Despite clear signals of regional impacts of the recent severe drought in California, e.g., within Californian Central Valley groundwater storage and Sierra Nevada forests, our understanding of how this drought affected soil moisture and vegetation responses in lowland grasslands is limited. In order to better understand the resulting vulnerability of these landscapes to fire and ecosystem degradation, we aimed to generalize drought-induced changes in subsurface soil moisture and to explore its effects within grassland ecosystems of Southern California. We used a high-resolution in situ dataset of climate and soil moisture from two grassland sites (coastal and inland), alongside greenness (Normalized Difference Vegetation Index) data from Landsat imagery, to explore drought dynamics in environments with similar precipitation but contrasting evaporative demand over the period 2008–2019. We show that negative impacts of prolonged precipitation deficits on vegetation at the coastal site were buffered by fog and moderate temperatures. During the drought, the Santa Barbara region experienced an early onset of the dry season in mid-March instead of April, resulting in premature senescence of grasses by mid-April. We developed a parsimonious soil moisture balance model that captures dynamic vegetation–evapotranspiration feedbacks and analyzed the links between climate, soil moisture, and vegetation greenness over several years of simulated drought conditions, exploring the impacts of plausible climate change scenarios that reflect changes to precipitation amounts, their seasonal distribution, and evaporative demand. The redistribution of precipitation over a shortened rainy season highlighted a strong coupling of evapotranspiration to incoming precipitation at the coastal site, while the lower water-holding capacity of soils at the inland site resulted in additional drainage occurring under this scenario. The loss of spring rains due to a shortening of the rainy season also revealed a greater impact on the inland site, suggesting less resilience to low moisture at a time when plant development is about to start. The results also suggest that the coastal site would suffer disproportionally from extended dry periods, effectively driving these areas into more extreme drought than previously seen. These sensitivities suggest potential future increases in the risk of wildfires under climate change, as well as increased grassland ecosystem vulnerability.
The severe drought between 2012 and 2016 affected most of the state of California (USA), resulting in substantial impacts on water resources and ecosystems (NDMC, 2020; Prugh et al., 2018; Shukla et al., 2015; Williams et al., 2015), yet current understanding of the California drought's impacts is based on research within particular regions and biomes. Consecutive years of low precipitation, above-average temperatures, and extremely dry conditions (meteorological drought) over this drought period resulted in severely reduced snowpack, streamflow, and groundwater storage (hydrological drought); periods of increased soil moisture deficit; and elevated vegetative stress (agricultural drought), with dramatic effects on upland forest dieback and tree mortality (Berg and Hall, 2017; Diffenbaugh et al., 2015; Swain et al., 2014; Williams et al., 2015). Although, the entire state experienced drought effects to some degree, there were notable differences in vegetation responses between Northern California and Southern California (Dong et al., 2019). In upland forests within the Sierra Nevada, there was large-scale canopy water loss and forest dieback as a result of the accumulated precipitation deficits, increased evaporative demand, and soil moisture drying (Asner et al., 2016; Fettig et al., 2019; Goulden and Bales, 2019), while there was only a documented decline in vegetation greenness in Southern California (Dong et al., 2019). Little is known about the propagation of drought from the atmosphere into soil moisture or its associated effects on vegetation in lowland areas, especially within water-limited regions where grasses and shrubs dominate the landscape. These lowland water-limited grassland ecosystems exhibit complex relationships between vegetation and water availability that affect the spatial pattern and extent of different vegetation types, as well as the relative responses of different species to drought stress (Caylor et al., 2006, 2009; D'Odorico et al., 2007; Okin et al., 2018). The progression of climate change and its potential impacts on the water balance demand a better understanding of how mean climate (temperature, precipitation) and soil water availability drive vegetation dynamics in lowland grasslands. The increasing loss of grassland ecosystems increases the threat of overall land degradation and encroachment of invasive species, which ultimately feeds back into heightened vulnerability of these ecosystems to water deficits under climate change (Gremer et al., 2015; Lian et al., 2020). In this study, we explore the links between climate, soil moisture, and vegetation during the recent California drought and analyze the potential consequences of future climate scenarios to advance our understanding of dynamic drought responses within vegetation in lowland grassland ecosystems.
