assessing climate change impacts on live fuel moisture and

Biogeosciences 18 4005ndash4020 2021httpsdoiorg105194bg-18-4005-2021copy Author(s) 2021 This work is distributed underthe Creative Commons Attribution 40 License

Assessing climate change impacts on live fuel moisture and wildfirerisk using a hydrodynamic vegetation modelWu Ma1 Lu Zhai2 Alexandria Pivovaroff3 Jacquelyn Shuman4 Polly Buotte5 Junyan Ding6Bradley Christoffersen7 Ryan Knox6 Max Moritz8 Rosie A Fisher9 Charles D Koven6 Lara Kueppers10 andChonggang Xu1

1Earth and Environmental Sciences Division Los Alamos National Laboratory Los Alamos NM United States2Department of Natural Resource Ecology and Management Oklahoma State University Stillwater OK United States3Atmospheric Science and Global Change Division Pacific Northwest National Laboratory Richland WA United States4National Center for Atmospheric Research Climate and Global Dynamics Terrestrial Sciences SectionBoulder CO United States5Energy and Resources Group University of California Berkeley CA United States6Climate and Ecosystem Sciences Division Lawrence Berkeley National Laboratory Berkeley CA United States7Department of Biology University of Texas Rio Grande Valley Edinburg TX United States8UC ANR Cooperative Extension Bren School of Environmental Science amp ManagementUniversity of California Santa Barbara CA United States9Centre Europeacuteen de Recherche et de Formation Avanceacutee en Calcul Scientifique Toulouse France10Energy and Resources Group University of California Berkeley and Lawrence Berkeley National LaboratoryBerkeley CA United States

Correspondence Chonggang Xu (cxulanlgov)

Received 18 November 2020 ndash Discussion started 25 November 2020Revised 20 May 2021 ndash Accepted 14 June 2021 ndash Published 6 July 2021

Abstract Live fuel moisture content (LFMC) plays a criticalrole in wildfire dynamics but little is known about responsesof LFMC to multivariate climate change eg warming tem-perature CO2 fertilization and altered precipitation patternsleading to a limited prediction ability of future wildfire risksHere we use a hydrodynamic demographic vegetation modelto estimate LFMC dynamics of chaparral shrubs a dominantvegetation type in fire-prone southern California We param-eterize the model based on observed shrub allometry and hy-draulic traits and evaluate the modelrsquos accuracy through com-parisons between observed and simulated LFMC of threeplant functional types (PFTs) under current climate condi-tions Moreover we estimate the number of days per year ofLFMC below 79 (which is a critical threshold for wild-fire danger rating of southern California chaparral shrubs)from 1960 to 2099 for each PFT and compare the numberof days below the threshold for medium and high green-house gas emission scenarios (RCP45 and 85) We find thatclimate change could lead to more days per year (52 ndash

148 increase) with LFMC below 79 between the histor-ical (1960ndash1999) and future (2080ndash2099) periods implyingan increase in wildfire danger for chaparral shrubs in south-ern California Under the high greenhouse gas emission sce-nario during the dry season we find that the future LFMCreductions mainly result from a warming temperature whichleads to 91 ndash186 reduction in LFMC Lower precipita-tion in the spring leads to a 63 ndash81 reduction in LFMCThe combined impacts of warming and precipitation changeon fire season length are equal to the additive impacts ofwarming and precipitation change individually Our resultsshow that the CO2 fertilization will mitigate fire risk by caus-ing a 35 ndash48 increase in LFMC Our results suggest thatmultivariate climate change could cause a significant net re-duction in LFMC and thus exacerbate future wildfire dangerin chaparral shrub systems

Published by Copernicus Publications on behalf of the European Geosciences Union

4006 W Ma et al Assessing climate change impacts on live fuel moisture and wildfire risk

1 Introduction

Historical warming and changes in precipitation have alreadyimpacted wildfires at a global scale (eg Stocks et al 1998Gillett et al 2004 Westerling et al 2003 2006) and it isexpected that accelerating future warming will continue tosubstantially affect global wildfires (eg Flannigan et al2009 Liu et al 2010 Moritz et al 2012) So far prior stud-ies have mainly focused on the impacts of dead fuel mois-ture fuel loads and weather conditions on wildfires Limitedstudies have applied proxies of live fuel moisture in globalfire models For example dead fuel moisture is found to berelated to fire ignition and fire spread potential (or potentialarea burned) (Aguado et al 2007) specific weather condi-tions such as increased vapor pressure deficit (Williams et al2019) can lead to a vast increase in fire activity (Goss et al2020) and wildfire fuel loads are projected to increase underclimate change (Matthews et al 2012 Clarke et al 2016)In global fire models studies have used proxies of live fuelmoisture (Bistinas et al 2014 Kelley et al 2019) as wellas explicit representation of live fuels (Hantson et al 2016Rabin et al 2017) While previous studies provide great in-sights into fire risks with changes in climate dead fuel mois-ture fuel loads and representation of live fuel moisture thereis still limited understanding of how climate change influ-ences live fuel moisture content (LFMC) and the consequentwildfire risks This is particularly true for the combined im-pacts of warming temperature altered precipitation and in-creasing CO2 fertilization (Chuvieco et al 2004 Pellizzaroet al 2007 Caccamo et al 2012a b Williams et al 2019Goss et al 2020)

A measure of water content within living plant tissue inrelation to their dry weight LFMC has been found to beone of the most critical factors influencing combustion firespread and fire consumption (eg Agee et al 2002 Zarco-Tejada et al 2003 Bilgili and Saglam 2003 Yebra et al2008 Dennison et al 2008 Anderson and Anderson 2010Keeley et al 2011) This is because a low LFMC leadsto increased flammability and a higher likelihood of igni-tion (Dimitrakopoulos and Papaioannou 2001) For instanceLFMC was found to be a significant factor contributing to theoccurrence of wildfires in Australia (Plucinski 2003 Nolanet al 2016 Yebra et al 2018 Rossa and Fernandes 2018Pimont et al 2019) Spain (Chuvieco et al 2009) and Cali-fornia (Santa Monica Mountains Dennison et al 2008 Den-nison and Moritz 2009 Pivovaroff et al 2019) Dennisonand Moritz (2009) found strong evidence of a LFMC thresh-old (79 ) for southern California chaparral shrubs whichmay determine when large fires can occur in this region

Vegetation moisture content is dependent on both eco-physiological characteristics of the species and environmen-tal conditions including both climatic variables and soil wa-ter availability (Rothermel 1972 Castro et al 2003 Pel-lizzaro et al 2007 Pivovaroff et al 2019 Nolan et al2020) So far little is known about the relative importance

of different climate variables to future LFMC dynamicsOn one hand warming could contribute to a higher atmo-spheric demand and higher evapotranspiration (Rind et al1990) and thus lead to a lower LFMC On the other handhigher CO2 concentration will decrease stomatal conduc-tance (Wullschleger et al 2002) and plant water loss andthus lead to a higher LFMC The impacts of CO2 and warm-ing could be complicated by local changes in precipitationpatterns and humidity (Mikkelsen et al 2008)

The sensitivity of LFMC to climate change is likely to beaffected by plant hydraulic traits (the plant properties thatregulate water transport and storage within plant tissues)which affect plant water regulation (Wu et al 2020) Vari-ations in hydraulic traits reflect contrasting plant droughtadaptation strategies when responding to dry conditions Twocontrasting overall strategies are (1) water stress avoidersand (2) water stress tolerators (Tobin et al 1999 Wei et al2019) The ldquoavoidersrdquo are generally characterized by a moreconservative hydraulic strategy under water stress by closingtheir stomata early dropping leaves or accessing deep waterto avoid more negative water potentials and therefore xylemcavitation Meanwhile the ldquotoleratorsrdquo typically build xylemand leaves that are more resistant to cavitation so that theycan tolerate more negative water potential and continue toconduct photosynthesis under water stress Therefore com-pared with the tolerators the avoiders normally have a lowersapwood density and higher plant water storage capacity intheir tissues to avoid cavitation (Meinzer et al 2003 2009Pineda-Garcia et al 2013) Because the avoiders rely on wa-ter storage capacity as one way to avoid cavitation therebymaintaining a relatively high LFMC and because water lossfrom storage should increase with warming LFMC could bemore sensitive to climate change in avoiders relative to toler-ators

While over half of terrestrial landscapes on Earth are con-sidered fire-prone (Krawchuk et al 2009) Mediterranean-type climate regions are routinely impacted by fire often onan annual basis This is partly because Mediterranean cli-mate regions are characterized by winter rains followed byan annual dry season when little to no rainfall occurs forseveral months Multiday periods of extreme high tempera-tures as well as katabatic hot dry and intense winds oftenpunctuate the annual drought leading to some of the worstfire weather in the world (Schroeder et al 1964) This canresult in wildfires that are large high intensity and stand re-placing (Keeley 1995 Keeley and Zedler 2009 Balch et al2017) Globally Mediterranean climate regions are charac-terized by evergreen sclerophyllous-leaved shrublands TheMediterranean climate region in California is dominated bychaparral which is adapted to the periodic fire regime in Cal-ifornia (Venturas et al 2016) Previous studies have pro-posed a variety of relationships between chaparral LFMCand fire danger in southern California (Dennison et al 2008Dennison and Moritz 2009) but less is known about how cli-mate changes could alter LFMC and fire danger In chaparral

Biogeosciences 18 4005ndash4020 2021 httpsdoiorg105194bg-18-4005-2021

W Ma et al Assessing climate change impacts on live fuel moisture and wildfire risk 4007

LFMC is usually high during the winter and spring (wet sea-son) and then gradually declines during the dry season (sum-mer and fall) which leads to a typical fire season approxi-mately six months long in southern California (Pivovaroff etal 2019) One key risk is that severe drought conditions arebecoming exacerbated under climate change which mightlead to the occurrence of larger and higher-intensity firesin chaparral (Dennison et al 2008 Dennison and Moritz2009)

