Table Of ContentFLORAL BIOLOGY OF NORTH RobertA.Raguso,3AlmutKelber,4MichaelPfaff,4
AMERICAN OENOTHERA SECT. Rachel A. Levin,5 and Lucinda A. McDade6
LAVAUXIA (ONAGRACEAE):
ADVERTISEMENTS, REWARDS,
AND EXTREME VARIATION IN
FLORAL DEPTH1,2
ABSTRACT
WestudiedthefloralbiologyoffiveNorthAmericanmembersofOenotheraL.sect.Lavauxia(Spach)Endl.(OnagraceaeL.)
infieldandcommongreenhousesettings.Oenotherasect.Lavauxiafloralmorphologyrangesfromsmall,cleistogamousflowers
(O.flavasubsp.flava(A.Nels.)GarrettinGarrett)tosomeofthelongest-tubedflowersinNorthAmerica(O.flavasubsp.
taraxacoides (Wooton & Standl.) W. L. Wagner). Our goal was to compare qualitative and quantitative aspects of floral
advertisementandrewardamongtaxainsectionLavauxia.Alltaxaarenight-bloomingandself-compatible,haveyellowpetals
with ultraviolet reflectance, and produce floral scents dominated by nitrogenous compounds and monoterpenes. Methyl
nicotinate is present in the fragrances of all taxa of sectionLavauxia regardless of flower size or putative mating system.
BecausethisfloralvolatileislargelyabsentfromotherOenotheraspecies,wehypothesizethatitisasynapomorphyforsection
Lavauxia. The rare O. acutissima W. L. Wagner, which is endemic to the Uintah Mountains, is polymorphic for odors
dominatedbylinalool-orocimene-derivedcompounds.FieldobservationsinitstypelocalityinnortheasternUtah,U.S.A.,
revealedfrequentfloralvisitationbycrepuscularhawkmothsduringthefirst1.5hoursafteranthesis,apatterncommontoO.
flavasubsp.taraxacoidesandotherlarge-floweredOenotherathroughoutwesternNorthAmerica.Quantitativeaspectsoffloral
advertisement(flowersize,scentemission)andreward(nectarvolume)aredramaticallyreducedinputativelyautogamoustaxa
(O. flava subsp. flava, O. triloba Nutt.), whereas qualitative aspects (flower color, scent, and nectar chemistry) remain
comparable.Alltaxa couldbedistinguishedthroughordinationofcharactersrelatedtoflowersize,herkogamy,andscent
chemistry. Extreme nectar tube length variation across the range of O. flava renders this an excellent model system for
measuringthecostsandmechanismsofshiftsbetweenoutcrossingandautogamy.
Keywords: biogeography, floralscent, fragrance, nectar, night-blooming, Oenothera, Onagraceae, pollination.
The genus Oenothera L., with about 120 species numerousevolutionaryshiftsbetweenoutcrossingand
native to the Americas, has long served as a model autogamy,nocturnalanddiurnalanthesis,annualand
system for the study of evolutionary pattern and perennial habit,and xericand mesic habitat special-
process in flowering plants (Raven, 1979, 1988, and izationinthisgenus.Recenteffortstoestablishwell-
references therein). The basic floral ground plan is supportedphylogenetichypothesesforthisgenusand
fairly conserved across the genus with tetramerous itsclosestrelativesintheOnagraceaeL.(Levinetal.,
flowers, white, pink, or yellow-colored petals, and 2003a,2004)havebeenmotivatedinpartbythegoal
along,tubularhypanthium.However,therehavebeen of understanding the frequency of these shifts and
1TheauthorsthankWarrenWagnerforhisgenerosity,encouragement,andknowledgeofOenothera;DonaldMillerIIIfor
generouslyprovidingO.trilobaplants;RobertBellseyforseveralcollectingtripstotheWhiteandSacramentoMountains;and
Amanda Labadie for preparing the figures and plates. Raguso, Levin, and McDade were supported by NSF grants
DEB9806840 and DEB-0317217 and an Andrew W. Mellon Foundation grant to Warren Wagner, Peter Hoch, Elizabeth
Zimmer,JorgeCrisci,andKenSytsma.KelberandPfaffweresupportedbytheSwedishResearchCouncil(VR)inStockholm,
the Swedish Foundation for International Cooperation in Research and Higher Education (STINT) in Stockholm, and the
Science Faculty of Lund University. We thank Helena Kelber and Sebastian Pfaff for assistance in the field and Terry
GriswoldandJimCanefortheirexpertisewithLasioglossumbeesandhospitalityatUtahStateUniversity,Logan.Special
thanksareduetoMikeSingerforpermissiontousethephotographforFig.1E,RickClinebell,PeterHoch,andVictoria
Hollowell for editorial suggestions, and Roman Kaiser for providing authentic aldoxime standards and for discussing the
unique organoleptic properties of 1-pyrroline. This paper is dedicated to the memory of Rick Clinebell, whose boundless
enthusiasmforprairieecologyremainsaninspirationtoallwhoknewhim.
