water transport in vesselless angiosperms

14 Pages·2009·0.67 MB·English

Int.J.PlantSci.168(8):1113–1126.2007.
!2007byTheUniversityofChicago.Allrightsreserved.
1058-5893/2007/16808-0001$15.00 DOI:10.1086/520724
WATER TRANSPORT IN VESSELLESS ANGIOSPERMS: CONDUCTING
EFFICIENCY AND CAVITATION SAFETY
U. G.Hacke,*J. S.Sperry,1, T. S.Feild, Y.Sano,§E.H. Sikkema, andJ. Pittermann
y z y k
*DepartmentofRenewableResources,UniversityofAlberta,EdmontonT6G2H1,Canada; DepartmentofBiology,UniversityofUtah,
y
257S1400E,SaltLakeCity,Utah84112,U.S.A.; DepartmentofEcologyandEvolutionaryBiology,UniversityofTennessee,
z
Knoxville,Tennessee37996,U.S.A.;§LaboratoryofWoodyPlantBiology,GraduateSchoolofAgriculture,
HokkaidoUniversity,Sapporo060-8589,Japan;and DepartmentofIntegrativeBiology,
k
UniversityofCalifornia,Berkeley,California94720,U.S.A.
Twostructure-functionhypothesesweretestedforvessellessangiospermwood.First,vessellessangiosperm
wood should have much higher flow resistance than conifer wood because angiosperm tracheids lack low-
resistancetorus-margopits.Second,vessellesswoodoughttobeexceptionallysafefromcavitationifthesmall
cumulative area of pits between tracheids confers safety (the pit area hypothesis). Data were obtained from
branch wood of 19 vesselless angiosperms: Amborella trichopoda, Trochodendron aralioides, Tetracentron
sinense,and16WinteraceaefromTasmannia,Zygogynum,Bubbia,Pseudowintera,andDrimys.Contraryto
thefirsthypothesis,vessellessandconiferspecieswithnarrowtracheids(belowca.18mm)hadsimilararea-
specific resistivities. The reason was that vesselless angiosperms had an intertracheid pit resistance (mean
1662MPasm!1)thatwasnearlyaslowasthatofconifers(661MPasm!1)andmuchlowerthanthatof
eudicotintervesselpits(336681MPasm!1).Lowpitresistancewasassociatedwithgreaterpitmembrane
porosityinferredfromscanningelectronmicroscopyobservationsandsiliconepenetrationandmayrepresent
incipientpitmembraneloss.Pitresistancewasoftengreaterinwiderangiospermtracheidsandobscuredany
dropinwoodresistivitywithtracheidwidth.Insupportofthesecondhypothesis,vessellesswoodsaverageda
cavitation pressure of 3:460:3 MPa, which is low for their wet habitats. In agreement with the pit area
!
hypothesis, resistance to cavitation increased with decreasing total pit area between conduits. However,
vesselless angiosperms were more vulnerable for a given pit area than eudicots, consistent with their more
permeablepitmembranes.Smalltotalpitareabetweenconduitsmayallowangiospermtracheidstohavemore
porousmembranes forconducting efficiencywithoutcreating a cavitation problem.
Keywords:basalangiospermphysiology,cavitation,ecologicalwoodanatomy,vessellessangiosperms,xylem
andvessel evolution.
Introduction until relatively recently, its functional properties have attracted
less attention (Feild and Holbrook 2000; Feild et al. 2000,
The ancestral angiosperm is thought to be vesselless, as is 2002;FeildandArens2005).
Amborellatrichopoda(Amborellaceae),atthebaseoftheex- Our previous work on eudicots and conifers (Pittermann
tantangiospermtree(SoltisandSoltis2004).Thesubsequent et al. 2005, 2006a, 2006b; Hacke et al. 2006; Sperry et al.
