A Stage-Based Study of Drought Response in Cryptantha Flava (Boraginaceae): Gas Exchange, Water Use Efficiency, and Whole Plant Performance Author(S): Brenda B

Home , Cryptantha

A Stage-Based Study of Drought Response in Cryptantha flava (Boraginaceae): Gas Exchange, Water Use Efficiency, and Whole Plant Performance Author(s): Brenda B. Casper, I. N. Forseth, D. Alexander Wait Reviewed work(s): Source: American Journal of Botany, Vol. 93, No. 7 (Jul., 2006), pp. 978-987 Published by: Botanical Society of America Stable URL: http://www.jstor.org/stable/4125585 . Accessed: 23/01/2012 09:21

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Botanical Society of America is collaborating with JSTOR to digitize, preserve and extend access to American Journal of Botany.

http://www.jstor.org AmericanJournal of Botany93(7): 977-987.2006.

A STAGE-BASED STUDY OF DROUGHT RESPONSE IN CRYPTANTHA FLAVA (BORAGINACEAE): GAS EXCHANGE, WATER USE EFFICIENCY, AND WHOLE PLANT PERFORMANCE1

BRENDA B. CASPER,2'5 I. N. FORSETH,3 AND D. ALEXANDER WAIT4

2Departmentof Biology,University of Pennsylvania, Philadelphia, Pennsylvania 19104 USA; 3DepartmentofBiology, Universityof Maryland,College Park, Maryland 20742 USA; and4Department of Biology,Missouri State University, Springfield,Missouri 65897 USA

Models of globalclimate change predict an increasein thefrequency of majordroughts, yet we knowlittle about the consequencesof droughtfor the demography of naturalpopulations. This studyexamined a populationof thesemi-desert perennialCryptantha flava (Boraginaceae) to determine how plants of different developmental stages respond to droughtthrough changesin leafgas exchange,leaf water potential, water use efficiency,growth, and reproduction.In two of thefour years, droughtwas appliedusing rainout shelters, and a severenatural drought occurred in another.Small, presumably younger, plants sometimeshad lower rates of maximum photosynthesis, lower leaf water potentials, and lower instantaneous orintegrated water- useefficiency than large plants. Small plants also had higher relative growth rates and lower reproductive effort. Large plants with evidenceof shrinkagefrom a previouslylarger size oftenproduced less growthand reproductionthan large healthy plants, suggestinga decline in plant vigor with age. Drought depressed gas exchangeand leaf water potentials equally in all plantstages. Thus,leaf-level physiological attributes provide no cluesfor why drought reduces growth more strongly in largeplants. The resultspoint to severaladditional avenues of researchrelevant to understandingstage-dependent or age-dependentplant performanceunder drought conditions.

Key words: Boraginaceae;Cryptantha flava; drought; plant aging; relative growth rate; water use efficiency.

Global climate change models predict an increasing toward understandinghow climate change will alter the frequencyin the occurrenceof extremeweather events, structureof desertvegetation. includingdrought (Palmer and Rilsinen, 2002; Ackerly, Consistentwith the observations of C. flava,several types of 2003; Diffenbaughet al., 2005). Yet, surprisinglyfew studies evidencesuggest that susceptibility to droughtvaries with plant have determinedhow droughtimpacts the demographyof size or stageof development.These rangefrom size-specific naturalpopulations. Far moreis knownabout morphological growthduring drought in otherspecies (Francoet al., 1994; and physiologicaladaptations of plantsto consistentlyarid Toft, 1995) to stage-specificdifferences in photosynthesis, environments(Larcher, 2003) thanabout plantresponses to transpiration,or water use efficiency(Nobel, 1987; Knappand temporalvariation in wateravailability. If droughtresponse Fahnestock,1990; Donovan and Ehleringer,1991; Anderson varies with plant size (Casper, 1996), droughtcan have and Tomlinson,1998; Cavender-Baresand Bazzaz, 2000). importantbut predictable repercussions for the size structureof Stage-specificdrought responses are also demonstratedin thepopulation and/or the relationship between plant size and annualcrops when final seed yieldis affectedby thetiming of fecundity(Caswell, 2001; Gurevitchet al., 2002; Brunaand waterstress within the growing season (Blum,1996; Lopez et Oli, 2005). al., 1996; Winkelet al., 1997). common studies have linked This study builds on the observationthat in natural Moreover, garden plant under conditions to leaf-level populationsof theherbaceous perennial Cryptantha flava (A. performance drought gas or water Such studies often Nels.) Payson shrink exchange potential. explore (Boraginaceae) large plants during variationin and selectionon attributes droughtwhile small plantscontinue to grow(Casper, 1996). genotypic physiological and Geberand Griffen, in The goals were to determinewhether plants of different (Arntz Delph,2001; 2003), usually to annuals,and the experimental approach maintains different but developmentalstages respond differently droughtthrough constantwater levels the season changesin growthor reproductionand to examineleaf-level throughout growing (Dudley, 1996a,b; Hescheland 2005). How of different physiologicalattributes that might explain such differences. Riginos, plants stagesrespond to droughtis notnormally a focus. Basal-rosetteforming perennials, like C. flava,are particularly Here,we examinethree developmental stages of C. flava for commonin cold deserts(McLaughlin, 1996). Understanding differencesin leaf waterpotential, leaf level photosynthesis how plantsof thisgrowth form respond to droughtis a step (Amax),stomatal conductance (gmax), instantaneouswater use efficiency (Amax/gmax), and integratedwater use efficiency, received25 October revision 11 2006. I Manuscript 2005; accepted April determinedfrom carbon isotopic ratios. We assesswhether there The authorsthank the Bureau of Land for to Management permission are stage-specificresponses to droughtin thesephysiological use thefield site and the Uintah Basin Branch Campus of theUtah State and L. for facilities.M. Peek and H. traitsor in growthand reproduction.Over this 4-year study, we University Squires laboratory created Kempenichassisted with data collection,and R. Lucas assistedwith droughtexperimentally by installingrainout shelters statisticalanalyses. The MarylandAgricultural Experiment Station duringthe growing season in 2 years,and in anotheryear we providedsummer support for I. Forseth.The workwas supportedby followedplants through a severenatural drought. NSF grantIBN95-27833 to B.B.C. andI.N.F. We used the physiological measurementsand plant 5 Authorfor correspondence (e-mail: [email protected]) performanceto test specific hypothesesregarding water 978 July2006] CASPER ET AL.-STAGE-BASED DROUGHT RESPONSE 979