2TheeditorsoftheAnnalsthankSophiaBalcombforhereditorialcontributiontothisarticle.
3DepartmentofNeurobiologyandBehavior,CornellUniversity,Ithaca,NewYork14853-2702,U.S.A.Authorforcorres-
pondence:[email protected].
4DepartmentofCellandOrganismBiology,LundUniversity,22362Lund,Sweden.
5DepartmentofBiology,AmherstCollege,Amherst,Massachusetts01002,U.S.A.
6RanchoSantaAnaBotanicGarden,1500N.CollegeAve.,Claremont,California91711,U.S.A.
ANN. MISSOURI BOT. GARD. 94: 236–257. PUBLISHEDON 26 APRIL 2007.
Volume94, Number 1 Raguso etal. 237
2007 Oenotherasect.LavauxiaFloral Biology
their underlying processes. In parallel, we and our short-tubed (less than 6 cm), often cleistogamous O.
colleagues have initiated studies of evolutionary flava subsp. flava to the extremely long-tubed (up to
transitions in reproductive biology and life history 18cm)O.flavasubsp.taraxacoides,whichareamong
strategies in several monophyletic sections of Oe- the deepest flowers in the North American flora
nothera, including section Anogra (Spach) Engl. (Gregory, 1964;Grant, 1983;see Fig. 1).
(Evansetal.,2005),sectionCalylophus(Spach)Torr. We studied floral traits (i.e., color, morphology,
&A.Gray,sectionPachylophus(Spach)Endl.(R.A. scent,nectar)associatedwithpollinatorattractionand
Raguso, in prep.), and in Gaura L. (Clinebell et al., rewardinonlytheNorthAmericanspeciesofsection
2004). Lavauxia, because the South American species
Here we examine the floral biology of the North (Oenothera acaulis and O. centauriifolia) were not
AmericanspeciesofOenotherasect.Lavauxia(Spach) available to us at the time of study. Given the
Endl. The recognition of Lavauxia as a distinctive biogeographic,morphological,andtaxonomicpatterns
entity dates from early in the study of Onagraceae, described above, we sought to address several
eitherasagenus(Spach,1835:367;Raimann,1893) questionswith such data.
or a subgenus of Oenothera (Endlicher, 1840: 1190; (1) Are floral scent and nectar (or components
Jepson, 1923–1925: 680). As monographed by Munz thereof) reduced or absent in the small-flowered,
(1930) and amended by Wagner (1981, 1986), putatively autogamous flowers of O. triloba and O.
Oenothera sect. Lavauxia presently consists of three flava subsp. flava? The dramatic difference in visual
yellow-flowered North American species, O. acutis- display between these and outcrossing taxa (O.
sima W. L. Wagner, O. flava (A. Nels.) Garrett in acutissima, O. flava subsp. taraxacoides) suggests
Garrett (including O. flava subsp. flava and O. flava reduced selective pressure for floral advertisement
subsp. taraxacoides (Wooton & Standl.) W. L. and reward when self-pollination is the predominant
Wagner),andO.trilobaNutt.,andtwowhite-flowered strategy.
South American species, O. acaulis Cav. and O. (2) Regarding components of floral fragrance, are
centauriifolia (Spach) Steud. Oenothera triloba is plantsofthetwosubspeciesofO.flavamoresimilarto
a small-flowered, self-compatible spring annual or each other than to plants of O. triloba and O.
biennial herb native to south-central U.S.A. This acutissima or does fragrance chemistry mirror polli-
species is broadly parapatric with the western nationstrategy, rather thanphylogenetic affinity?
members of section Lavauxia, which are perennial
(3)Doindividualsfromtwoisolatedpopulationsof
herbs occurring throughout montane western North
O. flava subsp. taraxacoides share a uniform pheno-
AmericaintocentralMexico(Wagner,1986).Typical
type or do floral traits vary distinctively between
O. flava is a small-flowered, self-compatible plant
populations? The mountaintop distribution of these
withaverybroaddistribution,whereasO.flavasubsp.
plants suggests that their floral traits may have
taraxacoidesisalarge-floweredplantknownonlyfrom
diverged, either through genetic drift or local
disjunct, high-elevation populations in southwestern
adaptation(Slentz et al.,1999;Boyd,2002).
