Differing Selection on Plant Physiological Traits in Response to Environmental Water Availability: A Test of Adaptive Hypotheses Author(s): Susan A. Dudley Source: Evolution, Vol. 50, No. 1 (Feb., 1996), pp. 92-102 Published by: Society for the Study of Evolution Stable URL: http://www.jstor.org/stable/2410783 Accessed: 28/06/2010 08:47 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=ssevol. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. 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 support@jstor.org. Society for the Study of Evolution is collaborating with JSTOR to digitize, preserve and extend access to Evolution. http://www.jstor.org Evolution, 50(1), 1996, pp. 92-102 DIFFERING SELECTION ON PLANT PHYSIOLOGICAL TRAITS IN RESPONSE TO ENVIRONMENTAL WATER AVAILABILITY: A TEST OF ADAPTIVE HYPOTHESES SUSAN A. DUDLEY1 Departmentof Ecology and Evolution,University of Chicago, Chicago, Illinois 60637 Abstract.-I used phenotypic selection analysis to test the predictionfromfunctionaland comparative studies of plants that smaller leaves and more efficientwater use are adaptive in drier environments.I measured selection gradients on leaf size and instantaneous water-use efficiency(a measure of carbon gain per unit water loss) in experimental in thefield.Linearand nonlinear populationsof Cakile edentulavar.lacustrisplaced intowet and dryenvironments as predicted.Water-useefficiency was selectedto be selectiondifferedsignificantly betweenthe two environments There was also higher,and leaf area was selectedtowarda small intermediate optimum,in the dryenvironment. selectionon water-useefficiency and leaf size, suggesting thattheoptimumleaf size significant positivecorrelational in thedryenvironment is greaterforplantswithhigherwater-useefficiency. In contrast, neitherleaf size norwateruse efficiency wereselectedin thewetenvironment, thoughlargerleaves resultedin greatervegetativebiomass.Path analysisof the linearselectiongradientsfoundthatwater-useefficiency affectedplantfitnessprimarily because it increasedvegetativebiomass,as suggestedby thehypotheses aboutthefunction of physiologicaltraits.These results in waterwerenotonlyconsistentwiththefunctional hypothesesbut also withtheobservedgeneticdifferentiation use efficiency and leaf size betweenwet and drysitepopulations. selectionanalysis,pathanalysis,water-use Keywords.-Adaptivephenotypic plasticity, Cakile,leafsize, multivariate efficiency. ReceivedAugust30, 1994. AcceptedApril25, 1995. The evolutionary responseto selectioncan be predicted bythephenotypic selectionin one generation andthegenetic variancesand covariances(Lande and Arnold 1983). Phenotypicselectionstudies,whichdescribetherelationship betweentraitsand fitness, can be used notonlyto predictthe evolutionof the traitsstudiedbut also as an empiricaltest To supportthehypothesis of adaptivehypotheses. thata trait is adaptivein a givenenvironment, demonstrating thatthe traitis correlated withfitnessin thatenvironment is notsufficient.The correlation of traitand fitnessmustalso be lessened or absentin an environment wherethetraitis notexpectedto be adaptive(Wade and Kalisz 1990). Manystudies have demonstrated strongnaturalselectionin thewild (Endler 1986), butfewstudieshave attempted to testevolutionary predictionsand adaptivehypotheses.Here I describea testof adaptivehypotheses forplantphysiologicaltraitsthat affectplantcarbonuptakeand waterloss. A large body of functionalanalyses and comparativestudiesprovideshypotheseson how selectionon thesetraitsshoulddependon environmental wateravailability(i.e., Ehleringer1975; Givnish1986,Cowan 1986). In thispaper,I describea fieldstudy in wetand of selectionon leaf size and water-useefficiency and comparethe resultswiththe predicdryenvironments tionsof adaptivehypotheses. In a companionpaper(Dudley in these traits 1996), I measurethe geneticdifferentiation betweenpopulationsfromwetand dryenvironments as well as the geneticcovariancematrixand comparetheseresults withthefieldstudiesof selection. The adaptivehypotheses I examinein thispaperarebased on thephysiologyof photosynthesis. Whenplantsopentheir stomates(poresin theleafepidermis)to allowcarbondioxide to diffuseintotheleafforphotosynthesis, waterdiffusesout waterloss is potentiallycostlyfor rapidly.Transpirational away morewaterthantheplantcan acplants.Transpiring quiremaycause droughtstress,whichcan compromisethe abilityto growand to acquirecarbon,and in theextremity is fatal (Givnish 1986; Schulze 1982). Thus, althoughincreasednet carbonacquisitiongives plantsmoreresources (Ehto compete,grow,and reproducein all environments leringer1975), theevolutionof traitsthataffectcarbonacin drierhabitatsbecausecarbon quisitionmaybe constrained gain is accomplishedat thecost of waterloss. One waythat is by reducingwaterloss plantsadaptto dryenvironments (Givnish1986). Waterloss can be reducedbyhavingsmaller leaf surface(Givnish1979) leaves to reducethetranspiring and thuslower and in some cases to reduceleaftemperature thewaterpotentialat theleafsurface(Nobel 1991). Another potentialadaptationis to change the relativerates of gas exchangeto maximizethecarbonassimilationto water-loss efficiency (Cohen 1970,Cowan ratio,definedas thewater-use 1986). These adaptivehypothesesare supportedby comparative in populations studiesshowinghigherwater-useefficiencies and speciesfromdrierenvironments (Gurevitchet al. 1986; and Cooper1988). Studies Kalisz and Teeri1986,Ehleringer strumarium selectionin Xanthium grownin a of phenotypic cultivatedfield(Farrisand Lechowicz 1990), Prunellavulgaris grownin the greenhousein highand low lightenvironment(Winnand Evans 1991), and Plantago lanceolata have foundsegrownin an old field(Tonsor,manuscript) beThe relationship lectionforhigherwater-useefficiency. tweenbiomassand water-use efficiency (measuredas carbon isotope ratio) has been exploredin both well-wateredand in a commongardenstudyon the environments water-limited desert shrub Chrysothamnusnauseosus (Donovan and Ehler- testof the hypothesisthat inger1994). However,a further and smallerleaves are selected higherwater-useefficiency to droughtstressis to compare 1 Presentaddress:Department of Biology,McMasterUniversity, because theyare adaptations and thestrength of naturalselectionon water-useefficiency 1280 Main St. West,Hamilton,OntarioL8S 4K1, Canada. 92 ? 