requirementsand droughtresponses for the differentde- 2001), butmost growth occurs in spring.Seedlings emerge either in October velopmentalstages of C. flava. Our firsthypothesis dealt with and overwinteras greenplants, or in thespring. water with have A vegetativemeristem may continueto produceleaves forseveral years uptake: increasingdevelopmental age, plants until it bolts,producing an erect stalk with evenlydistributed leaves and greaterdifficulty meeting their water requirements and have a terminalinflorescence, usually bearing more than 30 flowers.Flowering is lowerrates of photosynthesis during drought. This is consistent initiatedin spring,with the first flowers opening by mid-May.Flower stalks die withthe root/shootratio declining in manyspecies as plants when seeds ripenin mid-Juneto earlyJuly, and the remainingleaves on grow(Coleman et al., 1994; Gedrocet al., 1996; Mulleret al., vegetativerosettes senesce a fewweeks later. The shrinkageof plantsthrough Directmeasurements of root are notfeasible themortality of vegetativerosettes is particularlycommon in olderplants (L. 2000). responses of and B. and in forC. becausethe diffuse root extends as Spindler,University Pennsylvania, Casper,unpublished data) flava system laterally dryyears (Casper, 1996). much as a meter(Peek and Forseth,2005) and because the The studypopulation is locatedadjacent to Red Fleet StatePark in Uintah ephemeralfine roots, which provide most of the absorptive Countyin northeasternUtah at 1730 m a.s.l. (40030' N, 109022'30"E), where surfacearea, are difficultto recover.Alternatively, leaf-level vegetationis dominatedby sagebrush,Artemisia tridentata Nutt., rabbitbrush, measurementsshould reveal whether Chrysothamnusnauseosus (Pallas) Britt.,and the small tree Juniperus physiological largeplants Little. have more difficultymeeting their water requirements than osteosperma(Torr.) small plants.If so, largeplants should have lowerleaf water and lower rates of and Climatic data-Precipitation and temperaturesduring the studywere potentials possibly transpiration comparedto long termclimatic means from data collectedat the Maeser 9 photosynthesis,and/or experience greater decline in these weatherstation, located -18 km SW of the site,at 1950 m a.s.l. (Western physiologicalparameters under drought conditions. RegionalClimate Center, http://www.wrcc.dri.edu). Our secondhypothesis involved water use efficiency,often considereda measureof droughttolerance (Donovan and Experimentaldesign-In early spring1997, 18 5 X 5 m studyplots, Ehleringer,1994; Dudley, 1996a; Heschel et al., 2002). By spatiallyarranged in blocks of threeplots, were demarcatedwithin a natural bothinstantaneous water use and carbon populationof C. flava usingat least 1 m betweenplots and severalmeters examining efficiency betweenblocks. Within each one selected was covered ratio,which wateruse over the block, randomly plot isotopic integrates efficiency witha rainoutshelter during the 1998 growingseason and anotherwas covered periodthe carbonwas assimilated,we testedthe hypothesis in 1999. The six remainingplots received ambient precipitation all years. thatwater use efficiencyincreases with increasing develop- In each plot,at leastthree small, young plants (<12 rosettes)and at leastsix mentalage. We based thishypothesis on researchwith this largeplants (>12 rosettes)were selected arbitrarily from among those growing species (Casper et al., 2005) and others(Knapp and Fahne- in full sun (>0.5 m froma shrub)for measures of gas exchangeand plant werefurther on the or stock,1990; Donovan and Ehleringer,1991, 1992; Cavender- performance.Large plants separated,depending presence Bares and in whicholder or had absence of dead vegetativerosettes, which is evidence of shrinkagefrom Bazzaz, 2000) largerplants a previouslylarger size. Thus, there were three size/condition categories: small, greaterwater use efficiency.However, an increasein wateruse largehealthy, or largewith dieback. Although absolute plant age couldnot be efficiencyin larger,older plants is inconsistentwith increasing known,small plants were essentially juveniles, developmentally younger than susceptibilityto droughtwith age. largeplants, and easilydistinguished from large plants that had shrunkin size. Finally,we examinedplant growth and reproductionto test We judged thatplants with evidence of diebackwere, on average,the oldest the that reducesthe overall of because a 20-yrstudy of a singlecohort showed that most individuals shrink hypothesis drought performance before die of and B. thanof small We thatif they (L. Spindler,University Pennsylvania, Casper, largeplants more plants. recognized unpublisheddata), and therehad been no recentsevere drought to cause the relativegrowth rate normallydeclines with plant size, then shrinkage.A plantwas moved fromthe adulthealthy to the adultdieback droughtcould resultin negativegrowth in large plantsand categorywhen it lost morethan 30% of its vegetativerosettes. When large positivegrowth in small plants even if droughtdepresses plantsshrank to <12 rosettesthey were removed from the study.Additional small wereselected in 1999 to thosethat had out of the growthequally in plantsof all sizes.We evaluatedhow drought plants replace grown affected and both for size smallsize categoryor had died.Plants were marked with flags on shortstakes growth reproduction, adjusted plant and buriedmetal tags that could be relocatedwith a metaldetector. (relative growthrate and reproductiveeffort), and used to the between regressionanalyses compare relationship Droughttreatments-Two different types of rainoutshelters were used. In relativegrowth rate and plantsize amongyears differing in 1998,six plotswere covered only during rain events between 1 Marchand 17 precipitation.The latterapproach allowed us to determine Juneby unrolling6 m X 6 m (0.5 m plot overhangto reduceedge effects) whetherdrought alters the slope of therelationship in a way opaque canvastarpaulins over metal frames slanted from 1 m to 2.5 m in height thatreflects disproportionately less growthin largerplants. to allow runoff.Covers were extendedonly duringrainfall to minimizethe effectson plotmicroclimate. In 1999,the frames were moved to six different plots, and the tarpaulinwas replacedwith stationarypolyethylene roofing materialthat transmitsboth photosyntheticallyactive (400-700 nm) and MATERIALS AND METHODS infraredradiation (Reynolds et al., 1999).The roofingmaterial was leftin place continuouslybetween 1 March and 23 May. Ambientprecipitation was The studysystem-Cryptantha flava growsin sandysoils fromcentral measuredat the fieldsite duringthe growingseasons of 1998 and 1999, in Wyoming,throughout eastern Utah, and into northernArizona and New orderto calculatethe amount of rainfall excluded by theshelters (Casper et al., Mexico,USA. A plantconsists of one to >70 basal leafrosettes supported by 2001). Airtemperatures were measured continuously under the shelters in 1999 a branchedwoody underground stem (caudex) connectedto a singletaproot. and in unshelteredareas nearby. Smallerroots branch off the taprootand may extendlaterally as much as a meterbefore turning downward (Peek andForseth, 2005; B. Casper,personal Gas exchange-Maximum dailyphotosynthetic rates (Amax) and stomatal observation).The suberizedcovering of the woody root system is interruptedat conductances(gmax) were measuredon fullyexposed leaves frommarked intervalswhere ephemeral,fine roots are produced in clusters. This plantsbetween 1000 to 1400 hoursMST approximatelybiweekly from mid- arrangementallows fora rapidincrease in absorptiveroot surface following May to Julyin all 4 yearsof thestudy. Instantaneous water use efficiencywas precipitationbut minimizes surface exposure to drysoils. The narrow,nearly calculatedas theratio of Amax/gmax. Photosynthesis was measuredby enclosing vertical,oblanceolate leaves (6.0-9.0 cm long) firstappear in mid-April.New one to threefully expanded, healthy leaves fromthe same vegetative rosette in leaves are producedand old ones die throughoutthe spring and earlysummer a 0.25-L chamberof a LiCor 6200 closed photosynthesissystem (LiCor, growingseason (Casper et al., 2001), andnew rosettes arise from axillary buds. Lincoln,Nebraska, USA) and calculatingCO2 depletionin thechamber over Additionalleaves mayappear in responseto late summerrains (Casper et al., time.Leaf areawas determinedfrom length X widthof theenclosed portion of 980 AMERICAN JOURNALOF BOTANY [Vol. 93 each leaf.It was notpossible to sampleplants in all 18 plotson a singleday, but all markedplants from at leastone plotfrom each droughttreatment (sheltered in 1998,sheltered in 1999,and never sheltered) were measured on thesame day to maintainbalanced comparisons. For each year,the threemeasures, Amax, 0 gmax, and Amax/gmax,were separatelyanalyzed as a functionof the size/ conditioncategory and droughttreatment, both fixed effects, using ANCOVA withdate as thecovariate and blockas a randomfactor. in We expectedeither that rates of gas exchangewould be inherentlylower t-- largeplants or thatour drought treatments would depress gas exchangemore in large plantsthan in small plants,which would resultin a significantsize/ \Plant size conditioncategory X droughttreatment interaction in 1998 and 1999-the 0 years when the droughttreatments were applied. No differencesin gas exchangeas a functionof droughttreatment should be evidentin 1997,prior to thefirst drought application. B