originofvesselsisusuallyconsideredtobeanimportantpart 2006) provides an explicit hypothesis for the magnitude of
oftheangiospermsuccessstory.However,vesselsinangiosperms flow resistance in vesselless angiosperms. We predicted that
apparently arose long before major radiations, and the ves- angiosperm tracheids should be ca. 38 times more resistant
selless condition persists in two otherwise derived lineages, to flow per diameter and per length than either conifer tra-
theWinteraceae(fivegenera,65species)andTrochodendrales cheidsoreudicotvessels(Pittermannetal.2005).Thepredic-
(twomonotypicfamilies, Trochodendraceae andTetracentra- tionwasbasedonangiospermtracheidshavingthesameshape
ceae; Young 1981; Doyle 2000; Feild et al. 2002; Feild and andpitfrequencyasconifertracheidsbuthavingangiosperm
Arens 2005). This pattern suggests that the vesselless con- pit membranes instead of the torus-margo membranes of
ditionmaynotbeinallwayshydraulicallyinferiortoatleast conifers. The homogenous,microporous pit membranes be-
themore‘‘primitive’’vessel-bearingsituation.Justhowineffi- tweenangiospermvesselsofeudicotshadca.60timeshigher
cient is vesselless angiosperm wood for transporting water? area-specificflowresistancethanthetorus-margomembrane.
Andhowvulnerableisittocavitationbywaterstress?Although Assuming the same high resistance for intertracheid pits pre-
the anatomy of vesselless wood has been well studied (Bailey dictsthe38-foldgreaterresistanceoftheangiospermtracheid.
1953;Carlquist1975,1983,1987,1988,1990,1992a,1992b), Thehighresistancehypothesizedforangiospermtracheidssug-
gests a tremendous advantage to vessel evolution—decreasing
1Authorforcorrespondence;[email protected].
resistance per unit length by spacing high-resistance endwall
ManuscriptreceivedDecember2006;revisedmanuscriptreceivedApril2007. pitting much farther apart. In contrast, conifers achieved the
1113
Table 1
Species,CollectingLocale,Habitat,GrowthForm,andtheFractionoftheTracheidWallAreaOccupiedbyIntertracheidPits(F )
P
Species Abbreviation Region,elevation,andlat.,long. Habitatandgrowthform F 6SE
P
Amborellaceae:
AmborellatrichopodaBaill. ATR NewCaledonia,nearSarramea, Tropicalmontanecloudforest; .0726.005
slopesofPlateaudeDogny; understory-subcanopyshrub
650m;21"37913.40S,
165"529.590E
Trochodendrales:
TetracentronsinenseOliver TSI Berkeley,California,Berkeley NativetoSWChina;broad-leaved .1596.009
BotanicalGarden;small evergreen-mixeddeciduous
open-growntree forests;1100–3500m,tree
to40m
Trochodendronaralioides
Seibold&Zucc. TAR Japan,vicinityofKyoto,Kuta Temperatebroad-leafforest; .1806.006
experimentalforest;500m evergreenripariantree
Winteraceae:
BubbiaqueenslandianaVink BQU Australia,northernQueensland, Tropicalmontanecloudforest; .0576.006
MainCoastRange;867m; understory-subcanopytreelet
16"35911.70S,145"17950.60E tosmalltree
Bubbiasemecarpoides
(F.Muell.)B.L.Burtt BSE Australia,northernQueensland, Tropicalmontanecloudforest; .0646.007
KurandaRange;500m; understory-subcanopytree
16"50913.40S,145"40920.60E
DrimysgranadensisL.f. DRI CostaRica,HerediaProvince, Tropicalmontanecloudforest; .0636.004
VolcanBarva;2300m; normallyanunderstoryto
10"06945.70N,84"07916.60W subcanopytreelet;fencerow
treeswerecollected
DrimyswinteriJ.R.Forst.
&G.Forst DWI Chile,CoquimboRegion;550m; SameasD.granadensis,except .0776.004
30"409300S,71"409300W temperateforest
Pseudowinteraaxillaris
(J.R.Forst.&G.Forst.)