1996 The Society forthe Study of Evolution. All rightsreserved. SELECTION ON PLANT PHYSIOLOGICAL TRAITS 93 Both siteshad onlysparsely to thebeach,facingnorthwest. distributed annual vegetation,and consequentlyplantsrewateravailabilitywas ceived full sunlight.Environmental determined by observation.The wet site containeda permanentseep of wateralongthebase of thedune.Waterwas thesummer, and fromtheseep throughout observedtrickling plantsneartheseep grewin a slurryof sand and water.The drysite lacked any watersourceotherthanthe lake itself. in exchange traits in Cakile edentula var. lacustris (Brassica- Naturalpopulationsof plantsalso indicateddifferences ceae), a succulentC3 beach annualfoundon sandybeaches favorableness betweensites,withverylargeplantsfoundat alongtheGreatLakes oftheNorthAmerica(Rodman1974). thewet site and smallplantsat thedrysite,suggestingthat This species is consideredto be subjectto edaphicdrought the plantsdo not reach the watertable. Nutrientmeasurebecause its beach sand substratedoes not retainwater,and mentsindicatedthatlevelsweresimilarand verylow forthe Lake Michiganpopulationsof C. edentulavar.lacustrismay twosites.The twopopulationsfromcontrasting habitatswere populationused in this study experienceseveraldays or weeks withoutrainfall.Popula- pooled into the experimental variationand ensurethat tionsof C. edentulavar.lacustrisaroundLake Michigangrow to increasetheexpectedphenotypic populationwas notmoreadaptedto one of in sites thatdiffergreatlyin wateravailabilitybecause of theexperimental rainfall, topography, seeps,andothersourcesofgroundwater.theexperimental environments. To testtheadaptivehypothesessuggestedby thephysiolog- Because the breedingsystemis believedto be predomiwere genI measuredphenotypic progenyforexperiments selectionon instanta- nantlyself-pollinating, ical literature, neous water-useefficiency and leaf size forplantsof known eratedby allowingplantsto self-fertilize. Maternalsibships in wa- fromthefield-collected plantswereraisedin thegreenhouse ancestryplaced intonaturalenvironments contrasting ter availability.These measurements allow me to ask the and allowedto selfpollinate.One plantfromeach maternal followingquestions:(1) Does selectionfavorgreaterwater- sibshipwas randomlyselectedto be the parentof the ex10 offspring fromeach in thedryenvironment Approximately generation. use efficiency thanin the wet envi- perimental ronment?(2) Does selectionfavorsmallerleaf size in the of the 23 originalfamilieswere assignedto each of two thanin thewet environment? treatments: wet and dry. dryenvironment leafsize in environments thatdiffer onlyin wateravailability (WadeandKalisz 1990). Measuresofnaturalselectioncannot provea causal relationship betweentraitsand fitness(Mitchell-Oldsand Shaw 1987; Rausher1992),butfinding thatthe associationsbetweentraitsand fitnesschange as predicted betweenenvironments wouldprovidestrongsupportfortheir hypothesized adaptivevalue (Wade and Kalisz 1990). Here I presenta studyof naturalselectionforleafand gas MATERIALS AND METHODS StudySpecies and Sites Field Environments tookadvantageof thenaturalpatchThe fieldexperiment iness in wateravailabilityfoundat one site. At the Mount genusnativetothesandybeachesoftheNorthAtlanticOcean Baldy site,a naturalseep has createda narrowstrip(from wet sand and adjoiningbodies of water.All membersof thegenusare 1-5 m in width)parallelto thebeach of extremely Therewas abundantavailable thewetenvironment. succulentherbaceousannuals.Leaves of Cakilespecieshave providing have waterduringthe entireseason near the seep. One transect been reportedto rangefrom0.5-1.5 mmin thickness, stomateson bothleafsurfaces,and haveundifferentiated pa- was laid outalongthelengthof theseep for50 m. The slope 4 m above the seep providedthe dry envirenchyma(Rodman 1974). Physiological measurements approximately made on greenhouse-grown Cakile maritimafromnorthern ronment.The sand on the slope was well drained,and it Californiaindicaterelativelyhighphotosynthetic capacities receivedwateronlyfromrainfall.The wetanddrytreatments and bothhad soil thatwas 99.9 % coarseand finesand,0.1% silt withmaximumphotosynthetic ratesof 37 pwM-m-2-s-' In the conductancesof 12 mm-s-1(De Jong1978). Cakileedentula and clay particles,and had similar,low soil nutrients. twoparalleltransects(each of 25 m) were var.lacustrisis an indeterminately annualwithcon- dryenvironment, flowering 2 m above the otherrunning siderablevariationin flowering schedule.Breedingsystems laid out, one approximately In bothenvironof Cakile species rangefrompredominantly self-pollinatingparallelto thebeach and theothertransect. in areas fromperennials away transects were placed ments, topredominantly Flowers are on born outcrossing. elongating terminalracemes.Fruitsconsistof two segments,theupper and shrubsgrowingon the seep such thattheexperimental to only Limitingtheexperiment being deciduous and the lower remainingattachedto the plantsreceivedfullsunlight. raceme.Each segmentcontainsusuallyone seed,withupper two siteswas necessarybecause of practicalconsiderations oftheenvironmental water seeds beingslightlylargerthanlowerseeds. Cakileedentula butdidresulta lack ofreplication However,althoughthe sites differed var. lacustrisis nativeto the GreatLakes and is the only availabilitytreatment. theyweresimilarin nutrient memberof thegenusfoundon freshwater beaches.It is re- markedlyin wateravailability, gardedas a primarily autogamoustaxon,withsmallflowers levels and in lightavailability. Seeds of theexperimental populationwereinducedto gerand highratesof spontaneousfruitset (Rodman1974). and removingtheirseed Seeds were collectedfromthe Indiana Dunes National minateby soakingthemovernight Lakeshorefrom23 randomlychosen naturallypollinated coats. Seeds were plantedMay 7, 1990 into50:50 promix plants,12 froma drysite (WestBeach), and 11 froma wet Turfacein plugtraysand lightlycoveredwithsoil mix.The site(MountBaldy) approximately a 26-kmdistancefromone plug trayswerekepton dryheat undernaturallightingfor in 1 wk and thenin the greenhouseforthe 2-day week. All another.These two populationsare further characterized within4-8 days. The seedlingswerethen Dudley(1996). Bothsiteslay alongthebase ofdunesparallel seeds germinated The study plant, C. edentula var. lacustris, belongs to a 94 SUSAN A. DUDLEY transferred to a cold frameto acclimate.Later,theywere labeled withbird-bandtags and transplanted to thefieldon June12, 1990. At thattime,theyhad twoto six trueleaves. The seedlingswereplantedin randomizedorderin thetransects every20 cm. Ratherthanplantinga straightline of plants,theseedlingswerealternately offseteither5 cmabove or 5 cm belowthetransect to presenta slightlymorenatural appearance. humidity spans.The relativehumidity of0 was verified using driedair and the PPM CO2 was verifiedby measuringair passed through soda limeto removeCO2. Leaf temperatures in thechamberwas estimatedfromtheenergy-balance equation (Parkinson1984; AnalyticalDevelopmentCompany 1992). Boundary-layer resistancesforCakileleaves in a Parkinsonleaf chamberwere estimatedon filter-paper mimics (Parkinson1985; AnalyticalDevelopmentCompany1992). Because