Waterpotential-Predawn and midday leaf water potentials were measured Fig. 1. The expectedrelationship between relative growth rate and size usinga pressurechamber (Plant Moisture Systems, Corvallis, Oregon, USA) of plants(solid line) and two hypotheticalchanges in therelationship after and/ora Peltier-typethermocouple psychrometer (Tru-Psi Water Potential drought.Line A indicatesthat growth is depressedequally in plantsof Meter, Model SC10X, Decagon Devices, Pullman, Washington,USA). differentsizes, while line B illustratesa disproportionatereduction in Because theseare destructive measurements, leaves were taken from unmarked growthfor larger plants. In bothcases, largerplants have negativegrowth. smalland largeplants within the study plots. Three to fourtimes each yearwe collectedone randomlyselected, fully expanded, healthy leaf from at leastthree RGR normallydeclines with plant size (Evans, 1972),we expecteda negative to fiveplants per size categoryin each treatment.We did notdistinguish large slopefor the regression (Fig. solidline). Should all plantsbe equallyaffected healthyplants and diebackplants, and thetreatment categories were either the 1, by drought,a dryyear should produce the same slopebut a smallery-intercept droughttreatment of thecurrent year or controls,i.e., two drought treatments in (Fig. 1,Line A), withlarger plants potentially having negative growth. If large 1998 and 1999but none in 1997 and 2000. Foreach year,predawn and midday leaf water were as a functionof size plantswere more severely affected by drought,then the negative slope of the potentials separatelyanalyzed category linewould be of in line We vs. and if dateas a covariate regression greatermagnitude dryyears (Fig. 1, B). (large small) droughttreatment, applicable, using includedall marked in thecontrol treatment and usedANOVA to in ANCOVA. plants only compareslopes of significant regressions. We testedour specific prediction that As soil waterpotentials decreased in thelatter part of thegrowing season, the regressionof RGR on plantsize would producea negativeslope of the leavesoften broke in thepressure chamber prior to waterpotential equilibrium, greatestmagnitude 1999-2000 because 2000 was thedriest year. preventingan accuratedetermination of middayvalues. Therefore, in 1999and 2000, a psychrometerwas used in conjunctionwith the pressure chamber after firstusing the two methodson differentleaves of thesame plantto checkfor Statisticalanalyses-Physiological data involving ANOVA andANCOVA instrumentalbias, whichwas nonsignificant.For psychrometermeasurements, were analyzedusing STATISTICA (version6.0, StatSoft,Tulsa, Oklahoma, Post-hoc were the Fisherleast leafsurfaces were abraded with fine grit (600) siliconcarbide paper, and leaves USA). comparisons made using significant test < Data were when to meet wereplaced in a stainlesssteel sample cup. Leafwater potential was plottedvs. difference (P 0.05). transformed, necessary, ofvariance. For some data that were still not theKruskal- timeto determinesuitable equilibration for these measurements. homogeneity normal, Wallis testwas applied.Regression analyses of RGR were performedusing ANOVA in SAS (version8.2, SAS Institute,Cary, North Carolina, USA). Integratedwater use efficiency-The carbon isotopicratio 6 ('1CO2/ 12CO2)of leaftissue was usedas an integratedmeasure of wateruse efficiency (Farquharet al., 1989).Each June,two to fourleaves were collected from every livingmarked plant, dried, and groundin liquid nitrogenbefore analyzing RESULTS isotopicratios relative to standardsby mass spectrometryat the Stable Isotope Ratio Facilityfor Ecological Research,University of Utah,Salt Lake City, Climatepatterns-At the Maeser 9 NW climatestation, Utah,or at theDepartment of Geology,University of Maryland,College Park, annualprecipitation averaged 357 mmfrom 1971 to 2003,and Maryland.Fifty samples were used to crosscalibrate measurements from both in to facilitiesto ensure consistentresults. Carbon ratio values were meanmonthly temperature ranged from -6.2'C January isotope 20.40C From1997 to in winterand convertedto discriminationvalues (A) by theequation (Farquhar et al., 1989): in July. 1999,precipitation spring,when 60% of the annualprecipitation normally falls, - A = (6a 6p)/(1 + 6p/1000) was above the long-termmean (Fig. 2). Precipitationfrom October 1999 to 2000 was well below where,6a = carbon ratioof CO2 in the (assumedto be February normal, isotope atmosphere in soils for the initiationof -8/o/oo)and 68 = measuredcarbon isotope ratio of theplant tissue. Lower values resulting dry springgrowth; of A indicatehigher water use efficiency.For each year,A was examinedas precipitationwas also below normalin May and June2000 a functionof droughttreatment and plantsize/condition category, with block as (Fig. 2). a randomfactor, using ANOVA. Temperaturesduring the study differed from normal in two obvious ways. From 1998 through2000, mean monthly Wholeplant performance-Vegetative and floweringrosettes on marked temperatureswere generally warmer than long-term means in plantswere counted once annually,in the middleof the growingseason, to thewinter and springmonths but especially so in January1999, determinethe effect of droughttreatment on growthand reproduction.Growth whenthe exceededthe mean 40C was expressedas relativegrowth rate (RGR), (totalno. rosettesin yeart+l- averagetemperature 30-yr by total no. rosettesin year,)/totalno. rosettesin yeart.Reproduction was (Fig. 2). An exceptionto the warmertrend occurred between expressedas reproductiveeffort (RE), thenumber of floweringrosettes divided April and June 1999, when temperatureswere considerably by thetotal number of rosettes.For each yeart,RGR and RE wereseparately below the long-termmean. There were 13 dates between1 analyzedas a functionof drought treatment and plant size categoryin yeartwith Apriland 15 May 1999 when ambientnighttime air temper- blockas a randomfactor in ANOVA. aturesat the studysite dropped below Thento assesswhether the severe natural drought in 2000 reducedgrowth in O0C. largeplants differentially, we comparedRGR graphedas a functionof plant detailelsewhere size (actual numberof rosettes)among differentyears. For each year,we Effectsof shelters-As reportedin (Casper graphedRGR betweenyeart and yeart+las a functionof actualplant size in et al., 2001), in 1998 theretractable rainout shelters reduced yeartand used linearregression to testfor a significantrelationship. Because precipitationbetween 1 Marchand 17 June,1998 by roughly July2006] CASPER ET AL.-STAGE-BASED DROUGHT RESPONSE 981 U/