Dandy PAX NewZealand,NWendofS.island, Warmtemperatelowlandand .0456.004
WakamaramaRange;350m; lowermontaneforest;
40"48928.80S,172"28953.50E understoryshrub
Pseudowinteracolorata
(Raoul)Dandy PCO NewZealand,westcoastof Warmtemperatelowlandtohigher .0536.004
S.island;140m;42"52930.80S, montaneforest,commonin
171"109.140E disturbedareas;smalltree
Pseudowinteratraversii
(Buchan.)Dandy PTR NewZealand,NWendofS.island, Montaneforestmarginsandscrub; .0626.004
WakamaramaRange;900m; smallcompactshrub
40"48901.60S,172"27947.10E
Tasmanniainsipida
R.Br.exDC. TIN Australia,northernQueensland, Broad-leavedtropicalrainforest; .0636.006
MainCoastRange;1080m; understorytree
16"31921.90S,145"15956.70E
Tasmanniainsipida
R.Br.exDC. TIS Australia,northernNewSouth Broad-leavedtropicalrainforest; .0606.006
Wales,MainRangeNat’lPark. understorytree
700m.28"17932.10S,
152"26938.30E
Tasmanniastipitata
(Vick.)A.C.Sm. TST Australia,northernNewSouth Tallmoisteucalyptforestandrain .0766.006
Wales,ForestlandStateForest; forest;erectunderstoryshrub
1050m;29"12947.00S,
152"079.670E
Tasmanniamembranea
(R.Muell.)A.C.Sm. TME Australia,northernQueensland, Tropicalmontanecloudforest; .0636.004
BartleFrerepeak;1500m; understory-subcanopytreelet
17"2399.90S,145"48923.40E tosmalltree
HACKE ETAL.—WATERTRANSPORT IN VESSELLESS ANGIOSPERMS 1115
Table 1
(Continued)
Species Abbreviation Region,elevation,andlat.,long. Habitatandgrowthform F 6SE
P
Tasmanniasp.(probably
Tasmanniapurpurescens) TSP Australia,greaterSydney Cultivatedtree,campusof .0776.006
MacquarieUniversity,Sydney
ZygogynumbailloniiTiegh. ZBA NewCaledonia,nearNoumea, Tropicalmontanecloudforest; .0626.004
MontDzumac;900m; understory-subcanopysmalltree
22"01947.10S,166"2899.50E
ZygogynumbicolorTiegh. ZBI NewCaledonia,nearSarramea, Tropicalmontanecloudforest; .0766.007
slopesofPlateaudeDogny; understory-subcanopysmalltree
860m;21"37913.40S,
165"529.590E
Zygogynumcrassifolium
(Baill.)Vink ZCR NewCaledonia,southofNoumea, Canopyshrubinriparianmaquis .0536.004
RivieredesLacs;180m; scrub
22"10951.90S,166"50949.50E
Zygogynumpancherii
(Baill.)Vinkssp.pancherii ZPA NewCaledonia,nearNoumea, Tropicalmontaneforest; .0596.004
Mt.Koghisslopes;400m; subcanopysmalltree
22"10939.60S,166"30922.90E
Zygogynumpommiferum
Baill.ssp.pommiferum ZPO NewCaledonia,nearSarramea, Subcanopytropicalmontanecloud .0746.008
TableUnio(PicVincent);680m; forest;subcanopysmalltree
21"35938.90S,165"46926.70E
sameendbyevolvingtorus-margopitmembraneswithmuch narrowporesofthemembrane.Thesafetyofaconduitfrom
lowerflowresistance(Pittermannetal.2005). cavitation depends on the single largest membrane pore that
The conifer and eudicot comparison groups also predict a willformtheweakestcapillaryseal.Accordingtothepitarea
specific allometry between flow resistance and conduit diam- hypothesis, the size of this largest pore is influenced by the
eter (Sperry et al. 2006). In these groups, conduit resistance total area of pit membrane in a conduit: the more pores that
standardized for length and lumen cross-sectional area (con- are present, by chance the larger will be the single largest
duitarearesistivity)decreasedwithdiametertothe 2power, pore(Hargraveetal.1994;Choatetal.2005).Insupportof
!
ascalingthatisconsistentwiththeHagen-Poiseuilleequation. thehypothesis,thereisarelativelytightrelationshipbetween
Consistent with the theory (Lancashire and Ennos 2002), this increasing pit area per vessel and increasing vulnerability to
was associated with conduit length increasing with diameter cavitationintheeudicotdataset(Hackeetal.2006).
