Prairie plant phenology driven more by temperature than moisture in climate manipulations across a latitudinal gradient in the Pacific Northwest, USA

Abstract Plant phenology will likely shift with climate change, but how temperature and/or moisture regimes will control phenological responses is not well understood. This is particularly true in Mediterranean climate ecosystems where the warmest temperatures and greatest moisture availability are seasonally asynchronous. We examined plant phenological responses at both the population and community levels to four climate treatments (control, warming, drought, and warming plus additional precipitation) embedded within three prairies across a 520 km latitudinal Mediterranean climate gradient within the Pacific Northwest, USA. At the population level, we monitored flowering and abundances in spring 2017 of eight range‐restricted focal species planted both within and north of their current ranges. At the community level, we used normalized difference vegetation index (NDVI) measured from fall 2016 to summer 2018 to estimate peak live biomass, senescence, seasonal patterns, and growing season length. We found that warming exerted a stronger control than our moisture manipulations on phenology at both the population and community levels. Warming advanced flowering regardless of whether a species was within or beyond its current range. Importantly, many of our focal species had low abundances, particularly in the south, suggesting that establishment, in addition to phenological shifts, may be a strong constraint on their future viability. At the community level, warming advanced the date of peak biomass regardless of site or year. The date of senescence advanced regardless of year for the southern and central sites but only in 2018 for the northern site. Growing season length contracted due to warming at the southern and central sites (~3 weeks) but was unaffected at the northern site. Our results emphasize that future temperature changes may exert strong influence on the timing of a variety of plant phenological events, especially those events that occur when temperature is most limiting, even in seasonally water‐limited Mediterranean ecosystems.

plant phenological responses at both the population and community levels to four climate treatments (control, warming, drought, and warming plus additional precipitation) embedded within three prairies across a 520 km latitudinal Mediterranean climate gradient within the Pacific Northwest, USA. At the population level, we monitored flowering and abundances in spring 2017 of eight range-restricted focal species planted both within and north of their current ranges. At the community level, we used normalized difference vegetation index (NDVI) measured from fall 2016 to summer 2018 to estimate peak live biomass, senescence, seasonal patterns, and growing season length. We found that warming exerted a stronger control than our moisture manipulations on phenology at both the population and community levels.
Warming advanced flowering regardless of whether a species was within or beyond its current range. Importantly, many of our focal species had low abundances, particularly in the south, suggesting that establishment, in addition to phenological shifts, may be a strong constraint on their future viability. At the community level, warming advanced the date of peak biomass regardless of site or year. The date of senescence advanced regardless of year for the southern and central sites but only in 2018 for the northern site. Growing season length contracted due to warming at the southern and central sites (~3 weeks) but was unaffected at the northern site. Our results emphasize that future temperature changes may exert strong influence on the timing of a variety of plant phenological events, especially those events that occur when temperature is most limiting, even in seasonally water-limited Mediterranean ecosystems.
Shifts have been observed at the individual species level (Fitter & Fitter, 2002;Whittington, Tilman, Wragg, & Powers, 2015), as well as for entire plant communities (Sherry et al., 2007;Theobald, Breckheimer, & Hille Ris Lambers, 2017), through both observational and manipulative studies. At the plant population level, the first appearance of flowers as well as the timing of peak flowering has important consequences for reproductive success and population viability. Phenological shifts in flowering may create asynchronies among interacting species (Yang & Rudolph, 2010), potentially disrupting mutualisms such as pollination or seed dispersal (Rafferty, Caradonna, & Bronstein, 2015), or result in mismatches with favorable environmental conditions, increasing the potential for detrimental events such as frost damage (Inouye, 2008). Shifts in phenology may also alter demographic vital rates and influence range distributions, which in turn can have large implications for patterns of biodiversity and species extinctions or persistence (Chuine & Beaubien, 2001;Miller-Rushing, Høye, Inouye, & Post, 2010;Parmesan & Yohe, 2003). At the community level, changes to the timing of seasonal biomass growth and senescence can affect processes such as primary productivity, carbon cycling, and competition (Cleland et al., 2007;Tang et al., 2016).
