Interactive effects of water limitation and elevated

Interactive effects of water limitation and elevated temperatureon the physiology, development and fitness of diverse accessionsof Brachypodium distachyon

David L. Des Marais1,4, Jesse R. Lasky2, Paul E. Verslues3, Trent Z. Chang3 and Thomas E. Juenger1

1Department of Integrative Biology and Institute for Cell and Molecular Biology, The University of Texas at Austin, Austin, TX 78712, USA; 2Department of Biology, Pennsylvania State

University, University Park, PA 16802, USA; 3Institute of Plant and Microbial Biology, Academia Sinica, Taipei 11529, Taiwan; 4 Present address: Arnold Arboretum and Department of

Organismic and Evolutionary Biology, Harvard University, Cambridge, MA 02138, USA

Author for correspondence:David L. Des MaraisTel: +1 617 384 5495

Email: [email protected]

Received: 27 June 2016Accepted: 3 October 2016

New Phytologist (2017) 214: 132–144doi: 10.1111/nph.14316

Key words: abiotic stress, Brachypodiumdistachyon, genotype-by-environment inter-action (G9 E), local adaptation, phenotypicplasticity, proline, soil drying, temperature.

Summary

� An enduring question in plant physiology and evolution is how single genotypes of plants

optimize performance in diverse, often highly variable, environments.� We grew 35 natural accessions of the grass Brachypodium distachyon in four environments

in the glasshouse, contrasting soil water deficit, elevated temperature and their interaction.

We modeled treatment, genotype and interactive effects on leaf-level and whole-plant traits,

including fecundity. We also assessed the relationship between glasshouse-measured traits

and parameters related to climate at the place of origin.� We found abundant genetic variation in both constitutive and induced traits related to

plant–water relations. Most traits showed strong interaction between temperature and water

availability, and we observed genotype-by-environment interaction for several traits. Notably,

leaf free proline abundance showed a strong effect of genotype9 temperature9water. We

found strong associations between phenology, biomass and water use efficiency (WUE) with

parameters describing climate of origin.� Plants respond to multiple stressors in ways not directly predictable from single stressors,

underscoring the complex and trait-specific mechanisms of environmental response. Climate–trait correlations support a role for WUE and phenology in local adaptation to climate in

B. distachyon.

Introduction

Plants have evolved diverse strategies to optimize fitness understressful and unpredictable environmental conditions. The extentto which plant populations are adapted to local climates, includ-ing their ability to acclimate to within-generation variation in cli-mate and resource limitation via plastic responses, will be a keydeterminant of their ability to persist in the face of climatechange (Jump & Penuelas, 2005). Classic theory on plantresponse to resource limitation suggests that a plant will respondlinearly to the increasing availability of a single limiting resource– for example, water, light, nutrients – and will not respond tothe addition of other resources until deficiency in the limitingresource is met (Bloom et al., 1985). However, manipulative fieldand glasshouse experiments have revealed that interaction (i.e.nonadditivity) between plant responses to limiting resources iscommon (e.g. Peace & Grubb, 1982; Coleman & Bazzaz, 1992;Ackerly & Bazzaz, 1995; Elser et al., 2007; Xu et al., 2013).Although research in the ecological tradition has long consideredthe multifactorial environment (Chapin et al., 1987; Mooneyet al., 1991), a functional understanding of how environmental

factors act interactively – particularly at the molecular level – toaffect plant fitness remains elusive (Suzuki et al., 2014; Anderegget al., 2015).

A conceptually related, but largely independent, body ofresearch has revealed that seemingly disparate abiotic and bioticcues often elicit common molecular signaling pathways(Yoshioka & Shinozaki, 2009). One example of this molecular‘cross-talk’ is the observation that both cold and drying stressinduce responses via abscisic acid (ABA) signaling (Yamaguchi-Shinozaki & Shinozaki, 2006), suggesting synergism betweenplant responses to multiple stressors and echoing ecologicalresearch on plant responses to multiple limiting resources. Quan-titative genetic experiments assessing patterns of genetic diversityin environment-by-environment interactions will be an impor-tant first step in determining the functional relationship betweenmolecular cross-talk and plant performance in a multivariateenvironment. Moreover, because genotype-by-environmentinteraction (G9 E) is the substrate on which artificial and natu-ral selection may operate, species-wide patterns of diversity ofresponse can also reveal plant functional traits which are involvedin local adaptation to climate (Kawecki & Ebert, 2004).

132 New Phytologist (2017) 214: 132–144 � 2016 The Authors

New Phytologist� 2016 New Phytologist Trustwww.newphytologist.com

Research

Temperature and water availability are two critical determi-nants of plant distributions and crop yield (Stebbins, 1952;Boyer, 1982), making a detailed understanding of their indepen-dent and interactive effects on plant cell function and organismalperformance essential. Intuition, and limited evidence frommanipulative experiments, suggests that some responses to ele-vated temperature might be maladaptive in the presence of soildrying stress, and vice versa (reviewed in Suzuki et al., 2014). Forexample, antagonism might arise because sustained transpirationcools leaves under thermal stress, but could be deleterious – orsimply biophysically impossible – when soil moisture is limiting(Gates, 1968). Similarly, free proline has been observed to accu-mulate in Arabidopsis thaliana exposed to dry soil, which is gener-ally regarded as an adaptive cellular response, although not whensoil drying is paired with elevated temperature (Rizhsky et al.,2004). These authors hypothesized that proline might be cyto-toxic at high temperatures, although their analysis of a singlegenotype of A. thaliana precludes generalization. Are similar pat-terns observed in other species or among individuals withinspecies?

Brachypodium distachyon is a small, diploid, annual grassspecies found in diverse habitats in its native and invasive rangesthroughout temperate regions (Schippman, 1991; Catalan et al.,2012). A recent study assessed the transcriptional response of asingle genotype of B. distachyon to several abiotic stressors, andfound that thousands of transcripts responded to elevated tem-perature and to drying stress (Priest et al., 2014). Nine hundredand ninety one genes responded to both elevated temperatureand drying, suggesting possible synergism between these twostress response pathways. However, many genes responded toonly one stress, and several genes were upregulated by one stressand downregulated by the second. Cao et al. (2016) recentlyidentified several key signaling genes in Brachypodium that like-wise showed contrasting transcriptional responses to temperatureand osmotic stress. Additional studies have documented whole-plant responses to soil drying – at a single temperature – indiverse accessions of B. distachyon. These studies found putativelyadaptive responses to drying that included increased water useefficiency (WUE) (Manzaneda et al., 2015) and soluble sugarconcentration (Luo et al., 2011). How might these plantsrespond to elevated temperature experienced concurrently withsoil drying? And are these responses, whether additive or interac-tive, conserved among genotypes of B. distachyon?

Here, we use a quantitative genetic approach to test thehypothesis that the plant response to multiple stressors is an addi-tive function of response to each stress experienced singly againstthe alternative hypothesis that the plant response to these stressorsis interactive. We also test the hypothesis that three of the planttraits measured here (WUE, days to flowering following vernal-ization and total biomass) are associated with local adaptation toclimate. Our results demonstrate, for the first time inB. distachyon, strong interactive effects of elevated temperatureand water limitation on key aspects of plant performance. Fur-ther, we found significant genetic diversity in this interactiveresponse (G9 E9 E) in leaf free proline abundance. Our find-ings, combined with the extensive genetic and genomic resources

available for B. distachyon (Brutnell et al., 2015; Kellogg, 2015),will facilitate detailed molecular functional analysis of plant geno-type interactions with multidimensional environments.

Materials and Methods

Plant material and growth

We selected 35 diploid natural accessions (‘genotypes’) ofBrachypodium distachyon (L.) P. Beauv. (Table 1) that spannedthe southern Eurasian range of the species (Supporting Informa-tion Fig. S1). The accessions descend from field-collected seedinbred for more than five generations in the laboratory (Vogelet al., 2006, 2009; Filiz et al., 2009), and genome sequences areavailable for each of these accessions (Gordon et al., 2014;S. Gordon et al., unpublished). We bulked seed from each acces-sion for one generation in a glasshouse to ensure that the seedsused in our experiments were of common age and quality, and toreduce maternal effects.

We staged our experiment across three harvests to accommo-date destructive sampling needed to score multiple traits oneach genotype in each environment. Each harvest consisted of16 individuals of each accession subdivided into two split-plotexperimental blocks. All plants in a given plot were planted ina single day, treated as a unit during cold stratification, growthand the application of treatment, and harvested on the sameday. The total experiment consisted of 1680 plants (35 geno-types by four environments by four technical replicates by threeharvests).