U.S.A. and northern Mexico. Wagner (1986) com-
(4)AreflowersofthenarrowendemicO.acutissima
binedthesetwotaxaassubspeciesofO.flavaonthe
visited by the same spectrum of medium- to long-
grounds that small-flowered, autogamous individuals
tongued hawkmoths that pollinate other Oenothera
and large-flowered, modally outcrossing individuals
throughout western North America (e.g., Wagner et
representextremephenotypesofasinglespeciesthat
al., 1985)? Pollinator data are especially needed for
intergradeextensivelyinzonesofcontact.Incontrast,
this geographically restricted entity, which could be
Wagner (1981) segregated large-flowered, self-com-
vulnerable to extinction due to habitat loss or
patiblebutpresumablyoutcrossingplantsendemicto
modification.
the mountains of northwestern Colorado and north-
Theresultsofthisstudywillprovideabaselinefor
easternUtah,U.S.A.,asanewspecies,O.acutissima
quantitative investigations of the costs, benefits, and
(Wagner, 1981), citing several vegetative and re-
evolution of floral advertisement and reward in this
productiveautapomorphiesthatdistinguishitfromO.
fascinating lineage.
flava, with which it co-occurs. Subsequently, O.
acutissima was treated as a variety of O. flava by
Welsh (1986), but the poor viability (and sterility) of METHODS
artificial hybrids between O. acutissima and O. flava
PLANTCARE
(Wagner,1981)compelsustorecognizeO.acutissima
sensu Wagner (1986). Regardless of nomenclatural Tento60individualsofeachtaxonwerecultivated
treatment, the western members of Lavauxia are in a greenhouse at the University of South Carolina,
remarkably variable in floral characters, from the Columbia,SouthCarolina,wheretheywerestudiedin
238 Annalsof the
MissouriBotanical Garden
Figure 1. Floral variation in Oenothera sect. Lavauxia. —A. Newly opened flowers of O. triloba; note lack of strong
herkogamy.—B.Nectartube(hypanthium)lengthvariationfromputativeautogamy(O.triloba,left,O.flavasubsp.flava,
center)toputativeoutcrossing(O.flavasubsp.taraxacoides,right).—C.Side-by-sidecomparisonofnectarydepthbetweenO.
flavasubsp.taraxacoides(right)andAngraecumsesquipedaleThouars(left),Darwin’sMalagasyStarOrchid.—D.Differences
infloraldiameterbetweenO.flavasubsp.taraxacoides(left)andO.flavasubsp.flava(right).—E.Hyleslineataleavingflower
ofO.flavasubsp.taraxacoidesinSacramentoMtns.,OteroCo.,NewMexico.PhotosinA–DbyR.A.Raguso.PhotoinEbyM.
S.Singer,withpermission.Allscalebars51cm.
Volume94, Number 1 Raguso etal. 239
2007 Oenotherasect.LavauxiaFloral Biology
March–April 2002. Oenothera acutissima and O. herkogamy (difference in length between stigma and
trilobaplantsweretransplantedfromthefieldinJuly longest anther), to the nearest 0.5 mm. Fresh mass
2001andMarch2002,respectively.Plantsfromthree was recorded after nectar removal using a Mettler
populations of O. flava were grown from seed analytical balance, to the nearest 0.001g. Flowers
germinated in September 2001. Plants were grown thenweredriedat50uCfor24hr.,anddrymasswas
in 60:40 potting soil:sand mix, bottom-watered daily, recordedonthesamebalance.Becauseplantsofeach
and fertilized with Miracle-Gro (15% N:30% P:15% species opened one to two flowers per night, most
K) once per month for the duration of the study. floral measures included flowers from different
VoucherspecimensweredepositedatARIZ,US,and individuals.
USCH; collection numbers and source localities are
provided inAppendix1.
FLORALSCENTCOLLECTIONANDANALYSIS
Floral odors were collected using two complemen-
VISUALREFLECTANCE
tary methods. The first method, solid-phase micro-
The spectral properties of newly opened flowers extraction (SPME) (Zhang & Pawliszyn, 1993), was
weremeasuredusingaSpectralInstruments(Tucson, combined with gas chromatography–mass spectrome-
Arizona) SI-440 CCD array UV-VIS Spectrophotom- try(GC-MS)tooptimizethequalityofmassspectrafor
eter, connected by a 400mm fiber optic probe to scent compound identification. Headspace bags were
aLabsphere(NorthSutton,NewHampshire)9 cmID preparedfromReynolds(nylonresin)ovenbagsusing
integration sphere. We used a 10W tungsten light animpulseheatsealer,asdescribedbyRagusoetal.
source to collect reflectance data from 350 (ultravi- (2003a). Bags were placed over living, uncut flowers
olet) to 700nm (infrared) wavelengths from freshly andcinchedwithplasticties.Parallelcollectionswere
excised flowers placed over a black felt cloth made from Oenothera leaves and greenhouse air in
background.Spectralreflectancefromupper(adaxial) ordertoidentifyvegetativeandambientcontaminants
petal surfaces was measured at both distal and infloralsamples.Inotheranalyses,5–10mlsamples
proximalpositions,asweanticipatedacentral‘‘target ofnectarwerespottedontofilterpaperwedges,sealed
pattern’’ of contrasting ultraviolet-absorbance and withinheadspacebags,andanalyzedfornectarodors
reflectance across the flower (Dement & Raven, as described by Raguso (2004). All samples were
1974). Data were collected as percent reflectance equilibrated for 15minutes, then allowed to adsorb
relative toa whitepigment (Duraflect)standard,with onto a 100mm polydimethylsiloxane (PDMS) SPME
negligible (0.5%–1% of standard) signal contributed fiber for an additional 15minutes, immediately
by theblack clothbackground. followed by GC-MS analysis (see below). Results
were unchanged when longer equilibration or expo-
suretimes wereused.