There is a long history of wildfire modeling with threetypes of models (1) fine-scale fire behavior models (egFIRETEC by Linn et al 2002) (2) landscape-scale fire dis-turbance models (eg LANDIS-II by Sturtevant et al 2009)and (3) global-scale fire dynamics models (eg Hantson etal 2016 Rabin et al 2017 SPITFIRE by Thonicke et al2010) While these models focus on simulation at differentscales fire measures of the simulation are mainly calculatedfrom climate and dead fuel moisture and currently lack pre-diction of LFMC dynamics One key limitation is that mostprevious models have not yet considered plant hydrodynam-ics (Holm et al 2012 Xu et al 2013 Seiler et al 2014)which is integral to LFMC prediction Recently there havebeen important improvements to global dynamic and demo-graphic vegetation models by incorporating plant hydrody-namics (McDowell et al 2013 Xu et al 2016 Fisher et al2018 Mencuccini et al 2019) These models have been usedto study the interaction between elevated CO2 and drought(Duursma and Medlyn 2012) the impact of hydraulic traitson plant drought response (Christofferson et al 2016) therole of hydraulic diversity in vegetation response to drought(Xu et al 2016) and hydroclimate change (Powell et al2018) and vegetation water stress and root water uptake(Kennedy et al 2019) While the main purpose of the newhydraulic components is to improve the vegetation responseto drought the fact that hydrodynamic models consider tis-sue water content as a prognostic variable provides an oppor-tunity to assess the climate impacts on LFMC

The objective of this study is to quantify LFMC dynamicsand associated changes in fire season duration for a chaparralecosystem in southern California under climate change usinga vegetation demographic model (that resolves the size andage-since-disturbance structure of plant populations) (Xu etal 2016 Fisher et al 2018) that incorporates plant hy-draulics We test one overarching hypothesis future climatechange will decrease LFMC and consequently result in alonger fire season as determined by a critical threshold ofLFMC (H0) Specifically we test the following four sub-hypotheses (1) warming has a stronger impact on LFMCthan CO2 fertilization (H1) (2) the reductions in spring andautumn precipitation lead to a longer fire season as deter-mined by LFMC (H2) (3) the combined impacts of warmingand precipitation on fire season length are equal to the addi-tive impacts of warming and precipitation change individu-ally (H3) and (4) LFMC for plants with more conservative

hydraulic strategies (ldquoavoidersrdquo) will be more vulnerable towarming (H4)

2 Materials and methods

To understand climate change impacts on LFMC for the cha-parral ecosystem we applied the Functionally AssembledTerrestrial Simulator (FATES Fisher et al 2015 Massoudet al 2019 Koven et al 2020) coupled with a hydrody-namic vegetation module (FATES-HYDRO Christoffersenet al 2016) in the Santa Monica Mountains in California Wevalidated the model using the observed LFMC for three cha-parral shrub plant functional types (PFTs) Then we appliedFATES-HYDRO to estimate long-term dynamics of leaf wa-ter content (LWC) during 1960ndash2099 for each PFT usingdownscaled Earth system model (ESM) climate scenariosWe converted simulated LWC to LFMC within leaves andshoots Based on the simulated LFMC we evaluated wild-fire danger based on the number of days per year of LFMCbelow the critical value of 79 from 1960 to 2099 for eachPFT under RCP45 and 85 Finally we assessed the relativeimportance of changes in individual and combined climatevariables including CO2 temperature and precipitation andtested the corresponding hypotheses

21 Study site

The study site is located at the Stunt Ranch Santa MonicaMountains Reserve in the Santa Monica Mountains in Cal-ifornia USA (3405prime N 11839primeW) Stunt Ranch is dom-inated by chaparral vegetation with an elevation of approxi-mately 350 m a west-facing slope and a Mediterranean-typeclimate The study site harbors an abundance of fauna par-ticularly birds and reptiles The mean annual temperature is181 C The mean annual precipitation is 478 mm occurringmostly during the wet season (ie NovemberndashMarch) withalmost no rainfall during the dry season (ie AprilndashOctober)Stunt Ranch last burned in 1993 We focused on PFTs repre-senting 11 study species (Fig 1) including chamise (Adenos-toma fasciculatum ndash Af) red shank (Adenostoma sparsi-folium ndash As) big berry manzanita (Arctostaphylos glauca ndashAg) buck brush (Ceanothus cuneatus ndash Cc) greenbark cean-othus (Ceanothus spinosus ndash Cs) mountain mahogany (Cer-cocarpus betuloides ndash Cb) toyon (Heteromeles arbutifolia ndashHa) laurel sumac (Malosma laurina ndash Ml) scrub oak (Quer-cus berberidifolia ndash Qb) hollyleaf redberry (Rhamnus ilici-folia ndash Ri) and sugar bush (Rhus ovata ndash Ro) Detailed in-formation about the study site and species characterizationsfound at Stunt Ranch can be found in Venturas et al (2016)and Pivovaroff et al (2019)

22 FATES-HYDRO model

FATES is a vegetation demographic model (Fisher et al2015) that uses a size-structured group of plants (cohorts)

httpsdoiorg105194bg-18-4005-2021 Biogeosciences 18 4005ndash4020 2021


Figure 1 Hierarchical cluster analysis of allometry and hydraulic traits for 11 chaparral shrub species used to define three plant functionaltypes at Stunt Ranch The plant functional types with a low productivity and an aggressive drought tolerance hydraulic strategy (PFT-LA)were defined based on traits of red shank (Adenostoma sparsifolium ndash As) toyon (Heteromeles arbutifolia ndash Ha) chamise (Adenostomafasciculatum ndash Af) and big berry manzanita (Arctostaphylos glauca ndash Ag) The plant functional types with a high productivity and anaggressive drought tolerance hydraulic strategy (PFT-HA) were defined based on traits of mountain mahogany (Cercocarpus betuloides ndashCb) greenbark ceanothus (Ceanothus spinosus ndash Cs) buck brush (Ceanothus cuneatus ndash Cc) and hollyleaf redberry (Rhamnus ilicifolia ndashRi) The plant functional types with a medium productivity and an conservative drought tolerance hydraulic strategy (PFT-MC) were definedbased on traits of laurel sumac (Malosma laurina ndash Ml) scrub oak (Quercus berberidifolia ndash Qb) and sugar bush (Rhus ovata ndash Ro)

and successional trajectory-based patches based on theecosystem demography approach (Moorcroft et al 2001)FATES simulates the demographic process including seedproduction seed emergence growth and mortality (Koven etal 2020) Because the main purpose is to assess LFMC wecontrolled for variation in plant size structure that could arisefrom plant traits or climate differences between model runsby using a reduced-complexity configuration of the modelwhere growth and mortality are turned off and ecosystemstructure is held constant FATES has to be hosted by a landsurface model to simulate the soil hydrology canopy temper-ature and transpiration These host land models include theExascale Energy Earth System Model (E3SM Caldwell etal 2019) land model (ELM) as well as the Community EarthSystem Model (Fisher et al 2015) and the Norwegian EarthSystem Model (NorESM Tjiputra et al 2013) In this studywe used the DOE-sponsored ELM as our host land modelThe time step of FATES to calculate carbon and water fluxesis 30 min and it can downscale the data from 6-hourly climatedrivers

A key component of FATES the plant hydrodynamicmodel (HYDRO based on Christoffersen et al 2016) sim-ulates the water flow from soil through the roots stem andleaves to the atmosphere In this model water flow is cal-culated based on water pressure gradients across differentplant compartments (leaf stem transporting roots absorbingroots and rhizosphere) Specifically flow between compart-ment i and i+ 1 (Qi) is given by

Qi =minusKi1hi (1)

where Ki is the total conductance (kg MPaminus1 sminus1) at theboundary of compartments i and i+ 1 and 1hi is the totalwater potential difference between the compartments

1hi = ρwg (zi minus zi+1)+ (ψi minusψi+1) (2)

where zi is compartment distance above (+) or below (minus)the soil surface (m) ρw is the density of water (103 kg mminus3)g is acceleration due to gravity (98 m sminus2) and ψi is tissueor soil matric water potential (MPa)Ki is treated here as theproduct of a maximum boundary conductance between com-partments i and i+ 1 (Kmaxi) and the fractional maximumhydraulic conductance of the adjacent compartments (FMCior FMCi+1) which is a function of the tissue water contentA key parameter that controls FMC is the critical water po-tential (P50) that leads to 50 loss of hydraulic conductivityThe tissue water potential is calculated based on pressurendashvolume (PV) theory (Tyree and Hammel 1972 Tyree andYang 1990 Bartlett et al 2012) For leaves it is describedby three phases (1) capillary water phase with full turgor(2) elastic drainage phase before reaching turgor loss pointand (3) post-turgor loss phase For other tissues it only hasphases 2 and 3 Compared to a non-hydrodynamic modelthis formulation allows the simulation of plant water trans-port limitation on transpiration For the non-hydrodynamicversion of FATES the water limitation factor for transpira-tion (Btran) is calculated based on the soil moisture potential(Fisher et al 2015) For the hydrodynamic version Btran iscalculated based on the leaf water potential (ψl) (Christof-fersen et al 2016) as follows

Btran =

[1+

(ψl

P50_gs

)al]minus1

(3)

where P50_gsis the leaf water potential that leads to 50 lossof stomatal conductance and al is the shape parameter Pleaserefer to Christoffersen et al (2016) for details of formula-tions of FMC for different plant tissues