tothe1.5poweracrossspecies.Theevidentcontrolofconduit The tracheids of vesselless angiosperms lack a torus, and
shape resulted in an approximately constant contribution of they should obey the pit area hypothesis as eudicot vessels
endwallstothetotalflowresistanceregardlessofconduitsize. do. Accordingly, the small pit area to be expected in these
Theendwallcontributionwas64%64%inconifertracheids tracheids should confer very high resistance to cavitation. If
and 56%62% in eudicot vessels. These percentages come the pore size distribution in the intertracheid pit membranes
closetominimizingtheconduitarearesistivityifoneassumes isapproximately similar to that in eudicot intervessel pits, we
species-specificconstraintsonconduitlength.Dovessellessan- ought to be able to predict the cavitation pressure from their
giospermsfollowthisseeminglyoptimalallometry,andifnot, pitareabasedontherelationshipseenineudicots.
whatpreventsthemfromdoingso? To test these hypotheses of flow resistance and cavitation
Theeudicotdata(Hackeetal.2006)alsosuggestanexplicit pressureinvessellessangiosperms,wemeasuredtheseparam-
hypothesis for the cavitation properties of vesselless angio- eters on the branch wood of 19 vesselless species from three
sperms.Vessellessangiospermsshouldbemoreresistanttocavi- groups:Amborellaceae(onespecies),Trochodendrales(twospe-
tation than eudicots because tracheids have much less total cies), and Winteraceae (16 species). We also evaluated whether
interconduitpitareaperconduitthanvessels.Thisprediction these vesselless species show a trade-off between increasing
isbasedonthepitareahypothesis,whichstatesthatthevul- resistance to cavitation on the one hand and increasing flow
nerabilitytocavitationbywaterstressincreaseswiththetotal resistanceorwooddensityontheother.
areaofinterconduitpittingperconduit(Wheeleretal.2005).
The pit area hypothesis is based on the air-seeding mecha-
Material and Methods
nismforcavitationbywaterstress(Zimmermann1983).Neg-
ativesappressureduringwaterstresscancausecavitationby
Plant Material
pulling air from embolized conduits into functional conduits
through the interconduit pits (Tyree et al. 1994). Homoge- Species, collecting locale, and habitat are listed in table 1.
nouspitmembranessealtheairoutbycapillaryforcesinthe Collection of the material was often assisted by an expert in
1116 INTERNATIONAL JOURNAL OF PLANTSCIENCES
theregionalflora(see‘‘Acknowledgments’’).Inallbuttwocases as the flow resistance (pressure difference per volume flow
(Tasmanniasp.,Tetracentronsinense),plantswerefromtheir rate,symbolizedbylowercaser)perlength(pressuregradient
native habitats. Stem pieces at least 20 cm long and ca. 0.8– per volume flow rate). Resistivity is the reciprocal of con-
1.5 cm thick and lacking major side branches were cut from ductivity;we use bothtermsas a matter of mathematicalor
one or more trees from a single population. Care was taken graphical convenience. Resistivity was based on the net flow
to avoid senescing or otherwise unhealthy material. Herbar- rate,whichwasthetotalflowunderahydraulicheadof4–5
ium vouchers were made for most species and filed in J. S. kPaminustheusuallynegative‘‘background’’flowintheab-
Sperry’s collection. Where necessary, identification was con- senceofadifferenceinhydraulicheadacrossthestem.Back-
firmed by botanical experts (‘‘Acknowledgments’’) and the ground flow was measured before and after the pressurized
herbarium study. The Tasmannia sp. collected from the cam- flow,withthe averageusedto calculate the netflow.Eachof
pusofMacquarrieUniversitywastentativelyidentifiedasTas- thesethreeflowmeasurementswastheaverageofsixsuccessive
manniapurpurescens,butthiswasnotconfirmedbecausethe 10-smeasurements, with the flowrate monitored gravimetri-
tree died in a heat wave. One species, Tasmannia insipida, is callyusinganelectronicbalance.Asingleresistivitymeasure-
represented twice, but from very distant populations: one ment, including the installment of the stem, took less than
samplewasfromanapparentlydisjunctpopulationinnorth- 15 min. Resistivity was corrected to 20"C to standardize for
ernQueensland(Australia),andtheotherwasfromtheheart temperature-dependent viscosity effects (solution temperature
of its range in New South Wales. By necessity, most of our was a few degrees higher). After the resistivity measurement,
vesselless species (16 of 19) came from the most diverse ves- safranin dye (0.1%–0.05% w/w) was siphoned through the
sellessfamily,theWinteraceae. stems tostain functionalxylem.Inmoststems, all xylem was
After collection, stems were wrapped in plastic bags and functional.Thecross-sectionalareaofstainedxylemwasmea-
either Express Mailed or hand-carried to J. S. Sperry’s labo- suredatthecenterofthestemandmultipliedbytheresistivity
ratory at the University of Utah, where their hydraulic prop- to give a ‘‘sapwood area resistivity’’ symbolized as R (see
Xa
ertiesweremeasuredassoonaspossible.Ingeneral,thiswas table2forlistofparameterdefinitions,abbreviations,andunits).