Z2_ L/

c- I r- J.I i 0? C r- 0 E 19971)98/9 0

Na.~Na. ~ ~ ij.Ia ~ Na . a. I 97 I 19 99 I20

Fig. 2. Monthlymean temperaturesand precipitationat the Maeser, Utah, weatherstation in relationto the 30-yrmean for the period from Fig. 3. Values of Amax (mean + SE) forplants of Cryptanthaflava of November1996 to July2000. differentsizes/conditions (sm = small, Ig = large, Ig db= large with dieback) and droughttreatments (she198 = shelteredin 1998, she199= shelteredin 1999, or not shel = controls)for each of the 4 years. 50%, while in 1999 shelterseliminated all precipitation between1 March and 23 May 1999. In 1999, sheltersalso had a moderatinginfluence on nighttimeair temperatures, whichwere warmer than ambient by 2.6 + 0.1VC(mean + SE). Whenair temperatures under shelters dropped below freezing, theyremained so forshorter periods (1.5 + 0.6 h) comparedto thosein theopen (4.5 + 0.8 h). Daytimeair temperatures under shelters,in contrast,were always within 0.50C of ambient.

Gas exchange-Gas exchange rates were reduced in shelteredplants and, in some years differedamong plant size/conditioncategories, but the consistent lack of a treatment X categoryinteraction for Amax, gmax or Amax/gmaxmeans that gas exchangein differentplant categories was equallyaffected our treatments.The values cn by drought gas exchange confirm r thatthe droughttreatment was moreeffective in 1999, when r o c shelterssubstantially reduced both Amax andgmax, than in 1998, r whensheltering plants reduced only gmax (Figs. 3-5, Table 1). Instantaneouswater use differed efficiency among drought r?J. treatmentsonly in 1999,when surprisingly, plants sheltered in 1998 exhibiteda higherAmax/gmax than those sheltered in 1999 and thosenever sheltered. In yearswhen there was a differenceamong size/condition categoriesin gas exchange,small plants had thelowest rates, with significantlylower values of Amax than large healthy plantsin 1997 and lowerAmax/gmax than both categoriesof large plantsin 1999 and 2000 (Figs. 3-5, Table 1). Large diebackplants did notdiffer from large healthy plants except in 1999, whenthey had lowergmax thaneither of the othertwo size categories. There was more variationamong years in gas exchange Fig. 4. Values of gmax (mean + SE) forplants of Cryptanthaflava of parametersthan between treatments. The naturaldrought of differentsizes/conditions and droughttreatments for each of the 4 years. 982 AMERICAN JOURNALOF BOTANY [Vol. 93

TABLE 1. ANOVA mean square values forgas exchange parametersin Cryptanthaflava analyzed as a functionof currentsize, drought treatment,and date as a covariate.In some years,sufficient data were collectedin fewerthan six blocks.