23 Allometry and trait data for modelparameterization

FATES-HYDRO has a large number of parameters (gt 80see Massoud et al 2019 for a complete list except for hy-draulic parameters) Based on a previous sensitivity analy-sis study (Massoud et al 2019) we focused our parameterestimation efforts on the most influential parameters for al-lometry leaf and wood traits and hydraulic traits from ob-servations of 11 chaparral shrub species (see Table S2 inthe Supplement) collected from Jacobsen et al (2008) andVenturas et al (2016) For this study we assumed that theallometry of a shrub is analogous to that of a small treeHowever we did make several important modifications toaccommodate the allometry of a shrub as their height andcrown area relationships to diameter could be different fromtrees First instead of using the diameter at breast height asthe basis for allometry to calculate the height crown areaand leaf biomass we used the basal diameter as the basisfor shrubs Second in the allometry of trees the diameterfor maximum height (d1 Fates_allom_dbh_maxheight Ta-ble S1) is the same as the diameter for maximum crown area(d2 Fates_allom_d2ca_max Table S1) As our data showedthat d1 and d2 are different for shrubs we have modified thecodes so that d1 and d2 can be set for different values It ispossible that different branching and path length patterns forstems of chaparral species could impact the hydraulics com-pared to trees however FATES-HYDRO treats all the above-ground xylem as a single pool and thus it should not affectour model simulation results

Based on a hierarchical cluster analysis (Bridges 1966)of allometry and trait data there is a clear separationamong the shrub species First the dendrogram is builtand every data point finally merges into a single clus-ter with the height shown on the y axis Then we cutthe dendrogram in order to create the desired numberof clusters determined by a pragmatic choice based onhydraulic traits of 11 chaparral shrub species (Fig 1)Rrsquos recthclust function (httpswwwrdocumentationorgpackagesstatsversions362topicsrecthclust last access25 April 2021) was used to see the clusters on the dendro-gram All parameters of allometry leaf and wood traits andhydraulic traits were collected from observations shown inTables S2 and S3 According to the principle of model parsi-mony we do not want to classify the species into more than3 PFTs Meanwhile we also want to differentiate the funda-mental plant growth and water use strategies that will deter-mine plant transpiration rate and the corresponding LFMCIf we choose to classify the species into two PFTs (basedon the solid horizontal line in Fig 1) then we will not beable to differentiate species with aggressive and conserva-tive hydraulic strategies in the second group and not be ableto test H4 Therefore the chaparral shrub species were clas-sified into three PFTs (based on the dotted horizontal linein Fig 1 and Table S3) that are able to differentiate plant

growth and hydraulic strategy The three PFTs include a lowproductivity aggressive drought tolerance hydraulic strategyPFT (PFT-LA) with a relatively low Vcmax25 (the maximumcarboxylation rate at 25 C) and a very negative P50 (the leafwater potential leading to 50 loss of hydraulic conductiv-ity) a medium productivity conservative drought tolerancehydraulic strategy PFT (PFT-MC) represented by a mediumVcmax25 and a less negative P50 turgor loss point and waterpotential at full turgor and a high productivity aggressivedrought tolerance hydraulic strategy PFT (PFT-HA) with arelatively high Vcmax25 and a very negative P50 The meanof the species-level trait data weighted by species abundanceat the site were used to parameterize FATES-HYDRO

24 Model initialization

Our model simulation is transient in terms of soil watercontent leaf water content and carbon and water fluxesThe forest structure (plant sizes and number density) isfixed and is parameterized based on a vegetation inven-tory from Venturas et al (2016) The soil texture anddepth information are parameterized based on a nationalsoil survey database (httpswebsoilsurveyscegovusdagovAppWebSoilSurveyaspx last access 28 March 2021 Ta-ble S1) The soil moisture is initialized with 50 of the sat-uration and the tissue plant water content is initialized so thatit is in equilibrium with the soil water potential We run themodel for 10 years based on 1950ndash1960 climate so that thesimulated soil moisture leaf water content and carbon andwater fluxes are not dependent on their initial conditions

25 Live fuel moisture content for model validation

In this study we used measured LFMC to validate simu-lated LFMC FATES-HYDRO does not directly simulate theLFMC Thus we estimated the LFMC based on simulatedLWC The LWC in the model is calculated as follows

LWC=fwminus dw

dwtimes 10 (4)

where fw is the fresh weight and dw is the dry weight whichare both simulated within FATES-HYDRO Then we esti-mated the LFMC within leaves and shoots (lt 6 mm diame-ter) using the empirical equation derived from shrub LFMCand LWC data including the three regenerative strategies(seeder (S) resprouter (R) and seederndashresprouter (SR)) insummer autumn and winter from Figs 4 and 5 in the studyby Saura-Mas and Lloret (2007) as follows (Fig S4)

LFMC= 31091+ 0491LWC (5)

The climate in Saura-Mas and Lloretrsquos study is Mediter-ranean (northeast Iberian Peninsula) which is consistentwith the climate of our study area LFMC was measuredon our site approximately every three weeks concurrentlywith plant water potentials in 2015 and 2016 LFMC mea-surement details can be found in Pivovaroff et al (2019) For



comparison with our model outputs we calculated the meanLFMC within leaves and shoots for each PFT weighted bythe species abundance (Venturas et al 2016) Species abun-dance was calculated by dividing mean density of a specificspecies by the mean density of all species

26 Climate drivers

We forced the FATES-HYDRO model with 6-hourly tem-perature precipitation downward solar radiation and windcomponents Historical climate data during 2012ndash2019which were used for FATES-HYDRO calibration were ex-tracted from a local weather station (httpsstuntranchucnrsorgweather-date last access 26 April 2021) Historical andfuture climate data during 1950ndash2099 which were used forsimulations of LFMC by the FATES-HYDRO model weredownloaded from the Multivariate Adaptive ConstructedAnalogs (MACA) datasets (Abatzoglou and Brown 2012httpmacanorthwestknowledgenet last access 25 April2021) The MACA datasets (124 or approximately 4 kmAbatzoglou and Brown 2012) include 20 ESMs with his-torical forcings during 1950ndash2005 and future Representa-tive Concentration Pathways (RCPs) RCP45 and RCP85scenarios during 2006ndash2099 from the native resolution ofthe ESMs The gridded surface meteorological dataset MET-DATA (Abatzoglou 2013) were used with high spatial res-olution (124) and daily timescales for near-surface min-imum and maximum temperature precipitation downwardsolar radiation and wind components Then we downscaledthe MACA daily data to 6-hourly based on the temporalanomaly of the observed mean daily data to the hourly datafor each day during 2012ndash2019 The model is driven byyearly CO2 data obtained from Meinshausen et al (2011)

27 Hypothesis testing

To test H0 (future climate change will decrease LFMC andconsequently result in a longer fire season as determined bya critical threshold of LFMC) we compared the simulatedmean LFMC derived from modeled leaf water content un-der the climate projections from 20 ESMs under RCP45 and85 We then tested if the LFMC during the AprilndashOctoberdry season in the historic 960ndash1999 period is significantlyhigher than that in the future 2080ndash2099 period For thefire season duration we estimated the number of days peryear below a critical threshold of LFMC (79 ) Similarlywe tested if the number of days per year below the criticalthreshold of LFMC during the historical period are signifi-cantly different from that during the future period We useda bootstrapped approach (Jackson 1993) to test if the meanof LFMC or fire season duration are significantly differentbetween these two periods Specifically we randomly draw10 000 samples from the simulated residuals of LFMC or fireseason durations estimated by 20 ESMs for these two peri-ods under the null hypothesis that there is no difference in the

mean We then calculated p-values by comparing the simu-lated mean difference to the empirical distribution of differ-ence estimated from these 10 000 samples (see supplemen-tary Sect 52 within Xu et al (2019) for details)

To test H1 (warming has a stronger impact on LFMCthan CO2 fertilization) we compared mean simulated LFMCand fire season length for three PFTs with and without CO2changes (fixed CO2 at 367 ppm vs dynamic CO2 concentra-tions from RCP45 or RCP85) and warming To remove thefuture warming trend future temperature was replaced withhistorical (1986ndash2005) temperature data for every 20 yearperiod Similarly to test H2 (the reductions in spring and au-tumn precipitation lead to a longer fire season as determinedby LFMC) we compared the model outputs of LFMC andfire season length for three PFTs with and without precipita-tion changes To test H3 (the combined impacts of warmingand precipitation on fire season length are equal to the ad-ditive impacts of warming and precipitation change individ-ually) we compared model outputs of LFMC and fire sea-son length for three PFTs under three scenarios (1) withoutwarming (2) without precipitation changes and (3) with-out warming and precipitation changes Finally to test H4(LFMC for plants with more conservative hydraulic strate-gies will be more vulnerable to warming) we comparedmodel outputs of LFMC and fire season length across thethree different PFTs with different hydraulic strategies

3 Results

31 Comparison between simulated and measuredLFMC

Our results showed that FATES-HYDRO was able to cap-ture variation in the LFMC for different PFTs and soil wa-ter content at 5 cm depth (Figs 2 and S3) as well as forchamise in 2018 (Fig S5) although we had limited observedLFMC data Specifically the model was able to capture 96 86 and 80 of the variance in observed LFMC for the2015ndash2016 period for three PFTs respectively (Fig 2b df) The model was also able to capture the seasonal dynamicsof soil water content LFMC and LFMC below the thresh-old of 79 in comparison to observed data (Figs 2a c eand S3) To validate that FATES-HYDRO is able to capturethe interannual variability of LFMC we compared the simu-lated LFMC for PFT-LA with the long-term observations ofLFMC for the chamise species (Adenostoma fasciculatumFig S5) Our results showed that the model is able to rea-sonably capture the seasonal and interannual variability forthe 2006ndash2019 period (R2

= 07) although it underestimatespeaks in LFMC in 4 of 14 years



Figure 2 Simulated and observed monthly live fuel moisture content and related R2 values for three PFTs (see Fig 1 for an explanation ofthe PFTs)