Plants were grown in 600 ml of Profile porous ceramic rootingmedium (Profile Products, Buffalo Grove, IL, USA) in DeepotD40H pots (650 ml; Stuewe & Sons, Tangent, OR, USA). Thedry weight of each pot was recorded to provide a baseline for thecalculation of soil water content (WC). Pots were saturated witha 1 : 50 dilution of Liquid Grow Plant Food (Dyna-Gro,Richmond, CA, USA) by bottom watering and allowed to dripovernight to field capacity (FC). Two seeds were sown per potand the pots were weighed to determine FC. The WC of each potwas calculated as (FC) – (dry weight); these WCs provided dailywatering targets during the dry-down (see later). Pots were coldstratified at 6°C for 14 d to ensure synchronized germination.

Plants were moved as plots on sequential days from the cold tothe glasshouse at Brackenridge Field Laboratory of the Universityof Texas at Austin, TX, USA. All plants germinated within 3 d,after which each pot was thinned to a single plant. During theinitial growth period of 21 d, all plants were exposed to ambientglasshouse temperatures with daily highs ranging from 23°C to28°C and night-time lows from 14°C to 18°C. Natural autumnsunlight was supplemented by artificial lighting to ensure lightlevels of 400–1000 lmol m�2 photosynthetically active radiation(PAR; mean of 825 lmol m�2) for 10 h d�1 (short-day condi-tions to prevent rapid flowering). Plants were bottom wateredevery second day with fresh water and once per week with fertil-izer-supplemented water, as described earlier.

Following 21 d of initial growth, each plant received one offour treatments for 10 d in a split-plot design: Cool Wet (CW),

� 2016 The Authors

New Phytologist� 2016 New Phytologist TrustNew Phytologist (2017) 214: 132–144

www.newphytologist.com

NewPhytologist Research 133

Cool Dry (CD), Hot Wet (HW) or Hot Dry (HD), imple-mented as two plots of each temperature treatment with soilwater treatment and genotype fully randomized within each plot.On the 22nd day of growth, each block of plants from harvests 1and 2 was transferred to one of four insulated open-top boxesmeasuring 100 cm wide by 240 cm long by 61 cm high. Weplaced Redi-Heat seedling mats (Phytotronics Inc., Earth City,MO, USA) in the bottom of two of these boxes, which resultedin an increase of c. 10°C above ambient glasshouse temperature;these plants constituted the ‘Hot’ treatment in which daytimehighs were c. 35°C. (Because the pots hung from racks, the soilwas not directly warmed by the seedling mats.) The two plots ofplants without seedling mats constituted the ‘Cool’ treatment, inwhich daytime highs were c. 25°C. Within each plot, plants wererandomly assigned to irrigation control (‘Wet’) or restriction(‘Dry’) treatments. Wet plants were watered to FC every secondday with fresh water. Dry plants were hand watered daily bypipette such that the soil water was reduced by no more than 5%each day (i.e. all Dry treatment plants experienced approximately

the same soil moisture despite differences in water use caused byplant size, leaf area or transpiration variation). Harvests 1 and 2began on the 11th day after the beginning of treatment and con-sisted of 8 d of phenotyping; for example, on day 1, Hot plot 1was harvested, on day 2 Cool plot 1 was harvested, and so on.Pot weights were recorded at harvest and showed that the pots ofDry plants averaged 45.7% soil WC (equivalent to 0.29 gH2O g�1 soil, SE = 0.3%), whereas Wet plants averaged 85.2%soil WC (0.54 g H2O g�1 soil, SE = 0.3%), regardless of temper-ature treatment. We have shown previously that 40–45% WC inProfile medium corresponds to soil water potentials of c.�1.2MPa (Des Marais et al., 2012) and results in a significantreduction in leaf relative water content (RWC) in B. distachyon(Des Marais et al., 2016). Blocks were harvested on sequentialdays, such that plants had spent the same amount of time fromgermination through treatment.

Harvest 3 plants did not receive this initial period of stress treat-ment. On the 22nd day of growth, harvest 3 plants were placed ina 6°C walk-in growth chamber with 270 lmol m�2 PAR for12 h d�1 to stimulate flowering (Schwartz et al., 2010). After12 wk of vernalization, plants were returned to the glasshouseunder identical conditions as the initial growth period, allowed toacclimate for 5 d, and assigned to temperature and water treat-ments, as earlier. On the 11th day of treatment, the seedling matswere turned off and water was withheld from all plants until senes-cence; for most plants, this occurred within 14 d.

Scoring leaf-level traits

Leaf traits were scored on harvest 1 plants. Plants from both irri-gation treatments in each plot were harvested together. Two fullyexpanded leaves were excised at the base of the lamina with arazor blade and imaged on a flat-bed scanner (Seiko Epson Inc.,Tokyo, Japan). To determine leaf WC and RWC, these leaveswere weighed on a fine balance, providing a measure of FW, andtransferred to a 15-ml tube half-filled with deionized water withthe cut edge submerged in water. Tubes were left overnight inthe dark at 4°C and then the leaves were removed, spotted dry ona tissue and weighed to determine the turgid weight (TW).Leaves were then dried for 48 h at 80°C and weighed again todetermine DW. Leaf WC was calculated as 1009 ((FW�DW)/DW) and leaf RWC was calculated as 1009 ((FW�DW)/(TW�DW)) (Boyer, 1995). The projected area of the leaveswas determined from the scanned images of fresh leaves usingIMAGEJ. Leaf mass area (LMA) was calculated as DW/(leaf area),expressed as g m�2.

The remaining leaves on each plant were excised with a razorblade, flash-frozen in liquid nitrogen and stored at �80°C. Thistissue was subsequently lyophilized in a freeze-drying system(Labconco, Kansas City, MO, USA) and ground to a fine pow-der. Lyophilized samples (typically 50–100 mg) were extractedwith 1 ml of 80% methanol. For Wet samples, the entire extractwas dried in a Speed-Vac (Thermo Fisher, Waltham, MA, USA)and resuspended in 200 ll of water for proline assays. For Drysamples, a portion (200 ll) of the extract was dried in a Speed-Vac and resuspended in water for proline assay. Proline was

Table 1 Collection information for the natural accessions of Brachypodiumdistachyon used in the experiments

Line Locale Country Latitude Longitude

ABR2 Octon, Herault France 43°39018″N 3°18012″EABR3 Huesca, Aisa Spain 42.679°N 0.621°WABR4 Huesca, Aren Spain 42°160N 0°440EABR5 Huesca, Jaca Spain 42°330N 0°330WABR6 Navarra, Los Arcos Spain 42.567°N 2.183°WABR7 Valladolid, Otero Spain 39°59057″N 4°30046″WABR8 Siena, Italy Italy 43°19007″N 11°19050″EABR9 Ljubljana Slovenia 46°03020″N 14°30030″EAdi-10 Adiyaman Turkey 37°46014.5″N 38°2108.2″EAdi-12 Adiyaman Turkey 37°46014.5″N 38°2108.2″EAdi-2 Adiyaman Turkey 37°46014.5″N 38°2108.2″EBd1-1 Soma, Manisa Turkey 39°110N 27°370EBd18-1 Kaman Turkey 39°210N 33°430EBd2-3 Uncertain IraqBd21 Near Irbil Iraq 36°11028″N 44°0033″EBd21-3 Near Irbil Iraq 36°11028″N 44°0033″EBd3-1 Uncertain IraqBd30-1 Dilar, Spain Spain 37°40N 03°360WBdTR10c Turkey 37°46041.64″N 31°5305.68″EBdTR11g Turkey 41°25017.86″N 27°28036.81″EBdTR11i Turkey 39°44017.39″N 28°2024.71″EBdTR12c Turkey 39°44053.45″N 34°3901.15″EBdTR13A Turkey 39°45023.35″N 32°25056.46″EBdTR1i Turkey 38°5035.03″N 28°34059.02″EBdTR2b Turkey 40°4055.55″N 31°19052.01″EBdTR2g Turkey 40°23037.13″N 32°5907.32″EBdTR3c Turkey 36°46058.92″N 32°57046.71″EBdTR5i Turkey 40°23037.13″N 32°5907.32″EBdTR9k Turkey 39°45010.62″N 30°47019.07″EBis-1 Bismil Turkey 37°52035.6″N 41°0054.3″EGaz-8 Gaziantep Turkey 37°7039.8″N 37°23026.9″EKah-1 Kahta Turkey 37°4402.3″N 38°3200.2″EKah-5 Kahta Turkey 37°4402.3″N 38°3200.2″EKoz-1 Kozluk Turkey 38°908.2.6″N 41°36034.8″EKoz-3 Kozluk Turkey 38°908.2.6″N 41°36034.8″E

Data compiled from Jenkins et al. (2003), Vogel et al. (2009) andD. Garvin (pers. comm.).

New Phytologist (2017) 214: 132–144 � 2016 The Authors


Research

NewPhytologist134

quantified by the acid-ninhydrin assay in a 96-well plate format(Bates et al., 1973; Verslues, 2010). The proline content of leavesis presented here as lmol g�1 dry leaf weight.