Year Effect df Amax gmax Amaxgmax

&6 1997 Date 1 177.7 1.075*** 1219.4 Category1997 (Cat97) 2 224.2* 0.033 44.5 Treatment 2 148.6 0.057 1060.4 o (Treat) E Block 4 319.7 0.590* 3555.5 o Cat97 x Treat 4 121.2 0.081 115.0 E Cat97 X Block 8 43.2 0.076 714.6 TreatmentX Block 8 80.0 0.058 378.0 Cat97 X TreatX Block 16 67.5 0.030 751.4 Error 149 47.5 0.052 743.0 1998 Date 1 459.4*** 0.095 2079.8*** Category1998 (Cat98) 2 68.1 0.168 154.4 Treatment(Treat) 2 96.0 0.311* 236.2 Block 5 36.4 0.080 567.0 ? _..,...' Cat98 x Treat 4 8.5 0.013 70.8 TreatX Block 10 97.2* 0.091 297.7 Cat98 X Block 10 49.6 0.051 103.7 .~ = Cat98 x TreatX Block 12 33.7 0.052 155.3 Error 466 31.3 0.044 141.8 1999 Date 1 1396.8*** 4.748*** 919.9** Category1999 (Cat99) 2 14.1 0.627** 853.1** Treatment(Treat) 2 1178.9*** 1.680** 786.9* Block 5 30.2 0.479 692.4* Cat99 x Treat 4 2.5 0.057 84.0 X 10 29.3 0.051 96.9 5. Values of ? for of Cat99 Block Fig. Amax/gmax (mean SE) plants Cryptantha Treatmentx Block 8 46.0 0.158 174.3 flava of differentsizes/conditions and droughttreatments for each of the4 Cat99 x TreatX Block 14 24.4 0.085 76.1 years. Error 355 22.9 0.066 119.2 2000 Date 1 2931.8*** 1.584*** Category2000 (CatOO) 2 28.5 0.011 P < 0.05a Treatment(Treat) 2 10.9 0.002 NS 2000 reduced Amax and gmax slightlymore than did our Block 5 140.4 0.181 experimentaldrought treatments in 1998 or 1999 (Figs. 3, 4). Cat00 x Treat 4 34.6 0.018 Date as a significantcovariate indicated a declinein Amax over CatOOx Block 10 42.4 0.029 the season in and 2000 and a similar Treatmentx Block 10 90.3 0.057 growing 1998, 1999, CatOOx Treatx Block 16 32.6 0.020 declinein gmax in 1998, 1999,and 2000. The changesin these Error 462 28.3 0.018 two resultedin a increasein Amax/gmax parameters significant *** duringthe 1998 and 1999 seasons. Block was significantas Note: *P < 0.05, ** P < 0.01, P < 0.001, NS = not significant. a H = a main effect,indicating spatial variation independent of our Kruskal-Wallistest, 8.64. droughttreatments, only for gmax in 1997 and forAmax/gmax in 1999.A significanttreatment x block interaction term for Amax in 1998 meansthat the 1998 droughttreatment reduced Amax lower values for shelteredin 1999. morein some blocksthan others. significantly plants Otherwise,there were no otherdifferences in A amongsize/ condition or treatments.Plants Water water were reduced categories among drought potential-Plant potentials by in tendedto have lower A values that treatmentand in some were lower in small sheltered 1999 year drought years thanthe other two treatment butthe difference was not plants.Sheltering plants reduced both predawn and midday leaf groups, (P < 0.11). waterpotentials to a greaterextent in 1999 thanin 1998 (Fig. 6, significant Table 2). Small had lower water than plants predawn potentials Whole the RGR in 1999 and 2000 and lower midday water plant performance-On average, greatest large plants occurred1999-2000 and the lowestfrom 1998-1999. Small in 2000. In most years,both measuresof water potentials had RGR than potentialdeclined over the growingseason, exceptpredawn plantsalways higher largehealthy plants. Large values did nothave a seasonaldecline in 1998. healthyplants had higherRGR thanlarge plants with dieback in all yearsexcept 1998-1999, when they did notdiffer (Fig. 8, Integratedwater use efficiency-Carbonisotopic ratios (A) Table 3). In both 1997-1998 and 2000-2001, the two weremuch lower in 2000, theyear of naturaldrought, than in categoriesof large plantshad, on average,zero or negative otheryears (Fig. 7). Therewere differences in A amongplant RGR, while small plants produced positive growth.The size/conditioncategories only in 1997, whensmall plants had experimentaldrought treatments did notaffect RGR anyyear. significantlyhigher (F2,10o = 7.7; P < 0.01) values,indicating Therewere differencesamong size/condition categories in lowerwater use efficiency,than the othertwo size/condition RE in 1997, 1999, and 2000 withsmall plantsconsistently categories.The blockX treatmentterm was significant(F10,16 - havinglower RE thanlarge healthy plants (Fig. 9; Table 4); 3.22; P < 0.02) forvalues in 2000, indicating,in someblocks, manysmall plants did notflower at all. Large diebackplants July2006] CASPERET AL.-STAGE-BASED DROUGHTRESPONSE 983

05 I

C O ) 0_

C , ,J

Fig. 7. Carbon isotopicratios (A, means ? SE) for plantsof Pre-dawn Midday Cryptanthaflava of differentsize/conditions and drought treatments. Fig.6. Predawnand midday leaf water potential values (means ? SE) forsmall and largeplants of Cryptanthaflava, separately presented for DISCUSSION plantssheltered and those not sheltered ineach of the four years. Hypothesis1-We foundno supportfor our hypothesis that of have had intermediatelevels of fromboth plants increasingdevelopmental age greaterdifficulty RE, differingsignificantly their water and lower leaf-level in < meeting requirements small and largehealthy plants 1997 (P 0.05), differing ratesunder conditions.In when < photosynthetic drought years onlyfrom small plants in 1999 (P 0.01) and differingfrom leaf water differed neithersmall nor in 2000. potentials among size/developmental largehealthy plants smallplants had lowervalues, suggesting that they treatmentaffected RE in 1999 when that categories, Drought only plants adjust leaf water potentialswhile maintainingthe same had been shelteredin 1998 had RE thanthose sheltered higher transpirationrate as large plants.This adjustmentcould be in 1999 (P < 0.001) or controls(P < 0.05). Overall,RE in 1999 was lowerthan in otheryears, and manyflowering stalks on unshelteredplants were visibly frost damaged while those TABLE2. ANCOVAmean square values for predawn and midday (1200- undershelters were not. 1400 hours)water potential for Cryptantha flava, with date as the covariateand size and coveras fixedeffects. Error df in RGR between t and t + 1 plant plot Graphing years providessupport parentheses. forthe idea thatdrought affects large plants more severely than small plants (Fig. 10). The regressionrelationships are Year Effect Predawn Midday significantfor each ofthe four years but explain more variation 1997 Date 11.45*** 1.293*** in 1999-2000 (r2 - 0.24, P < 0.001) and 2000-2001 (r2 - PlantSize 11.43 0.003 0.27, P < 0.001) thanin 1997-1998 (r2 0.16, P < 0.01) or Error 0.17 (62) 0.047(42) 1998-1999 (r2 = 0.12, P < 0.05), whichcould reflectthe 1998 Date 0.014 2.853*** PlantSize 0.021 0.006 severe2000 drought.The slopesof therelationships, although PlotCover 0.113* 0.310** alwaysnegative, differ across the four years (F3,173 =10.44; P Size x Cover 0.001 0.025 < 0.001). The steepestslope occurredfor RGR from1999 to Error 0.018(67) 0.030(50) 2000,but this is partlyaccounted for by greatergrowth of some 1999 Date 21.032*** 36.792*** small to other The of the PlantSize 0.463 0.502* plants compared years. slope PlotCover 1.715*** 1.069*** regressionfor RGR 1999-2000 is only marginallydifferent Size X Cover 0.024 0.026 fromRGR 2000- 2001 (F1,91- 3.20; P < 0.08), whichhas the Error 0.138(139) 0.090(175) nextsteepest slope, but significantly different from RGR 1997- 2000 Date 9.692*** 5.307*** Size 0.199* 0.296** 1998 (F1,96 8.80; P < 0.001); 1997 and 1998 areboth years Plant Error 0.033(172) 0.029(117) withouteither natural drought or exceptionallycold spring temperatures. Note:* P < 0.05, ** P < 0.01, *** P < 0.001. 984 AMERICAN JOURNALOF BOTANY [Vol. 93

LL, Lk.