32 Changes in the LFMC and fire season length fromhistorical to future periods

Using the validated model driven by climate projections from20 ESMs under greenhouse gas emission scenarios RCP45and RCP85 we found that the daily mean LFMC during the

future 2080ndash2099 period was projected to become signifi-cantly lower than that during the historical 1960ndash1999 pe-riod for all three PFTs (Fig 3 plt 0000001) Our resultsalso showed that the spread among models increases withtime suggesting a larger uncertainty in the future projectionSpecifically the histogram of daily mean LFMC during the



AprilndashOctober dry season showed that there was a higherprobability of low LFMC under future climate conditions(Fig S1) The daily mean LFMC decreased from 847 1013 and 784 during the historical 1960ndash1999 periodto 810 ndash828 963 ndash988 and 748 ndash766 dur-ing the future 2080ndash2099 period under both climate scenar-ios for PFT-LA PFT-MC and PFT-HA respectively (Fig 3)

Based on the projected LFMC there was a significant in-crease in the fire season length with the critical threshold ofLFMC from the historical 1960ndash1999 period to the future2080ndash2099 period for three PFTs With the critical thresh-old of 79 LFMC the fire season length was projected toincrease by 20 22 and 19 d under RCP85 (Fig 4 and Ta-ble S4) and to increase by 9 11 and 8 d under RCP45 (Fig 4and Table S4) Our results also showed that the spread amongmodels increases with time suggesting a larger uncertaintyin the future projection The above results for mean LFMCand fire season length support hypothesis H0 that future cli-mate change will decrease LFMC and consequently result ina longer fire season as determined by critical thresholds forLFMC for all three PFTs

33 Relative effects of individual climate changes on thelength of the fire season

In order to better understand the relative contribution to fireseason length of different climate variables we ran FATES-HYDRO for three PFTs using meteorological forcings thatisolated and removed changes in individual specific vari-ables Our results showed that the increase in fire seasonlength mainly resulted from warming which led to a 16ndash23 d(91 ndash186 ) per year increase in fire season length for thecritical threshold of 79 LFMC under RCP85 (Fig 5) Thisis because warming is pushing vapor pressure deficit (VPD)higher resulting in increased fire season length For RCP45the warming contributed to a 5ndash6 d (38 ndash43 ) per yearincrease in fire season length (Fig 5) We also found that el-evated CO2 concentrations decreased fire season length witha 6ndash7 d (35 ndash48 ) per year decrease in fire season lengthunder RCP85 (Fig 5) Under RCP45 CO2 increases led toa 2ndash3 d (15 ndash22 ) per year decrease in fire season length(Fig 5) Because the impact of warming on fire season lengthwas stronger than the mitigation from CO2 enrichment ourresults support hypothesis H1 (warming has a stronger im-pact on LFMC than CO2 fertilization)

Even though total precipitation was projected to increasein the future lower precipitation in the spring and autumn(Fig S2a b) led to an 8ndash10 d (63 ndash81 ) per year in-crease in fire season length with the critical threshold of 79 LFMC under RCP85 (Fig 5) Under RCP45 the precipita-tion changes contributed to a 1ndash3 d (08 ndash16 ) increasein fire season length (Fig 5) This result supported hypothe-sis H2 that the reductions in spring and autumn precipitationlead to a longer fire season as determined by LFMC

Our results showed that the combined impacts of warmingand precipitation on fire season length were equal to the ad-ditive impacts of warming and precipitation change individ-ually This supported hypothesis H3 Specifically the com-bined changes in temperature and precipitation caused a 24ndash33 d yrminus1 (156 ndash268 ) increase in fire season length withthe critical threshold of 79 LFMC under RCP85 (Fig 5)Under RCP45 the combined changes in temperature andprecipitation caused a 6ndash9 d yrminus1 (48 ndash61 ) change infire season length

34 Comparison of changes in fire season length amongthree PFTs under climate change

Regarding three PFTs under both climate scenarios the fireseason length for PFT-HA was the longest (167ndash176 d yrminus1)while fire season length for PFT-MC was the shortest (114ndash124 d yrminus1) during 2080ndash2099 (Fig 4) However the re-sponse of fire season length to warming was strongest forPFT-MC Specifically for PFT-MC warming under RCP85led to an increase of 216 (22 d) in fire season length(Fig 5b) and warming under RCP45 led to an increase of108 (11 d) in fire season length For PFT-LA warmingunder RCP85 led to an increase of 147 (19 d) in fire sea-son length (Fig 5a) while warming under RCP45 led to anincrease of 74 (9 d) in fire season length Finally for PFT-HA warming under RCP85 led to an increase of 102 (18 d) in fire season length (Fig 5c) and 53 (8 d) in fireseason length under RCP45 Because PFT-MC has a moreconservative hydraulic strategy with a less negative P50 tur-gor loss point and water potential at full turgor this resultsupported hypothesisH4 that the LFMC for plants with moreconservative hydraulic strategy will be more vulnerable towarming

To validate our classification scheme we compared thesePFT-level results to those obtained with single-PFT and 2-PFT simulations and found that using the three PFTs definedby our cluster analysis gives a qualitatively different view ofLFMC change than a single- or 2-PFT simulation We foundsignificant differences in the percentage changes of LFMCand fire season length between the future period (2080ndash2099)and historical period (1960ndash1999) using three distinct PFTsbut no significant differences between PFTs in 2-PFT simu-lations under the different climate scenarios (Fig S6)

4 Discussion

Low LFMC within shrub leaves and shoots increases theflammability and likelihood of combustion making it vi-tally important to monitor temporal variations in LFMCespecially during the dry season (Dennison et al 2008)The strong relationships between observed and simulatedLFMC of all PFTs (Fig 2) suggested that the plant hydro-dynamic model FATES-HYDRO could accurately estimate



Figure 3 Temporal changes in daily mean live fuel moisture content (black solid line) and 95 confidence interval (black dash-dot line)from 1960 to 2099 for three PFTs (see Fig 1 for an explanation of the PFTs) under climate scenarios RCP45 and 85 with 20 Earth systemmodels considering all climatic variables changes The p values were calculated using bootstrap sampling to test whether the daily mean livefuel moisture content across different models during the future period (2080ndash2099) was significantly lower than that during the historicalperiod (1960ndash1999) The gray horizontal dotted line represents the ensemble mean for 2080ndash2099

LFMC seasonal dynamics as a function of modeled leaf wa-ter content and consequently be useful to predict fire risks inMediterranean-type climate regions although only a smallamount of validation data were used and the underlying as-sumption that a shrub was analogous to a small tree wasmade Based on the simulated monthly mean LFMC during2006ndash2019 for PFT-LA which includes the chamise specieswe found that our model can capture the seasonal variationand interannual variability but underestimates the highestwet season peaks in LFMC in 4 of 14 years (Fig S5) Whileit would not highly affect the long-term trend of LFMC andfire season length this may cause biases for future projec-

tions During both the future period (2080ndash2099) and his-torical period (1960ndash1999) lower values in the dry season(AprilndashOctober) were displayed which is consistent withlower LFMC during the summerndashfall dry season rather thanthe winterndashspring wet season (Chuvieco et al 2004 Pel-lizzaro et al 2007 Pivovaroff et al 2019) Extremely lowdaily LFMC was more likely to occur during the future pe-riod which had higher temperature than the historical periodFrom the historical to the future period fire season lengthcould increase by 52 ndash148 under climate change forchaparral shrub ecosystems (H0) The fire season length was



Figure 4 Temporal changes in average number of days per year of live fuel moisture content below 79 (black solid line) and 95 confidence interval (black dash-dot line) from 1960 to 2099 for three PFTs (see Fig 1 for an explanation of the PFTs) under climate scenarioRCP45 and 85 with 20 Earth system models considering all climatic variables changes The p values were calculated using bootstrapsampling to test whether the number of days across different models during the future period (2080ndash2099) was significantly higher than thatduring the historical period (1960ndash1999) The gray horizontal dotted line represents the ensemble mean for 2080ndash2099

not validated rather it was defined as the number of dayswith LFMC below 79

Quantifying influences of climatic variables on LFMC iscrucial to predicting future fire risks (Dennison and Moritz2009) Our results showed that future warming was themost important driver of LFMC This finding suggestedthat warming would substantially push vapor pressure deficit(VPD) higher decrease LFMC and strongly increase the fireseason length which may greatly increase fire risks in thefuture (eg Dennison et al 2008 Chuvieco et al 2009Pimont et al 2019) CO2 fertilization is expected to re-duce stomatal conductance (Pataki et al 2000 Tognetti et

al 2000) and thus could mitigate the impacts of warm-ing on LFMC Our results illustrated that even though theCO2 impact did cause a 35 ndash48 reduction in fire seasonlength the impact of warming on fire season length is about56 ndash138 larger than the CO2 effect (H1 warming has astronger impact on LFMC than CO2 fertilization) This resultsuggests that CO2 fertilization cannot offset the LFMC im-pacts from warming The FATES-HYDRO model assumes aconsistent stomatal sensitivity to CO2 concentration acrossMediterranean shrub species While Mediterranean shrubfunctional types in arid and semi-arid systems would vary intheir stomatal response in the real world (Pataki et al 2000)



Figure 5 Differences in number of days per year of live fuelmoisture content below 79 from 2080 to 2099 for three PFTs(see Fig 1 for an explanation of the PFTs) under climate scenarioRCP45 and 85 between considering all climatic variable changesand removing the CO2 precipitation temperature and precipitationand temperature changes

Therefore our model may overestimate or underestimate theCO2 effect on stomatal conductance and its mitigating influ-ence might be smaller in reality for some species