Scoring whole-plant traits

Whole-plant traits were scored from harvest 2 plants. In this har-vest, 10 ABR8 plants, four Bd3-1 plants, two Bd21-3 plants, oneBd2-3 plant and one Bd18-1 plant flowered during the treatmentperiod. Flowering plants occurred in all four treatments (5 HW,2 HD, 3 CD, 8 CW). Complete above-ground (AbvGrd) tissuewas excised at soil level and dried at 80°C for 48 h. Roots (be-low-ground, BlwGrd) were washed of potting medium and thendried. AbvGrd and BlwGrd tissues were weighed separately.AbvGrd tissues were ground to a fine powder and analyzed ford13C (as a proxy for lifetime integrated WUE), carbon (C) con-tent and nitrogen (N) content using mass spectrometry at theUniversity of California at Davis Stable Isotope Facility (http://stableisotopefacility.ucdavis.edu/13cand15n.html).

Date of first flower and fitness measures

Harvest 3 plants were used to measure the date of first flower andlifetime fecundity. Plants were scored daily for flowering, reportedas the number of days between removal from the cold room and thefirst visual sign of an awn emerging from a tiller. Four Bd3-1 indi-viduals, one ABR8 individual and one ABR7 individual floweredbefore the beginning of the stress period. Seeds were harvested onceplants had senesced and dried in their pots. In our experience, viableseeds can be readily identified by their hardness scored by pressing asingle seed between the thumb and forefinger; we counted all suchseeds from each plant as a measure of lifetime fitness.

Quantitative genetic analyses

We fitted linear mixed models to partition variance in each traitby the experimental factors genotype, water treatment, tempera-ture treatment, all two- and three-way interactions, and glasshousebench (block). Unless otherwise noted, all statistical analyses wereperformed using PROCMIXED in SAS v.9.3 (SAS Institute, Cary,NC, USA). To account for our split-plot design (one Hot plotand one Cool plot were sited on each of two glasshouse benches),we also fitted a bench and bench-by-temperature treatment inter-action (Littel et al., 2006). For the traits scored by mass spectrom-etry, we fitted a term to account for two sessions on thespectrometer. Water treatment (‘W’), temperature treatment (‘T’)and their interaction (‘T9W’) were modeled as fixed effects, andthe significance for these factors was determined using F-ratiotests. Variance components for genotype (‘G’), G9W, G9 T,G9 T9W, bench, bench by temperature interaction and, whererelevant, mass spectrometer batch were modeled as random effectsusing restricted maximum likelihood. Significance for randomeffects was determined by removing each factor from the fullmodel, refitting the model and performing likelihood ratio tests.In this context, a significant genotypic effect indicates the presenceof genetic variation in that trait, a significant effect of temperature

or water treatment indicates trait plasticity, a significant interac-tion between genotype and either treatment factor indicatesgenetic diversity in plastic response (G9 E), and an interactionbetween temperature and water treatment indicates that theresponse to the treatments was nonadditive. Broad-sense heritabil-ity for each trait was estimated separately in each environment asthe proportion of total phenotypic variance (Vp) in a traitexplained by genotype (Vg) in the mixed model. We used z-teststo determine whether Vg was greater than zero.

Past research in herbaceous plants has identified a significantnegative correlation between days to flowering and WUE(McKay et al., 2003; Sherrard & Maherali, 2006; Kenney et al.,2014; Manzaneda et al., 2015). To test whether these traits werecorrelated in our dataset, we used the approach of Fry (2004) toestimate genetic covariances between them and to test whethertheir covariance was greater than zero.

Climate associations

We tested for a significant correlation between the cross-environment means and plasticities of days to flowering, totalplant biomass and d13C with parameters describing climate atthe site of origin for each accession. We estimated genotypicmeans for each accession in each environment as LSMeansusing PROCMIXED with Satterthwaite degrees of freedom andincluding experimental plot as a random effect. We usedthree cross-treatment contrasts to encompass the phenotypicdiversity of response observed in our experiments: CW vsCD, HD vs CW and HW vs CW. The results of all threecontrasts were similar, and so we present here only the resultsof the HW vs CW contrast. Interannual variability in precipi-tation for the years 1948–2009 was calculated using NCAR/NCEP Reanalysis data (Kalnay et al., 1996). Long-term aver-age (1950–2000) climate data were taken from Worldclim(Hijmans et al., 2005).

For annuals, such as B. distachyon, that occupy climates with highseasonality, climate conditions during the growing periods may bemost relevant for explaining trait divergence among populations.We used Worldclim data to predict the growing season for eachaccession. Here, we defined the growing season as the months inwhich the mean temperature was at least 4°C and when sufficientmoisture was available following the United Nations Food andAgriculture Organization (FAO) model for dryland crops (http://www.fao.org/geonetwork/srv/en/metadata.show?id=73). The FAOmodel requires precipitation in a given month to be at least0.49 potential evapotranspiration (PET). Furthermore, precipita-tion greater than monthly PET can be stored in soil up to a maxi-mum capacity (smc) for each accession’s home environment(http://daac.ornl.gov/SOILS/guides/DunneSoil.html). Stored soilmoisture (ssm) was transferred to the following month as:

ssmtþ1 ¼ maxð0;minðsmc; ssmt þ rainfallt � PETtÞÞ

As a result, months in which rainfall + ssm is at least0.49monthly PET are also considered growing season months,that is, when:





http://stableisotopefacility.ucdavis.edu/13cand15n.html

http://stableisotopefacility.ucdavis.edu/13cand15n.html

http://www.fao.org/geonetwork/srv/en/metadata.show?id=73

http://www.fao.org/geonetwork/srv/en/metadata.show?id=73

http://daac.ornl.gov/SOILS/guides/DunneSoil.html

ðssmt þ rainfalltÞ� 0:4PETt

as months in which the mean temperature is > 4°C and for whichPET is at least 0.4.

We summarized multivariate climate using principal compo-nent (PC) analysis and found that the first three PCs explained86% of climate variation; therefore, we focused on the identifica-tion of trait associations with these first three climate PCs. Cli-mate PC loadings are listed in Table S1.

For bivariate comparisons between climate of origin and ourcommon garden phenotypes, we calculated nonparametric Spear-man’s rank correlations to accommodate nonlinear andheteroscedastic data. For comparisons between climate and plas-ticity phenotypes, we used linear models to incorporate bothmean trait and plasticity associations with climate. That is, we fit-ted models of climate of origin where covariates were mean traitand plasticity. Here, plasticity was calculated from the genotypicLSMean values for the three contrasts as (CD–CW), (HD–CW)and (HW–CW).

Controlling for multiple testing

We adjusted all P values obtained from statistical tests in theANOVA, heritability estimates and climate correlations using afalse discovery rate with the p.adjust call in R.

Results

Response of B. distachyon to water deprivation, tempera-ture and their interaction

Mixed-model ANOVAs revealed that all traits, except days tofirst flower, responded to our water treatment, temperaturetreatment or the interaction between the two (Table 2).Notably, most traits displayed an interactive effect of the treat-ments. Temperature and water level interacted to affect plantlifetime fitness, expressed either as total seed count on the abso-lute scale (F1,429 = 6.97, P = 0.020; Fig. 1a) or log-transformedseed count (F1,429 = 4.00, P = 0.046). On either scale, plants inthe HW treatment tended to be most fit, whereas plants in theHD treatment were the least fit; there was, however, consider-able variance among accessions in these effects (see later). Thechange in trait values under drying as a function of temperaturewas highly variable among traits. Proline showed the moststriking pattern (F1,70.3 = 37.75 and P < 0.0001; Fig. 1b), withno difference between the HW and CW treatments, but signifi-cantly higher proline in the HD treatment compared with CD(result of slice test, P = 0.033). LMA (F1,476 = 8.43 andP = 0.010 for T9W; Fig. 1c) and root mass ratio (RMR;F1,34.7 = 6.69 and P = 0.031 for T9W; Fig. 1d) also showedstrong interaction between temperature and water treatments.Generally, there were stronger differences in these traitsbetween temperatures in the Dry environment, although thespecific contrasts of effects were nonsignificant in our experi-ment: plants in the Hot environments tended to have higher

LMA (denser leaves; results of slice test, P = 0.34), but lowerRMR (as a result of generally larger AbvGrd tissue; results ofslice test, P = 0.17), relative to plants in the Cool environments.By contrast, WUE (F1,68.4 = 5.75 and P = 0.041 for T9W;Fig. 1e) and leaf N (F1,468 = 11.38 and P = 0.002 for T9W;Fig. 1f) tended to show a stronger effect of temperature in theWet environments, although, again, the specific contrasts werenot significant at a = 0.05 (slice test for WUE, P = 0.25; slicetest for leaf N, P = 0.24). It should be noted that our split-plotexperimental design provides greater power to detect watertreatment effects than temperature treatment effects because thewater treatment was completely randomized within plots whichvaried the temperature levels.