NECTARANDFLORALMORPHOLOGY
The second method, dynamic headspace trapping
Nectarvolumewascollectedfromatleast10newly (Raguso & Pellmyr, 1998), was used to quantify
opened flowers of each species at dusk, using 5 or volatilecompound emission ratesduringthefirstfew
10ml glass capillary micropipettes. Because of the hoursafteranthesis.Floralvolatileswereconcentrat-
unusual length and narrowness of the floral tubes, ed within headspace bags (ca. 500ml volume) and
nectar collection required floral dissection with trapped on adsorbent cartridges using Supelco (Ber-
a razor, but care was taken to avoid contaminating wick, Pennsylvania, U.S.A.) personal air sampler
or diluting nectar with other plant fluids (Cruden et vacuum pumps. Pasteur pipettes were packed with
al., 1983). We measured sugar concentration in 5ml 100mgofSuperQ(80–100mesh)adsorbent(Alltech
aliquots from each sample with a hand-held re- Associates, Waukeegan, Illinois, U.S.A.) between
fractometer(Leica,Brix50)designedtomeasure0%– plugs of quartz wool, and headspace air was pulled
50%sucroseequivalentsbyweight,inunitsof0.25%. overtheflowersandintotheadsorbenttrapataflow
Floral morphological measurements were taken on rateofca.250ml/min.Fragrancewascollectedfor6–
freshly excised flowers. We used a metric ruler to 8 hr. after anthesis (dusk in all taxa) in a protected
measure floral diameter (the greatest distance across areaoutsidethegreenhouse.Additionalscentcollec-
the open corolla limbs, perpendicular to the nectar tion was performed for Oenothera acutissima plants
tube), tube flare (diameter across the mouth of the growingatthetypelocality(seeAppendix1).Trapped
nectar tube, not including corolla limb), floral depth fragrancewaselutedwith3 mlofhigh-purityhexane
(length of nectar tube from mouth to ovary), stamen and stored at –20uC in Teflon-capped borosilicate
lengthandstylelength(distancefromtheovarytothe glassvials.BeforeGC-MSanalysis,weusedaflowof
longest stamen and to the stigma, respectively), and gaseous N to concentrate samples to 75ml, then
2
240 Annalsof the
MissouriBotanical Garden
added 5ml of 0.03% toluene (16ng) as an internal measurements including floral diameter, tube flare,
standard. One ml aliquots of each sample were floral depth, herkogamy, dry mass, and total number
injected into a Shimadzu GC-17A equipped with of scent compounds were normally distributed and
a Shimadzu QP5000 quadrupole electron impact MS were not transformed. Scent emission rates (ng scent
(Shimadzu Scientific Instruments, Columbia, Mary- perflowerperhour)werecalculatedforthesumsofall
land, U.S.A.) as a detector. All analyses were done (1)nitrogenous,(2)ocimene-derived,and(3)linalool-
using splitless injections on a polar GC column derivedcompounds;theserateswerethennaturallog-
(diameter 0.25 mm, length 30m, film thickness transformed (y 5 ln(x +1)) for analysis. These data
0.25mm (Econo Cap’s carbowax coating, known as werecombinedforprincipalcomponentanalysisusing
EC WAX); Alltech Associates), as described by the varimax rotation option (SPSS 11.5), in which
Raguso et al. (2003a). SPME fibers were directly factors with eigenvalues greater than unity were
injectedintotheGCinjectionportat240uCandwere retained. Discriminant function analysis (SPSS 11.5)
analyzed using the GC-MS parameters described was then used to determine whether factor loadings
above. Compounds were tentatively identified using could be used to correctly assign individuals to their
computerized mass spectral libraries (Wiley Registry source populationor taxon.
ofMassSpectralData,NationalInstituteofStandards
and Technology, and Robert Adams’ libraries (. POLLINATOROBSERVATIONSFOROENOTHERAACUTISSIMA
120,000 mass spectra)). Chiral GC was not available
Floral visitation to a natural population of
to us, so compounds with chiral carbons (e.g., a-
Oenotheraacutissimawasobservedatthetypelocality
pinene) were assumed to represent racemic mixes.