Previous studies implied that the timing of precipitationmay have a strong impact on subsequent LFMC (eg Veblenet al 2000 Westerling et al 2006 Dennison and Moritz2009) In this study precipitation was also a key driver ofLFMC under future climate conditions Our results showedthat even though total precipitation was projected to increasethe reduction in spring and autumn precipitation (Fig S2)was projected to cause a longer fire season length (H2 the re-ductions in spring and autumn precipitation lead to a longerfire season as determined by LFMC Fig 5) This result was

in agreement with a prior study indicating that spring pre-cipitation particularly in the month of March was found tobe the primary driver of timing of LFMC changes (Denni-son and Moritz 2009) We also found that the combined im-pacts of warming and precipitation on fire season length wereequal to the linearly additive impacts of warming and precip-itation change individually (H3) Our results suggested thatwhen evaluating future fire risks it is critical that we con-sider the seasonal changes in precipitation and its interactionwith the warming impact

Modeled vegetation responses to environmental changes isa function of variation in plant functional traits (Koven et al2020) The three PFTs represented in this study have simi-lar patterns in LFMC in response to climate change during1960ndash2099 but we did see some critical differences Specif-ically the plant functional type PFT-MC with more conser-vative hydraulic strategy had the strongest responses to cli-mate change (Fig 5) This could be related to the fact thatthe PFT-MC is a more conservative drought tolerant PFT interms of hydraulic strategy with less negative P50 turgor losspoint and water potential at full turgor The PFT-MC plantshad a relatively high saturated water content based on ob-served data (Fig 2) and the water within plant tissues thuschanges more quickly in response to the environmental con-dition changes (H4 LFMC for plants with more conservativehydraulic strategies will be more vulnerable to warming)However the three different PFTs coexisted at the same loca-tion in model simulations therefore coexistence and hetero-geneity in LFMC might impact fire behavior and fire seasonlength

Because the moisture content of live fuels (sim 50 ndash200 )is much higher than that of dead fuels (sim 7 ndash30 ) leafsenescence induced by drought stress and subsequent mortal-ity are potentially vital factors to cause large wildfires (Nolanet al 2016 2020) Thus drought-induced canopy diebackand mortality could largely increase surface fine fuel loadsand vegetation flammability which can increase the proba-bility of wildfires (Ruthrof et al 2016) Since growth andmortality are turned off in model runs by using a reduced-complexity configuration it is possible that vegetation den-sity might decrease and LFMC could be conserved underfuture scenarios In addition potential vegetation transitions(eg shrubs to grassland and species composition changes)might substantially affect flammability and thus fire inten-sity and frequency In this study we used the static mode ofFATES-HYDRO to simulate LWC dynamics under climatechange If we need to assess how the leaf senescence andvegetation dynamics will impact the fire behavior we can usethe same model with dynamic mode to assess their impact onfire behavior under future drought and warming conditions

Application of a hydrodynamic vegetation model to es-timate LFMC dynamics could potentially benefit wildfiremodeling at the fine scale landscape scale and global scaleLFMC is a potentially critical factor influencing fire spreadand consumption Many previous wildfire models focus on



the impacts of dead fuel moisture weather conditions andfuel loads rather than the representation of live fuel moisture(Anderson and Anderson 2010 Keeley et al 2011 Jollyand Johnson 2018) The implications of including LFMCare that fire potential will change with plant water potentialand uptake from soils photosynthetic and respiratory activ-ity carbon allocation and phenology with variability acrossspecies over time (Jolly and Johnson 2018) Therefore workto incorporate LFMC dynamics in models could play a vi-tally important role in projections of wildfire behavior andeffects under current and future climate

5 Conclusions

A hydrodynamic vegetation model FATES-HYDRO wasused to estimate leaf water status and thus LFMC dynam-ics of chaparral shrub species in southern California un-der historical and future conditions The FATES-HYDROmodel was validated using monthly mean LFMC for threePFTs The fire season length was projected to substantiallyincrease under both climate scenarios from 1960ndash1999 to2080ndash2099 This could increase wildfire risk over time forchaparral shrubs in southern California Our results showedthat temperature was the most important driver of LFMCamong all climatic variables The LFMC estimated by theFATES-HYDRO model offered a baseline of predicting planthydraulic dynamics subjected to climate change and pro-vided a critical foundation that reductions in LFMC fromclimate warming may exacerbate future wildfire risk Longerfire seasons might have a significant impact on overall publichealth and quality of life in the future

Data availability LFMC measurement data can be found in Pivo-varoff et al (2019) Species abundance data can be found in Ven-turas et al (2016) All other data are available within this paper andin the Supplement

Supplement The supplement related to this article is available on-line at httpsdoiorg105194bg-18-4005-2021-supplement

Author contributions WM LZ and CX were involved in designingthe study WM and CX conducted the model simulations and dataanalysis in consultation with JS PB JD MM CDK and LK WMand CX wrote the first draft of the manuscript LZ AP JS PBJD BC RK MM RAF CDK and LK reviewed and improved themanuscript

Competing interests The authors declare that they have no conflictof interest

Disclaimer Publisherrsquos note Copernicus Publications remainsneutral with regard to jurisdictional claims in published maps andinstitutional affiliations

Acknowledgements We thank the editor Martin De Kauwe and thethree referees for their help improving this paper

This project is supported by the University of California Office ofthe President Lab Fees Research Program and the Next GenerationEcosystem Experiment (NGEE) Tropics which is supported by theUS DOE Office of Science Charles D Koven and Junyan Dingare supported by the DOE Office of Science Regional and GlobalModel Analysis Program Early Career Research Program

Financial support This research has been supported by the Uni-versity of California Office of the President Lab Fees ResearchProgram and the Next Generation Ecosystem Experiment (NGEE)Tropics which is supported by the US DOE Office of Science theDOE Office of Science Regional and Global Model Analysis Pro-gram Early Career Research Program Los Alamos National Lab-oratory High Performance Computing and the USDepartment ofEnergy Office of Science Office of Biological and EnvironmentalResearch

Review statement This paper was edited by Martin De Kauwe andreviewed by Douglas I Kelley and two anonymous referees

References

Abatzoglou J T Development of gridded surface meteorologicaldata for ecological applications and modelling Int J Climatol33 121ndash131 httpsdoiorg101002joc3413 2013

Abatzoglou J T and Brown T J A comparison of statisticaldownscaling methods suited for wildfire applications Int J Cli-matol 32 772ndash780 httpsdoiorg101002joc2312 2012

Agee J K Wright C S Williamson N and Huff M H Fo-liar moisture content of Pacific Northwest vegetation and its re-lation to wildland fire behavior For Ecol Manag 167 57ndash66httpsdoiorg101016S0378-1127(01)00690-9 2002

Aguado I Chuvieco E Boren R and Nieto H Estimation ofdead fuel moisture content from meteorological data in Mediter-ranean areas Applications in fire danger assessment Int J Wild-land Fire 16 390ndash397 httpsdoiorg101071WF06136 2007

Anderson S A and Anderson W R Ignition and fire spreadthresholds in gorse (Ulex europaeus) Int J Wildland Fire 19589ndash598 httpsdoiorg101071WF09008 2010

Balch J K Bradley B A Abatzoglou J T Nagy R C FuscoE J and Mahood A L Human-started wildfires expand thefire niche across the United States P Natl Acad Sci USA 1142946ndash2951 httpsdoiorg101073pnas1617394114 2017

Bartlett M K Scoffoni C and Sack L The determinants of leafturgor loss point and prediction of drought tolerance of speciesand biomes a global meta-analysis Ecol Lett 15 393ndash405httpsdoiorg101111j1461-0248201201751x 2012



Bilgili E and Saglam B Fire behavior in maquis fuels in TurkeyFor Ecol Manag 184 201ndash207 httpsdoiorg101016S0378-1127(03)00208-1 2003

Bistinas I Harrison S P Prentice I C and Pereira J MC Causal relationships versus emergent patterns in the globalcontrols of fire frequency Biogeosciences 11 5087ndash5101httpsdoiorg105194bg-11-5087-2014 2014

Bridges Jr C C Hierarchical cluster analysis Psychol Rep 18851ndash854 httpsdoiorg102466pr01966183851 1966

Caccamo G Chisholm L A Bradstock R A and Puotinen ML Using remotely-sensed fuel connectivity patterns as a toolfor fire danger monitoring Geophys Res Lett 39 L01302httpsdoiorg1010292011GL050125 2012a

Caccamo G Chisholm L A Bradstock R A Puotinen ML and Pippen B G Monitoring live fuel moisture contentof heathland shrubland and sclerophyll forest in south-easternAustralia using MODIS data Int J Wildland Fire 21 257ndash269httpsdoiorg101071WF11024 2012b

Caldwell P M Mametjanov A Tang Q Van Roekel L P Go-laz J C Lin W Bader D C Keen N D Feng Y Jacob Rand Maltrud M E The DOE E3SM coupled model version 1Description and results at high resolution J Adv Model EarthSyst 11 4095ndash4146 httpsdoiorg1010292019MS0018702019

Castro F X Tudela A and Sebastiagrave M T Modeling moisturecontent in shrubs to predict fire risk in Catalonia (Spain) AgricFor Meteorol 116 49ndash59 httpsdoiorg101016S0168-1923(02)00248-4 2003

Christoffersen B O Gloor M Fauset S Fyllas N M Gal-braith D R Baker T R Kruijt B Rowland L Fisher RA Binks O J Sevanto S Xu C Jansen S Choat B Men-cuccini M McDowell N G and Meir P Linking hydraulictraits to tropical forest function in a size-structured and trait-driven model (TFS v1-Hydro) Geosci Model Dev 9 4227ndash4255 httpsdoiorg105194gmd-9-4227-2016 2016

Chuvieco E Cocero D Riano D Martin P Martınez-VegaJ de la Riva J and Peacuterez F Combining NDVI and surfacetemperature for the estimation of live fuel moisture content inforest fire danger rating Remote Sens Environ 92 322ndash331httpsdoiorg101016jrse200401019 2004