within 2–3 wk of collection. During storage, stem surfaces The value of R depends on the number of conduits per
Xa
were kept dry to avoid fungal growth, and material was re- sapwood area, conduit diameter and length, and the proper-
frigerated when possible. On two occasions, the USDA in- ties of interconduit pitting. To break down R into these
Xa
spection process created unusually long delays and the stems components, we adopted the same methods used for the eu-
were not measured until nearly 4 wk after collection. How- dicot and conifer studies. These studies should be consulted
ever,wenoticednoanomaloushydraulicbehaviorinanystems formoredetailedpresentationsofequationsandassumptions
regardlessofstoragetime.Dyeperfusionsshowedcontinuous (Hacke et al. 2006; Pittermann et al. 2006a). Measurements
circumferential staining, with sapwood extending from cam- of the lumen number and cross-sectional area per sapwood
bium to pith in nearly every stem and species. In all species,
conductivity showed similar time courses and vulnerability
Table 2
curves showed similar variability between stems of a species.
Extensive anatomical observation of each species indicated DefinitionofParameters,Abbreviations,andUnitsUsedinText
noblockageoftracheidsfromfungalgrowthorothercauses.
Abbreviation Definition Unit
The consistent behavior and appearance of the material sug-
gestednoeffectofdifferentstoragetimesontheresults. AP Totalareaofinterconduitpitsper mm2
Most data points represent species means based on mea- conduit
A * A wheretracheidlengthisfrom mm2
surements of six stems per species. Minimum flow resistance P P
silicone/Uvitexinjections
(in absence of any reversible embolism) and anatomical pa-
D Averageconduitdiameterofa mm
rametersweremeasuredonthesamesetofsixstems.Another
species,correspondstoR
L
set of six stems was required for cavitation ‘‘vulnerability
F Fractionofconduitwallsurfacearea None
P
curves,’’ and a final set of five to six stems was used for sili- occupiedbyinterconduitpits
cone injections to estimate tracheid lengths. This made for a L Averageconduitlength mm
minimum collection size of 17 stems per species. Additional rP Flowresistancethroughpitson MPasm!1
material was required for air-permeability experiments and membranearea–specificbasis
scanningelectronmicroscopy(SEM)observationsmadeona rW Averageflowresistanceacrossone MPasm!3
subsetofspecies. endwall
RC Averageconduitresistivity MPasm!4
RW Averageconduitendwallresistivity MPasm!4
Flow Resistance and Anatomy RL Averageconduitlumenresistivity MPasm!4
RCa Lumenarea–specificflowresistivity MPasm!2
Six stems per species were cut to 14.5-cm lengths under-
ofxylem
water and immediately flushed for 30–45 min with 20 mM RXa Sapwoodarea–specificflowresistivity MPasm!2
KCl solution indeionized water at ca.75 kPa to remove any ofxylem
reversibleembolism.Stemswerefittedtoatubingsystemwhere
Note. Resistanceisdefinedasthepressuredifferencepervolume
hydraulicresistivityofthesame20mMKClsolutionflowing
flowrateandisnotstandardizedforlengthoftheflowpath.Resistiv-
through the stem was measured. The KCl solution was used ityispressuregradientpervolumeflowrateandisstandardizedfor
tocontrolforioniceffectsonflowresistance(Zwienieckietal. length. Both resistance and resistivity can be expressed on a cross-
2001). Resistivity (symbolized by uppercase R) was defined sectional-areabasis.