in Daggett Co., Utah, U.S.A. (Appendix 1). Several
Whenever possible, GC peak identities were verified
hundred plants were found in bloom along with
using co-retention with known standards on both EC
flowering individuals of Achillea L., Geum L., Iris L.,
WAX and EC-5 GC columns. Peak areas were
Opuntia Mill., and Potentilla L. species in a moist
integrated usingShimadzu’s Class-5000 software and
meadow at the margin of a Pinus ponderosa Douglas
were quantified by comparison with the internal
forest. Individual flowers were watched on the
standard. Total scent emission rates (per hour) were
evenings of 11 June 2001 and 25–28 June 2003, for
calculated as sums of all peak areas, converted to
a total of 19 observer hours. Insect visitors were
nanogramsusingtheinternalstandard,andexpressed
photographed and collected for identification when
perflower, andper gram freshand dryfloral mass.
necessary. Voucher specimens remain in the posses-
sion of the lead author. To relate flower visitation to
COMPARISONOFSCENTCOMPOSITION ambient conditions, light levels and temperatures
were measured at the study site at one- to three-
Wecomparedvariationinfragranceprofileswithin
minuteintervals.Lightlevelsweremeasuredincd/m2
and among taxa using the relative amounts of floral
with a highly sensitive silicon detector attached to
volatiles. For each individual, we calculated the
a radiometer (International Light, IL1700). The time
proportion of total scent contributed by each com-
for sunset was obtained from the NASA database
poundandstandardizedthesedatatozscores,which
(http://aa.usno.navy.mil/).
havemeansofzeroandvariancesofone.Thesevalues
were used to compute a matrix of pairwise dissimi-
larity values between all individuals in the study RESULTS
(Euclidean distance). Pairwise dissimilarity values
FLOWERSIZEANDCOLOR
amongconspecific(andco-occuring)individualswere
comparedtothosebetweenheterospecific(ordisjunct) Flower size varied dramatically among taxa. The
individuals using the Wilcoxon rank sum test (SPSS small, short-tubed flowers of Oenothera flava subsp.
11.5). This constituted a two-tailed test of the null flavadidnotopenfullyinthegreenhouseorthefield.
hypothesis that mean ranks of pairwise dissimilarity Whenbudswereforcedopenontheeveningoffloral
valuesbetweentaxaorpopulationsdidnotdifferfrom maturity(determinedbycolorandfirmness),themean
mean ranks of pairwise dissimilarity values within floraldiameterwasonly25mm(Table 1).Flowersof
groups (see Levinetal.,2001,2003b). O. triloba were twice this large on average, whereas
flowers of O. flava subsp. taraxacoides and O.
acutissima measured 60–85mm in diameter. The
FACTORANALYSISOFFLORALPHENOTYPE
nectartubes(hypanthia)ofalltaxaarelongcompared
Ordination was used to determine whether combi- withotherNorthAmericanflowersandvarymarkedly
nations of floral attributes were characteristic for among taxa and populations (Fig.1). Flowers of O.
different section Lavauxia taxa. Floral morphological flavasubsp.taraxacoidesfromArizonawere1.1–1.7-
Volume94, Number 1 Raguso etal. 241
2007 Oenotherasect.LavauxiaFloral Biology
Table 1. Flowermorphologicalmeasurementsandnectarcharacteristics;plantsfromlocalitiesinAppendix1weregrown
undercommongreenhouseconditions.AZ,Arizona;NM,NewMexico;OA,Oenotheraacutissima;OFF,Oenotheraflavasubsp.
flava;OFT,Oenotheraflavasubsp.taraxacoides;OT,Oenotheratriloba;v/v,volumepervolume.