Despite ample evidence of recent phenological shifts, the controls of future shifts are not well understood. Temperature is typically viewed as one of the strongest controls of plant phenology, although other abiotic factors such as photoperiod and moisture can also exert influences (Moore, Lauenroth, Bell, & Schlaepfer, 2015;Rathcke & Lacey, 1985). Phenological events tend to advance with warming and are generally thought to be delayed with drought (Menzel et al., 2006;Wolkovich et al., 2012), although there are conflicting reports regarding the latter (Bernal, Estiarte, & Peñuelas, 2011;Cui, Martz, & Guo, 2017). While most studies have focused on temperature, moisture may be a stronger control than temperature for late-flowering species (Moore & Lauenroth, 2017) and is especially critical in water-limited ecosystems (Crimmins, Crimmins, & David Bertelsen, 2010;Diez et al., 2012). In Mediterranean climate regions, which are characterized by pronounced cool/wet and warm/dry seasons, moisture becomes increasingly limiting during the latter part of the growing season. Water availability thus becomes a critical factor, and moisture manipulation has been shown to affect plant phenology within Mediterranean regions (Bernal et al., 2011;Hänel & Tielbörger, 2015). Moisture has even been shown to have greater influence on phenology than temperature in some cases, depending on the phenological event .
However, a 60-year observational study of 29 plant species in Spain suggests temperature is the primary driver of changes in phenology in that Mediterranean region (Gordo & Sanz, 2010). The influence of biotic interactions (e.g., competition) on phenology is largely unknown outside the findings of Wolf, Zavaleta, and Selmants (2017) that plant diversity can affect phenology through its effects on soil temperature, nutrients, and moisture.
Globally, Mediterranean regions contain some of the most imperiled habitats and have among the greatest risks for biodiversity loss (Klausmeyer & Shaw, 2009;Sala et al., 2000). Much of the US Pacific Northwest (PNW) has a Mediterranean climate (Kottek, Grieser, Beck, Rudolf, & Rubel, 2006), and models for the PNW predict ~3°C temperature increases by the end of the 21st century, with increasingly warmer, wet winters and hotter, drier summers including greater drought potential during the growing season (Jung & Chang, 2012;Mote & Salathé, 2010). Native prairie ecosystems in this region have dwindled to <10% of their historic extent and most are highly degraded (Crawford & Hall, 1997;Noss, Laroe, & Scott, 1995;UFWS, 2010) because of land-use change, altered fire regimes, and invasive species (Bachelet et al., 2011). Climate change may further exacerbate the perturbations affecting these ecosystems, causing species range shifts or contractions, declining populations, or altering biogeographic patterns (Pfeifer-Meister et al., 2013, 2016. Considering the vulnerability of prairie species and communities within this region, it is thus imperative to explore the implications of changing temperature and moisture patterns on prairie plant phenology and abundances, so land managers can plan and adapt appropriate practices. Several studies have demonstrated the robustness of integrating manipulative experimentation with natural climate gradients to identify climate change effects on species, communities, and ecosystems (Dunne, Harte, & Taylor, 2003;Dunne, Saleska, Fischer, & Harte, 2004;Frenne et al., 2013;Pfeifer-Meister et al., 2013). However, this approach has been underutilized for phenological studies (but see Henry & Molau, 1997;Dunne et al., 2003;Prieto et al., 2009), especially considering that latitude may influence the magnitude or sensitivity of responses to climate change (Parmesan, 2007;Prevéy et al., 2017). Additionally, manipulative experiments designed to study climate change effects on phenology often impose extensions of the growing season via snow removal, temperature increases, or moisture manipulations (Bernal et al., 2011;Peñuelas et al., 2004;Rosa et al., 2015;Tielbörger et al., 2014;Whittington et al., 2015) but rarely are designed to manipulate both temperature and moisture, despite potentially confounding effects (Wolkovich et al., 2012). Here, we manipulated both temperature and soil moisture in three prairies across a 520 km latitudinal Mediterranean climate gradient within the PNW to examine the responses of plant phenology at both the population and community levels. At the population level, we focused on the flowering times of eight native, range-restricted focal species that we planted within and beyond their current ranges. Additionally, as we discovered that many species had very low survival (limiting our sample size for the flowering phenology data), we also examined how site and climate impacted their abundances. At the community scale, we focused on the seasonality of growth and senescence of canopy biomass. We asked: (a) How will the phenology of individual species, as well as prairie plant communities, respond to climate change across a latitudinal gradient? (b) Will range-restricted species' phenological responses and flowering abundances differ in direction and/or magnitude when planted within versus beyond their current northern range limits? And, (c) will changes to soil temperature or moisture be more predictive of phenological responses?