Fig.8. Relativegrowth rate from year t, the year shown in panel, to Fig. 9. Annualreproductive effort presented separately for plants of yeart + 1 (means? SE) presentedfor plants of Cryptanthaflava of Cryptanthaflava of differentsizes/conditions and drought treatments. differentsizes/conditions anddrought treatments. whenwe replacedsome plants that had outgrownthe small size necessaryif theirroots are less developed or draw from category.Differences in wateruse efficiencyamong plants of shallower,drier soil layers.In someyears, small plants also had differentstages might also depend on soil water status, lowerAmax or lowerAmax/gmax, indicatinggreater water loss occurringonly in some years(Cavender-Bares and Bazzaz, per unit of carbon gained. Importantly,our experimental 2000). It is noteworthythat in our study plants of different sizes droughttreatments reduced gas exchangeequally in plantsof did notdiffer in A in 2000, thedriest year. differentsizes and developmentalstages. Changesin water-use physiology with ontogeny is a common patternin perennialspecies. Larger or olderplants often have Hypothesis2-Differences in A as a functionof plantsize higherinstantaneous water use efficiencyor integratedwater existedonly in the first year of the study, when small plants had use efficiencyas measuredby A (Knappand Fahnestock, 1990; higherA values, indicatinglower wateruse efficiency,than Donovan and Ehleringer,1991; Donovan and Ehleringer, largeplants. This resultsupports our second hypothesisthat 1992). In red oaks (Cavender-Baresand Bazzaz, 2000), wateruse efficiencyincreases with increasing developmental droughtcauses a greaterincrease in wateruse efficiencyfor age, butfails to explainwhy large plants shrink during drought. adultsand a greaterdepression of photosynthesisin seedlings. The factthat stage-specific differences in A werenot maintained Those differencessuggest that older plants are more competent in subsequentyears could be due to thesmall plants selected in thanyounger plants to cope withtemporal variation in water 1997 becomingmore similar in size to largeplants over the availability. courseof the4 years.Consistent with this interpretation, there was a slight,nonsignificant trend for differences in A among Hypothesis3-We examinedwhole plantperformance to plantsof differentsizes and conditions(P < 0.11) in 1999, testthe specifichypothesis that drought reduces the overall

TABLE3. ANOVAmean square values for relative growth rate (RGR) (Year,to Year,+I) in Cryptantha flava analyzed as a functionof size category and droughttreatment by year. Error df is in parentheses.

Effect df 1997-98 1998-99 1999-2000 2000-01

Size Category(Cat) 2 6.189*** 1.480*** 5.731** 5.681** Treatment(Treat) 2 1.165 1.019 0.289 0.406 Block 5 0.674 0.248 0.532 1.095 Cat X Treat 4 0.210 0.291 0.071 0.664 Cat X Block 10 0.331 0.095 0.678 0.622 TreatX Block 10 0.424 0.367 1.060 0.863 Cat X Treatx Block 20 0.326 0.139 0.867 0.354 Error 0.246(98) 0.219(74) 0.250(95) 0.339(83) July2006] CASPER ET AL.-STAGE-BASED DROUGHTRESPONSE 985

TABLE 4. ANOVA meansquare values for reproductive effort (RE) (Yeartto Yeart+l)in Cryptanthaflava analyzed as a functionof size categoryand droughttreatment in Yeart. Error df in parentheses.

Effect df 1997-98 1998-99 1999-2000 2000-01

Size Category(Cat) 2 4.424** 0.022 P < 0.001a P < 0.01c Treatment(Treat) 2 0.001 0.039 P < 0.001b NS Block 5 0.005 0.121 Cat X Treat 4 0.016 0.013 Cat x Block 10 0.020 0.019 TreatX Block 10 0.034 0.029 Cat x Treatx Block 20 0.015 0.024 Error 0.016(103) 0.018(92) Note:NS = notsignificant. a Kruskal-Wallistest, H = 43.31. b H-- 18.58. cH = 12.42.

performanceof largeplants more strongly than small plants. overall was not the lowest of the study. The regression Even in the absence of drought,RGR decreasedand RE analysis,therefore, provides limited support for our hypothesis increasedwith plant size, which is typical(Evans, 1972; Franco thatdrought reduces growthmore stronglyin large plants. et al., 1994). RGR and RE are presumablyrelated because Because each regressionline essentially represents a single data reproductionshould reduce the amount of energyavailable for point,these analyses should be replicatedacross more years in growth.The negativerelationship between plant size and RGR orderto evaluate more fullywhether drought differentially may also reflectthe fact that canopy level assimilation, affectsthe growth of smalland largeplants. expressedon a leaf area basis, cannot scale directlywith Should the patternhold that droughtreduces growth in measurementsof photosynthesismade on individualleaves disproportionately large plants despite size-independent effectsof on leaflevel therewill be heldperpendicular to incidentradiation. This is due to variable drought assimilation,Amax, theneed to understandcarbon and water in leaf and leaf and self-shadingwithin the budgets acquisition angles temperatures whole as a functionof size and environmental leafcanopy (Forseth and Norman,1993; Larcher,2003). plants plant conditions.Further should include (1) whole Because RGR normallydeclines with plant size, a changein investigations the of RGR as functionof size is plant carbon assimilation,carbon allocation(Franco et al., slope graphed plant required and round demands to demonstratethat more affects in 1994), year respiratory (Wyka,1999), (2) drought strongly growth the that shrinkdue to failurein For the of the did possibility plants hydraulic large plants. C. flava, slope relationship a localized of the et Cruiziatet The was forRGR 1999- portion plant(Sperry al., 1991; changeamong years. slope steepest al., 2002; et al., 2002; Vesk and and 2000 with2000 thedriest of the but Sperry Westoby,2003), being year study, overall, (3) the thatleaf surface area is to maintain RGR 1999-2000 was thanin other and possibility adjusted actuallyhigher years, favorableroot to shootratios regardless of hydraulicfailure the was driven RGR steeperslope largely by generallyhigher (Sharpand Davies, 1985; Sperry,2000; Chaves et al., 2002). of small to other Some of plantscompared years. repercussions Thereis evidencefrom agricultural species that roots can detect 2000 were in RGR the drought likelyexpressed 2000-2001, drysoils and signalthe shoot, which can adjustdevelopment, the year withthe next steepestslope, but even thenRGR evenbefore unfavorable leaf water potentials occur (Sharp and Davies, 1985; Zhangand Davies, 1987). It is also possiblethat the dailyor seasonal durationof assimilationdiffers between small and large plants in ways that are not indicatedby instantaneousgas exchangemeasurements made at theheight of thegrowing season as theywere here.