Chuvieco E Gonzaacutelez I Verduacute F Aguado I and Yebra MPrediction of fire occurrence from live fuel moisture contentmeasurements in a Mediterranean ecosystem Int J WildlandFire 18 430ndash441 httpsdoiorg101071WF08020 2009

Clarke H Pitman A J Kala J Carouge C Haverd Vand Evans J P An investigation of future fuel load andfire weather in Australia Clim Change 139 591ndash605httpsdoiorg101007s10584-016-1808-9 2016

Dennison P E and Moritz M A Critical live fuel moisture in cha-parral ecosystems a threshold for fire activity and its relationshipto antecedent precipitation Int J Wildland Fire 18 1021ndash1027httpsdoiorg101071WF08055 2009

Dennison P E Moritz M A and Taylor R S Evaluating pre-dictive models of critical live fuel moisture in the Santa Mon-ica Mountains California Int J Wildland Fire 17 18ndash27httpsdoiorg101071WF07017 2008

Dimitrakopoulos A P and Papaioannou K K Flammability as-sessment of Mediterranean forest fuels Fire Technol 37 143ndash152 httpsdoiorg101023A1011641601076 2001

Duursma R A and Medlyn B E MAESPA a model to studyinteractions between water limitation environmental drivers andvegetation function at tree and stand levels with an example ap-plication to [CO2]times drought interactions Geosci Model Dev5 919ndash940 httpsdoiorg105194gmd-5-919-2012 2012

Fisher R A Muszala S Verteinstein M Lawrence P Xu CMcDowell N G Knox R G Koven C Holm J RogersB M Spessa A Lawrence D and Bonan G Taking off thetraining wheels the properties of a dynamic vegetation modelwithout climate envelopes CLM45(ED) Geosci Model Dev8 3593ndash3619 httpsdoiorg105194gmd-8-3593-2015 2015

Fisher R A Koven C D Anderegg W R ChristoffersenB O Dietze M C Farrior C E Holm J A HurttG C Knox R G Lawrence P J and Lichstein J WVegetation demographics in Earth System Models A reviewof progress and priorities Glob Change Biol 24 35ndash54httpsdoiorg101111gcb13910 2018

Flannigan M D Krawchuk M A de Groot W J WottonB M and Gowman L M Implications of changing climatefor global wildland fire Int J Wildland Fire 18 483ndash507httpsdoiorg101071WF08187 2009

Gillett N P Weaver A J Zwiers F W and Flanni-gan M D Detecting the effect of climate change onCanadian forest fires Geophys Res Lett 31 L18211httpsdoiorg1010292004GL020876 2004

Goss M Swain D L Abatzoglou J T Sarhadi A KoldenC A Williams A P and Diffenbaugh N S Climate changeis increasing the likelihood of extreme autumn wildfire con-ditions across California Environ Res Lett 15 094016httpsdoiorg1010881748-9326ab83a7 2020

Hantson S Arneth A Harrison S P Kelley D I Prentice I CRabin S S Archibald S Mouillot F Arnold S R ArtaxoP Bachelet D Ciais P Forrest M Friedlingstein P HicklerT Kaplan J O Kloster S Knorr W Lasslop G Li F Man-geon S Melton J R Meyn A Sitch S Spessa A van derWerf G R Voulgarakis A and Yue C The status and chal-lenge of global fire modelling Biogeosciences 13 3359ndash3375httpsdoiorg105194bg-13-3359-2016 2016

Holm J A Shugart H H Van Bloem S J andLarocque G R Gap model development validationand application to succession of secondary subtropicaldry forests of Puerto Rico Ecol Modell 233 70ndash82httpsdoiorg101016jecolmodel201203014 2012

Jackson D A Stopping rules in principal components analysisa comparison of heuristical and statistical approaches Ecology74 2204ndash2214 httpsdoiorg1023071939574 1993

Jacobsen A L Pratt R B Davis S D and Ewers FW Comparative community physiology nonconvergence inwater relations among three semi-arid shrub communitiesNew Phytol 180 100ndash113 httpsdoiorg101111j1469-8137200802554x 2008

Jolly W M and Johnson D M Pyro-ecophysiology shift-ing the paradigm of live wildland fuel research Fire 1 8httpsdoiorg103390fire1010008 2018

Keeley J E Future of California floristics and systematics wildfirethreats to the California flora Madrono 42 175ndash179 availableat httpswwwjstororgstable41425064 (last access 25 March2021) 1995



Keeley J E and Zedler P H Large high-intensity fire eventsin southern California shrublands debunking the fine-grain agepatch model Ecol Appl 19 69ndash94 httpsdoiorg10189008-02811 2009

Keeley J E Bond W J Bradstock R A Pausas J Gand Rundel P W Fire in Mediterranean ecosystems ecol-ogy evolution and management Cambridge University Presshttpsdoiorg104000mediterranee6936 2011

Kelley D I Bistinas I Whitley R Burton C Marthews T Rand Dong N How contemporary bioclimatic and human con-trols change global fire regimes Nat Clim Change 9 690ndash696httpsdoiorg101038s41558-019-0540-7 2019

Kennedy D Swenson S Oleson K W Lawrence DM Fisher R Lola da Costa A C and Gentine PImplementing plant hydraulics in the community landmodel version 5 J Adv Model Earth Syst 11 485ndash513httpsdoiorg1010292018MS001500 2019

Koven C D Knox R G Fisher R A Chambers J Q Christof-fersen B O Davies S J Detto M Dietze M C Fay-bishenko B Holm J Huang M Kovenock M KueppersL M Lemieux G Massoud E McDowell N G Muller-Landau H C Needham J F Norby R J Powell T RogersA Serbin S P Shuman J K Swann A L S VaradharajanC Walker A P Wright S J and Xu C Benchmarking andparameter sensitivity of physiological and vegetation dynamicsusing the Functionally Assembled Terrestrial Ecosystem Simula-tor (FATES) at Barro Colorado Island Panama Biogeosciences17 3017ndash3044 httpsdoiorg105194bg-17-3017-2020 2020

Krawchuk M A Moritz M A Parisien M A VanDorn J and Hayhoe K Global pyrogeography the cur-rent and future distribution of wildfire Plos One 4 e5102httpsdoiorg101371journalpone0005102 2009

Linn R Reisner J Colman J J and Winterkamp J Studyingwildfire behavior using FIRETEC Int J Wildland Fire 11 233ndash246 httpsdoiorg101071WF02007 2002

Liu Y Stanturf J and Goodrick S Trends in global wildfire po-tential in a changing climate For Ecol Manag 259 685ndash697httpsdoiorg101016jforeco200909002 2010

Massoud E C Xu C Fisher R A Knox R G Walker A PSerbin S P Christoffersen B O Holm J A Kueppers L MRicciuto D M Wei L Johnson D J Chambers J Q KovenC D McDowell N G and Vrugt J A Identification ofkey parameters controlling demographically structured vegeta-tion dynamics in a land surface model CLM45(FATES) GeosciModel Dev 12 4133ndash4164 httpsdoiorg105194gmd-12-4133-2019 2019

Matthews S Sullivan A L Watson P and Williams R J Cli-mate change fuel and fire behaviour in a eucalypt forest GlobChange Biol 18 3212ndash3223 httpsdoiorg101111j1365-2486201202768x 2012

McDowell N G Fisher R A Xu C Domec J C Houmllttauml TMackay D S Sperry J S Boutz A Dickman L GehresN and Limousin J M Evaluating theories of drought-inducedvegetation mortality using a multimodelndashexperiment frameworkNew Phytol 200 304ndash321 httpsdoiorg101111nph124652013

Meinshausen M Smith S J Calvin K Daniel J S KainumaM L Lamarque J F Matsumoto K Montzka S A RaperS C Riahi K and Thomson A G The RCP greenhouse

gas concentrations and their extensions from 1765 to 2300Clim Change 109 213ndash241 httpsdoiorg101007s10584-011-0156-z 2011

Meinzer F C James S A Goldstein G and Woodruff DWhole-tree water transport scales with sapwood capacitance intropical forest canopy trees Plant Cell Environ 26 1147ndash1155httpsdoiorg101046j1365-3040200301039x 2003

Meinzer F C Johnson D M Lachenbruch B McCulloh KA and Woodruff D R Xylem hydraulic safety margins inwoody plants coordination of stomatal control of xylem ten-sion with hydraulic capacitance Funct Ecol 23 922ndash930httpsdoiorg101111j1365-2435200901577x 2009

Mencuccini M Manzoni S and Christoffersen B Modellingwater fluxes in plants from tissues to biosphere New Phytol222 1207ndash1222 httpsdoiorg101111nph15681 2019

Mikkelsen T N Beier C Jonasson S Holmstrup M SchmidtI K Ambus P Pilegaard K Michelsen A Albert K An-dresen L C and Arndal M F Experimental design of multi-factor climate change experiments with elevated CO2 warmingand drought the CLIMAITE project Funct Ecol 22 185ndash195httpsdoiorg101111j1365-2435200701362x 2008

Moorcroft P R Hurtt G C and Pacala S W A method forscaling vegetation dynamics the ecosystem demography model(ED) Ecol Monogr 71 557ndash586 httpsdoiorg1018900012-9615(2001)071[0557AMFSVD]20CO2 2001

Moritz M A Parisien M A Batllori E Krawchuk M AVan Dorn J Ganz D J and Hayhoe K Climate changeand disruptions to global fire activity Ecosphere 3 1ndash22httpsdoiorg101890ES11-003451 2012

Nolan R H Boer M M Resco de Dios V CaccamoG and Bradstock R A Large-scale dynamic trans-formations in fuel moisture drive wildfire activity acrosssoutheastern Australia Geophys Res Lett 43 4229ndash4238httpsdoiorg1010022016GL068614 2016