OFT
Measurement,mean6SEM
(range,N) OA OT OFF NM AZ
Floraldiameter(mm) 86.361.1 42.460.9 24.460.8 65.062.2 82.962.2
(80–98,21) (32–53,28) (19–32,19) (49–78,14) (65–101,18)
Corollaflare(mm) 7.760.2 4.260.1 2.860.1 7.360.3 7.460.2
(6.5–9.0,21) (3.0–5.5,28) (1.5–3.5,19) (6.0–8.0,8) (6–8.5,18)
Floraldepth(mm) 96.764.0 82.361.7 57.861.7 166.163.8 177.563.6
(75–135,21) (66–97,28) (43–71,19) (150–186.5,14) (155–203,18)
Stamenlength(mm) 128.164.6 97.962.1 66.761.9 190.464.3 201.864.1
(100–171,21) (80–116,28) (51–87,19) (170–216,14) (176–238,18)
Stylelength(mm) 141.564.7 98.462.1 63.761.9 196.664.0 212.363.9
(113–184,21) (82–116,28) (51–72,19) (177–227,14) (184–247,18)
Herkogamy(mm) 13.460.7 0.660.3 22.960.7 6.261.1 10.561.0
(10–21,21) (22–4,28) (29–3,19) (0–13,14) (0–16,18)
Nectarstanding 9.761.3 11.562.0 2.360.6 11.261.3 40.962.9
crop(mL) (3.5–15,8) (4.3–22,10) (0.8–5.0,11) (6.2–15.9,6) (26–56,9)
Nectarconcentration 25.460.8 27.460.8 25.960.9 26.660.7 25.260.9
(%v/v) (19–30,13) (22–31,10) (23.5–32,8) (23–30,12) (16.5–31.5,17)
Freshmass(g) 1.2660.02 0.3260.02 0.1760.01 1.5560.07 1.9260.07
(1.21–1.29,3) (0.19–0.44,14) (0.14–0.25,12) (1.22–1.81,8) (1.63–2.38,9)
Drymass(g) 0.1460.01 0.0460.003 0.0260.001 0.1860.006 0.2360.01
(0.12–0.16,3) (0.02–0.06,14) (0.02–0.03,12) (0.16–0.21,8) (0.16–0.35,9)
foldlargerthanO.flavasubsp.taraxacoidesfromNew NECTARVOLUMEANDCONCENTRATION
Mexico in corolla diameter, nectar tube length, and
Nectar volumes per flower varied considerably
anther-stigma separation. Stigmas of O. acutissima
between taxa and populations, with means ranging
flowers were exserted the greatest mean distance
from 2.3 ml in the cleistogamous Oenothera flava
beyond the anthers (positive herkogamy: 13.4 mm),
subsp. flava to more than 40ml in O. flava subsp.
whereas flowers of O. flava subsp. flava were
taraxacoidesfromArizona(Table1).Surprisingly,the
consistently negatively herkogamous, with the open
flowers of O. triloba produced nectar volumes
stigma lobes positioned 3 mm below the dehiscing
comparable to those of the much larger flowers of O.
anthers, on average. Flowers of O. triloba varied
acutissimaandO.flavasubsp.taraxacoidesfromNew
continuously from negative to positive herkogamy
Mexico.Ontheotherhand,flowersofO.flavasubsp.
(Table 1).
taraxacoidesfromArizonaproducedatleastfour-fold
All taxa produced flowers that were yellow to the
more nectar than those of any other taxon studied
human eye. Flowers ofOenothera triloba appeared to
(Table 1), despite being grown on the same green-
betheleastsaturatedyellow(Fig. 1)butwerenotless
housebench.Nectarsugarconcentrationsandranges
reflective than other taxa in yellow wavelengths
were nearly identical (25%–27.5% volume per
(Fig.2). The inner petal surfaces of O. acutissima
volumedissolvedsucroseequivalents)amongalltaxa,
flowers were uniformly reflective above 500nm but
including O. flavasubsp. flava.
absorbed light of shorter wavelengths, including
ultraviolet (UV) (Fig. 2). In all other taxa, the
FLORALSCENTEMISSIONRATESANDCHEMICALCOMPOSITION
proximalportionofeachpetal(i.e.,thecentralregion
of the corolla) absorbed UV, whereas the distal petal Odoremissionratesvarieddramatically,withlarge-
regions reflected strongly from 350 to 380nm, floweredtaxaproducingupto20-foldmorescentper
creating a pattern of UV contrast at the flower’s flower than small-flowered taxa (Fig.3). Flowers of
center (Fig. 2). Dissection of buds revealed this Oenothera flava subsp. flava and O. triloba had the
pattern to be present at least 24hr. before anthesis lowest emission rates and were only half as strongly
(datanotshown). scented as large-flowered taxa even when standard-
242 Annalsof the
MissouriBotanical Garden
eachblackubsp.
Inheas
hitestandard.emarkedbytOenotheraflav
wv
dtoecurway.
mpareandthnthis
ntreflectance,cothecorollalobe,smalltodissecti
ntsperceregionofweretoo
y-axisrepreseoximal(basal)ava(panelA)
eprfl
Thhesp.
nminwavelength.area)discwithintenotheraflavasub
surfaces,from350–70021cmdiameter(0.78cmofthepetal.PetalsofOC,andD,respectively.