HACKE ETAL.—WATERTRANSPORT IN VESSELLESS ANGIOSPERMS 1117
area were made on the same stems measured for R (mean Standarderrorswereestimateddirectlyfrommultiplesam-
Xa
of 10 subsamples per stem). The R value multiplied by the ples for most parameters (R , R , R , R , R , F , L, and
Xa Xa Ca C L W P
average conduit number per sapwood area gave the average D). For those parameters that could be calculated only from
conduit resistivity, symbolized as R . The average lumen re- species means (r , A , r ), errors were propagated through
C P P W
sistivity(R ,at20"C)forthestemwasdeterminedfromlumen thevariousequations.Allparametersareaveragesforallfunc-
L
diametermeasurementsaccordingtotheHagen-Poiseuilleequa- tionalconduitswithinthebranchxylemofthespecies.
tion.Weusedthediameter(D)correspondingtoR torepresent
L
theaveragetracheiddiameterperstem.TheRCmultipliedby Silicone Injections
thecorrespondinglumenarea(fromD,assumingasquarelu-
To check for ‘‘cryptic vessels’’ in these vesselless woods
men)gavetheaverageconduitarearesistivity,orR .
Ca
(Feild et al. 2000), we also measured conduit lengths by the
Assuming that lumen and endwall resistances are in a se-
silicone injection method as described in detail previously
ries, we estimated the average endwall resistivity (R ) as the
W
(Wheeler et al. 2005). Cryptic vessels would result if inter-
conduit resistivity minus the Hagen-Poiseuille lumen resistiv-
tracheidpitmembranesweremissingorcontainedexception-
ity(R R R ).Eachparameterwasdeterminedindivid-
W C L
¼ ! ally large pores. The silicone has been shown to penetrate
ually for the six stems to estimate a standard error for the
margo-sized pores in conifers, which often exceed 0.3 mm in
species.
span, but not the much smaller pores of intervessel pit mem-
Theendwallresistivity,R ,wasbrokendownintoitsana-
W
branes (Andre 2005). Stems were flushed to remove embo-
tomical components. The endwall was defined as the area of
lism and then injected at 25–50 kPa overnight with a 10 :1
overlap between tracheids that links them into longitudinal
silicone-hardenermixture(RTV-141,Rhodia,Cranbury,NJ).
files. We assumed that, on average, one-half of the tracheid
Initially, we colored the silicone with red pigment (Silastic
overlaps with upstream tracheids and the other half overlaps
LSPRD11,DowCorning,Kendallville,IN)asinourprevious
withdownstreamones.Accordingly,theaverage distancebe-
work. However, we found that the average particle size was
tween endwalls equaled half of the average tracheid length
toolarge(0:6060:11mm)topotentiallytravelwith thesili-
(Lancashire and Ennos 2002). The average resistance of an
conethroughtracheidendwallswithlarge-pitmembranepores.
endwall, r , was estimated as R multiplied by the average
w W
Species were reinjected with a soluble fluorescent whitening
distance between endwalls. Tracheid length was measured in
agent (CibaUvitex OB, CibaSpecialty Chemicals,Tarrytown,
macerations, averaging from ca. 80 to 140 tracheids per spe-
NY).TheUvitexwasdissolvedinchloroform(1%w/w),and
cies.Theendwallresistanceinturnwasexpressed ontheba-
one drop of silicone-hardener mix per gram was added. The
sis of the endwall pit dimensions. The pit area resistance (r )
P
solutionwasmixedquicklytoavoidevaporationandprecipi-
wasdefinedasthepitresistanceonamembranesurfacearea
tationoftheUvitex.Thesiliconefluorescedbrightlyandserved
basis.Eventhoughthepitresistancewasstandardizedforarea,
asaconvenientmarkerforfilledconduits,whichwereother-
itwasnotstandardizedforthepathlengthfromlumentolu-
wiseclearinmicroscopesections.
men,soitisaresistanceratherthanaresistivityinourtermi-
Injectedstemsweresectionedat1,2,3,6,and10mmand
nology (table 2). The endwall resistance is the r divided by
P
occasionallyatlongerintervalsfromtheinjectedsurface,and
the pit membrane area per endwall: r 2r =A , where A
w ¼ P P P the number of filled conduits per cross-sectional area were
representsthetotalinterconduitpitareaperconduit(bothend-
counted. These counts formed the basis for calculating the
wallstogether).WemeasuredA andcalculatedr fromr .