Soil moisture is essential for plant growth and health; accordingly, there are strong seasonal responses of vegetation to temperature and precipitation changes (Coates et al., 2015; Roberts et al., 2010). Grassland ecosystems throughout Southern California naturally exhibit green and senescent (brown) periods each year, due to the region's strong Mediterranean climate, which makes these ecosystems naturally fire prone during the dry season. Although such fires are part of the natural ecosystems of Southern California, they are also capable of encroaching on inhabited areas with disastrous effects (e.g., huge areas are currently burning due to fires spreading through grasslands in many western states at the time of submitting this article). Rising soil moisture deficits due to meteorological droughts can cause early senescence of vegetation and thus priming grasslands for intense wildfires while also modifying species composition, runoff responses, and nutrient dynamics (Lian et al., 2020; Ludwig et al., 2005; McDowell et al., 2008; Michaelides et al., 2009). In recent decades, wildfire extent has increased substantially in Southern California, due to increased evaporative demand, reduced snowpack in mountainous areas, and loss of dry season precipitation. Under these conditions native grasslands become more susceptible to non-native species invasion, and native sage scrub is lost (Singh and Meyer, 2020; Williams et al., 2019). The most destructive fires often occur at the end of the dry season when moisture content of live and dead fuels is severely reduced after months of warm and dry weather (Keeley and Syphard, 2016; Williams et al., 2019). One example is the cascading effects of wildfire, subsequent rains, and debris flows that devastated Montecito in Santa Barbara County in 2018 (Oakley et al., 2018). Significant changes in rainfall intensity are expected around the globe (Trenberth, 2011; Westra et al., 2014), even in dryland areas (Singer and Michaelides, 2017; Singer et al., 2018), where we might expect drier spring and fall periods and an increase in subsequent dry years throughout many locations in California (Pierce et al., 2018). Such climatic conditions would likely further increase fuel aridity and wildfire potential and lead to a shift in future fire regimes with more frequent and intense wildfires throughout the western US (Abatzoglou and Williams, 2016; Williams et al., 2019) and thus potentially increasing the overall vulnerability of grasslands and surrounding communities.
Advances in remote sensing have provided new, spatially explicit observations of vegetation dynamics and moisture availability (Coates et al., 2015; Liu et al., 2012; Small et al., 2018). Additionally, the Food and Agriculture Organization (FAO) developed a well-established approach to estimate soil moisture for agricultural purposes (Allen et al., 1998), which has also proven to be useful for other non-agricultural applications (Cuthbert et al., 2013, 2019). This simple soil moisture balance approach, combined with remote sensing data, shows promise for understanding drought propagation into soil moisture. Soil moisture is our key drought metric of interest, as it inherently links precipitation, evaporative demand, and vegetation greenness as measured by Normalized Difference Vegetation Index (NDVI). The timing of vegetation growth and die-off is strongly related to seasonal fluctuations in water availability to plants, especially in annual grasslands, so the assessment of soil moisture and greenness is essential for vegetation drought monitoring (Liu et al., 2012; Small et al., 2018).
Currently, the vulnerability of California grasslands to future climate change is classified as “moderately high”, with some studies estimating a substantial loss of grassland habitats by the end of the 21st century (Thorne et al., 2016; Wilkening et al., 2019). The greater vulnerability of vegetation to drought in Southern California (compared to Northern California) and a continuing trend of aridification in this region will likely pose a compounding challenge to lowland vegetation and water resources throughout the entire US southwest (Dong et al., 2019). Increases in temperatures and evaporative demand may shift soil moisture conditions towards drier conditions, thereby increasing the risk of extreme droughts and stronger summer heat waves (Ault et al., 2016; Lian et al., 2020). Although many grass species are adapted to dry periods, a better understanding of the responses of lowland grassland vegetation to time-varying soil moisture stress associated with precipitation variability induced by climate change is essential to advance our knowledge and capabilities to mitigate the potential negative impacts of drought on these ecosystems.