boundary-layer resistancewas correlatedwiththe size of the "leaf," a regressionequationestimating boundTraits ary-layerresistanceas a functionof leaf size was used in Leaf size and gas-exchangerates were measuredin late subsequentcalculations. summeron survivingplants,whichwereproducingflowers Photosynthetic ratesand stomatalconductanceswerecaland fruits.For each plant,one recentlyproducedfullyex- culatedaccordingto von Caemmererand Farquhar(1981). panded leaf was sampled,and measuresmade of net pho- Measuresof gas exchangewerecovariatecorrectedfordate tosynthetic rate (assimilationof CO2 in FM C02-m-2s-1), and time (Winn and Evans 1991; Harris 1975; Farrisand therateat whichthe leaf surfaceacquirescarbonforpho- Lechowicz1990). Lightlevelandtemperature werenotfound stomatalconductance(moles H20-m-2s-1), the to explainanysignificant tosynthesis; variancewhendateand timewere leafsurface'sinfluence on therateof waterloss; andthearea includedin themodel.Plantsthatappearedto be wiltedor and weightof thedried,pressed,leaf. Instantaneous water- to have extremely aberrant water-useefficiencies wereomituse efficiency was thencalculatedas theratioof photosyn- ted fromsubsequentanalysis(N = 3). theticrate to stomatalconductance.Such single-leafmeaA sensitivity analysisconductedon thedatafoundthatthe sures of gas exchangeand leaf size are heritable(Dudley conclusionsabouttherelativedifferences amongindividuals 1996). Because gas-exchangeratesweremeasuredon a dry- and populationswere robustto the potentialerrorin calileaf area basis whichis extrapolatedto give wet-leafarea brationoftheequipmentor estimation of theboundary-layer measures(see below),theseresultsaremoreaccurateas com- conductance. parativemeasuresamongplantsratherthanas estimatesof I harvestedplantson September15, 1990 to avoid loss of fieldgas-exchangerates.Because fewplantsdied aftermea- the studypopulationto burial by sand, waves fromlarge the data were used foranalysisof fecundity se- storms,or movementof beach sand offshoreby changesin surement, lection. currents. At thistime,mostof theplantsin thedryenvironMeasurements ofphotosynthesis andstomatalconductance menthad ceased flowering, but the plantsin the wet enviweredone on August23 and 24, 1990, usingan Analytical ronment variedfrommostlyflowering to senescent.The meaDevelopmentCompanyLCA-2 portablephotosynthesis sys- surements I obtainedestimateplantfitnessesgivenan early tem(Long and Hallgren1985; Field et al. 1989). Measure- end to thegrowingseason.For an indeterminately flowering mentsweremade underambientlight;bothdays wereclear annual,therankingof fitnessesin a populationmaydepend withoccasional clouds. The mean lightlevel was 990 pLM on season lengthbecausehighearlyreproduction maycome photons-m-2-s'.Thoughlightlevels variedfrom330-1280 at thecost of loweredvegetativegrowthreducingresources 90% of measureswere made at light forlaterreproduction FLMphotons-m-2-s-1, (Geber 1990). levels above 810 FLMphoton-m-2-s-1. Cuvettetemperatures The collectedplantswere air driedat roomtemperature. variedfrom25-33?C. Plantsweremeasuredbetween10:00 Sand was scrubbedofftheplantswhennecessary.The airA.M. and 3:00 P.M. in a predetermined orderthatwas random driedplantswerepartitioned intovegetativebiomass(stems withrespectto environment, positionwithinenvironment,and largeroots)and reproductive biomass(fruits)and then and theplant'sparentage.Gas-exchangetraitscould notbe weighed. measuredon 50% of the survivingdry-siteplantsbecause I used the reproductive biomass allocatedto fruitsas a theywere moribundor defoliated.Wet-environment plants fitnessmeasurebecause it is the hypothesizedcausal link thatweretemporarily inaccessiblebecauseofsandmovement betweencarbonacquisitiontraitsand fitness(Fig. 1). Most weremade on fruitscontaintwoseeds (Rodman1974); and because Cakile (N = 38) werenotmeasured.Measurements all dry-environment plantsthatpossessed leaves (N = 62) flowersindeterminately, higher reproductivebiomass is and on 102 wet-environment plants.For each plant,one re- largelytheresultofmorefruits rather thanlargerfruits. Using centlyexpandednewleafwas detachedand measuredwithin fruitbiomass for fitnessmeasuresmaternalinvestment in 90 sec. Detached leaves of this species maintainconstant reproduction, but ignoresthe potentialforvariancein biogas-exchangeratiosformorethan5 min(Dudley,pers.obs.). mass allocatedper fruitand per seed. The sampleleaves were subsequently pressedand driedto providea measureof thesize of a recentfullyexpandedleaf Data Analysis and leaf drymass. Areas of dryleaves wereestimatedwith a LI-COR leaf-areameter.Wetanddryleafareasfromleaves Means and standarddeviationswere calculatedfor leaf sampledfromseveralgreenhousegrownplantsprovideda size,photosynthetic water-use efrate,stomatalconductance, conversionfactorforcalculatinggas-exchange rateson a wet- ficiency, vegetative biomass,andfruitbiomassin thewetand leaf area basis (wet area = 1.90 dryarea in chamber). the dry environments forthe subsetof plantson whichI The LCA-2 was factory calibratedfortheCO2 andrelative measured gas exchange (PROC MEANS, SAS Institute SELECTION ON PLANT PHYSIOLOGICAL Vegetative biomass Reproductive biomass 95 TRAITS - numberof fruits numberof seeds Leaf traits 1. Path model of the hypothesized relationships between physiological traitsand fitness.Only those paths indicated by thick lines will be measured in this study. FIG. in regressioncoefficients be1989). Thedifference inmeansbetweenthestudypopulations ease in testingfordifferences is presentedanalogouslyto tweenenvironments. grownin thetwo environments high withextremely One datapointin thewetenvironment thegeneticdifferences betweenpopulations(Lande and Arlarge effect nold 1983) as thedifference in meansfora traitbetweenthe fitnesswas foundto have a disproportionately dividedbythepooledphenotypic standard on theregressionanalysisand was omittedas an outlierfoltwoenvironments, lowingthe criterionof Sokal and Rohlf(1981). In the dry deviationforeach environment. which include both environment, the nonlinearselectionanalysiswas runwith Standardizedselectiondifferentials, directand indirectselectionon a trait,were calculatedfor and withoutone outlyingdatapointwithverylargeleafsize, selectiongradientsreas thecovariancebetweenthestandardized and the stabilizingand correlational each environment traitand fitness(PROC CORR) (Lande and Arnold1983). mainedsignificant. wereconsideredsignificantly different I performed a path analysisto examinethe linearcomSelectiondifferentials dependencybetweenthewater-use coefficient betweenthetrait ponentsof thefunctional from0 ifthePearsoncorrelation from0 (Lande and efficiency and leaf size and fitness(Arnold1983). Path diand relativefitnessdiffered significantly