P P W
conduitlengthdistribution.
Theintertracheidpitareaperconduit(A )usedtocalculate
P
Forvessels,whichhaveamuchmorespread-outlengthdis-
r wasestimatedfromthesamemacerationsusedtomeasure
P
tributionthantracheids,wehaveusedasingle-parameterex-
tracheid length. From the macerated tracheids,wedetermined
ponential decay function to fit the decrease in the density of
the average fraction of tracheid wall surface area occupied
silicone-filled vessels (N ) with distance (L) from injection:
byintertracheidpits(F ).Intertracheidpitswererareontan- L
P N N e kL,whereN N atL 0andkisthebest-fit
gentialwallsand essentiallyconfined to theradialwalls. The L ¼ 0 ð! Þ 0¼ L ¼
extinction coefficient. The second derivative of this equation
fraction of the radial wall occupied by intertracheid pits was
multiplied by L/N gives the fraction of injected conduits of
measured on typically 12–15 tracheids per species. The frac- 0
lengthL(Cohenetal.2003).Tracheids,however,haveamuch
tionwas dividedby2(toaccountforthenonpittedtangential
shorter length distribution, and an artifact of this function
walls) to estimate F . The average A per species was then
P P
wastogivesignificantfractionsoftracheidsthatwereunreal-
calculatedasF timestheaveragesurfaceareaofatracheid
P
isticallyshort—ofjustafewtenthsofamillimeter,forexample.
perspecies( 4DL,whereLisaveragetracheidlength).Inone
¼ The artifact arises because the rate of decay for the function
species, Trochodendron aralioides, where latewood and early-
increasesmonotonicallyasLgoesto0.Thefunctionalsoforces
wood were rather distinct and latewood was abundant, the
the distribution to be strongly short skewed, which is appro-
pitting pattern differed so dramatically between the corre-
priateforvessels(ZimmermannandJeje1981)butnotneces-
sponding tracheids that we weighted the F value according
P
sarilysofortracheids.Tobetterrepresenttracheiddistributions,
to the relative fraction of intertracheid contact in earlywood
we used a Weibull function to fit the decline in silicone-filled
and latewood. Intertracheid contact was quantified in cross
tracheidswithdistanceL:
sectionasthetotalperimeteroftracheidsincontactwitheach
olattheewroiondawgarsocwatlhcurlaintegd,.aTnhdetwheeigfrhatcetdioFn ienqueaalreldywthoeodeaarlnyd- NL¼N0eð!kLÞc; ð1Þ
P
wood F times the early-wood contact fraction plus the late- wherecisasecondcurve-fittingparameter.Forbest-fitc>1
P
woodF timesthelatewoodcontactfraction. (nearly always the case), the rate of decay does not increase
P
1118 INTERNATIONAL JOURNAL OF PLANTSCIENCES
monotonically as L goes to 0. The second derivative multi- Cavitation ‘‘Vulnerability Curves’’ and Evaluation
pliedbyL/N givesthefractionofvesselsoflengthL: of the Pit Area Hypothesis
0
FL¼ckLc!1eð!kLÞcðckLcþ1!cÞ; ð2Þ ityVwulintherianbcirleitaysicnugrlvyensesghaotwivethxeylleomssporfeshsyudrrea.uCliucrvcoesndfourctsivix-
andtheintegralfromL 0to‘ofF gives1.However,for stems per species were measured with the standard centrifu-
¼ L
theusualcasewherec>1,F becomesnegativebelowamin- galforce method (Alder et al. 1997). Stems were flushed and
L
imum length L c 1 =ck1=c. We used L to repre- measuredforconductivity(reciprocalofresistivity)aspreviously
min¼½ð ! Þ ’ min
sent the minimum tracheid length and adjusted F from described. Measurements were repeated after stems were spun
L
equation(2)accordinglybydividingitbytheintegralofequa- in a custom-designed centrifuge rotor to progressively more
tion (2) from L to ‘. The Weibull cannot be integrated negative xylem pressures. Stems were held at each pressure
min
analytically,sonumerical methods were used.Typicaltracheid- for 15 min to ensure complete cavitation before being taken
length distributions using this method were still short skewed, outoftherotoranddirectlyremeasuredforconductivity(Al-
although much less so than vessel distributions. Except for deretal.1997).Thepercentagelossinconductivityfromthe
comparisonwithmacerationdata(fig.6),werepresentedthe original value was plotted versus the negative pressure to
silicone-based length distribution using the log-transformed give the vulnerability curve. Curves were fit with a Weibull
mean. function for each stem, and mean cavitation pressure was
Fig.1 Xylemflowresistivityonasapwood(A)andconduitlumen(B)areabasisversusaverageconduitlumendiameter.Eachdatapointisa
species mean; species abbreviations from table 1. Conifer and eudicot branch wood data are from Hacke et al. (2006) and Pittermannet al.