In this study we build upon the FAO soil moisture modeling approach by including dynamic interactions between vegetation and climate through the incorporation of remotely sensed data. We use the model to investigate the evolution of soil moisture during the recent California drought and under several potential future drought scenarios. Our primary objective was to understand the broader patterns in the soil moisture and vegetation responses to climate forcing and to advance the understanding of how drought propagates through shallow soil moisture to affect lowland grassland vegetation. We investigated (i) how local soil moisture evolved over the recent California drought; (ii) how changes in precipitation amounts and timing affected soil moisture dynamics and grassland vegetation; and (iii) how soil moisture might respond to more prolonged dry periods under plausible climate scenarios. We employed NDVI from Landsat alongside long-term high-resolution meteorological and soil moisture data from two distinct grassland locations in Santa Barbara County with contrasting climate conditions due to orography and air flow affecting evaporative demand: a coastal and an inland site. We used these data to parameterize a simple parsimonious single-layer soil moisture balance model for generalizing the impact of climate on plant-available water in grassland ecosystems. We also developed a leading indicator of greenness based on available precipitation, which is used in our modeling framework to explore the effects of plausible climate change scenarios.
In this study, we focused on two grasslands sites in Santa Barbara County in Southern California. The natural geography of this region is characterized by coastal plains, oak woodlands, and a rugged mountain range (Roberts et al., 2010). Two sites were chosen from a network of several sites as they had the best data availability spanning over 10 years, while also representing the diverse geography of the region: a coastal grassland plain and an inland grassland site, north of the Santa Ynez Mountains (Fig. 1). Both sites are characterized by a Mediterranean climate, with strongly seasonal precipitation during the winter and prolonged dry periods in the summer. The majority of precipitation falls between November and March, with an average of 352 mm (coastal) and 314 mm (inland) per water year (October–September). Previous studies have shown that growing season water availability strongly controls annual growth cycles and senescence of vegetation at these sites (Liu et al., 2012; Roberts et al., 2010).
Location of stations in Santa Barbara County showing the coastal grassland site (COPR, green), with a marine microclimate, and the semiarid inland grassland site (AIRS, blue) north of the Santa Ynez mountain range.
The coastal site is located at the Coal Oil Point Reserve at an elevation of 6 m a.s.l. The dominant vegetation at this site is classified as introduced European grassland with several non-native species, including a range of annual grasses and forbs. Species vary significantly between years, due to rainfall variability, however wild oat grass (
The United States Drought Monitor (USDM;
Time series of
We used meteorological and soil moisture data from a network of several sites where data have been continuously recorded at 15 min resolution since 2007 by UCSB for educational purposes (Roberts et al., 2010). The data are publicly available and continuously updated (
Volumetric soil water content and soil temperature were measured using in situ probes (Stevens Hydro Probe II, Stevens Water Monitoring Systems Inc., Portland) at three different depths (10, 20, and 50 cm at the coastal site and 15, 23, and 46 cm at the inland site) (Roberts et al., 2010). For the purposes of this study, we use the shallowest soil moisture at each site, in order to capture the precipitation and evapotranspiration dynamics of the shallow soil horizon we are investigating, which comprises the majority of the moisture availability to grasses. We present historical soil moisture as relative saturation levels, ranging from dry (0 %) to fully saturated (100 %), defined as the ratio of volumetric moisture content to the volume of pore space (porosity). This allows for a direct comparison of soil moisture between the two sites, considering the differing soil textural properties. While the data recovery for both meteorological stations was continuous for the period of interest, the soil moisture probes at the inland site experienced significant data loss between 2016–2018, due to battery and sensor failure; these gaps in the data are indicated in our results.