Arnold1983). Selectiondifferentials forthesametraitin the agramscan be used to factorlinearselectiongradientson traitsintotheeffectof a traiton some aspectof differentphenotypic two environments were consideredsignificantly Pearsoncorrelation co- performance fromeach otherif thez-transformed (e.g., leaftraitson vegetativebiomass),and the traitson fitness(e.g., vegetativebioefficients significantlyeffectof performance betweenthetraitand fitnessdiffered fromeach other(Sokal and Rohlf1981). Phenotypiccorre- mass on fruitbiomass) (Arnold1983). I used a pathmodel to affect lationsamongtraitswere obtainedfromthePearsoncorre- thatallowed forleaf size and water-useefficiency fitnessthroughvegetativebiomassand thatallowedforleaf lationmatrixof traits(PROC CORR). to affectfitnessvia an alternate efficiency To testthe adaptivehypothesesaboutdirectselectionon size and water-use I calculatedlinear and path (Fig. 1; Ehleringerand Clark 1988). Path coefficients leaf size and water-useefficiency, nonlinear(stabilizingand correlational)selectiongradients wereestimatedusingPROC GLM to obtaintheappropriate variablesandrelon thestandardized followingLande and Arnold(1983). Directionalselection regressioncoefficients gradientswereobtainedfroma linearregressionof relative ativefitness.The sumof all pathsleadingfromtheleaftraits fitnesson the traits,and quadraticselectiongradientswere to relativefitnessequals the directionalselectiongradients obtainedfromthenregressing relativefitnesson linearand obtainedabove (Li 1975; Arnold1983). quadraticterms(PROC GLM in SAS). I estimatedselection efficiency and leafsize to avoid gradientsonlyforwater-use RESULTS from"spuriouscorrelations"(Sokal and Rohlf collinearity had greaterfruit The plantsgrownin thewetenvironment func1981) caused by includingtraitsthatare mathematical The plants tionsof each other.Thus,photosynthetic rateand stomatal biomassthandid plantsin thedryenvironment. and had veryhigh conductancecould notbe includedin thesameregressionas grewmuchbiggerin thewetenvironment (Table 1). In thewetsite,21% oftheoriginal water-use efficiency, andbecauseleafsizes weresmallerthan fruitproduction thecuvettesamplingarea,thecalculationsof photosyntheticplantswerewashedaway because of shiftsin sand location rateand stomatalconductanceincludedleafsize as a divisor. and storms.Of theremainingwet-siteplants(N = 183), an In the fromhumantrampling. is estimatedindependently of additional23% died,primarily However,water-useefficiency in drysite,50% oftheplantsdied,eitherfromapparentdrought leaf size. I testedfordifferences betweenenvironments Although standardizedselectiongradientsthrougha testof heteroge- stress,insectdamage,or fromhumantrampling. neityof slopes in an analysis of covariance(ANCOVA) the sites were near each other,the presenceof insectsap(PROC GLM). Because the residualsfromthe regression pearedto varybetweenthesites.More herbivoreswereobI also calculatedthe servedin the dry site, and possible pollinatorswere more analysiswerenotnormallydistributed, maybe theresult levelsofthestandardized coefficients prevalentin thewetsite.Such differences regression significance in theplantsratherthanthesitesthemselves. fromjackknifeestimatesofthevariances(FreeStat,Mitchell- of differences amongfamilies in both Because mortality did notdiffer significantly Olds 1989). I foundthatsignificance changedslightly = directionsbutthe overalltrendwas similar,suggestingthat or betweenpopulationseitherin thedryenvironment'(P betheparametric (P = 0.58), differences analysiswas robustto the lack of normality 0.22) or in thewetenvironment in traitsindicatetheaveragephenotypic of theresidualsin thisdata set. I therefore presentthe sig- tweenenvironments nificancelevels fromtheparametric analysisbecause of the plasticityforthepopulation. 96 SUSAN A. DUDLEY 1. Mean traitvaluesin each environment withthestandarddeviationsin parentheses. Predicteddifferences in traitmeansbetween in traitmeansbetweenenvironments environments are fromtheselectionanalysis(Table 4). The observeddifference was standardized by a pooled phenotypicstandarddeviationto facilitatecomparisonsamongtraits.Observeddifferences thatagree withthepredicted N = 101; dryenvironment, N directionare indicatedin bold. Probabilitiesare takenfroman analysisof variances.Wetenvironment, 60. TABLE Dry environment Trait Photosynthetic rate(A) P.mol/m2/s Stomatalconductance(g) mol/m2/s Water-useefficiency (A/g) means 24.0 (9.75) 0.55 (0.24) 45.9 0.54 (0.41) 1.20 (0.99) 0.36 (0.46) Leaf size cm2 Vegetativebiomass grams Reproductive biomass grams P < 0.10; * P < 0.05; ** P < 0.01; * means 27.5 (8.63) 1.10 (0.40) 27.4 (11.50) P'mol/mol Wet environment (13.56) 1.89 (1.21) 55.63 (49.57) 13.48 (14.91) Predicted differences Observed differences (ZDRY - ZWET)/IP Dry = wet 0.37* Dry < wet 1.77*** Dry > wet -1.44*** Dry < wet P < 0.001. mearate,a short-term of carbonuptake,and photosynthetic sureofcarbonuptake,werepositivelyassociatedwithfitness, hypotheses.The associa resultpredictedby thefunctional beationsbetweenstomatalconductanceand fitnessdiffered hypotheses: tweenenvironments as predictedby functional andthushigherrates plantswithgreaterstomatalconductance of waterloss per unitleaf area were less fitin the dryenvironment, and therewas no associationof stomatalcon(Table 2). The asductanceand fitnessin wet environment and fitnessdiffered sociationsbetweenwater-useefficiency hypotheses: byfunctional as predicted betweenenvironments whichgainedmorecarplantswithhigherwaterefficiency, bon forthe same waterloss, weremorefitin thedryenviand fitronment, but no associationof water-useefficiency larger Unexpectedly, ness existedin the wet environment. leaf sizes werenotassociatedwithhigherfitnessin thewet environment but were associatedwithhigherfitnessin the (Table 2). This resultwas notpredictedby dryenvironment the hypothesisthatreductionof waterloss throughhaving smallerleaves is adaptivein drierenvironments. The phenotypic correlations amongtraitswerestrongand (Table 3, Fig. 2). Photosyndependedon the environment showeda strongpositive stomatal conductance thetic rate and TABLE 2. Standardized directionalselectiondifferentials in each in theirrelationship was the wet environment, but correlation environment. from0 in each environment Significancedifferences Stomatalconductancewas and thet-testof differences betweenselectiondifferentials on the weakerin the dryenvironment. same traitin thedifferent environments are takenfromsignificance negativelycorrelated withleaf size in thedryenvironment. levelsforthecorresponding correlation coefficient betweenthetrait Photosynthetic withleaf size. Leaf rates were notcorrelated = N 57, wet environment N and relativefitness.Dry environment in with water-use efficiency correlated size was