conditions promoted the rapid biomass growth of introduced annuals, they formed a dense cover early in the season (Vulpia and Bromus spp. often reaching 20+ hits per pin) coupled by a thick litter layer that remained the following season. Such conditions shade out young seedlings of slowergrowing perennials during their critical establishment phase. Thus, an additional mechanism mediated by introduced annuals seems to be through increasing competition for space and light.
There was also a strong direct effect of moisture on introduced perennials. In California, introduced perennial grass cover is strongly correlated with proximity to the moist coastline (Clary, 2012).
Likewise, in PNW prairies, this functional group was highly successful at invading cool, moist sites (Sinclair et al., 2006). Although the land surrounding our plots at the driest (southern) site is dominated by introduced perennial grasses, the property is regularly irrigated from spring to summer (P. Reed, pers. obs.). The strong effect of moisture on introduced perennials implies that, even in the absence of competitive pressure by introduced annuals, future moisture limitation may prove detrimental for this group. From a management perspective, any decline in a non-native functional group may appear desirable on the surface. However, some non-natives can play an important role relative to management objectives (Dunwiddie & Rogers, 2017). For example, in PNW prairies, the introduced perennial forb Plantago lanceolata serves as a critical host for the endangered Taylor's checkerspot butterfly (Dunwiddie & Rogers, 2017), and several other introduced perennial forbs provide important nectar resources for a variety of pollinators (Lindh, 2018). Introduced perennial grasses provide important forage value for livestock throughout the region. Given the potential undesirability of a perennial-to-annual state transition (D'Antonio & Vitousek, 1992), any perennial cover (whether native or introduced) may be worth preserving to some degree.
As for native annuals, the direct effects of temperature and moisture strongly support our parallel demography study, which suggests that climate change will cause many native annuals to decline, regardless of competition .
Although several species in that study were strongly disturbancedependent, population growth rates rapidly deteriorated with warming even under favourable disturbance regimes (i.e. low competition). Importantly, the two most abundant native annuals in this study (Collinsia grandiflora and Plectritis congesta; Figure S10) were also the two least affected by disturbance in that demography study . Thus, their relatively high abundances may have influenced the negligible direct effect from introduced annuals.
Surprisingly, there was a positive effect from introduced perennials to native annuals (0.36 std. est., Figure 2). We included this pathway as introduced perennials have long been directly outcompeting native species in this system (Sinclair et al., 2006;Stanley et al., 2011). However, the positive relationships that we found between native annual, native perennial and introduced perennial cover suggest that these three functional groups can coexist under common favourable conditions without inherent competitive exclusion ( Figure S5). Therefore, we suspect that the significant positive arrow from introduced perennials to native annuals is correlative, rather than causative, as both functional groups were most common at the northern site, where cooler and moister conditions are the true underlying causes. Throughout PNW prairies, native annuals have already declined to an alarming degree (Dunwiddie et al., 2014). Our results suggest that this trend may continue with climate change, whether introduced annuals are present or not.
We found support for the hypothesis that diversity will decline with warmer and drier conditions (Harrison, 2020). This result was driven mostly by changes in the two annual functional groups.
Similar to California annual grasslands (Harrison et al., 2018), native annual cover was a strong factor contributing to richness. Soil moisture was also a strong driver of richness ( Figure S9b), consistent with several grassland studies that find positive relationships between moisture and richness (Adler & Levine, 2007;Hallett et al., 2014;Harrison et al., 2018). If soil water availability declines in the future (due to more frequent droughts, higher evaporative demand, and/or greater competition), native annuals may drop out of these communities, thereby reducing overall diversity through a decline in species richness.
Intriguingly, introduced annuals played a large role in affecting diversity not by their effects on richness, but by their strong negative effects on evenness ( Figure S9). This result is consistent with a study conducted in a more arid sagebrush-grassland ecosystem, in which communities with more introduced annuals (e.g. Bromus tectorum) had lower evenness (despite higher richness) as only a few species dominated (Allen & Knight, 1984). This finding is also consistent with the idea of the "invasion paradox," in which native and non-native species richness can have both a negative or positive relationship, depending on spatial scale (Fridley et al., 2007).
With plant communities, complete local extinctions resulting from invasive competitors are generally rare (Gurevitch & Padilla, 2004), especially when the natives and invaders are of different functional groups. Consequently, it is possible that a time-lag to local extinction could explain why the effects of the introduced annuals on diversity are driven by evenness rather than richness (Fridley et al., 2007).
In our study, this phenomenon seems to be driven by fundamental differences between the central and southern sites.
Introduced annuals were the most species-rich functional group ( Figure S10), and most of their richness occurred at the central site. Vulpia spp. and Bromus hordeaceus tended to dominate in these plots (hence low evenness; Figure S8i), but the central plots were moist enough to maintain relatively high richness (Figure 4e).
Richness plummeted under the driest conditions of the southern plots (Figure 4e). These conditions so highly favoured Vulpia spp. and Bromus hordeaceus that there actually became a tipping point for evenness: cover was almost exclusively these two species, often in a roughly equal split. This may explain why evenness was actually highest in the driest (southern) plots ( Figure 4f).
However, we caution that this evenness-moisture bivariate relationship was weak (R 2 m = 0.03; R 2 c = 0.23), and the total proportion of variance explained for evenness in its SEM remained relatively low (R 2 m = 0.22; R 2 c = 0.41). Overall, diversity appears poised to decline with climate change due to an increasing dominance by only a handful of introduced annuals.
While our large-scale experiment and SEM approach were powerful ways to identify the pathways by which climate alters plant communities, there are a few caveats. Separate from its effects on the aboveground community, climate change can also directly and/or indirectly affect the soil seed bank (Ma et al., 2020), which we do not consider in this study. However, seedbank responses to climate change may parallel or even amplify aboveground responses, despite being independent (Basto et al., 2018). Additionally, there are likely bidirectional, causative relationships between variables (i.e. feedback loops) and missing pathways not considered in these models. However, given our previous evidence (Pfeifer-Meister et al., 2016), we feel our hypothesized model structure represents the most likely reality.
Finally, our plots were relatively early-successional, "restored" communities that were established following a clearing disturbance. It is quite possible that effects would be different in established, undisturbed communities, where priority effects exist (Corbin & D'Antonio, 2004). However, this makes our results highly relevant for local prairie restoration efforts, which must consider the complex relationships between disturbance, invasives and climate if they are to be successful.
In conclusion, while we found direct negative effects of warming and drying on extant vegetation (i.e. native annuals, native perennials and introduced perennials), we also found that these effects were commonly amplified by increasing introduced annuals. These invaders directly benefit from warmer temperatures and seem to out-compete other functional groups for limiting resources such as moisture, space and light. Thus, introduced annuals appear to be the only functional group in this system that are poised to prosper under warmer and drier conditions. Other groups and overall diversity will decline because of this shift in dominance and/or due to direct negative effects of climate change. This shift towards increasing introduced annual dominance may be akin to that previously experienced in California grasslands, resulting in the "Californication" of PNW prairies. Such changes will challenge local land managers in their efforts to maintain species-rich and functionally diverse prairie ecosystems in the future.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest to declare.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/ddi.13426.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data are archived at the Dryad digital data repository: https://doi.