Vegetation indices from remote sensing have been widely used to monitor the
effects of drought on vegetation, as well as the links between precipitation, soil moisture, and plant sensitivity (Dong et al., 2019; Gu et al., 2008; Small et al., 2018). Multispectral indices, such as NDVI, provide good spatial and temporal representation of drought conditions, which can be combined with in situ measurement of soil moisture for a more detailed understanding of drought propagation and drought stress on vegetation (Gu et al., 2008; Okin et al., 2018). To analyze the seasonality and relationship between soil moisture and vegetation for our study period, we used NDVI computed from red and near-infrared surface reflectance data distributed by the USGS for Landsat-5 (Thematic Mapper), Landsat-7 (Enhanced Thematic Mapper), and Landsat-8 (Operational Land Imager) – each with a 16 d acquisition interval and 30 m resolution. Because we are using multiple Landsat instruments, the data from Landsat-5, Landsat-7, and Landsat-8 were homogenized using the approach of Goulden and Bales (2019). If a pixel was cloudy, we removed the whole image to create a consistent time series of all pixels over the sampling area. We defined polygons around the measuring stations to capture a broader area of homogenous grassland vegetation and soil textural properties at the coastal (19 800 m
We developed a simple, parsimonious model to better understand the linkages
between climate, plant water availability, and plant health and include experimental manipulations of climate variables to explore plausible future
climate scenarios. Rather than attempting to model detailed soil moisture
processes, we used a simplified soil moisture balance model (SMBM) established by the FAO, which is based on a “bucket” approach (Allen et al., 1998) and is a variant of a code previously developed for estimating groundwater recharge (Cuthbert et al., 2013, 2019). Simple modeling frameworks capable of linking vegetation to water availability can be useful tools to assess past and future ecohydrological dynamics in a range of water-limited environments (Caylor et al., 2009; D'Odorico et al., 2007; Evans et al., 2018; Quichimbo et al., 2020). Therefore, model inputs are kept as simple as possible and include information on soil properties, vegetation cover, and climate (precipitation and the meteorological variables required to estimate reference evapotranspiration (ET
Simple conceptual design of a homogenous soil column with incoming and outgoing fluxes and relevant soil parameters defining the amount of available water.
Within the SMBM actual evapotranspiration (AET) is estimated using a crop
coefficient (
The data were separated into calibration and validation sets, and model
performance in each period was evaluated for acceptance or rejection of models. During calibration, model performance was optimized using data from
1 January 2008 to 31 December 2014. This time frame was chosen to include the natural variation of soil moisture dynamics, including non-drought and drought period. The model was then tested against data from 1 January 2015
to 30 September 2019. This period also includes natural variations in soil
moisture, including the drought, and individual very wet and dry years to account for the possibility of different combinations of parameter values that may all be equally successful at reproducing the observed soil moisture data. We defined the quantitative measures of acceptance or rejection criteria using Kolmogorov–Smirnov (goodness-of-fit) testing to identify parameter combinations that achieve statistically similar (
Projections of future climate change in California suggest that there will be shifts in precipitation frequency and variability during the dry season, with an increased number of dry days and increased evaporative demand, thus partly offsetting any increases in winter precipitation and possibly shifting towards more extreme events (Aghakouchak et al., 2018; Berg and Hall, 2015; Cook et al., 2015; Pierce et al., 2018). A rise in temperature is expected throughout the southwest and across the entire continent (Diffenbaugh et al., 2015). Furthermore, trends in emissions for California point towards a higher emissions scenario of RCP8.5, where annual maximum temperatures are projected to increase by more than 4
We used the SMBM model to explore the possible effects of such variations in Scenario A simulates the effects of a truncated rainy season (November–February) that reflects a loss of spring rains. This scenario represents an extreme decline in annual precipitation totals (average Scenario B simulates redistribution of lost spring rains from scenario A into the truncated rainy season from November–February, thus increasing the precipitation intensity and frequency during the compressed rainy season, combined with an increase in dry season length. Projections of CMIP5 indicated an increase in the number of dry days combined with increased frequencies of heavy precipitation, overall increasing interannual precipitation variability over California (Berg and Hall, 2015). Scenario C simulates the effects of extreme drought. It uses scenario A's loss of spring rains, along with increased evaporative demand combined with a 25 % reduction in winter rainfall totals. Annual evaporative demand was increased to represent an average 4
We retained dynamic vegetation responses in our investigation of the climate
scenarios. To replace historic NDVI values (which do not exist for potential
future scenarios), we developed a heuristic relationship between NDVI and
available precipitation (aP) as aP
Violin plots showing historic climate variables.