positively = 99. bothenvironments (Table 3). Standardizedselection I performed a multivariate fecundity selectionanalysison differentials in to size each environment efficiency and leaf water-use s/U estimatethedirectselectionon each trait(Table 4, Fig. 3). Wet Dry Because of thehighmortality beforetraitsand fitnesswere Trait environment t-value environment of selection measured,thepowerto testforthesignificance rate 0.41* Photosynthetic 0.18* 0.666 only signifIn environment, the low. the wet gradients was Stomatalconductance 0.10 2.24* -0.33* icant selectiongradientwas weak stabilizingselectionon Water-useefficiency 0.66*** 3.23** 0.01 Leaf size 0.54*** 0.00 2.56* In thedryenvironment, therewas strong efficiency. water-use 0.39*** Vegetative biomass 0.89*** 2.06* stadirectionalselectionforincreasedwater-useefficiency, * P < 0.05; ** P < 0.01; **P < 0.001. leaf size, and positive bilizingselectionforan intermediate In these results,extrapolatedfromdryleaf areas (Table 1), and in greenhouse measures with fresh leaf areas (Dudley, pers. obs.), C. edentula var. lacustris exhibited moderately high photosyntheticrates and very high stomatal conductances (Korner et al. 1979). High gas-exchange rates are thoughtto be typical for amphistomatousthick leaves found in high light environments(Mott et al. 1982). As predicted by functionalhypothesesforadaptive plasticity,plants grown in the dry environmenthad smaller leaves than in the wet environment(Table 1). Contraryto adaptive-plasticitypredictions, stomatalconductances were much higherin the dry environment.Though photosyntheticrates were also somewhat higherin the dry environment,the water-useefficiency was lower in the dryenvironment.Specific leaf weight (ratio of leaf weight to leaf area) was higher in the dry site (dry 0.0123g/cm2,wet = 0.0074 g/cm2,P < 0.0001), and leaves were noticeably thickerin the dry environment. The fecundityselection differentials,which include direct and indirect selection, differedsignificantlybetween environmentsfor all traitsbut photosyntheticrate (Table 2). In both environments,vegetativebiomass, a long-termmeasure SELECTION ON PLANT PHYSIOLOGICAL 97 TRAITS 3. Phenotypic correlations among physiological traits.Wet environmentcorrelations are in the lower diagonal, dry environment correlations are in the upper diagonal. TABLE Stomatal conductance Photosynthetic rate rwet/rdry Photosyntheticrate Stomatal conductance Water-use efficiency Leaf size 0.23t 0.82*** 0.08 0.11 -0.37*** -0.08 Water-use efficiency 0.57*** -0.58** 0.22* Leaf size 0.14 -0.48*** 0.65*** t P < 0.10; * P < 0.05; ** P < 0.01; *** P < 0.001. correlationalselection between water-useefficiencyand leaf size (Table 4, Fig. 3). As the fitnesssurface (Fig. 3) shows, the correlational selection caused the optimum leaf size to increase with increased water-use efficiency.There was no significantdirectional selection gradientin the dry environment for leaf size. The significantdirectional selection differentialfor leaf size in the dry site (Table 2) appears to be an indirectresponse to selection on water-useefficiencyand to the correlational selection on leaf size and water-use efficiency. Because of the small sample size, the power to test for differencesbetween environmentsin selection on water-use efficiencyand leaf size was low. Nonetheless, the differences in directional selection on water-use efficiency,stabilizing selection for an intermediateleaf size, and positive correlational selection between water-useefficiencyand leaf size between the wet and dry environmentswere all highly significant(Table 4, Fig. 4). These differencesare consistent with functionalhypotheses that suggest that environmental water availability causes the selection on leaf size and wateruse efficiency. Path-analysis models using leaf size and water-use efficiency as carbon uptake traitswere used to analyze the di10090 0 0 O 80- o m3 0 O Q > Cc _70- 0~~~ D 60 00 E1 W CD q 0 * 00 go o~ 0 0 E ~~3~~30 ~00 20 100 0 3 0 20- 0 o 0 eDry (Fig. rectionalselectiongradientsin the two environments 5). The resultsagreedwiththe expectationthatleaf traits affectfitnessthroughtheireffecton vegetativebiomass. Greaterleaf size increasedvegetativebiomass in the wet higherwater-useefIn the dryenvironment, environment. ficiencyled to greatervegetativebiomass.The analysesdid directeffectof water-useefshow a marginallysignificant ficiencyon fruitbiomass in the dry site. In bothenvironments,highervegetativebiomass caused increasedfitness, buttheincreasein fruitbiomasswithvegetativebiomasswas greater(P < 0.05, analysisofcovariance)in the significantly (Fig 5). thanin thewetenvironment dryenvironment DISCUSSION selection in fecundity This studyexaminedthedifferences Significant on leaftraitsbetweenwetand dryenvironments. were found in selectionbetweenenvironments differences despitetherelativelysmallsamplesizes. The results,as predictedby physiologyof carbonacquisitionand waterloss, the importanceof net carbonuptakeforplant demonstrate fecundityand show thatfecundityselectionon traitsthat as predictedbetweenwet and dry affectwaterloss differed environments. As predicted,selectionforgreaterwater-use comparedwith was foundin thedryenvironment efficiency leaf thewetenvironment. Selectionfavoredan intermediate whereaslargerleaves wereassize in thedryenvironment, sociatedwithgreatervegetativebiomassin thewetenvironment.This studycomparedselectionin two stronglyconenvironments forone species butfoundresultsthat trasting Further studiesin other supportadaptivehypotheses. strongly are necenvironments, species,replicatedacrosscontrasting essaryto see if these resultscan be generalized.In interpretingtheseresults,I considerthreequestions:thesupport becausal relationships of the resultsforthe hypothesized fortheevolutionof tweentraitsand fitness;theimplications and the description thetraitsin the different environments; of theadaptivevalue of thetraitsin each environment. Causal Interpretationof Selection Analyses in selectionobservedbetweenwetand dry The difference environments providesdirectevidenceforthe selectiveimpact of wateravailabilityon carbonacquisitiontraits(Wade andKalisz 1990). The pathanalysisoftheselectiongradients providesa testof thehypothesesaboutwhywater-useeffiLeafSize (cm2) ciencyand leafsize are selected(Arnold1983). It supported FIG.2. The relationship andleafsize efficiency betweenwater-use and leaf hypothesisthatwater-useefficiency in bothenvironments. A nonlinearanalysisofcovariancefoundthat the functional becausetheyaffectnetcarbonacquisition. the regressionsof water-useefficiency on leaf size did differbe- size affectfitness in thedryenselectionon water-useefficiency tweenenvironments The stronger (P < 0.001). 