Two of the studied traits did not display significant interactiveeffects of the two treatments, but did show a significant maineffect of one or both of the treatments. Plant AbvGrd tissue inthe Dry treatments was 23.6% lower than that in the Wet treat-ments (F1,33.7 = 81.8, P < 0.0001). AbvGrd tissue showed a sig-nificant response to the temperature treatment, with plantsbeing, on average, 6.3% larger at the higher temperature(F1,34.5 = 8.28, P = 0.016). The interaction between temperatureand water was nonsignificant for AbvGrd tissue (F1,34.6 = 0.05,P = 0.904). Leaf C content was slightly higher in the Dry envi-ronment (400.2 mg g�1 dry leaf in Dry treatment vs390.8 mg g�1 in Wet treatment; F1,472 = 187.5, P < 0.0001), andthe interaction between water and temperature was nonsignifi-cant (F1,471 = 3.19, P = 0.137).

Diversity of response

We detected genetic variation in response to soil drying (G9 E;Table S1). Most notably, there was significant G9 E for fitnessin our water treatment, observed whether fitness was expressed astotal seed count on the raw scale (likelihood ratio test for geno-type9 water term G = 14.8, P < 0.0001; Fig. 2) or log-transformed (G = 7.7, P = 0.006). Considerable difference ingenetic variance for fitness among environments was apparent; insome cases, the rank of genotypic fitness changed between envi-ronments (Fig. 2). We also detected a significant G9W term forleaf N (G = 15.3, P < 0.0001; Fig. S2k) and the ratio of leaf C toN (G = 9.9, P = 0.005; Fig. S2l). Leaf N was higher in the Wettreatment in all genotypes, suggesting that G9W for the C : Nratio is driven by genetic variation in the magnitude of theincrease in leaf N. No traits showed significant G9 E for temper-ature at a = 0.05.

Proline was the only trait with significant G9W9 T, that is,genetic variation in the interactive effects of temperature andwater availability (G9 E9 E; likelihood ratio test G = 20.8,P < 0.0001). Genotypes differed in both the magnitude and, insome cases, the direction of their proline response to water andtemperature (Fig. S2d). Generally, genotypes accumulated moreproline in response to the drying treatment, but varied consider-ably in the extent to which temperature enhanced this response:Adi-10 produced 3.29 more proline and BdTR2g produced2.89 more proline when drying was combined with high tem-perature, as compared with drying in the cool environment. By



Research

NewPhytologist136

contrast, BdTR9k showed no difference in proline contentamong the four environments (Tukey–Kramer honestly signifi-cant difference (HSD) test at a = 0.05). Notably, no genotypeproduced less proline in the HD environment when comparedwith the CD environment.

Several interesting patterns were observed with respect to heri-tability estimates across the four treatments, probably reflectingdifferences in genetic variance among environments (Table 3).Proline had very low heritability in the CW (0.04) and HW(0.05) environments, but moderate heritability in the CD (0.33)and HD (0.27) environments. The opposite pattern was observedfor LMA: heritability in the CW (0.23) and HW (0.18) environ-ments was much higher than that in the CD (0.00) and HD(0.09) environments.

Contrary to previous research (McKay et al., 2003; Kenneyet al., 2014; Manzaneda et al., 2015), we found no significantgenetic correlation between days to flower and WUE in any ofthe four sampled environments (data not shown).

Correlations between measured traits and climate of origin

We tested for the role of the traits measured here in local adapta-tion to climate by regressing parameters describing each acces-sion’s climate of origin (Garvin, 2016) on the genotypic meansof WUE, total plant biomass and days to flowering. We focusedon these three plant traits because of their high heritabilitiesobserved across our experimental treatments and because of pre-vious work suggesting a key role for the traits in determiningplant fitness in natural and agricultural settings (McKay et al.,2003; Condon et al., 2006; Kooyers et al., 2014). Here, we pre-sent the results of associations with the cross-environment traitmeans and plasticities contrasting our CW and HD treatments(i.e. the most divergent pair of conditions); the results of the con-trasts between (CW and HW) and (CW and CD) were similar(data not shown). We also observed strong correlations betweenclimate variables and genotypic means of traits averaged across allfour treatments (Fig. S3).

We observed a significant negative correlation between days toflowering and climate PC1 (F = 11.15, P = 0.017; Fig. 3a). Cli-mate PC1 loads positively with summer temperatures and nega-tively with several climate parameters describing early summerprecipitation (Table S1). These associations indicate thatearly-flowering accessions of B. distachyon come from habitatswith relatively warm, dry summers. At a = 0.1, we also observeda negative correlation between d13C and climate PC3 (results ofANOVA: F = 6.53, P = 0.067; Fig. 3b), which was positivelyassociated with isothermality (a measure of low annual variabilityin temperature) and winter temperatures, and negatively associ-ated with early spring precipitation. (It should be noted thatmore negative values of d13C indicate lower WUE.) The associa-tion between WUE and climate PC3 indicates that low WUEplants are found in climates with relatively mild winters. Thisnegative regression was partly influenced by the very high WUEand high latitude accession ABR9. However, the rank correlationbetween d13C and climate PC3 was also significant (rho =�0.38,P = 0.0287).T

able2Resultsofmixed

-modelan

alyses

forea

chtrait

Trait

Fixedeffects

W9T

Ran

dom

effects

G9T

G9T9W

Ben

chBen

ch9T

Mass

Spec

Residual

Water

(W)

Tem

p.(T)

Gen

otype(G

)G9W

Leaf

RWC(%

)F 1

,33.6=177.58**

F 1,1=2.09

F 1,477=38.45**

0.20(0.66)

2.44(0.88)**

––

–0.96(1.44)**

na

9.08(0.59)**

Leaf

WC(%

DW)

F 1,33.8=187.4**

F 1,1=5.38

F 1,477=1.93

174.23(56.27)**

12.36(25.70)

––

–79.42(119.64)**

na

711.01(46.09)**

LMA(g

m�2)

F 1,34.2=117.49**

F 1,1=1.29

F 1,476=8.43*

1.35(0.50)**

0.05(0.31)

––

–2.13(3.11)**

na

9.40(0.61)**

Proline(µgg�1

dry

weight)

F 1,33.2=118.78**

F 1,2.14=5.19

F 1,70.3=37.75**

6.44(13.30)

35.11(19.37)~

–22.70(13.75)**

–17.54(13.75)**

na

209.09(14.97)**

AbvG

rdtissue(g)

F 1,33.7=81.8**

F 1,34.5=8.28~

F 1,34.6=0.05

764.44(219.39)**

106.91(67.51)

9.45(47.70)

53.89(66.84)

––

na

855.71(59.53)**

BlwGrd

tissue(g)

F 1,34.2=3.69

F 1,1.22=0.02

F 1,34.5=1.36

116.98(37.62)**

14.32(14.18)

17.76(15.00)

8.24(16.79)

5.19(21.01)

10.64(17.41)~

na

234.03(16.32)**

RMR(%

)F 1

,34.7=300.37**

F 1,1.13=2.38

F 1,34.7=6.69~

6.69(2.13)**

0.52(0.88)

0.70(0.93)

0.17(1.16)

1.26(2.07)

0.579(0.999)

na

17.61(1.23)**

d13C

F 1,35.3=239.04**

F 1,2.17=0.23

F 1,68.4=5.75~

0.11(0.03)**

0.01(0.01)

–0.01(0.01)

–0.01(0.01)~

0.02*

0.23(0.02)**

Leaf

C

(mgg�1leaf)

F 1,472=187.5**

F 1,3.22=1.25

F 1,471=3.19

12.60(4.84)*

–4.48(3.08)

––

0.308(0.835)

2.29~

61.87(4.05)**

Leaf

N

(mgg�1leaf)

F 1,34.5=156.73**

F 1,2.01=1.18

F 1,468=11.38*

2.38(0.99)~

1.87(0.70)**

––

–2.31(2.37)**

0.43*

7.62(0.50)**

C:N

ratio

F 1,34.5=201.37**

F 1,2.01=1.06

F 1,68.8=8.17~

0.24(0.10)~

0.17(0.07)*

–0.01(0.04)

–0.32(0.33)**

0.04~

0.84(0.06)**

Flowering(d)

F 1,437=0.09

F 1,2.14=0.04

F 1,437=0.65

7.69(1.92)**

0.01(0.09)

0.11(0.11)

––

0.16(0.18)

na

2.57(0.18)**

Fitness(absolute)

F 1,35=5.38

F 1,2.01=0.02

F 1,429=6.97~

1131.56(314.75)**

176.79(75.27)**

––

–653.53(660.11)**

na

959.84(65.75)**

Dashes

indicatethat

nova

rian

cewas

attributedto

that

random

effect.Values

inparen

theses

indicatestan

darderrorofthemea

n.Bold

text

indicates

P<0.05;~,

P<0.05.*,

P<0.01.**

,P<0.001.