The 2012–2019 drought in Southern California was marked by several years of
above-average temperatures, high evaporative demand, and low precipitation.
The seasonal temperature differences during the March–October dry season
between drought periods were
The drought was expressed differently in the soil moisture at each site. Soil moisture observations showed increased drying of soils during drought periods at both sites compared to the non-drought period, reaching extremely low moisture levels in 2013 and 2014 (daily saturation fell below 5 % inland). Similar low soil moisture occurred at both sites in 2008, a particularly dry year for the Santa Barbara (SB) region (Fig. 2b). At both sites, monthly average saturation was significantly different between the non-drought and drought periods at both sites, with significantly lower levels during the drought at both sites (Fig. 6a). Average saturation was similar at both sites during the non-drought period (40 %) but decreased to an average of 30 % at the coastal site and 23 % at the inland site during the extreme drought. At both sites average monthly NDVI during the non-drought period was significantly higher than during the drought periods (Fig. 6b). Monthly NDVI values over selected non-drought and drought years illustrate the strong seasonality of annual grass cover in the region, with a marked green-up period after the winter rains, followed by a decline into brown conditions over the dry season (Fig. 6d and e). In particular, there was a rapid increase of greenness during the extreme drought, following the winter rains in 2015 and 2016 and the subsequent unusually rapid and early decline of greenness in spring. Surprisingly, NDVI reached maximum values at the height of the drought in 2015 that were nearly double the non-drought averages (0.70 and 0.77 for coastal and inland, respectively). It is notable that the NDVI peak values during drought were higher than those for the non-drought period at both sites but very short-lived as NDVI declines rapidly back to low
values, in contrast to the shoulder of greenness and slower decline of NDVI
that occurred in most non-drought years. During the extreme drought, NDVI
dropped rapidly below 0.3 in April at the inland site, which was also visible in webcam images and spatial NDVI imagery over the region (Figs. 5c and 6e). These differences in the seasonal variation of NDVI suggest a strategy of rapid grass green-up after winter rains, accelerated by mild winter temperatures during the drought and especially during the exceptionally warm winter in 2014–2015. The growth of additional vegetation under these conditions likely led to the observed rapid decline in moisture during spring, as vegetation quickly depleted any excess moisture, and subsequently experienced increased browning and senescence due to the early onset of the dry season (Fig. 5a). Correlation between NDVI and soil moisture of the concurrent month over our study period was strongly positive and statistically significant for both sites (
SMBM results for the
Given the simple structure of the SMBM, we were encouraged that the best
models at each site were effective at capturing and predicting the timing and magnitude of interactions between
Under historic drought conditions, simulations for both coastal and inland
sites reveal a clear seasonal pattern of time below the vegetative stress
threshold in the fall, prior to winter rainfall, which by extension represents the senescent periods typical for grasslands in Southern California (Fig. 8a and b). The differences in the extent of time below the threshold as well as the minimum saturation levels are visible between sites and can be attributed to differences in soil water-holding capacity and aridity. Inland, soil saturation is below the appointed threshold for more than half (64 %) the simulation time compared to about 47 % at the coastal site. Scenarios A and C noticeably shift soil moisture towards a drier baseline, leading to more extended periods of low saturation and the accumulation of an extreme soil moisture deficit extending over several
years (Fig. 8c, d, g and h). Under scenario C, for example, the time below the threshold would increase from the historical simulation by almost 50 % at the coastal site and only 25 % at the already dry inland site. This suggests that the previously buffered coastal locations would suffer disproportionally more from extended dry periods under extreme drought, as moisture reaches increasingly low levels previously unseen at this site. In contrast, the higher-intensity
Simulations of soil moisture for the coastal and inland site. Panels
The loss of spring rains, with precipitation limited between November–February, artificially extends the dry period to a total of 8 months of the year (Fig. 8c and d), resulting in a loss of
Cumulative water balance results for the coastal
Scenario B represents an exploration of climate projections that increase the intensity of winter rains in Southern California with no change in total wetness, expressed as an increased number of large daily