| Site descriptions
The study was conducted at three sites from southwestern Oregon to central-western Washington in the Pacific Northwest (PNW) (Supporting Information Figure S1, Table S1). The southern site is  Figure S2).

| Experimental design
At each site, 20 circular plots (7.1 m 2 ) were randomly assigned to one of four climate treatments with five replicates each: control (ambient temperature and precipitation), warming (canopy temperature raised by 2.5°C), warming with additional precipitation (warming + ppt; plots irrigated to fully offset a warming-induced drying effect), and drought (annual precipitation reduced by 40%). The southern and central sites were part of a previous experiment from 2010-2012 with a different set of treatments consisting of control, warming by 2.5°C, increased precipitation intensity by 20%, and warming by 2.5°C + increased precipitation intensity by 20% (Pfeifer-Meister et al., 2013, 2016Reynolds, Johnson, Pfeifer-Meister, & Bridgham, 2015). However, the precipitation intensity treatments had almost no effect on either plant or ecosystem responses since most of the additional water was applied during the wet season (Pfeifer-Meister et al., 2013, 2016Reynolds et al., 2015). Thus, the current experiment has the same control and warming treatments at the two southernmost sites, but the enhanced precipitation intensity plots became the drought plots, and the warming plus enhanced precipitation intensity plots became the warming + ppt plots of the current experiment. The northern site was newly established for this experiment.
Warming treatments were achieved using six 2000-W infrared heaters per plot, as described in Pfeifer-Meister et al. (2013). The warming + ppt plots used an automated sprinkler system (with rainwater collected on site) designed to irrigate these plots for 30 min each night that the volumetric water content was below 95% of the control plot average. The drought treatment used a common fixed rain-out shelter design, with clear acrylic shingles (MultiCraft Plastics, Eugene, OR) covering 40% of the plot area to prevent 40% of annual rainfall from reaching the plot. The acrylic material has high light transmittance, reducing microclimatic impacts such as shading concerns or temperature buffering (Gherardi & Sala, 2013;Yahdjian & Sala, 2002). The 40% reduction in annual precipitation represents an "extreme" drought, consistent with a one-in-100-year event for the three sites, determined using the Precipitation Trends and Manipulation tools from Drought-Net (Lemoine, Sheffield, Dukes, Knapp, & Smith, 2016). Drought treatments were installed in February 2016, all warming treatments initiated by summer 2016, and irrigation initiated during summer 2016. Heaters were turned off in August and September 2017 at all three sites due to fire hazard. We used dataloggers to record continuous canopy temperature, soil temperature (at 10 cm depth), and volumetric water content (to 30 cm depth) within each plot. To compare soil moisture across sites with considerably different soil characteristics, we calculated soil matric potentials as described in Saxton and Rawls (2006). See Supporting Information Figure S2 for data on soil temperature and matric potential in plots during the study. Due to heater malfunctions in one of the central-site warming plots for a period of the 2017 growing season, we excluded data from this plot for phenological analyses occurring during that time.