AbvG

rd,ab

ove

-ground;B

lwGrd,below-ground;RW

C,relativewater

content;W

C,water

content;LM

A,leaf

massper

area

;RMR,rootmassratio;C,carbon;N,nitrogen

;na,

notap

plicab

le.





Climate PC2 showed two interesting trait associations anddescribed an axis positively loaded with autumn precipitation, aswell as the length and temperature of predicted growing seasons,and negatively loaded with altitude and PAR. Climate PC2 wasnegatively correlated with total plant biomass (F = 13.63,P = 0.067; Fig. 3c), indicating that larger plants are derived frompopulations at higher elevations with short growing seasons andhigh PAR. Climate PC2 was negatively correlated with plasticityin d13C after accounting for climate mean d13C effects (F = 4.87,P = 0.095; Fig. 3d), indicating that larger accessions found athigher elevation where PAR is relatively high are also more plasticin their WUE.

Discussion

Three major conclusions arise from our analysis of natural varia-tion in B. distachyon: first, there is considerable genetic diversityin the magnitude and, in some cases, the direction of traitresponse to changes in abiotic conditions among natural

accessions; second, there is a strong interaction between the effectof temperature and water availability on traits; third, there is evi-dence that natural variation in flowering time, plant biomass andWUE are associated with local adaptation to climate. We discussthese conclusions in the context of the evolution and study ofenvironmental responses in plants, and as they relate toB. distachyon as a model for understanding plant–environmentinteractions.

Temperature-by-water interaction

The mechanisms of the plant response to environmental cues act-ing singly or in combination are of tremendous interest to ecolo-gists, evolutionary biologists and plant breeders. Thepreponderance of empirical studies have imposed independentstresses, whilst acknowledging that the field environment presentsplants with a complex mix of biotic and abiotic factors (Pas-sioura, 2007; Mittler & Blumwald, 2010). Several studies haveexplored the interaction between water limitation and moderately

Leaf

mas

s ar

ea (g

m–2

)

Wet Dry22.5

32.5

27.5

P = 0.039

Root

mas

s ra

tio

Wet Dry25

35

30

P = 0.014

P = 0.019

Leaf

nitr

ogen

(mg

g–1dr

y)

Wet Dry30

40

35

P = 0.0024

Cool treatment Hot treatment

Leaf

pro

line

(μm

ol g

–1dr

y)

Wet Dry0

40

20

P < 0.0001

Seed

cou

nt

35

95

65

Wet Dry

P = 0.019

–32

–31

δ 13

c

Wet Dry

(a) (b)

(c) (d)

(e) (f)

Fig. 1 Trait responses of Brachypodiumdistachyon to temperature and soil watercontent. (a) Total seed count. (b) Leaf freeproline abundance. (c) Leaf mass area. (d)Root mass ratio (below-ground dry biomass/total plant dry biomass). (e) Water useefficiency, estimated as d13C. (f) Leafnitrogen content. n = 120 for each point. Thepresented P values are tests for temperatureby water interaction. Trait LSMeans in eachenvironment and the statistical significanceof temperature by water interactions weredetermined using linear mixed models, asdescribed in the main text. Vertical barsindicate � 1 SEM.



Research

NewPhytologist138

elevated temperature, with results spanning beneficial interactiveeffects on photosystem II (PSII) function in tomato (Havaux,1992), to strictly additive responses for traits related to growthand biomass in A. thaliana (Vile et al., 2012) to strongly interac-tive and antagonistic effects on proline accumulation inA. thaliana (Rizhsky et al., 2004) and on photoinhibition indiverse species (reviewed by Chaves, 1992).

Here, we investigated the interaction between soil WC andtemperature variability in a factorial glasshouse experiment usinga model grass species. We found that most traits respond to thecombined treatment in a manner not directly predicted fromresponses to individual stressors, varying in magnitude and, insome cases, direction. Contrary to our expectations, most geno-types realized highest seed yield and biomass in our HW (35°Cand ample soil moisture) treatment, suggesting that our high-temperature treatment was not stressful when water was plentifulfor most genotypes. A practical interpretation of this result is thatfuture studies of B. distachyon – particularly the reference geno-type Bd21 (see bold line in Fig. 2) – may benefit from growth athigher temperatures than is currently common (Hennig, 2010).Plants had lowest fitness in our HD treatment, although the largedifferences in fitness among accessions (both G and G9 E)obscured clear patterns between treatments (i.e. fitness in HDwas not significantly lower than fitness in CD).

Traits showed differing patterns of temperature-by-waterinteraction (Fig. 1). In some cases, the interaction suggested thathigh temperature exacerbated the effects of drying. This patternwas seen for proline (Fig. 1b), as plants generally accumulatedmore proline when confronted with soil drying and elevated

temperature as compared with soil drying alone. This leaf-levelpattern indicates that the HD treatment is analogous to astronger drying treatment and may have increased the experi-enced cellular osmotic or redox stress as compared with CD (itshould be noted that our experimental procedure ensured that allDry pots had the same soil WC, regardless of temperature andgenotypic factors). Of note is the high proline abundance in theHD environment, contrasting with the findings of Rizhsky et al.(2004) that proline or one of its biosynthetic intermediates istoxic under heat stress in A. thaliana. Several other recent studieshave likewise failed to replicate this earlier result in tobacco(Demirevska et al., 2010; Cvikrova et al., 2013) and two grassspecies (Gargallo-Garriga et al., 2015).

In contrast with proline, RMR and WUE showed steeperresponses to drying under cool temperatures than at high temper-ature (Fig. 1d,e). For some accessions, the moderate change inRMR was caused by increased AbvGrd tissue; this pattern ofincreased above-ground biomass in response to elevated tempera-ture was observed previously in natural accessions of A. thaliana(Vile et al., 2012). Increased RMR under soil drying is frequentlyobserved and is interpreted as a dehydration avoidance strategy toincrease the extraction of soil water (Kramer & Boyer, 1995).

Increased WUE indicates greater C fixation per unit waterconsumed and is observed during soil drying in many species(Van den Boogard et al., 1997; Heschel et al., 2002; Donovanet al., 2007; Des Marais et al., 2012). We observed a moderateeffect of temperature on WUE response to drying; the tempera-ture9 water term was significant, but specific contrasts betweentemperatures in each moisture treatment were not. Temperature

Environment

Cool dry Hot wetHot dryCool wet

Seed

cou

nt

0

100

150

50

200 BdTR13aBdTR2gBdTR1iBd18-1Bis-1BdTR2bGaz-8BdTR10cAdi-12Kah-1Koz-1BdTR5iBdTR11gBdTR3cABR7BdTR11iAdi-10Koz-3Bd2-3Kah-5Bd3-1BdTR12cBdTR9kABR8ABR2ABR6Bd21Bd30-1Adi-2Bd21-3ABR5ABR4ABR3ABR9Bd1-1

Fig. 2 Genotypic means (LSMeans) of totalseed count for each Brachypodium

distachyon accession in each experimentaltreatment. Accession legend is presented inthe same order as seed count in the CoolDry environment. For reference, the Bd21accession is presented with a bold blackline. Variance estimates for these LSMeansare presented in Supporting InformationFig. S2(m).





interacts with soil drying to affect stomatal conductance and pho-tosynthesis in complex ways. The relationship between d13C andWUE can be temperature dependent as a result of the effects oftemperature on the carboxylation and diffusion of CO2 (Brooks& Farquhar, 1985). Stomatal conductance remained high inlupins grown at elevated temperature, regardless of soil WC, pos-sibly as a result of the benefits of transpirative cooling at elevatedtemperatures (Correia et al., 1999). Havaux (1992) likewisereported a possible beneficial interaction between elevated tem-perature and dehydration in tomatoes: irreversible inhibition ofPSII was observed in plants exposed to elevated temperature inthe absence of water stress, but PSII recovered in plants experi-encing both elevated temperature and dehydration. Havaux spec-ulated that drying stress leads plants to protect PSII fromadditional stressors. In other species, however, elevated tempera-ture exacerbates the effect of soil drying as a result of enhancedphotoinhibition (Chaves, 1992). The rank-changing tempera-ture-by-water interaction for WUE, observed here, suggests thatthere may be some constraint in acclimation to both high tem-perature and soil drying in B. distachyon, albeit a fairly small con-straint under the conditions tested here. Additional research isnecessary to dissect the complex effects of this temperature andwater interaction on stomata and photosynthesis in B. distachyon;Bd2-3, which was one of the few accessions showing a WUEtemperature response, and ABR2, which showed nearly no suchresponse, might make an interesting contrast in this regard(Fig. S2j).