ancecurvesforupper(adaxial)petalegreyarrowshowsreflectanceoveraasimilarareawithinthedistalregionO.acutissimaareshowninpanelsB,
eflectbythoverand
SpectralrvemarkedeflectanceO.triloba,
Figure2.panel,thecurarrowshowsrtaraxacoides,
Volume94, Number 1 Raguso etal. 243
2007 Oenotherasect.LavauxiaFloral Biology
izedforfreshanddryfloralmass.Althoughflowersof Floral scent blends generally were taxon-specific
O. flava subsp. taraxacoides from Arizona were 20% (Table 2). Fragrance composition between isolated
larger than those from New Mexico, the latter were populations of Oenothera flava subsp. taraxacoides
twice as strongly scented as the former, both per wassimilar,exceptforthepresenceoftraceamounts
flowerandperunitfloralmass(Fig. 3).Emissionrates (,1%oftotalemissions)oflinaloolanditsfuranoid
of greenhouse-grown O. acutissima plants were oxidesinplantsfromArizonaandsubstantialamounts
comparable to those of O. flava subsp. taraxacoides (nearly 30%) of trans-b-ocimene in those from New
fromNewMexicobutwerelessthanhalfasstrongas Mexico.Theexceptiontothepatternoftaxon-specific
those measured from O. acutissima in the field odorswasO.acutissima,inwhichthepresenceoftwo
(Fig.3). distinctscentphenotypesor‘‘chemotypes’’resultedin
Fifty-four volatile organic compounds were de- sufficient within-taxon variation such that the floral
tectedinfloralheadspace,ofwhich32wereidentified fragranceproducedbyconspecificindividualswasnot
using known standard compounds or commercially significantly more similar than that produced by
available essential oil blends. Scent compounds flowersofdifferentspecies(Table2).Thefloralscent
representedseveralbiosyntheticcategories,including of individual plants was dominated either by linalool
mono- and sesquiterpenoids, benzenoid (aromatic) or by trans-b-ocimene, to the near exclusion of the
compounds, fatty acid derivatives and, especially, other compound. Three of eight field-collected
nitrogenous compounds derived from amino acids fragrance samples were linalool-dominated (mean 6
(Appendix 2). The nitrogenous aldoximes, nitro- and SEM 5 40.7 6 4.2% of total emissions) with little
nitrile compounds derived from valine, leucine, trans-b-ocimene (6.0 6 3.0%), whereas trans-b-
isoleucine, and phenylalanine (Fig.4), were present ocimene was the dominant component of the remain-
in all taxa, whereas sesquiterpenoids were limited to ing five plants (62.4 6 2.6%), in which linalool was
Oenothera acutissima. As a class, the nitrogenous nearly absent (0.2 6 0.06%). Similar patterns were
compounds accounted for 67%–96% of total scent observed in greenhouse-grown plants, none of which
emissions in all O. flava populations, but only 9.3% emitted large amounts of both compounds. Composi-
and 37% of emissions from O. triloba and O. tional differences between greenhouse- and field-
acutissima,respectively. The scents of O. triloba and grown O. acutissima were restricted to minor, de-
O. acutissima were dominated (62%–72% of emis- rivative components. Field-grown plants produced
sions) by monoterpenoids (Fig.5). Two additional unique linalool-derived compounds, whereas green-
nitrogenous compounds, nicotinic acid methyl ester house-grown plants produced putative ocimene-deri-
and 1-pyrroline (Fig.4), were present in all taxa vatives, increased amounts of b-caryophyllene, and
surveyed. The monoterpene trans-b-ocimene consti- relatedsesquiterpenesnotdetectedinthefragranceof
tuted 30%–44% of total emissions in most taxa but field-grown plants.
was a minor scent component (, 4%) in O. flava
subsp. flava and O. flava subsp. taraxacoides from
FACTORANALYSISOFFLORALADVERTISEMENTSANDREWARDS
Arizona(Fig.5).SPMEanalysisrevealedthatmethyl
benzoate and 1-pyrroline were present exclusively in Principalcomponentanalysisidentifiedtwofactors
the floral nectar of all taxa, whereas most other with eigenvalues greater than unity, accounting for
compounds were emitted by petals and other flower 85.8% and 12.5% of total sample variance, re-
parts (datanotshown). spectively. Floral dimensions above the nectar tube
Taxa with large, putatively outcrossing flowers (floral diameter, corolla flare, and herkogamy) were
produced the most complex fragrances, with 24–27 highly correlated and loaded positively on factor 1,
compoundsinOenotheraflavasubsp.taraxacoidesand along with the total number of scent compounds and
34–37 compounds in O. acutissima, including the emissionratesoflinalool-andocimene-derivedscent
fruity-scented isoamyl alcohol and three of its esters compounds (Table 3, Fig.7). This dimension clearly
(Appendix2,Fig.6).Incontrast,thesmall,putatively separated Oenothera acutissima from O. flava subsp.
autogamous flowers of O. flava subsp. flava and O. flava and O. triloba in floral phenotype space, and
triloba produced the least chemically complex odors separated most individuals of O. flava subsp.