Between October 2014 and January 2015, all plots at the southern and central sites were mowed and raked while the new northern plots were treated with Glyphosate 2% (a total of three times) to remove standing biomass. By February 2015, all plots were seeded with a common mix of 29 native grass and forb species found in PNW prairies (Pfeifer-Meister et al., 2013). Additionally, in fall of both 2015 and 2016, we seeded between 80-200 seeds per species of 14 range-restricted species within each plot for the purposes of a separate demography experiment. These species were selected for having medium to high fidelities to upland prairies with geographic range distributions within the PNW (~41-50° latitude). Due to low establishment of six of these 14 species at all sites, only eight were used as focal species in this study (Table 1). For each species and site, we used seeds from the nearest available source population.
Four species (Collinsia grandiflora, Festuca roemeri, Microseris laciniata, and Plectritis congesta) had unique sources for each site; the remaining four species (Achyracheana mollis, Plagiobothrys nothofulvus, Ranunculus austro-oreganus, and Sidalcea malviflora) had single sources for all sites. For the date of senescence, we chose the first date following peak biomass at which the NDVI was ≤80% of the peak. We calculated the rate of senescence as the slope (ΔNDVI/days) for the three sampling points with the greatest decline in NDVI. For the southern site in 2018, we only used two sampling points because senescence was so rapid that three sampling points would not have been linear. Lastly, we calculated the length of the growing season as the difference in days between the fall 2017 green-up (the first date following the summer 2017 minimum at which the NDVI was ≥125% of the minimum) and the end of the season (the 2018 date of senescence). This timeframe represents a full growing season in this Mediterranean climate system, as vegetation growth commences with the return of the fall rains and ceases with the return of the summer drought.

| Statistical analyses
All analyses used R version 3.3.2 (R Core Team, 2016). Site and climate treatment effects on flowering phenology were determined by analysis of variance (ANOVA), whereas significant differences among sites and climate treatments within sites were tested using Tukey's post hoc comparisons. Because the control and drought treatments never differed for either FFD or PFD (p ≥ 0.19) and the warming and warming + ppt treatments only marginally differed for PFD for one species (p = 0.07; all other cases p ≥ 0.15), we collapsed the climate treatments into two temperature categories: ambient (control and drought) and warming (warming and warming + ppt) and reran analyses. Due to site × warming interactions, we tested for site effects using ambient plots only. Within sites, we tested for an effect of warming using two-tailed t tests. PFD data for C. grandiflora at the northern site were excluded due to TA B L E 1 Characteristics of the eight focal species analyzed for flowering phenology observations (Jaster, Meyers, & Sundberg, 2017) (Barton, 2018) to compare and rank each model using the smallsample-size corrected version of Akaike information criterion (AICc).
Here, we report models that would be deemed equivalent based on a δAICc <2. However, we do not report two parameter models if their AICc score was greater than a model that included only one of its parameters to maintain parsimony in interpretation.

| Reproductive plant abundances
Site had a strong effect on focal species' abundances. In general, the number of reproductive plants increased dramatically from south to north (Figure 1; Supporting Information Table S2). To a lesser extent, climate treatment also affected the number of reproductive plants, but effects varied considerably by species and were generally idiosyncratic within sites (Figure 1; Supporting Information Table S2).
Several species had small or nonexistent reproductive populations at certain sites, within certain climate treatments, or a combination; R. austro-oreganus and M. laciniata did not survive to reproduce at all at either the southern or central sites, nor did F. roemeri at the F I G U R E 1 Median abundances of reproductive plants across the four climate treatments at each site. Shading is scaled independently for each species; darker corresponds to greater median abundances under that treatment and site, and lighter corresponds to lesser median abundances. C: control; D: drought; W: warming; WP: warming + ppt. Northern range-limit groups: HRL: highest northern range limit (~50°N); IRL: intermediate northern range limit (~46°N); LRL: lowest northern range limit (~43°N; see Table 1); WS: Widespread southern site. These abundance constraints ultimately hindered our ability to analyze all aspects of the flowering phenology data.
Under ambient temperatures, FFD and PFD varied by species across the latitudinal gradient ( Figure 3). Of the annual species, C. grandiflora flowered earliest in the southern site, but A. mollis, P. nothofulvus, and P. congesta all flowered earliest at the central site.