. Additional observations-Whole plant performancealso . . . . varieswith the condition of theindividual. This was indicated by adults with previous shrinkagehaving RGR and RE *" * U U U d" intermediatein magnitudebetween small and large healthy " , plants.As relativemeasures of performance,RGR and RE rNru ?- , adjust for differencesin absoluteplant size, so plantswith ?' 4 0 ", '0 i' previousdieback produceless growthand fewerflowering stalksthan comparably sized plantswithout prior shrinkage. Data froma 20-yrlongitudinal study of a singlecohort of C. flava at this site show thatmost plantsshrink and decrease Nt,) o flowerstalk production for a few yearsbefore they die (L. N1ltmher of rosetles Spindler,University of Pennsylvania,and B. Casper, un- an declinein Fig. 10. Relativegrowth rate (year t to yeart +1) graphedas a function publisheddata), suggesting age-related physiol- of size in yeart forplants of Cryptanthaflava in the controltreatment. All ogy (Watkinson,1992; Partridgeand Barton, 1993) or plants,healthy and withdieback, are included.All regressionrelationships hydraulicconductivity. We have no direct evidence for are significant(P < 0.05). hydraulicfailure in C. flava,but we did observelower stomatal 986 AMERICAN JOURNAL OF BOTANY [Vol. 93 conductancefor plantswith dieback in 1999, which could CASWELL, H. 2001. Matrix population models. Sinauer, Sunderland, reflectimpedance of watertransport. Massachusetts,USA. AND F. A. 2000. in Interestingly,our study shows that the biological con- CAVENDER-BARES, J., BAZZAZ. Changes drought of extended the itself. responsestrategies with ontogeny in Quercus rubra: implicationsfor sequences drought beyond droughtyear from to maturetrees. 124: 8-18. Therewere two effectsof the 1998 treatment scaling seedlings Oecologia carryover drought CHAVES, M. M., J. S. PEREIRA,J. MAROCO, M. L. RODRIGUES,C. P. P. in the followingyear. In 1999, the plants that had been RICARDO,M. L. OSORIO, I. CARVALHO,T. FARIA,AND C. PINHEIRO.2002. shelteredin 1998 had higherAmax/gmax and higherRE than How plantscope withwater stress in the field.Annals ofBotany 89: plantssheltered in 1999 or controls.Drought has been shown 907-916. to trigger floweringin some species (Southwick and COLEMAN, J. S., K. D. M. McCONNAUGHAY, AND D. D. ACKERLY. 1994. Davenport,1986; Nakajimaet al., 1993), so the increasein Interpretingphenotypic variation in plants. Trends in Ecology and RE we observedthe year following the drought treatment is not Evolution9: 187-191. H. AND T. 2002. architectureof surprising.Any residual effectsfrom the 1999 drought CRUIZIAT, P., COCHARD, AMEGLIO. Hydraulic trees: main concepts and results.Annals of Forest Science 7: 723- treatmentmight have been masked by the severe natural 752. in 2000. drought DIFFENBAUGH, N. S., J. S. PAL, R. J.TRAPP, AND F. GIORGI.2005. Fine-scale Annual variationin floweringand growthappears to be processes regulatethe responseof extremeevents to global climate influencedby springtimetemperatures as well as precipitation. change.Proceedings of theNational Academy of Sciences, USA 102: RGR from1998 to 1999 andRE in 1999 werethe lowest of the 15774-15778. 4 years.Neither 1998 nor1999 wereparticularly dry years, but DONOVAN,L. A., ANDJ. R. EHLERINGER.1991. Ecophysiologicaldifferences thespring of 1999 was exceptionallycold. Thatyear, freezing among juvenile and reproductiveplants of several woody species. 86: 594-597. nighttimetemperatures occurred after the initiationof plant Oecologia DONOVAN,L. A., AND J. EHLERINGER.1992. water-use growth,visibly damagingdeveloping flowering stalks that R. Contrasting were not rainout which elevated patternsamong size and life-historyclasses of a semi-aridshrub. protectedby shelters, FunctionalEcology 6: 482-488. nighttimetemperatures. DONOVAN,L. A., AND J. R. EHLERINGER.1994. Carbon isotope discrimination,water-use efficiency, growth, and mortalityin a natural Conclusions-Our findingsthat small plants sometimes had shrubpopulation. Oecologia 100: 347-354. lowerwater potentials, lower rates of gas exchange,and lower DUDLEY, S. A. 1996a. Differingselection on plantphysiological traits in wateruse efficiencythan large plants were directly opposite the response to environmentalwater availability: a test of adaptive Evolution50: 92-102. resultsexpected, given our observationsthat drought reduces hypotheses. DUDLEY, S. A. 1996b. The to selection on growthmore stronglyin large plants. Moreover,drought response differing plant physiologicaltraits: evidence for local adaptation.Evolution 50: 103- affectedwater potentials,gas exchange, and water use 110. in of efficiencyequally plants differentsizes. Resultssuggest EVANS,G. C. 1972. The quantitativeanalysis of plantgrowth. University that a decline in overall vigor occurs with plant age. of CaliforniaPress, Berkeley,California, USA. Explanationsfor size- and age-relatedphenomena, including FARQUHAR,G. D., J. R. EHLERINGER,AND K. T. HUBICK. 1989. Carbon responseto drought,require a betterunderstanding of whole isotope discriminationand photosynthesis.Annual Review of Plant plant assimilation,respiration, and hydraulicconductivity Physiologyand Plant Molecular Biology 40: 503-537. undera varietyof environmentalconditions. The abilityto FORSETH,I. N., ANDJ. M. NORMAN.1993. Modeling solar irradiance,leaf and In D. J. M. predicthow naturalpopulations will responddemographically energybudget, canopy photosynthesis. O. Hall, O. Scurlock, H. Bolhar, R. C. Leegood, and S. P. Long [eds.], to increasinglyfrequent extreme weather events relies on this and in a environment:a field information. Photosynthesis production changing and laboratorymanual, 207-219. Chapmanand Hall, London,UK. FRANCO,A. C., A. G. DE SOYZA, R. A VIRGINIA, J.F. REYNOLDS,AND W. G. LITERATURE CITED WHITFORD.1994. Effectsof plant size and water relationson gas exchangeand growthof thedesert shrub Larrea tridentata.Oecologia ACKERLY, D. 2003. Canopy gaps to climate change--extremeevents, 97: 171-178. and evolution.New 160: 2-4. ecology Phytologist GEBER, M. A., ANDL. R. GRIFFEN.2003. Inheritanceand naturalselection P. ANDP. T. 1998. affects of ANDERSON, D., TOMLINSON. Ontogeny response on functionaltraits. International Journal of Plant Sciences 164: northernred oak seedlings to elevated CO2 and waterstress. New S21-S42. 140: Phytologist 477-491. GEDROC, J. J., K. D. M. McCONNAUGHAY, AND J. S. COLEMAN. A. ANDL. F. DELPH.2001. Patternand evidencefor the ARNTZ, M., process: 1996.Plasticityin root:shootpartitioning: optimal, ontogeneticor of evolution photosynthetictraits in naturalpopulations. Oecologia both? FunctionalEcology 10: 44-50. 127: 455-467. GUREVITCH,J., S. M. SCHEINER,AND G. A. Fox. 2002. The ecology of BLUM, A. 1996. Crop responses to droughtand the interpretationof plants.Sinauer, Sunderland, Massachusetts, USA. adaptation.Plant GrowthRegulation 20: 135-148. HESCHEL,M. S., AND C. RIGINOS.2005. Mechanisms of selection for BRUNA,E. M., ANDM. K. OLI. 2005. Demographic effectsof habitat drought stress tolerance and avoidance in Impatiens capensis fragmentationon a tropical herb: life-tableresponse experiment. (Balsaminaceae). AmericanJournal of Botany92: 3744. Ecology 86: 1816-1824. HESCHEL, M. S., K. DONOHUE, N. HAUSMANN, AND J. SCHMITT. 2002. CASPER, B. B. 1996. Demographic consequences of drought in the Population differentiationand natural selection for water-use herbaceous perennialCryptantha flava: effectsof density,associa- efficiencyin Impatiens capensis (Balsaminaceae). International tions withshrubs, and plant size. Oecologia 106: 144-152. Journalof Plant Sciences 163: 907-912. CASPER, B. B., I. N. FORSETH,H. KEMPENICH, S. SELTZER,AND K. XAVIER. KNAPP, A. K., AND J. T. FAHNESTOCK. 1990. Influence of plant size on the 2001. Drought prolongs leaf life span in the herbaceous desert carbon and water relations of Cucurbita foetidissima HBK. perennialCryptantha flava. FunctionalEcology 15: 740-747. FunctionalEcology 4: 789-797. CASPER,B. B., I. N. FORSETH,AND D. A. WAIT.2005. Variationin carbon LARCHER,W. 2003. Physiologicalplant ecology. Springer-Verlag,New isotope discriminationin relationto plant performancein a natural York, New York, USA. populationof Cryptanthaflava. Oecologia 145: 541-548. LOPEZ, F. B., C. JOHANSEN, AND Y. S. CHAUHAN. 1996. Effectsof timingof July2006] CASPER ET AL.-STAGE-BASED DROUGHT RESPONSE 987