0- o o Wet 98 SUSAN A. DUDLEY Significancelevels fortheselectiongradientsare 4. Linearand nonlinearstandardized selectiongradientsin each environment. takenfromthejackknifeestimatesof the varianceof the standardizedselectiongradients.The overallmodelforthedryenvironment is not statistically significant. The F-values forthe testof is significant at P < 0.0002. The overall model forthe wet environment in selectiongradientsbetweenenvironments data. differences are froman analysisof covarianceson thepooled standardized TABLE Trait Dry environment I' Linear selection gradients Water-useefficiency Leaf size tP Wet environment F-value P' 0.51** 0.20 0.01 0.00 Nonlinearselection gradients yI y' Water-useefficiency*water use efficiency Leaf size*leafsize Water-useefficiency*leaf size 0.12 - 1.13*** 0.71** -0.22* 0.02 0.12 6.93** 1.04 2.20 13.88*** 5.31* < 0.10; * P < 0.05; ** P < 0.01; *** P < 0.001. success was foundto be largely atum,earlyreproductive butlater by timingof allocationto reproduction, determined associatedwithvegetative successwas strongly reproductive then growth(Geber1990). If Cakilefollowsa similarpattern, in yearswitha longgrowingseason,selectionforlargerleaf butselection size wouldbe predictedin thewetenvironment, on water-useefficiency wouldstillbe expectedto be greater shrub Chrysothamnusnauseosus, Donovan and Ehleringer in thedryenvironment thanthewet environment. (1994) found,contrary to the resultsin thisstudy,thatthe phenotypiccorrelationbetweenwater-useefficiency (meaPredictingEvolution suredas carbonisotoperatio) and vegetativebiomass was siblingsfrom23 familiesfromtwo positivein well-wateredtreatments and 0 in water-limited In this experiment, Thoughthe in each treatment populationswereplantedin bothenvironments. treatments, thoughthegeneticcorrelation in each environhighmortality plantssuffered was positive.Theysuggestthatunmeasured, correlated traits experimental betweensourcepopulationsor did notdiffer ment,mortality such as carbonallocationmayexplaintheirresults. thisstudyprovidesstandardized The pathanalysisalso providesa way of evaluatingpo- amongfamilies.Therefore, selectiongradientson thesame populationin diftentialselectionin longerseasons. Selectionon a traitthat fecundity and can be used to predicthow that affectsvegetativebiomassdependson thestrength of selec- ferentenvironments tionon vegetativebiomass,as theresultsheredemonstrate. populationwouldevolvein responseto selection.One caveat theresponseto selectionis thathighmortality For one indeterminately flowering annual,Polygonum arun- in predicting vironment was caused by greatereffectsof water-useefficiencyon vegetativebiomass,a greatereffectof vegetative and a marginally biomasson fitness, significant directeffect on fitnesscomparedwiththewetenof water-useefficiency vironment. The directeffectmaybe explainedbyplantswith higherwater-useefficiency ratesbeing more successfulin provisioning fruits.In a commongardenstudyon thedesert WetEnvironment DryEnvironment 4 FITNESS 5 FII JESS 4 33 2 ~~~~~~~~~~2 0 60 *. 8 3 3. Quadraticunstandardized selectionsurfacefor water-useefficiency(WUE) and leaf size (LEAF) in the wet and the dry environments. Note thattheaxes have different scales in thetwo figures. FIG. SELECTION ON PLANT PHYSIOLOGICAL 99 TRAITS selectiongradients.However,this studydid not have the powerto identify smallerselectiongradientsand differences in selectionbetweenenvironments, whichcan haveimportant * evolutionary implications. The observedselectiongradients indicatethatpopulations in each environment will be under different selectionregimes. 4* The significant difference in directionalselectionon water,tL * use efficiency suggeststhatforthe same populationwaterU23 00 0o use will efficiency be selectedto increasein thedrysiteand 0 0) 0 0 toremainnearthemeaninthewetsite.The selectiongradient analysisshowsthatleaf size was selectedto be at an inter0 CE C mediateoptimumin thedrysite.The lack of significant directional selection on in leaf size the dry environment implies b 1 *e ~ thattheoptimumleaf size in thedryenvironment does not cz differsignificantly fromthemeanleaf size in thedryenvi~ Wae Use Efiiec 0 (A/g 0 In contrast, ronment. thoughtherewas no significant selec0 tionon leaf size in the wet environment, the path analysis 0 ~~~~~(0 suggeststhatbecauseincreasedleafsize was associatedwith 0 20 40 60 80 100 increased vegetative biomass,largerleafsize maybe selected WaterUse Efficiency (A/g) in some years.These resultssuggestthatforthesame pop(,umoles/mole) ulationleaf size will be selectedto remainnearthemeanin to increasein thewet site. FIG. 4. The relationship betweenrelativefitnessand water-use thedrysite and potentially The intermediate optimumforleaf size in the dryenviin bothenvironments. efficiency ronment dependedon water-use Plantswithlower efficiency. water-use weremorefitiftheyhad smallerleaves, efficiency weremorefitif occurredbeforethephysiologicaltraitsweremeasured.The and plantswithhigherwater-useefficiency mortality, whichmayhave been an episode of survivorship theyhad largerleaves. Such correlational selectionmay act selection,was greatestin thedryenvironment. As a conse- as a forceof selectionon thegeneticvariancesand covariquence,thestatisticalpowerfortestingthehypotheseswas ances thatconstrainevolution(Phillipsand Arnold1989). low. The extremecontrastin wateravailabilitybetweenthe These resultssuggestthatselectionwas actingat the level environments did increasethe abilityto finddifferences in of the leaf on the balance betweenleaf carbonuptakeand * Dry o Wet 5- Vegetative DRY Environment 57 n=57 ~~~~~~~~~~biomass 0.06fitness 0.1 leafsize 6 =0.65*** ~~~~~0.36*/ / 0.74*** Relative 0.25 wateruse efficiency WET Environment n=98 0.20* Vegetative 0.44** bioassRelative ~~~~~~~~~~~~fitness leafsize wateruse efficiency FIG. 5. Pathanalysisof standardizedlinearselectionon leaf size and water-useefficiency. Significant paths(P < 0.05) are indicated by thicklines and bold coefficients. *P < 0.05, **P < 0.01, ***P < 0.001. 