(13–21compounds).Thisresultisnotanartifactoftoo taraxacoidesfromArizonaandNewMexicointotheir
littlefloraltissuebeingusedforodoranalysis,assome respective populations (Fig.7). Floral depth, dry
samplesincludeduptofiveflowers.Whenthenumber floral mass, and the emission of nitrogenous scent
ofscentcompoundsofO.trilobawasregressedagainst compoundshadthestrongestloadingscoresonfactor
total floral mass persample, odor complexity was not 2,forwhichtheonlycharacterwithnegativeloadings
significantlycorrelatedwithsamplefloralmass(linear was the emission of linalool and related scent
regression,R250.001,P50.94). compounds (Table 3). Floral diameter, herkogamy,
244 Annalsof the
MissouriBotanical Garden
Figure 3. Hourlyemissionratesoffloralscent,perflower(stippledbars),pergramfreshfloralmass(dark),andpergram
dry floral mass (light). In order to accommodate the substantial range of variation, the y-axis is discontinuous. Note the
similarityin emissionratesbetweenthetwo putativelyautogamoustaxa (OenotheratrilobaandO. flavasubsp.flava) and
betweentheoutcrossingtaxa(O.acutissimaandO.flavasubsp.taraxacoides).Alsonotethetwo-foldincreaseinemissions
from O. acutissima collected in its native habitat. All other collections were made from plants grown in a common
greenhousesetting.
and corolla flare loaded positively on both factors, flowered taxa, O. triloba and O. flava subsp. flava,
with larger loadings on factor 1. Factor 2 clearly from each other (Fig. 7). Discriminant function
separated O. flava subsp. taraxacoides from all other analysiscorrectlyassigned95%ofthe44individuals
taxa, and the combined factors separated the small- sampledtothecorrecttaxonandpopulationexceptfor
Volume94, Number 1 Raguso etal. 245
2007 Oenotherasect.LavauxiaFloral Biology
Figure 4. NitrogenousvolatilecompoundsprominentinthefloralscentofLavauxiaspecies.Theverticalseriesinthe
upperpanelshowsthederivationofaldoximes,nitriles,andnitro-compounds(indescendingorder)fromtheirputativeamino
acid precursors, after Kaiser (1993). Numbered compounds are (1) 2-methylpropanaldoxime, (2) 2-methylpropylnitrile, (3)
nitro-2-methylpropane,(4)2-methylbutyraldoxime,(5)2-methylbutyronitrile,(6)nitro-2-methylbutyrate,(7)3-methylbutyr-
aldoxime,(8)3-methylbutyronitrile,(9)nitro-3-methylbutyrate,(10)phenylacetaldoxime,(11)phenylacetonitrile, and(12)
nitro-2-phenylethane.Notethateachaldoximeskeletonoccursinsyn-andanti-isomers(notshown),andthatcompounds2
and 12 were not detected in thisstudy. Lowerpanel shows structuresof methyl nicotinate, present in allOenothera sect.
LavauxiabutinfewotherOenotheraspecies,and1-pyrroline,presentinthenectarofallOenotherastudiedtodate.
O. flava subsp. taraxacoides, for which one plant in uals were not seen until 40minutes after sunset and
each population was incorrectly assigned to New continued visiting flowers until observations ceased.
Mexico orArizona populations. During hawkmoth visitation, ambient temperatures
dropped from 15uC to 10uC, and light intensity
decreased from 120 cd/m2 at sunset to 1.25 cd/m2 at
FLORALBIOLOGYOFOENOTHERAACUTISSIMA
first H. lineata visit, and 0.06 cd/m2 at first M.
Flowers of Oenothera acutissima opened in the quinquemaculata visit. Hyles lineata and Sphinx
evening, from 30minutes before to 15minutes after moths visibly removed large amounts of pollen on
sunset. In 2001 and 2003, flowers were visited their head and legs (Fig.8D), whereas M. quinque-
frequently at dusk by crepuscular hawkmoths, in- maculatacarriedpollenonthelegsandtheextended
cluding Hyles lineata Fab., Sphinx chersis Hubner, proboscis (Fig. 8F). Female H. lineata moths ovipos-
Sphinx vashti Strecker, and Manduca quinquemacu- ited on leaves of O. acutissima between flower visits,
lata Haworth (Table4, Fig. 8). On 25 June 2001, and several larvae of different developmental stages
a mean of 0.9 visits per flower (N 5 13 watched were observed eating flower buds in this population.
flowers) was observed during a 30-minute period, Flowers remained open and bright yellow until
whereas on 27 June 2001, a mean of 1.3 visits per 10:00hr. on the following day before losing turgor
flower(N514flowers)occurredduringa20-minute and turning deep brick red. Small halictid bees,
period(Table 4).HyleslineataandtheSphinxspecies Evyleus (Lasioglossum) aberrans Crawford, collected
first arrived at flowers of O. acutissima 30minutes pollen from individual anthers within an hour after
after sunset (20:45 MST) and continued to forage sunrise but did not frequently touch the extended
sporadicallyuntil22:00 hr.,whenobservationsended stigmas (Fig.8B). Mule deer (Odocoileus hemionus
due to darkness. Manduca quinquemaculata individ- Rafinesque) were abundant in 2003 and browsed
Description:Apr 26, 2007 Field observations in its type locality in northeastern Utah, U.S.A., revealed
antennal responses from Manduca sexta L. and other. Table 2.