Rainfall event size and antecedent conditions together control drainage in our model, but our results indicate approximate rainfall thresholds that need to be overcome on daily and monthly timescales for drainage to occur. For example, a monthly total of
In the extreme drought conditions of scenario C, the effects of the increased precipitation loss and heightened ET
In light of the progression of climate change in semiarid environments such as Southern California, a better understanding of drought propagation and the climatic drivers of shifts in soil moisture and water availability to grassland vegetation (and, correspondingly, to the health and functioning of grassland ecosystems) would enable anticipation of how soil moisture and grassland dynamics might respond to intensified moisture limitations under future scenarios of climate change across the region. The severity of the recent synoptic California drought and its effects on vegetation were most notably documented through upland forest canopy water stress and mortality (Asner et al., 2016; Fettig et al., 2019; Goulden and Bales, 2019), as well as through declining groundwater levels that heavily impacted agricultural production throughout the Californian Central Valley (Thomas et al., 2017; Xiao et al., 2017). Similarly, the intensified moisture loss and accelerated ET also impacted lowland vegetation in Southern California, including differential species responses within chaparral and grassland ecosystems (Breshears et al., 2005; Gremer et al., 2015; Okin et al., 2018; Wilson et al., 2018). While the landscape in Southern California is dominated by vast stretches of brown grasslands during the dry season, the 2012–2019 drought hit Santa Barbara Country with considerable intensity and persistence, compared to the rest of the state (Fig. 2), and propagated into multiple years of soil moisture deficits and early die-off of grasses (Figs. 4–6).
Our analysis revealed that winter or spring precipitation deficits, coupled with higher evaporative demand in Southern California, led to temporal shifts in the onset of the dry season, which in turn also led to increased soil drying in spring and summer. The loss of essential precipitation pulses in spring months generated large soil moisture deficits and induced a faster die-off (browning) of grasses, especially at the inland site. We explored this shift in dry season onset further by simulating soil moisture responses under an even shorter rainy season. Our findings suggest that arid sites such as our inland site with low water-holding capacities, which is widespread over the region and more broadly over the southwest and other Mediterranean climate systems, would become increasingly vulnerable to climate change that favors milder winter and hotter summer temperatures, as well as decreased precipitation in key months during spring. Sites with low moisture holding capacities due to sandy soils and more arid climate seem less resilient to the loss of rain at the time when plant development is about to start and moisture is needed for seed germination and plant growth. Interestingly, the potential for apparent local groundwater recharge seems to remain unaffected by the loss of spring rains, suggesting that drainage only occurs during the winter months; even under prolonged periods of drought, there is a potential for local groundwater recharge. Such changes to the seasonal delivery of precipitation would increase the soil moisture drought frequency and magnitude, leading to much earlier senescence of vegetation and widespread desertification of the landscape while selectively priming the landscape for large and destructive wildfires, thus suggesting that already arid ecosystems might be brought to their physiological limit. These results can be viewed alongside prior work in the southwest that suggested chaparral landscapes (Okin et al., 2018) and perennial (
With climate change projected to impact the temperate and precipitation regimes in California, as well as much of the southwestern US, the frequency and magnitude of droughts and drought-like conditions are expected to increase (Bradford et al., 2020; Diffenbaugh et al., 2015). Under a more severe emission scenario of RCP8.5, the frequency of extreme dry years is projected to almost triple with temperatures projected rise by up to 4
Another key finding of the climate simulations has revealed that the occurrence of extreme events after prolonged periods of drought, as simulated in scenarios A and B, would provide temporary relief to soil moisture and most likely support considerable green-up and production of biomass during that season. However, if climate conditions revert to extreme dryness and minimal precipitation input during the following year, the soil moisture deficit would increase again to a level unlikely to support the extensive growth from the previous season. Under these conditions the senescent vegetation would turn into easily ignitable fuel that, coupled with the dried-out soils, would prime the landscape for extensive wildfires, thus potentially creating a severe chain reaction of extreme events as previously seen during the Montecito fires and mudslides.