Brachypodium distachyon and the abiotic environment

The native range of B. distachyon encompasses considerableclimatic diversity, extending from the Iberian peninsula eastto Iraq, in both southern Europe and north Africa (Fig. S1).Recent environmental niche modeling has shown that precipi-tation most strongly defines the native geographical range ofB. distachyon in relation to its annual congeners Brachypodiumstacei and Brachypodium hybridum (L�opez-Alvarez et al., 2015).(It should be noted that all accessions in our study arediploid B. distachyon sensu Catalan et al. (2012)). Manzanedaet al. (2015) have recently assessed the genetic diversity ofB. distachyon under well-watered and uncontrolled soil dryingconditions. They used a panel of Iberian natural accessions,and found that most accessions reduce stomatal conductanceand C gain when exposed to drought stress, resulting in anincrease in lifetime-integrated WUE, as observed in the pre-sent study.

The results of our trait–climate association analysis suggestthat natural populations of B. distachyon are locally adapted to cli-mate gradients through mechanisms that include the timing offlowering and WUE. Total plant biomass likewise shows strongclimate associations. Days to flowering following vernalization isstrongly negatively associated with summer temperature and pos-itively associated with summer rainfall (Fig. 3a), suggesting thatrapid-flowering accessions may have adopted this phenology toavoid poor summer growing conditions at their place of origin.Accessions located at the two extremes of climate PC1 illustrateT

able3Heritab

ilities

ineach

environmen

t

Trait

CW

Mean(SE)

CW

Vg(SE)

CW

Vg/V

pCDMean(SE)

CDVg(SE)

CDVg/V

pHW

Mean(SE)

HW

Vg(SE)

HW

Vg/V

pHDMean(SE)

HDVg(SE)

HDVg/V

p

Leaf

RWC(%

)99.13(0.94)

0.17(0.12)

0.125

94.68(0.94)

4.79(1.72)

0.332

99.27(0.94)

0(0)

091.63(0.94)

6.93(2.80)

0.172

Leaf

WC(%

DW)

347.22(7.63)

178.14(86.14)

0.215

317.17(7.63)

202.46(88.29)

0.256

329.03(7.63)

148.03(109.38)

0.128

292.67(7.63)

147.76(67.53)

0.109

LMA(g

m�2)

27.80(1.14)

1.46(0.66)

0.234

27.93(1.14)

0.042(0.89)

0.003

26.73(1.14)

2.05(1.13)

0.179

30.38(1.14)

2.12(1.17)

0.087

Proline(µgg�1)

8.92(3.49)

0.64(1.30)

0.038

22.18(3.49)

59.05(28.16)

0.327

9.85(3.51)

2.60(4.60)

0.052

41.54(3.49)

201.7

(82.25)

0.272

AbvG

rdtissue(g)

138.76(5.73)

1470.37(442.26)

0.523

104.50(5.74)

538.32(153.37)

0.582

146.46(5.73)

1289.55(383.35)

0.541

113.38(5.74)

421.70(143.35)

0.411

BlwGrd

tissue(g)

50.97(3.82)

226.90(72.76)

0.443

55.78(3.82)

145.36(46.91)

0.443

53.09(3.82)

149.19(52.17)

0.344

54.65(3.82)

100.28(39.91)

0.286

RMR(%

)26.88(1.19)

9.66(3.47)

0.352

34.80(1.19)

5.78(2.91)

0.195

26.48(1.19)

5.98(2.26)

0.244

32.52(1.19)

10.56(3.64)

0.362

d�3

1.72(0.13)

0.14(0.05)

0.359

�30.84(0.13)

0.12(0.04)

0.374

�31.57(0.13)

0.13(0.05)

0.279

�30.89(0.13)

0.11(0.04)

0.355

Leaf

C(m

gg�1leaf)

389.62(1.50)

10.83(7.93)

0.127

400.21(1.50)

19.63(9.45)

0.232

392.00(1.50)

15.18(8.12)

0.184

400.16(1.52)

14.73(7.13)

0.218

Leaf

N(m

gg�1leaf)

39.81(1.25)

6.07(2.18)

0.351

32.92(1.25)

1.35(0.64)

0.193

36.33(1.25)

7.91(2.67)

0.328

32.81(1.25)

2.83(0.97)

0.252

C:N

ratio

10.15(0.44)

0.42(0.16)

0.307

12.23(0.44)

0.25(0.11)

0.226

10.98(0.44)

0.69(0.25)

0.279

12.59(0.45)

0.47(0.16)

0.256

Flowering(days)

12.73(0.57)

9.29(2.45)

0.750

12.58(0.57)

7.10(1.92)

0.706

12.71(0.57)

6.48(1.73)

0.707

12.78(0.57)

7.96(2.13)

0.784

Absolute

fitness

62.91(19.28)

1552.92(455.15)

0.439

60.46(19.30)

1097.24(300.53)

0.607

73.44(19.28)

1454.66(476.65)

0.319

56.24(19.24)

845.22(241.99)

0.517

Vg,gen

otypicva

rian

cecomponen

t;va

lues

inbold

differsignificantlyfrom

zero

ata=0.05aftercorrectingformultipletests.Vp,totalp

hen

otypicva

rian

ce.Values

inparen

theses

indicateSE

M.CW,

Cold

Wet;C

D,Cold

Dry;HW

,HotW

et;H

D,HotDry;AbvG

rd,ab

ove-ground;BlwGrd,below-ground;RW

C,relativewater

content;W

C,water

content;LM

A,leaf

massper

area

;RMR,rootmass

ratio;C,carbon;N,nitrogen

.



Research

NewPhytologist140

this pattern. The rapid-flowering accession Bd21 is native tonorthern Iraq, where hot, dry summers are observed, whereas theABR3 accession from the Pyrenees is slow to flower and experi-ences moderate summer temperatures and year-round

precipitation (Fig. 4). We also see a trend for larger accessionsfrom our common garden study to originate from higher eleva-tions where growing seasons are short and PAR is high (Fig. 3c).Interestingly, the diploid annual species B. stacei is smaller and

Climate PC1

1014

18

−10 0 5 10 15−5

Tim

e to

flow

erin

g fo

llow

ing

vern

aliz

atio

n (d

)

−31.

5−3

0.5

Climate PC3−5 0 5 10

δ 13

C

15−5 0 5 10

100

150

200

Climate PC2

Tota

l bio

mas

s

−5 0 5 10 15

00.

51.

0

Climate PC2

δ 13

C pl

astic

ity(H

otD

ry -

Cool

Wet

)

(a) (b)

(c) (d)

Fig. 3 Correlations between trait genotypicmeans (LSMeans) and climate at the locationof origin for Brachypodium distachyonaccessions. LSMeans are presented as themean value between the Cool Wet treatmentand Hot Dry treatment. (a) Significantnegative correlation between days toflowering following vernalization and climatePC1 (P = 0.017). (b) Significant negativecorrelation between d13C (a proxy for wateruse efficiency) and climate PC3 (P = 0.067).(c) Significant negative correlation betweentotal plant biomass and climate PC2(P = 0.067). (d) Significant negativecorrelation between plasticity in d13C andclimate PC2 (P = 0.095). Plasticity in d13C isreported as (Hot Dry –Cool Wet).

2 4 6 8 10 12

0

20

40

60

80

100

120

Mon

thly

mea

n te

mp.

(°C

)

Month

0

50

100

150

200

250

Mon

thly

mea

n pr

ec. (

mm

)

ABR3

�

2 4 6 8 10 12

0

20

40

60

80

100

120

Mon

thly

mea

n te

mp.

(°C

)

Month

0

50

100

150

200

250

Mon

thly

mea

n pr

ec. (

mm

)

Bd21

�

Fig. 4 Diagrams showing climate parameters for the geographical origin of the Bd21 and ABR3 accessions. Black dots indicate months of the predictedgrowing season for each accession.





found at lower elevations with more xeric conditions thanB. distachyon (L�opez-Alvarez et al., 2015).

WUE, approximated here by d13C, shows several interestingpatterns in our analyses. First, contrary to several studies in theherbaceous annual Arabidopsis thaliana (McKay et al., 2003;Kenney et al., 2014), and an earlier study in B. distachyon (Man-zaneda et al., 2015), we did not observe a genetic correlationbetween WUE and days to flowering. These different findingsmay be because we subjected our plants to a longer vernalizationtreatment than did Manzaneda et al., and thereby probablyreduced the expressed genotypic differences in flowering time,although other experimental differences between the studies (daylength, temperature, specific accessions included) may also haveled to different results.