There was no effect of site on FFD for the two perennial species, F. roemeri and S. malviflora (Figure 3a; Supporting Information Table   S3). Four species, A. mollis, F. roemeri, P. congesta, and S. malviflora, reached PFD latest in the northern site. C. grandiflora followed a similar trend, reaching PFD earlier at the southern site compared to the central site, but this could not be tested due to a lack of variance. One species, P. congesta, reached PFD earliest at the central site. Site did not significantly affect PFD for P. nothofulvus (Figure 3b; Supporting Information  Figure S3). PFD temperature sensitivity did not differ by site for any species (p > 0.10; Supporting Information Table S5, Figure S3).
We identified the most likely model (  F I G U R E 2 Mean difference ± standard error between warmed and ambient plots for (a) first flowering date (FFD) and (b) peak flowering date (PFD) at each site. Negative value indicates an advancement with warming. Significance codes: ns p > 0.1, †p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001; two-tailed t tests. P. congesta FFD could not be tested statistically because it did not exhibit any variability among the warming plots (n = 9) nor the ambient plots (n = 10) Information Figure S4). For PFD, the most important predictors were the same as for FFD for A. mollis, P. nothofulvus, and S. malviflora; for C. grandiflora and F. roemeri, the date of first wilting point, and for P. congesta, the mean annual temperature became the most important predictors (Table 2; Supporting Information Figure S5). For all species, temperature variables were negatively related to flowering dates; higher temperatures resulted in earlier flowering.

| Phenology of community biomass
Across all plots, peak biomass was reached 20.6 ± 4.3 days earlier under warming than under ambient temperatures, regardless of site or year (mean difference ± standard error; p < 0.001; Supporting Information Table S7). Peak biomass occurred earliest in the south, with the southern site reaching its peak 29.8 ± 5.3 days earlier than the central site and 38.6 ± 5.3 days earlier than the northern site For all four NDVI phenology variables, every candidate model (seven total) included mean annual temperature, while mean annual matric potential appeared in one candidate model for each variable (four models) (Supporting Information Table S6). Thus, mean annual temperature was the most important predictor for three of the four response variables (date of peak biomass, date of senescence, and 2018 growing season length), while mean annual matric potential was the most important for the rate of senescence (Table 2; Supporting Information Figure S6).

| D ISCUSS I ON
We found that changes in temperature are likely to be more impactful than changes in precipitation on many aspects of plant phenology in PNW prairies, given the expectations for potential future climatic conditions in the region. Our 2.5°C increase in temperature in the warming treatments reflects expected future temperatures for the region, with models projecting ~3°C increase by the end of the 21st century (Mote & Salathé, 2010). Precipitation projections for the PNW are less certain, but generally predict an enhanced seasonality of wetter autumns and winters and drier summers, with a small (1%-2%) overall increase in annual precipitation (Mote & Salathé, 2010). Thus, our 40% reduction in annual precipitation in the drought treatment is more extreme than current predictions, yet it did not affect the phenological variables we assessed. This may be at least partly because our drought treatment had little measurable impact on soil moisture except during the beginning (i.e., fall) and end (i.e., late spring) of the growing season (Supporting Information Figure S2; see discussion below). Furthermore, the fact that the effects of our warming + ppt treatment did not differ from those of warming alone directly implicates the importance of increasing temperature. While the warming treatment was accompanied by a strong decrease in soil moisture (Supporting Information Figure S2), the warming + ppt treatment F I G U R E 4 Normalized difference vegetation index (NDVI) of the ambient plots (control and drought) and warming plots (warming and warming + ppt) at each of the three sites from November 2016 to August 2018. Dates of peak biomass are shown with vertical dashed lines and dates of senescence with vertical dotted lines for both 2017 and 2018 (warming in red, ambient in blue). Significance codes: †p < 0.1, *p < 0.05, **p < 0.01, ***p < 0.001; two-tailed t tests following repeated measures ANOVAs to examine warming effects within each date decoupled warming from the indirect effect of reducing soil moisture, indicating that the experimental effects we observed were indeed due to increasing temperature.