The 2012–2019 drought in California had profound impacts on soil moisture and vegetation. Employing long-term monitoring data, we delineated the differential responses of soil moisture and vegetation dynamics of grassland ecosystems to this unprecedented, multiyear drought in Southern California. A temporal shift of dry season onset led to early senescence and browning of vegetation and rendered soil moisture resources prematurely exhausted, and the landscape was primed for easily ignitable and widespread wildfires. During the drought, temporal patterns of vegetation productivity changed, including increased greenness attributed to mild winter temperatures after prolonged dry periods. However, this new vegetation growth quickly reached a state of senescence due to the early onset of the dry season, exacerbating the soil moisture deficit.
Through a simple, parsimonious soil moisture water balance model, we further explored the moisture dynamics and water balance in terms of soil moisture for grasslands under different conditions that represent possible simplistic climate change scenarios. We linked soil moisture and vegetation response through NDVI and explored the effects of various changes to precipitation and evaporative demand. The results suggest that such changes could have unprecedented effects on soil moisture and water availability to grassland ecosystems, leading to rapid dieback and prolonged desiccation of the landscape. Our results highlighted the differential responses of moisture and vegetation over a small geographical area. In future, more extreme and prolonged droughts, characterized by a shorter rainy season, higher evaporative demand, and/or protracted dry periods, will likely lead to increased soil moisture deficits at sites with low water-holding capacities, as moisture levels are likely to drop to a level of elevated vegetative stress for much of the year. The combination of such climate-induced changes, loss of precipitation pulses in spring and summer, a continuing shift of early dry season onset, and increased evaporative demand are likely contributors to affect grassland ecosystems in future and drive even previously less affected coastal areas into more severe droughts, as well as induce widespread desertification of the landscape in semiarid environments. A shift to a drier moisture baseline of soils and vegetation could potentially have deleterious effects on species diversity, increase the risk of shrub encroachment and invasive species, and leave the region overall more prone to destructive and widespread wildfires.
Working code used for data analysis can be found under
All climate and soil moisture data are publicly available for download at
The supplement related to this article is available online at:
Conceptualization of the research was done by MBS, MOC, KCa, DR and JS. Methodology was developed by MMW, MBS and MOC. Data curation and formal analysis was done by MMW and RS. The original draft was written by MMW. Review and editing was done by MMW, MBS, MOC, KC, JS, DR and RS.
The authors declare that they have no conflict of interest.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work was supported by the National Science Foundation (BCS-1660490, EAR-1700517 and EAR-1700555) and the Department of Defense's Strategic Environmental Research and Development Program (RC18-1006). We thank Dar Roberts for providing the IDEAS data set, which is publicly available at
This research has been supported by the Strategic Environmental Research and Development Program (grant no. RC18-1006), the National Science Foundation (grant nos. BCS-1660490, EAR-1700517, and EAR-1700555), and the Natural Environment Research Council (grant no. NE/P017819/1).
This paper was edited by Nunzio Romano and reviewed by two anonymous referees.