A second interesting pattern with WUE is the observation thatit is correlated with climate: low WUE accessions are found inlocations with mild winters and dry springs (Fig. 3b). These envi-ronments may select for rapid growth as a result of unpredictablesoil moisture availability. We also found that plasticity in WUEis significantly associated with a climate PC describing variationin altitude, PAR and growing season length (Fig. 3d). Plasticityin WUE was recently highlighted as a possible key factor in pop-ulation persistence in the presence of climate change (Nicotraet al., 2010).

A considerable challenge with the interpretation of thesepatterns is that there are few published data on the naturalhistory of B. distachyon in its native habitat (Des Marais &Juenger, 2016). Although we can assess broad associative pat-terns of adaptation to climate, the value of a detailed under-standing of species phenology (e.g. timing of germination,growth and reproduction) is evident. Brachypodium distachyonfrequently occurs in highly seasonal environments (L�opez-Alvarez et al., 2015), where the selective effect of potentialabiotic stressors at a given time of year is conditional onplant developmental status. A better understanding of thetiming, frequency and duration of environmental stress dur-ing a growing season will illuminate the role of plasticityand the G9 E9 E observed herein for local adaptation toclimate.

Implications of environmental interactions for plantevolutionary and functional studies

Plants experience a complex multidimensional environment inthe field (Passioura, 2007; Suzuki et al., 2014; Anderegg et al.,2015), although plant function is rarely studied in ways thatallow the partitioning of single and interactive effects. Interac-tions are predominant in factorial studies that explicitly test forinteractions (Chapin et al., 1987; Des Marais et al., 2013) and,as such, researchers should exercise caution when interpretingsingle treatments or when deploying agricultural traits opti-mized for single environments without a better understandingof their mechanisms of interaction. Environment-by-environment interaction may, in part, explain the considerablechallenges in ‘scaling up’ our understanding of molecular andphysiological mechanisms – derived largely from studies

conducted in two-level environmental contrasts – to providemeaningful insights or agricultural improvement at the whole-plant scale (Passioura, 2010). Additional genetic studies, such asthose performed here, assessing multiple environmentalparameters in factorial experiments, may facilitate bettergenotype-to-phenotype models.

Acknowledgements

We thank E. Sukamtoh, R. Timmerman, J. Bonnette, R. Hop-kins and B. Whitaker for assistance with data collection. R. Hop-kins and three anonymous reviewers provided helpful commentson the manuscript. This work was supported by USDA (NIFA-2011-67012-30663) funding to D.L.D., Academia Sinica, Insti-tute of Plant and Microbial Biology core research budget toP.E.V. and National Science Foundation (IOS-0922457) fund-ing to T.E.J.

Author contributions

D.L.D. and T.E.J. designed the experiment. D.L.D. and T.Z.C.collected the data. D.L.D., J.R.L., P.E.V. and T.E.J. performedthe analyses and all authors interpreted the results. D.L.D. wrotethe manuscript with contributions from all authors.

References

Ackerly DD, Bazzaz FA. 1995. Leaf dynamics, self-shading and carbon gain in

seedlings of a tropical pioneer tree. Oecologia 101: 289–298.Anderegg WR, Hicke JA, Fisher RA, Allen CD, Aukema J, Bentz B, Hood S,

Lichstein JW, Macalady AK, McDowell N et al. 2015. Tree mortality from

drought, insects, and their interactions in a changing climate. New Phytologist208: 674–683.

Bates LS, Waldren RP, Teare ID. 1973. Rapid determination of free proline for

water-stress studies. Plant and Soil 39: 205–207.Bloom AJ, Chapin FS III, Mooney HA. 1985. Resource limitation in

plants – an economic analogy. Annual Review of Ecology and Systematics16: 363–392.

Boyer JS. 1982. Plant productivity and environment. Science 218: 443–448.Boyer JS. 1995.Measuring the water status of plants and soils. San Diego, CA,

USA: Academic Press.

Brooks A, Farquhar GD. 1985. Effect of temperature on the CO2/O2 specificity

of ribulose-1,5-bisphosphate carboxylase/oxygenase and the rate of respiration

in the light. Planta 165: 397–406.Brutnell TP, Bennetzen JL, Vogel JP. 2015. Brachypodium distachyon and Setariaviridis: model genetic systems for the grasses. Annual Review of Plant Biology66: 465–485.

Cao J, Jiang M, Li P, Chu Z. 2016. Genome-wide identification and

evolutionary analyses of the PP2C gene family with their expression profiling in

response to multiple stresses in Brachypodium distachyon. BMC Genomics 17:175.

Catalan P, Muller J, Hasterok R, Jenkins G, Mur LA, Langdon T, Betekhtin A,

Siwinska D, Pimentel M, Lopez-Alvarez D. 2012. Evolution and taxonomic

split of the model grass Brachypodium distachyon. Annals of Botany 109: 385–405.

Chapin FS III, Bloom AJ, Field CB, Waring RH. 1987. Plant responses to

multiple environmental factors. BioScience 37: 45–97.Chaves MM. 1992. Effects of water deficits on carbon assimilation. Journal ofExperimental Botany 42: 1–16.

Coleman JS, Bazzaz FA. 1992. Effects of CO2 and temperature on growth

and resource use of co-occurring C3 and C4 annuals. Ecology 73: 1244–1259.



Research

NewPhytologist142

Condon AG, Farquhar GD, Rebetzke GJ, Richards RA. 2006. The application

of carbon isotope discrimination in cereal improvement for water-limited

environments. In: Ribaut JM, ed. Drought adaptation in cereals. New York, NY,

USA: Haworth Press, 171–219.Correia MJ, Rodriques ML, Osorio ML, Chaves MM. 1999. Effects of growth

temperature on the response of lupin stomata to drought and abscisic acid.

Australian Journal of Plant Physiology 26: 549–559.Cvikrova M, Gemperlova L, Martincova O, Vankova R. 2013. Effect of

drought and combined drought and heat stress on polyamine metabolism

in proline-over-producing tobacco plants. Plant Physiology and Biochemistry73: 7–15.

Demirevska K, Simova-Stoilova L, Fedina I, Georgieva K, Kunert K. 2010.

Response of oryzacystatin I transformed tobacco plants to drought, heat and

light stress. Journal of Agronomy and Crop Science 196: 90–99.Des Marais DL, Hernandez KH, Juenger TE. 2013. Genotype-by-environment

interaction and plasticity: exploring genomic responses of plants to the abiotic

environment. Annual Review of Ecology, Evolution and Systematics 44: 5–29.Des Marais DL, Juenger TE. 2016. Brachypodium and the abiotic environment.

In: Vogel J, ed. Genetics and genomics of Brachypodium. Cham, Switzerland:

Springer, 291–312.Des Marais DL, McKay JK, Richards JH, Sen S, Wayne T, Juenger TE. 2012.

Physiological genomics of response to soil drying in diverse Arabidopsis

accessions. Plant Cell 24: 893–914.Des Marais DL, Razzaque S, Hernandez KM, Garvin DF, Juenger TE. 2016.

Quantitative trait loci associated with natural diversity in water-use efficiency

and response to soil drying in Brachypodium distachyon. Plant Science 251:2–11.

Donovan LA, Dudley SA, Rosenthal DM, Ludwig F. 2007. Phenotypic selection

on leaf water use efficiency and related ecophysiological traits for natural

populations of desert sunflower. Oecologia 152: 13–25.Elser JJ, Bracken MES, Cleland EE, Gruner DS, Harpole WS, Hillebrand H,

Ngai JT, Seabloom EW, Shurin JB, Smith JE. 2007. Global analysis of

nitrogen and phosphorus limitation of primary producers in freshwater, marine

and terrestrial ecosystems. Ecology Letters 12: 1135–1142.Filiz E, Ozdemir BS, Budak F, Vogel JP, Tuna M, Budak H. 2009.Molecular,

morphological and cytological analysis of diverse Brachypodium distachyoninbred lines. Genome 52: 876–890.

Fry JD. 2004. Estimation of genetic variances and covariances by restricted

maximum likelihood using PROCMIXED. In: Saxton AM, ed. Geneticanalysis of complex traits using SAS. Cary, NC, USA: SAS Institute Inc., 11–27.

Gargallo-Garriga A, Sardans J, Perez-Trujillo M, Oravec M, Urban O, Jentsch

A, Kreyling J, Beierkuhnlein C, Parella T, Penuelas J. 2015.Warming

differentially influences the effects of drought on stoichiometry and

metabolomics in shoots and roots. New Phytologist 207: 591–603.Garvin DF. 2016. Brachypodium distachyon genetic resources. In: Vogel J, ed.Genetics and genomics of Brachypodium. Cham, Switzerland: Springer, 183–196.

Gates DM. 1968. Transpiration and leaf temperature. Annual Review of PlantBiology 19: 211–238.

Gordon S, Priest H, Des Marais DL, Schackwitz W, Figueroa M, Martin J,

Bragg J, Tyler L, Lee C-R, Bryant D et al. 2014. Genome diversity in

Brachypodium distachyon: deep sequencing of highly diverse inbred lines. PlantJournal 79: 361–374.