droughtstress on phenology,yield and yield componentsof short- maize plants in dryingsoil. Journal of ExperimentalBotany 36: durationpigeonpea. Journal of Agronomyand Crop Science 177: 1441-1456. 311-320. SOUTHWICK,S. M., ANDT. L. DAVENPORT.1986. Characterizationof water McLAUGHLIN, S. P. 1986. Floristicanalysis of the southwesternUnited stressand low temperatureeffects on flowerinduction in citrus.Plant States. Great Basin Naturalist46: 46-65. Physiology81: 26-29. MULLER,I., B. SCHMID, AND J. WEINER.2000. The effectof nutrient SPERRY,J. S. 2000. Hydraulic constraintson plant gas exchange. Forest 104: 13-23. availabilityon biomass allocationpatterns in 27 species of herbaceous Agriculturaland Meteorology J. U. G. R. ANDJ. P. 2002. Water plants.Perspectives in Plant Ecology, Evolution,and Systematics3: SPERRY, S., HACKE, OREN, COMSTOCK. 115-127. deficits and hydrauliclimits to leaf water supply. Plant, Cell & Environment25: 251-263. NAKAJIMA,Y., S. SUSANTO,AND K. HASEGAWA.1993. Influenceof water SPERRY,J. S., A. PERRY,AND J. E. M. SULLIVAN. 1991. Pit membrane stressin autumnon flowerinduction and fruitingin young pomelo and air-embolismin ageing xylem vessels of Populus trees (Citrus grandis (L.) Osbeck). Journalof the Japanese Society degradation tremuloidesMichx. Journalof ExperimentalBotany 42: 1399-1406. for HorticulturalScience 62: 15-20. C. A. 1995. A 10-yeardemographic study of rabbitbrush(Chry- NOBEL,P. S. 1987. Waterrelations and plantsize aspects of floweringfor Tovr, sothamnus nauseosus): growth, survival and water limitation. Agave deserti.Botanical Gazette 148: 79-84. Oecologia 101: 1-12. PALMER,T. N., AND J.RALSANEN. 2002. therisk of extreme Quantifying VESK, P. A., ANDM. WESTOBY.2003. Drought damage and recovery- seasonal eventsin a climate.Nature 415: 512- precipitation changing a conceptualmodel. New Phytologist160: 1-19. 514. WATKINSON, A. 1992. Plant senescence. Trendsin Ecology and Evolution PARTRIDGE,L., ANDN. H. BARTON.1993. Optimality,mutation and the 7: 417-420. evolutionof Nature 362: 305-311. ageing. WINKEL, T., J.-F.RENNO, AND W. A. PAYNE. 1997. Effectof the timingof AND FORSETH. lateral PEEK, M. S., I. N. 2005. Non-destructiveestimation of water deficit on growth, phenology and yield of pearl millet root distributionin an aridlandperennial. Plant and Soil 273: 211- (Pennisetumglaucum (L.) R. Br.) grown in Sahelian conditions. 217. Journalof ExperimentalBotany 48: 1001-1009. REYNOLDS, J. F., R. A. VIRGINIA,P. R. KEMP, A. G. DE SOYZA, AND D. C. WYKA, T. 1999. Carbohydratestorage and use in an alpine populationof TREMMEL.1999. Impact of droughton desert shrubs: effectsof the perennialherb, Oxytropis sericea. Oecologia 120: 198-208. seasonalityand degree of resourceisland development.Ecological ZANG, J.,AND W. J.DAVIES. 1987. Increasedsynthesis of ABA in partially Monographs69: 69-106. dehydratedroot-tips and ABA transportfrom roots to leaves. Journal SHARP,R. E., ANDW. J. DAVIES.1985. Root-growthand water-uptakeby of ExperimentalBotany 38: 2015-2023.