It is important to place our results that changes in temperature are likely to be more impactful than changes in soil moisture under future climate in the context of a Mediterranean climate system. Despite this high interannual variability in precipitation, we still saw the strong influence of temperature on community-level phenology across years, even though annual mean temperatures during this time were no greater than ±0.5°C of the 30-year averages for each site (PRISM). Thus, climate change would need to considerably alter the timing of future wet/dry seasons (i.e., substantially delaying the first rains or advancing the summer drought), rather than simply the magnitude of total precipitation, for moisture regime changes to meaningfully impact the timing of many phenological events in this system.

Many prairie plants that are adapted to
It is also important to note that different phenological events are likely to have different mechanistic triggers, especially in a Mediterranean climate system in which high temperatures are asynchronous with the wet season. For example, Peñuelas et al. (2004) found precipitation to be less influential than temperature on leafunfolding and flowering date events yet found a stronger influence for precipitation on fruiting events in a Mediterranean shrubland in the Iberian Peninsula. Additionally, the timing of the fall green-up in PNW prairies appears to be strongly controlled by the return of the rainy season. Indeed, we did not analyze fall 2017 green-up with our own NDVI data because it occurred in most plots as soon as soil matric potential returned to above wilting point in mid-October 2017, so there was not enough variation to analyze (Supporting Information Figure S2; Figure 4). Thus, it is possible that some later phenological events could be influenced by changes in moisture, although most events in this system tend to occur at times when temperature is more limiting. Some studies have found that phenological temperature sensitivity is greater at higher latitude (Prevéy et al., 2017), whereas others have found the opposite (Wang, Ge, Dai, & Tao, 2015) or no effect (Parmesan, 2007;Wolkovich et al., 2012). We did not find consistent evidence for any type of latitudinal trend in temperature sensitivity across our sites. Of the four instances in which we found significant site effects, two cases (FFD for A. mollis and S. malviflora) exhibited greater sensitivity at the northern site, but two other cases (FFD for P. nothofulvus and P. congesta) exhibited greater sensitivity at the southern site.
We also did not find particularly strong evidence for a consistent directionality along the latitudinal gradient in the flowering times of these species under ambient temperatures. Latitude is known to impact flowering times, and we expected to see species reach flowering in ambient plots later moving from south to north, due to natural differences in the climate across this gradient. However, our environmental data indicate that the central site was slightly warmer than the southern site for much of the growing season between April and June 2017 (Supporting Information Figure S2), despite the southern site being warmer on average across the year. Thus, spring temperatures did not quite follow the latitudinal gradient, which may have contributed to these results. Furthermore, our findings across sites need to be interpreted cautiously since the southern site was not sampled with the same frequency as the central and northern sites, and we had unique seed sources across sites for C. grandiflora, F. roemeri, and P. congesta. Populations from different latitudes may differ in their phenologies based on unique evolutionary responses to growing season cues (Olsson & Agren, 2002), which may also contribute to the lack of a latitude effect on temperature sensitivity for at least the species with seeds sourced uniquely for each site.
Advances in flowering times have important implications for species' individual fitness, interactions with other species, and the assemblages of plant communities. Shifts in flowering times may desynchronize associations with pollinators, leading to lower reproductive capacity for the host plant and cascading effects at other trophic levels (Forrest & Miller-Rushing, 2010;Miller-Rushing et al., 2010;Rafferty et al., 2015). In our imperiled prairies, S. malviflora is a known nectar source for the Fender's blue butterfly (Icaricia icarioides fenderi [Macy]), and both C. grandiflora and P. congesta are known host plants of the Taylor's checkerspot butterfly (Euphydryas editha taylori), two federally listed endangered species (Schultz, Hammond, & Wilson, 2003;Schultz et al., 2011). If phenological shifts are strong enough to cause asynchronies between these butterflies' lifecycles and the growth and flowering of these and other key plant species, there could be dramatic implications for these butterflies' recovery.