Havaux M. 1992. Stress tolerance of photosystem II in vivo: antagonistic effectsof water, heat, and photoinhibition stresses. Plant Physiology 100: 424–432.

Hennig L. 2010. Growth protocols for model plants in developmental biology.

In: Hennig L, K€ohler C, eds. Plant developmental biology: methods and protocols.New York, NY, USA: Humana Press, 1–10.

Heschel MS, Donohue K, Hausmann N, Schmitt J. 2002. Population

differentiation and natural selection for water-use efficiency in Impatienscapensis (Balsaminaceae). International Journal of Plant Science 163: 907–912.

Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A. 2005. Very high

resolution interpolated climate surfaces for global land areas. InternationalJournal of Climatology 25: 1965–1978.

Jenkins G, Hasterok R, Draper J. 2003. Building the molecular cytogenetic

infrastructure of a new model grass. In: Zwierzykowski Z, Surma M, Kachlicki

P, eds. Applications of novel cytogenetic and molecular techniques in genetics andbreeding of the grasses. Poznan, Poland: Institute of Plant Genetics, PAS, 77–84.

Jump AS, Penuelas J. 2005. Running to stand still: adaptation and the response

of plants to rapid climate change. Ecology Letters 8: 1010–1020.Kalnay E, Kanamitsu M, Kistler R, Collins W, Deaven D, Gandin L, Iredell M,

Saha S, White G, Woollen J et al. 1996. The NCEP/NCAR 40-year reanalysis

project. Bulletin of the American Meteorological Society 77: 437–471.Kawecki TJ, Ebert D. 2004. Conceptual issues in local adaptation. Ecology Letters7: 1225–1241.

Kellogg EA. 2015. Brachypodium distachyon as a genetic model system. AnnualReview of Genetics 49: 1–20.

Kenney AM, McKay JK, Richards JH, Juenger TE. 2014. Direct and indirect

selection on flowering time, water-use efficiency (WUE, d13C), and WUE

plasticity to drought in Arabidopsis thaliana. Ecology and Evolution 23: 4505–4521.

Kooyers NJ, Greenlee AB, Colicchio JM, Oh M, Blackman BK. 2014. Replicate

altitudinal clines reveal that evolutionary flexibility underlies adaptation to

drought stress in annualMimulus guttatus. New Phytologist 206: 152–165.Kramer PJ, Boyer JS. 1995.Water relations of plants and soils. San Diego, CA,

USA: Academic Press.

Littel RC, Milliken GA, Stroup WW, Wolfinger RD, Schabenberger O. 2006.

SAS® for mixed models. Cary, NC, USA: SAS Institute Inc.

L�opez-Alvarez D, Manzaneda AJ, Rey PJ, Giraldo P, Benavente E,

Allainguillaume J, Mur L, Caicedo AL, Hazen SP, Breiman A et al. 2015.Environmental niche variation and evolutionary diversification of the

Brachypodium distachyon grass complex species in their native circum-

Mediterranean range. American Journal of Botany 102: 1073–1088.Luo N, Liu J, Yu X, Jiang Y. 2011. Natural variation of drought response in

Brachypodium distachyon. Physiologia Plantarum 141: 19–29.Manzaneda AJ, Rey PJ, Anderson JT, Raskin E, Weiss-Lehman C, Mitchell-

Olds T. 2015. Natural variation, differentiation and genetic tradeoffs of

ecophysiological traits in response to water limitation in Brachypodiumdistachyon and its descendent allotetraploid B. hybridum (Poaceae). Evolution69: 2689–2704.

McKay JK, Richards JH, Mitchell-Olds T. 2003. Genetics of drought adaptation

in Arabidopsis thaliana: I. Pleiotropy contributes to genetic correlations among

ecological traits.Molecular Ecology 12: 1137–1151.Mittler R, Blumwald E. 2010. Genetic engineering for modern agriculture:

challenges and perspectives. Annual Review of Plant Biology 61: 443–462.Mooney HA, Winner WE, Pell EJ, eds. 1991. Response of plants to multiple

stresses. Physiological Ecology. San Diego, CA, USA: Academic Press Inc..

Nicotra AB, Atkin OK, Bonser SP, Davidson AM, Finnegan EJ, Mathesius U,

Poot P, Purugganan DM, Richards CL, Valladares F et al. 2010. Plantphenotypic plasticity in a changing climate. Trends in Plant Science 15: 684–692.

Passioura J. 2007. The drought environment: physical, biological and agricultural

perspectives. Journal of Experimental Botany 58: 113–117.Passioura J. 2010. Scaling up: the essence of effective agricultural research.

Functional Plant Biology 37: 585–591.Peace WJH, Grubb PJ. 1982. Interaction of light and mineral supply in growth

of Impatiens parviflora. New Phytologist 90: 127–150.Priest HD, Fox SE, Rowley ER, Murray JR, Michael TP, Mockler TC.

2014. Analysis of global gene expression in Brachypodium distachyonreveals extensive network plasticity in response to abiotic stress. PLoSONE 9: e87499.

Rizhsky L, Liang H, Shuman J, Shulaev V, Davletova S, Mittler R. 2004.When

defense pathways collide. The response of Arabidopsis to a combination of

drought and heat stress. Plant Physiology 134: 1683–1696.Schippman U. 1991. Revision der europaischen Arten der Gattung BrachypodiumPalisot de Beauvois (Poaceae). Boissiera 45: 1–250.

Schwartz CJ, Doyle MR, Manzaneda AJ, Rey PJ, Mitchell-Olds T, Amasino

RM. 2010. Natural variation of flowering time and vernalization

responsiveness in Brachypodium distachyon. Bioenergy Research 3: 38–46.Sherrard ME, Maherali H. 2006. The adaptive significance of drought escape in

Avena barbata, an annual grass. Evolution 60: 2478–2489.Stebbins GL. 1952. Aridity as a stimulus to plant evolution. American Naturalist86: 33–44.

Suzuki N, Rivero RM, Shulaev V, Blumwald E, Mittler R. 2014. Abiotic and

biotic stress combinations. New Phytologist 203: 32–43.





Van den Boogard R, Alewijnse D, Veneklaas EJ, Lambers H. 1997. Growth and

water-use efficiency of 10 Triticum aestivum cultivars at different water

availability in relation to allocation of biomass. Plant, Cell & Environment 20:200–210.

Verslues PE. 2010.Quantification of water stress-induced osmotic adjustment

and proline accumulation for Arabidopsis thalianamolecular genetic studies.

Methods in Molecular Biology 639: 301–315.Vile D, Pervent M, Belluau M, Vasseur F, Bresson J, Muller B, Granier C,

Simonneau T. 2012. Arabidopsis growth under prolonged high temperature

and water deficit: independent or interactive effects? Plant, Cell andEnvironment 35: 702–718.

Vogel JP, Gu YQ, Twigg P, Lazo GR, Laudencia-Chingcuanco D, Hayden DM,

Donze TJ, Vivian LA, Stamova B, Coleman-Derr D. 2006. EST sequencing

and phylogenetic analysis of the model grass Brachypodium distachyon.Theoretical and Applied Genetics 113: 186–195.

Vogel JP, Tuna M, Budak H, Huo N, Gu YQ, Steinwald MA. 2009.

Development of SSR markers and analysis of diversity in Turkish populations

of Brachypodium distachyon. BMC Plant Biology 9: 88.Xu X, Sherry RA, Niu S, Li D, Luo Y. 2013. Net primary productivity and rain-

use efficiency as affected by warming, altered precipitation, and clipping in a

mixed-grass prairie. Global Change Biology 19: 2753–2764.Yamaguchi-Shinozaki K, Shinozaki K. 2006. Transcriptional regulatory

networks in cellular responses and tolerance to dehydration and cold stresses.

Annual Review of Plant Biology 57: 781–803.Yoshioka K, Shinozaki K. 2009. Signal cross talk in plant stress response. Ames, IA,

USA: Wiley-Blackwell.

Supporting Information

Additional Supporting Information may be found online in theSupporting Information tab for this article:

Fig. S1 Climate diagrams for the place of origin for each of thestudied accessions.

Fig. S2 LSMeans for each accession in each of the four studiedenvironments.

Fig. S3 Significant correlations between trait genotypic means(LSMeans, presented as the grand mean value across all fourexperimental treatments) and climate at the location of origin.

Table S1 Principal component loadings of climate parametersused in the tests for association between trait values and climateat the location of origin

Please note: Wiley Blackwell are not responsible for the contentor functionality of any Supporting Information supplied by theauthors. Any queries (other than missing material) should bedirected to the New Phytologist Central Office.



Research

NewPhytologist144

Documents

Interactive effects of water limitation and elevated