Nolan R H Blackman C J de Dios V R Choat B MedlynB E Li X Bradstock R A and Boer M M Linking forestflammability and plant vulnerability to drought Forests 11 779httpsdoiorg103390f11070779 2020

Pataki D E Huxman T E Jordan D N Zitzer S F Cole-man J S Smith S D Nowak R S and Seemann J RWater use of two Mojave Desert shrubs under elevated CO2Glob Change Biol 6 889ndash897 httpsdoiorg101046j1365-2486200000360x 2000

Pellizzaro G Cesaraccio C Duce P Ventura A andZara P Relationships between seasonal patterns of live fuelmoisture and meteorological drought indices for Mediter-ranean shrubland species Int J Wildland Fire 16 232ndash241httpsdoiorg101071WF06081 2007

Pimont F Ruffault J Martin-StPaul N K and Dupuy J L Whyis the effect of live fuel moisture content on fire rate of spread un-derestimated in field experiments in shrublands Int J WildlandFire 28 127ndash137 httpsdoiorg101071WF18056 2019

Pineda-Garcia F Paz H and Meinzer F C Drought resistancein early and late secondary successional species from a tropicaldry forest the interplay between xylem resistance to embolismsapwood water storage and leaf shedding Plant Cell Environ36 405ndash418 httpsdoiorg101111j1365-3040201202582x2013



Pivovaroff A L Emery N Sharifi M R Witter M Kee-ley J E and Rundel P W The effect of ecophysio-logical traits on live fuel moisture content Fire 2 28httpsdoiorg103390fire2020028 2019

Plucinski M P The investigation of factors governing ignition anddevelopment of fires in heathland vegetation PhD thesis Univer-sity of New South Wales Sydney Australia 2003

Powell T L Koven C D Johnson D J Faybishenko B FisherR A Knox R G McDowell N G Condit R Hubbell S PWright S J and Chambers J Q Variation in hydroclimate sus-tains tropical forest biomass and promotes functional diversityNew Phytol 219 932ndash946 httpsdoiorg101111nph152712018

Rabin S S Melton J R Lasslop G Bachelet D ForrestM Hantson S Kaplan J O Li F Mangeon S Ward DS Yue C Arora V K Hickler T Kloster S Knorr WNieradzik L Spessa A Folberth G A Sheehan T Voul-garakis A Kelley D I Prentice I C Sitch S HarrisonS and Arneth A The Fire Modeling Intercomparison Project(FireMIP) phase 1 experimental and analytical protocols withdetailed model descriptions Geosci Model Dev 10 1175ndash1197 httpsdoiorg105194gmd-10-1175-2017 2017

Rind D Goldberg R Hansen J Rosenzweig C and RuedyR Potential evapotranspiration and the likelihood of fu-ture drought J Geophys Res Atmos 95 9983ndash10004httpsdoiorg101029JD095iD07p09983 1990

Rossa C G and Fernandes P M Live fuel moisture content Theldquopea under the mattressrdquo of fire spread rate modeling Fire 143 httpsdoiorg103390fire1030043 2018

Rothermel R C A mathematical model for predicting fire spreadin wildland fuels (Vol 115) Intermountain Forest amp Range Ex-periment Station Forest Service US Department of AgricultureMinneapolis USA 1972

Ruthrof K X Fontaine J B Matusick G Breshears DD Law D J Powell S and Hardy G How drought-induced forest die-off alters microclimate and increases fuelloadings and fire potentials Int J Wildland Fire 25 819ndash830httpsdoiorg101071WF15028 2016

Saura-Mas S and Lloret F Leaf and shoot water content and leafdry matter content of Mediterranean woody species with dif-ferent post-fire regenerative strategies Ann Bot 99 545ndash554httpsdoiorg101093aobmcl284 2007

Schroeder M J Glovinsky M Henricks V F Hood F C andHull M K Synoptic weather types associated with critical fireweather USDA Forest Service Pacific Southwest Range and Ex-periment Station Berkeley CA USA 1964

Seiler C Hutjes R W A Kruijt B Quispe J Antildeez S AroraV K Melton J R Hickler T and Kabat P Modeling forestdynamics along climate gradients in Bolivia J Geophys Res-Biogeo 119 758ndash775 httpsdoiorg1010022013JG0025092014

Stocks B J Fosberg M A Lynham T J Mearns L Wotton BM Yang Q Jin J Z Lawrence K Hartley G R Mason JA and McKenney D W Climate change and forest fire poten-tial in Russian and Canadian boreal forests Clim Change 381ndash13 httpsdoiorg101023A1005306001055 1998

Sturtevant B R Scheller R M Miranda B R Shin-neman D and Syphard A Simulating dynamic andmixed-severity fire regimes a process-based fire exten-

sion for LANDIS-II Ecol Modell 220 3380ndash3393httpsdoiorg101016jecolmodel200907030 2009

Thonicke K Spessa A Prentice I C Harrison S P DongL and Carmona-Moreno C The influence of vegetation firespread and fire behaviour on biomass burning and trace gas emis-sions results from a process-based model Biogeosciences 71991ndash2011 httpsdoiorg105194bg-7-1991-2010 2010

Tjiputra J F Roelandt C Bentsen M Lawrence D MLorentzen T Schwinger J Seland Oslash and Heinze C Eval-uation of the carbon cycle components in the Norwegian EarthSystem Model (NorESM) Geosci Model Dev 6 301ndash325httpsdoiorg105194gmd-6-301-2013 2013

Tobin M F Lopez O R and Kursar T A Responsesof Tropical Understory Plants to a Severe Drought Toler-ance and Avoidance of Water Stress Biotropica 31 570ndash578httpsdoiorg101111j1744-74291999tb00404x 1999

Tognetti R Minnocci A Pentildeuelas J Raschi A and Jones MB Comparative field water relations of three Mediterraneanshrub species co-occurring at a natural CO2 vent J Exp Bot51 1135ndash1146 httpsdoiorg101093jexbot5134711352000

Tyree M T and Hammel H T The measurement ofthe turgor pressure and the water relations of plants bythe pressure-bomb technique J Exp Bot 23 267ndash282httpsdoiorg101093jxb231267 1972

Tyree M T and Yang S Water-storage capacity of Thuja Tsugaand Acer stems measured by dehydration isotherms Planta 182420ndash426 httpsdoiorg101007BF02411394 1990

Veblen T T Kitzberger T and Donnegan J Cli-matic and human influences on fire regimes in pon-derosa pine forests in the Colorado Front Range EcolAppl 10 1178ndash1195 httpsdoiorg1018901051-0761(2000)010[1178CAHIOF]20CO2 2000

Venturas M D MacKinnon E D Dario H L Jacobsen A LPratt R B and Davis S D Chaparral shrub hydraulic traitssize and life history types relate to species mortality duringCaliforniarsquos historic drought of 2014 Plos One 11 e0159145httpsdoiorg101371journalpone0159145 2016

Wei L Xu C Jansen S Zhou H Christoffersen B O Pock-man W T Middleton R S Marshall J D and McDowellN G A heuristic classification of woody plants based on con-trasting shade and drought strategies Tree Physiol 39 767ndash781httpsdoiorg101093treephystpy146 2019

Westerling A L Gershunov A Brown T J Cayan D Rand Dettinger M D Climate and wildfire in the west-ern United States Bull Am Meteorol Soc 84 595ndash604httpsdoiorg101175BAMS-84-5-595 2003

Westerling A L Hidalgo H G Cayan D R and Swet-nam T W Warming and earlier spring increase west-ern US forest wildfire activity Science 313 940ndash943httpsdoiorg101098rstb20150178 2006

Williams A P Abatzoglou J T Gershunov A Guzman-Morales J Bishop D A Balch J K and Letten-maier D P Observed impacts of anthropogenic climatechange on wildfire in California Earths Future 7 892ndash910httpsdoiorg1010292019EF001210 2019

Wu J Serbin S P Ely K S Wolfe B T Dickman L TGrossiord C Michaletz S T Collins A D Detto M Mc-Dowell N G and Wright S J The response of stomatal con-



ductance to seasonal drought in tropical forests Glob ChangeBiol 26 823ndash839 httpsdoiorg101111gcb14820 2020

Wullschleger S D Gunderson C A Hanson P J WilsonK B and Norby R J Sensitivity of stomatal and canopyconductance to elevated CO2 concentrationndashinteracting vari-ables and perspectives of scale New Phytol 153 485ndash496httpsdoiorg101046j0028-646X200100333x 2002

Xu C McDowell N G Sevanto S and Fisher R A Our limitedability to predict vegetation dynamics under water stress NewPhytol 200 298ndash300 httpsdoiorg101111nph12450 2013

Xu C McDowell N G Fisher R A Wei L Sevanto SChristoffersen B O Weng E and Middleton R S In-creasing impacts of extreme droughts on vegetation produc-tivity under climate change Nat Clim Change 9 948ndash953httpsdoiorg101038s41558-019-0630-6 2019

Xu X Medvigy D Powers J S Becknell J M andGuan K Diversity in plant hydraulic traits explains sea-sonal and inter-annual variations of vegetation dynamics inseasonally dry tropical forests New Phytol 212 80ndash95httpsdoiorg101111nph14009 2016

Yebra M Chuvieco E and Riantildeo D Estimation oflive fuel moisture content from MODIS images for firerisk assessment Agric For Meteorol 148 523ndash536httpsdoiorg101016jagrformet200712005 2008

Yebra M Quan X Riantildeo D Larraondo P R van Dijk AI and Cary G J A fuel moisture content and flammabilitymonitoring methodology for continental Australia based on op-tical remote sensing Remote Sens Environ 212 260ndash272httpsdoiorg104225415837cd92ada9f 2018