100 SUSAN A. DUDLEY waterloss. The correlationalselectionmay reflectthe de- in drierenvironments (e.g., Ehleringerand Cooper 1988; pendenceof whole-plant carbonacquisitionon multiplein- Donovan and Ehleringer1994) thoughdecreasesin waterterdependent leaf traits,whichhas been long recognizedby use efficiency withseasonalincreasesin droughtstresshave physiologists(e.g., Ehleringerand Clark 1988). Thus, this been observedin some species (DePuit and Caldwell 1975; resultfurther supportsthe argument thatstudiesof natural Smedleyet al. 1991). In thisstudy,thefunctional arguments, selectionmustbe informedby underlyingbiology of the the field-selection beresults,and the geneticdifferences studytraits. tweenpopulationsall suggestthatCakile exhibiteda malThe differences betweenselectiongradients in thewetand adaptiveplasticresponseto the environment. Such an apwereconsistentwitha geneticanalysisof parently dryenvironment maladaptiveplasticresponsesuggestseitherthatthe betweenthewetsiteand drysitepopulations. evolutionof adaptiveplasticityin water-useefficiency differentiation may A companionstudy(Dudley 1996) foundthatthepopulation be constrained or thattheexpression by geneticcorrelations fromthedrysitehad higherwater-use efficiency and smaller of adaptivetraitcombinationsmay be constrainedby the leaves comparedwiththepopulationfromthewet site.The phenotypic correlations amongtraits(Scheiner1993). of thetraitssuggeststhatdifferentiation One phenotypic geneticarchitecture theadaptive correlation thatmayconstrain betweenpopulationsfoundin thediffering is plasticityof physiologicaltraitsis thatbetweenwater-use environments constrained by thelowergeneticvariationforwater-useef- efficiency and leaf size. Water-useefficiency was positively ficiencyand thepositivegeneticcorrelation betweenwater- associatedwithleaf size, because stomatalconductancewas and leaf size (Dudley 1996). use efficiency negativelyassociatedwithleaf size. As Figure2 shows,the water-useefficiencies and leaf sizes of plantsin bothenviWhether ronments appeartofollowa curvilinear relationship. Describing Adaptation thiscorrelation a functional is notknown reflects relationship Measuringtheadaptivevalue of traitsrequirescomparing (Bhagsariand Brown 1986). It is plausiblethatsuch a corthefitnesssurfacesfortheunstandardized traitsin each en- relationcould be theresultof thepositivecorrelational sevironment, ratherthanthe standardizedselectiongradients. lectionforwater-useefficiency and leaf size (Philips and However,because plasticitybetweenthe two environmentsArnold 1989; Arnold 1992). Nonetheless,fora given leaf of all traitsbut photosynthesis was extreme,the measured size,thedry-site than plantshad a lowerwater-use efficiency selectiongradients areforplantsshowingverydifferent phe- the wet-siteplants(Fig. 2), suggestingthatthiscorrelation and leaf size (Fig. 2) and is not sufficient notypesforwater-useefficiency in to explainthelowerwater-useefficiency different correlations betweenthetraits(Table 3). Therefore, thedry-environment plants. thefitnesssurfacescannotbe statistically comparedbecause Anotherhypothesisforthe greaterstomatalconductance theyare valid over different rangesof the traits,but some andhigherphotosynthetic ratesseenin dry-site plantsis suginferences maybe made. gestedbythesignificantly higherspecificleafweightandthe The lack of significant directionalselectionon leafsize in observationthatleaves were thickerand moresucculentin thedryenvironment, coupledwithstabilizingselection,im- the plantsgrownin the dry site. Specificleaf weighthas did oftenbeen observedto be positivelycorrelatedwithphotoplies thattheoptimumleaf size in thedryenvironment not differsignificantly fromthe mean leaf size in the dry syntheticrate (Bhagsari and Brown 1986). A biophysical environment. Leaves ofthedryenvironment plantsweresig- constrainton thickerleaves is thatgreaterconductanceis smallerthanin thewetenvironment. nificantly Thoughleaf needed to permitCO2 diffusionto the interiorof thicker size did not affectfitnessin the wet environment, larger leaves (Mottet al. 1982). Greatersucculencein thedry-site leaves led to increasedvegetativebiomass.These resultsare plantsmay permitlong-term waterstorageand thusbe an withthehypothesis thata smallerleafsize is adap- adaptationto drought,but this hypothesisrequiresfurther consistent tivein thedrysite,and a largerleafadaptivein thewetsite, studiesof thephysiologyand ecologyof succulentleaves in in agreement withfunctional predictions. C3 species. The finding thatwater-use was understrongposefficiency A weaknessof selectionanalysesis thattheyare based itivedirectional selectionin thedrysiteand weakstabilizing upon correlationsamong traits(Mitchell-Oldsand Shaw selectionin the wet site agreeswithfunctionalhypotheses 1987; Wade and Kalisz 1990). Such correlations mayresult thathighwater-useefficiency is morecriticalwhenwateris froma causal relationship but betweenthetraitsand fitness, in averagewater-useeffi- theymay also resultfromenvironmentally limiting.However,the difference inducedcovariciencybetweenenvironments suggestsan alternatehypoth- ance betweenthetraitsand fitnessor fromindirectselection have a commonselectioncurve on unmeasured, esis, thatthe environments traits(Rausher1992). correlated phenotypic withthewet-siteplantshavingthe These possibilitiescannotbe analyzedfromthe selection forwater-useefficiency, optimumvalues and thedry-siteplantshavingvalues lower statistics studies. butmustbe answeredthrough independent thantheoptimum.Therefore theresultsdo notconclusively The extensiveliterature on plantwateruse suggeststhatthe supportthe hypothesisthathigherwater-useefficiencyis resultsI observedare more consistentwitha causal relamoreadaptiveinthedrysitethaninthewetsite.A conclusive tionshipbetweentraitsand fitness thanwithenvironmentally testwould requireexperimental to extendthe inducedcovariance.Variationwithinthe dry treatment manipulation in in each environment. rangeof phenotypes environmental wateravailabilitywouldbe expectedto cause The lowerwater-useefficiency exhibitedby plantsin the a positivecorrelation betweenstomatalconductanceand fitis very ness if wettersites permitmoregrowthand greaterwater dryenvironment comparedwiththewetenvironment withwater-use Variation puzzling.Commonly plantsshowgreater efficiency loss butnottheobservednegativecorrelation. SELECTION ON PLANT PHYSIOLOGICAL TRAITS 101 logicalecologyofNorthAmericanplantcommunities. Chapman inthedrytreatment in environmental wateravailabilitycould and Hall, New York. also be expectedto cause a positivecorrelation betweenleaf J.R., and C. Clark. 1988. Evolutionand adaptationin size and fitnessif wettersitespermitbothlargerleaves and Ehleringer, Encelia (Asteraceae).Pp. 221-248 in L. D. Gottlieband S. K. moregrowthbutnottheobservedintermediate optimumfor Jain,eds. Plantevolutionary biology.Chapmanand Hall, New leaf size. The significant pathbetweenwater-useefficiency York. J. R., and T. A. Cooper. 1988. Correlationsbetween and vegetativebiomassin thedryenvironment suggeststhat, Ehleringer, carbonisotoperatioand microhabitat in desertplants.Oecologia if selectionon unmeasuredcorrelatedcharactersis respon76:562-566. sibleforthesignificant selectionon water-use efficiency, then Endler,J. 1986. Naturalselectionin thewild.PrincetonUniversity the unmeasuredcharactersmustbe associatedwithcarbon Press,Princeton, NJ. uptake. Farris,M. A., and M. J.Lechowicz. 