Moreover, phenological shifts in plants of interest to prairie restoration may affect the ability of practitioners to successfully accomplish common activities such as burning (Hamman, Dunwiddie, Nuckols, & Mckinley, 2011) or targeted weed control (Dennehy et al., 2011). Conservation and restoration practitioners will likely need to develop adaptive strategies and plans that consider phenological shifts in order to continue meeting management goals (Bachelet et al., 2011).
While we show that the flowering times of native prairie species are likely to advance with warming, our finding that their abundances were higher at our northern site relative to our southern may be more critical. Even after multiple years of seeding identical quantities into our plots, only a few species were able to successfully establish populations across the entire gradient, a theme that we have observed in the past (Pfeifer-Meister et al., 2013)  and levels of productivity that do not differ from our central site (Pfeifer-Meister et al., 2013, 2016Reynolds et al., 2015). Instead, it seems likely that the extreme summer temperatures and the earlyonset of summer drought experienced in that region (Supporting Information Figure S2) make it exceptionally difficult for these spe- and this site is beyond these species' current northern range limits. Unexpectedly, these species struggled to achieve reproduction when planted at sites within their current ranges yet had no such constraints when planted north of their current range, suggesting they may need to shift their ranges northward to persist. In general, the less extreme climatic conditions and the longer growing seasons to the north seem to be more favorable for the fitness of all eight species, irrespective of their current ranges. These findings have implications for understanding species range distributions under future climates, and in a parallel demography experiment, we are actively assessing population projections for these and six additional species across this gradient. Furthermore, these findings confirm the importance of considering climate change when attempting to select proper seed sources for rare species restoration and recovery (Havens et al., 2015) and when selecting which species to include in restoration projects (Bachelet et al., 2011).
At the community scale, we found live plant biomass (NDVI) to be affected by warming in the following ways: consistent suppression at both the southern and central sites during the late spring to summer of both relatively wet (2017)  Shifts in the phenology of canopy biomass may have implications for community dynamics and ecosystem processes. Changes to growing season lengths are known to affect water cycling, rates and amount of carbon sequestration, and nutrient uptake from the soil (Ibáñez et al., 2010). Shorter growing seasons could reduce annual productivity, thus lessening current rates of CO 2 sequestration (Cleland et al., 2007). Additionally, shifting community biomass phenology may provide chances for exotic species to seize on resource opportunities previously unavailable to them, increasing the potential for community invasions (Prevéy & Seastedt, 2014). Moreover, these phenological shifts could lead to greater fire hazard during the dry season. In our experiment, we saw cases of biomass increasing with warming in the winter, meaning there could be a greater accumulation of herbaceous fuels. When this is followed by an earlier date of senescence, warming may be expected to cause both an earlier and more extreme fire season in the US west (Westerling, Hidalgo, Cayan, & Swetnam, 2006).
Overall, our study offers substantial evidence that future changes in temperature may have great influence on the timing of many key plant phenological events in a Mediterranean climate system and that effects due to changes in soil moisture may be buffered from even large changes in the amount of precipitation so long as the timing and duration of the rainy season are unchanged. We observed a strong influence of temperature on flowering phenology in eight native plant species both within and beyond their current geographic ranges, as well as for canopy biomass phenology at the community scale. Additionally, we found that the majority of our eight focal species are experiencing considerable reductions in their abundances near or south of their northern range limits, suggesting that the clock is ticking on their ability to persist within their current ranges.

ACK N OWLED G M ENTS
We thank the Siskiyou Field Institute, The Nature Conservancy,

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R S' CO NTR I B UTI O N S
PR analyzed the data and wrote the manuscript. SB, LPM, BR, BJ, and PR designed the study. All authors contributed to data collection and revisions, with substantial edits made by SB, LPM, BR, BJ, and SH.

DATA ACCE SS I B I LIT Y
Data are available at the Dryad Digital Repository: https://doi. org/10.5061/dryad.rg0n5t5.