Zarco-Tejada P J Rueda C A and Ustin S L Water con-tent estimation in vegetation with MODIS reflectance data andmodel inversion methods Remote Sens Environ 85 109ndash124httpsdoiorg101016S0034-4257(02)00197-9 2003


Abstract

Introduction

Materials and methods

Study site

FATES-HYDRO model

Allometry and trait data for model parameterization

Model initialization

Live fuel moisture content for model validation

Climate drivers

Hypothesis testing

Results

Comparison between simulated and measured LFMC

Changes in the LFMC and fire season length from historical to future periods

Relative effects of individual climate changes on the length of the fire season

Comparison of changes in fire season length among three PFTs under climate change

Discussion

Conclusions

Data availability

Supplement

Author contributions

Competing interests

Disclaimer

Acknowledgements

Financial support

Review statement

References


1 Introduction

Historical warming and changes in precipitation have alreadyimpacted wildfires at a global scale (eg Stocks et al 1998Gillett et al 2004 Westerling et al 2003 2006) and it isexpected that accelerating future warming will continue tosubstantially affect global wildfires (eg Flannigan et al2009 Liu et al 2010 Moritz et al 2012) So far prior stud-ies have mainly focused on the impacts of dead fuel mois-ture fuel loads and weather conditions on wildfires Limitedstudies have applied proxies of live fuel moisture in globalfire models For example dead fuel moisture is found to berelated to fire ignition and fire spread potential (or potentialarea burned) (Aguado et al 2007) specific weather condi-tions such as increased vapor pressure deficit (Williams et al2019) can lead to a vast increase in fire activity (Goss et al2020) and wildfire fuel loads are projected to increase underclimate change (Matthews et al 2012 Clarke et al 2016)In global fire models studies have used proxies of live fuelmoisture (Bistinas et al 2014 Kelley et al 2019) as wellas explicit representation of live fuels (Hantson et al 2016Rabin et al 2017) While previous studies provide great in-sights into fire risks with changes in climate dead fuel mois-ture fuel loads and representation of live fuel moisture thereis still limited understanding of how climate change influ-ences live fuel moisture content (LFMC) and the consequentwildfire risks This is particularly true for the combined im-pacts of warming temperature altered precipitation and in-creasing CO2 fertilization (Chuvieco et al 2004 Pellizzaroet al 2007 Caccamo et al 2012a b Williams et al 2019Goss et al 2020)

A measure of water content within living plant tissue inrelation to their dry weight LFMC has been found to beone of the most critical factors influencing combustion firespread and fire consumption (eg Agee et al 2002 Zarco-Tejada et al 2003 Bilgili and Saglam 2003 Yebra et al2008 Dennison et al 2008 Anderson and Anderson 2010Keeley et al 2011) This is because a low LFMC leadsto increased flammability and a higher likelihood of igni-tion (Dimitrakopoulos and Papaioannou 2001) For instanceLFMC was found to be a significant factor contributing to theoccurrence of wildfires in Australia (Plucinski 2003 Nolanet al 2016 Yebra et al 2018 Rossa and Fernandes 2018Pimont et al 2019) Spain (Chuvieco et al 2009) and Cali-fornia (Santa Monica Mountains Dennison et al 2008 Den-nison and Moritz 2009 Pivovaroff et al 2019) Dennisonand Moritz (2009) found strong evidence of a LFMC thresh-old (79 ) for southern California chaparral shrubs whichmay determine when large fires can occur in this region

Vegetation moisture content is dependent on both eco-physiological characteristics of the species and environmen-tal conditions including both climatic variables and soil wa-ter availability (Rothermel 1972 Castro et al 2003 Pel-lizzaro et al 2007 Pivovaroff et al 2019 Nolan et al2020) So far little is known about the relative importance

of different climate variables to future LFMC dynamicsOn one hand warming could contribute to a higher atmo-spheric demand and higher evapotranspiration (Rind et al1990) and thus lead to a lower LFMC On the other handhigher CO2 concentration will decrease stomatal conduc-tance (Wullschleger et al 2002) and plant water loss andthus lead to a higher LFMC The impacts of CO2 and warm-ing could be complicated by local changes in precipitationpatterns and humidity (Mikkelsen et al 2008)

The sensitivity of LFMC to climate change is likely to beaffected by plant hydraulic traits (the plant properties thatregulate water transport and storage within plant tissues)which affect plant water regulation (Wu et al 2020) Vari-ations in hydraulic traits reflect contrasting plant droughtadaptation strategies when responding to dry conditions Twocontrasting overall strategies are (1) water stress avoidersand (2) water stress tolerators (Tobin et al 1999 Wei et al2019) The ldquoavoidersrdquo are generally characterized by a moreconservative hydraulic strategy under water stress by closingtheir stomata early dropping leaves or accessing deep waterto avoid more negative water potentials and therefore xylemcavitation Meanwhile the ldquotoleratorsrdquo typically build xylemand leaves that are more resistant to cavitation so that theycan tolerate more negative water potential and continue toconduct photosynthesis under water stress Therefore com-pared with the tolerators the avoiders normally have a lowersapwood density and higher plant water storage capacity intheir tissues to avoid cavitation (Meinzer et al 2003 2009Pineda-Garcia et al 2013) Because the avoiders rely on wa-ter storage capacity as one way to avoid cavitation therebymaintaining a relatively high LFMC and because water lossfrom storage should increase with warming LFMC could bemore sensitive to climate change in avoiders relative to toler-ators

While over half of terrestrial landscapes on Earth are con-sidered fire-prone (Krawchuk et al 2009) Mediterranean-type climate regions are routinely impacted by fire often onan annual basis This is partly because Mediterranean cli-mate regions are characterized by winter rains followed byan annual dry season when little to no rainfall occurs forseveral months Multiday periods of extreme high tempera-tures as well as katabatic hot dry and intense winds oftenpunctuate the annual drought leading to some of the worstfire weather in the world (Schroeder et al 1964) This canresult in wildfires that are large high intensity and stand re-placing (Keeley 1995 Keeley and Zedler 2009 Balch et al2017) Globally Mediterranean climate regions are charac-terized by evergreen sclerophyllous-leaved shrublands TheMediterranean climate region in California is dominated bychaparral which is adapted to the periodic fire regime in Cal-ifornia (Venturas et al 2016) Previous studies have pro-posed a variety of relationships between chaparral LFMCand fire danger in southern California (Dennison et al 2008Dennison and Moritz 2009) but less is known about how cli-mate changes could alter LFMC and fire danger In chaparral









21 Study site









Qi =minusKi1hi (1)




Btran =

[1+

(ψl

P50_gs

)al]minus1

(3)












LWC=fwminus dw

dwtimes 10 (4)


LFMC= 31091+ 0491LWC (5)





26 Climate drivers






3 Results





















4 Discussion

























5 Conclusions











References







































































































Abstract

Introduction


Study site

FATES-HYDRO model




Climate drivers

Hypothesis testing

Results





Discussion

Conclusions

Data availability

Supplement


Competing interests

Disclaimer

Acknowledgements

Financial support

Review statement

References








21 Study site









Qi =minusKi1hi (1)




Btran =

[1+

(ψl

P50_gs

)al]minus1

(3)












LWC=fwminus dw

dwtimes 10 (4)


LFMC= 31091+ 0491LWC (5)





26 Climate drivers






3 Results





















4 Discussion

























5 Conclusions











References







































































































Abstract

Introduction


Study site

FATES-HYDRO model




Climate drivers

Hypothesis testing

Results





Discussion

Conclusions

Data availability

Supplement


Competing interests

Disclaimer

Acknowledgements

Financial support

Review statement

References





Qi =minusKi1hi (1)




Btran =

[1+

(ψl

P50_gs

)al]minus1

(3)












LWC=fwminus dw

dwtimes 10 (4)


LFMC= 31091+ 0491LWC (5)





26 Climate drivers






3 Results





















4 Discussion

























5 Conclusions











References







































































































Abstract

Introduction


Study site

FATES-HYDRO model




Climate drivers

Hypothesis testing

Results





Discussion

Conclusions

Data availability

Supplement


Competing interests

Disclaimer

Acknowledgements

Financial support

Review statement

References










LWC=fwminus dw

dwtimes 10 (4)


LFMC= 31091+ 0491LWC (5)





26 Climate drivers






3 Results





















4 Discussion

























5 Conclusions











References







































































































Abstract

Introduction


Study site

FATES-HYDRO model




Climate drivers

Hypothesis testing

Results





Discussion

Conclusions

Data availability

Supplement


Competing interests

Disclaimer

Acknowledgements

Financial support

Review statement

References



26 Climate drivers






3 Results





















4 Discussion

























5 Conclusions











References







































































































Abstract

Introduction


Study site

FATES-HYDRO model




Climate drivers

Hypothesis testing

Results





Discussion

Conclusions

Data availability

Supplement


Competing interests

Disclaimer

Acknowledgements

Financial support

Review statement

References

















4 Discussion

























5 Conclusions











References







































































































Abstract

Introduction


Study site

FATES-HYDRO model




Climate drivers

Hypothesis testing

Results





Discussion

Conclusions

Data availability

Supplement


Competing interests

Disclaimer

Acknowledgements

Financial support

Review statement

References























5 Conclusions











References







































































































Abstract

Introduction


Study site

FATES-HYDRO model




Climate drivers

Hypothesis testing

Results





Discussion

Conclusions

Data availability

Supplement


Competing interests

Disclaimer

Acknowledgements

Financial support

Review statement

References































































































Abstract

Introduction


Study site

FATES-HYDRO model




Climate drivers

Hypothesis testing

Results





Discussion

Conclusions

Data availability

Supplement


Competing interests

Disclaimer

Acknowledgements

Financial support

Review statement

References

assessing climate change impacts on live fuel moisture and

Documents