1990. Functionalinteractions among traitsthatdeterminereproductivesuccess in a native The selectionstudyfoundconsiderableagreement between annualplant.Ecology 71:548-557. naturalselectionin thefieldand predictions fromfunctional Field, C. B, J.T Ball, and J.A Berry.1989. Photosynthesis: prinanalysis of the traits.This agreementis evidence for the ciples and fieldtechniques.Pp. 209-254 in R. W. Pearcy,J. predictivepower of both functionalstudiesand studiesof Ehleringer, H. A. Mooney,and P. W. Rundel,eds. Plantphysiological ecology fieldmethodsand instrumentation. naturalselection,despitethe complexityof possible interChapman and Hall, New York. actionsamongcorrelatedtraitsin determining fitness.This M. A. 1990. The cost of meristem limitation in Polygonum studyshows thatthe overriding importanceof the benefits Geber, arenastrum: Negativegeneticcorrelations betweenfecundity and of carbonuptakeand costs of wateruse in predictingadgrowth.Evolution44:799-819. aptationsamong species in different environments (Horn Givnish,T J. 1979. On theadaptivesignificance of leaf form.Pp. 375-407 in0. T. Solbrig,S. Jain,G. B. Johnson, andP.H. Raven, 1979) may be reflectedin the microevolutionary processes eds. Topics in plantpopulationbiology.Columbia University withinspecies. Press,New York. . 1986. Optimalstomatalconductance, allocationofenergy ACKNOWLEDGMENTS betweenleaves and roots,and themarginalcostoftranspiration. Pp. 171-214 in T. J.Givnish,ed. On theeconomyof plantform I would like to thankM. Newtonforexcellenttechnical and function.CambridgeUniversity Press,New York. J.,J.A. Teeri,and A. M. Wood. 1986. Differentiation assistanceand S. Yamins,S. Suwanski,and J. Zdenek for Gurevitch, amongpopulationsofSedumwrightii (Crassulaceae)in response expertgreenhousecare of the plants.I thankK. Black, L. to limitedwateravailability:Waterrelations,CO2 assimilation, Birch,K. Karoly,K. Donohue,M. Wade,andP. Chu fortheir growthand survivorship. Oecologia, Berlin70:198-204. helpwithfieldwork.This paperhas benefited fromthecom- Harris,R. J. 1975. A primerof multivariate statistics.Academic mentsof M. Liebold, S. Arnold,E. Simms,J. Schmitt,J. Press,New York. Teeri,S. DeWalt,thereviewers, and especiallyfrommydis- Horn,H. S. 1979. Adaptationfromtheperspectiveof optimality. Pp. 48-62 in 0. T. Solbrig,S. Jain,G. B. Johnson,and P. H. sertationadvisor,M. Wade. Raven,eds. Topics in plantpopulationbiology.ColumbiaUniversityPress,New York. LITERATURE CITED Kalisz, S., and J. A. Teeri. 1986. Population-levelvariationin metabolismand growthin Sedumwrightii. photosynthetic EcolAnalyticalDevelopmentCompany.1992. Instruction manual,type ogy 67:20-26. LCA-3 carbondioxideleafchamberanalysissystem.Analytical Korner,C. H., J. A. Scheel, and H. Bauer. 1979. Maximumleaf Development,Hoddesdon,England. diffusiveconductancein vascularplants.Photosynthetica 13: Arnold,S. J. 1983. Morphology, performance, and fitness.Amer45-82. ican Zoologist23:347-361. Lande, R., and S. J.Arnold. 1983. The measurement of selection . 1992. Constraints on phenotypic evolution.AmericanNaton correlatedcharacters.Evolution37:1210-1226. uralist140:S85-S107. Li, C. C. 1975. Path analysis-a primer.Boxwood Press,Pacific Bhagsari,A. A., and R. H. Brown. 1986. Leaf photosynthesis and Grove,CA. its correlationwithleaf area. Crop Science 26:127-132. of CO2 S. von,and G. D. Farquhar.1981. Some relationships Long, S. P., and J-E. Hallgren. 1985. The measurement Caemmerer, assimilationbyplantsin thefieldand in thelaboratory. Pp. 62betweenthe biochemistry of photosynthesis and the gas ex94 in J.Coombs,D. 0. Hall, S. P. Long,and J.M. 0. Scurlock, changeof leaves. Planta 153:376-387. eds. Techniquesin bioproductivity and photosynthesis, 2d ed. Cohen, D. 1970. The expectedefficiency of waterutilizationin PergamonPress,New York. plantsunderdifferent competition and selectionregimes.Israel T. 1989. Free-statusersmanual.TechnicalBulletin Mitchell-Olds, Journalof Botany19:50-54. 101. Division of Biological Sciences, Universityof Montana, Cowan,I. R. 1986. Economicsof carbonfixationin higherplants. Missoula. Pp. 133-170 in T. J.Givnish,ed. On theeconomyof plantform Mitchell-Olds,T., and R. G. Shaw. 1987. Regressionanalysisof and function.CambridgeUniversity Press,New York. naturalselection:Statisticalinferenceand biological interpreDe Jong,T. M. 1978. Comparative gas exchangeoffourCalifornia tation.Evolution31:1149-1161. beach taxa. Oecologia 34:343-351. DePuit, E. J.,and M. M. Caldwell. 1975. Stem and leaf gas ex- Mott,K., A. C. Gibson,and J. W. O'Leary. 1982. The adaptive significanceof amphistomatous leaves. Plant,Cell, and Envichangeof twoaridlandshrubs.AmericanJournalofBotany62: ronment 5:455-460. 954-961. and environmental plantphysDonovan,L. A., and J.R. Ehleringer.1994. Potentialforselection Nobel,P. S. 1991. Physicochemical iology.AcademicPress,New York. on plantsforwater-use as estimated efficiency bycarbonisotope Parkinson, K. J. 1984. A simplemethodfordetermining thebounddiscrimination. AmericanJournalof Botany81:927-935. Dudley,S. A. 1996. The responseto differing arylayerresistanceinleafcuvettes.Plant,Cell,andEnvironment selectionon plant 8:223-226. physiologicaltraits:evidenceforlocal adaptation.Evolution50: 103-110. . 1985. Porometry. Pp. 171-191 in B. Marshalland E I. Ehleringer, J. R. 1975. Annualsand perennialsof warmdeserts. forenvironmental Woodward,eds. Instrumentation physiology. Pp. 162-180 in B. E Chabotand H. A. Mooney,eds. PhysioCambridgeUniversity Press,Cambridge. 102 SUSAN A. DUDLEY Philips,P. C., and S. J. Arnold. 1989. Visualizingmultivariate B. Osmond,and H. Ziegler [eds.I, Physiologicalplantecology New York. II. Waterrelationsand carbonassimilation.Springer, selection.Evolution43:1209-1222. Rausher,M. D. 1992. The measurement ofselectionon quantitative Smedley,M. P., T. E. Dawson, J.P. Comstock,L. A. Donovan,D. E. Sherrill,C. S. Cook, and J. R. Ehleringer.1991. Seasonal traits:Biases due to environmental covariancebetweentraits in a grasslandcommunity. Oeccarbonisotopediscrimination and fitness.Evolution46:616-626. Rodman,J.E. 1974. Systematics andevolutionofthegenusCakile ologia 85:314-320. Freeman,New York. (Cruciferae).Contributions fromtheGrayHerbarium, Harvard Sokal,R. R., and F J.Rohlf. 1981. Biometry. University 205:3-146. Wade,M. J.,and S. Kalisz. 1990. The causes of naturalselection. Evolution44:1947-1955. SAS Institute, Inc. 1989. SAS/STATuser's guide,Version6, 4th Winn,A. A., and A. S. Evans. 1991. Variationamongpopulations ed. SAS Institute, Cary,NC. of PrunellavulgarisL. in plasticresponsesto light.Functional Scheiner,S. M. 1993. Geneticsand evolutionof phenotypic plasEcology 5:562-57 1. ticity.AnnualReview of Ecology and Systematics24:35-68. Schulze,E.-D. 1982. Plantlifeformsand theircarbon,water,and Editor:D. Waller Corresponding nutrient relations.Pp. 615-676 in 0. L. Lange, P. S. Nobel, C.