(2006a).PredictedvaluesforvessellessangiospermsinA(crosses,dashedline)arebasedonconifertracheidgeometryandeudicotintervesselpit
resistance(Pittermannetal.2005).Incontrasttoconifersandeudicots,vessellessangiospermsshownodecreaseinresistivitywithwiderdiameter
(species means6SE). Solid diagonal in B is the Hagen-Poiseuille area resistivity for an unobstructed conduit lumen. Dashed diagonal in B
representsconduitswhereendwallscontribute67%ofthetotalresistivity.Thisapproximatesthescalingseeninconifersandeudicotsandalso
minimizesthearearesistivity,assumingeachspecieshasafixedconduitlength(Sperryetal.2006).
HACKE ETAL.—WATERTRANSPORT IN VESSELLESS ANGIOSPERMS 1119
Air-Permeability Experiments
Totestforalinkbetweenpitmembranepermeabilitytoair
andcavitationpressure,wedidair-injectionexperimentsona
subsetofspecies(Zygogynumpancherii,Zygogynumbicolor,
Zygogynumpommiferum, Amborellatrichopoda,Tasmannia
insipida [New South Wales population], Pseudowintera axil-
laries,Pseudowinteratraversii,Bubbiasemecarpoides,Bubbia
queenslandiana,Trochodendronaralioides,andDrimysgrana-
densis). Dowelsofsecondaryxylemwerewhittledfromstems
andattachedatoneendtoacompressednitrogensource.Dowels
were ca. 2.2–2.5 cm long and ca. 3–5 mm in diameter. Gas
pressurewasraisedstepwiseandtheflowrateofgasmeasured
by its rate of displacement of water from a reservoir onto an
electronicbalance.Theflowratewasconvertedtoanapproxi-
matepneumaticconductivitybydividing itby theair-pressure
gradient across the dowel. The conductivity was approximate
because it did not take into account the compressibility of air.
Fig. 2 Log-log scaling of average conduit length with average Dowelswereusedratherthanentirestemstoavoidairpenetra-
conduitdiameter.Datapointsarespeciesmeans.Eudicotandconifer
tionthroughthepith.Typically,therewasapressurethreshold
branch wood data from Hacke et al. (2006) and Pittermann et al.
belowwhichnomeasurableairflowoccurredandabovewhich
(2006a). Vessel lengths are log-transformed means of short-skewed
the airflow increased abruptly. An ‘‘air-entry’’ pressure across
vessel-length distributions. Tracheid lengths are arithmetic means
thedowelwascalculatedfromplotsofpneumaticconductivity
from macerations. Reduced major axis regression lines are shown.
Slopesarenotdifferentfromthe1.5valuerequiredforproportional- versus gas pressure using the extrapolation method of Sparks
ityofendwallandlumenresistivity,assumingthatpitarearesistance and Black (1999). Theair-entry pressure represents the mini-
andpitareaperconduitsurfaceareaareindependentofconduitsize mumair-seedingpressurefortheintertracheidpitmembranes.
(LancashireandEnnos2002). Theaverageair-entrypressurecalculatedfrominjectionswas
comparedwiththeaverageair-entrypressurecalculatedfrom
stemvulnerabilitycurvesusingthesameextrapolationmethod.
calculatedfromthebestfit.Averagemeancavitationpressure Toafirstapproximation,thetwoshouldagreeifcavitationis
wasbasedonn 6stemmeansperspecies. causedbyair-seedingacrosspitmembranes.
¼
To evaluate the pit area hypothesis,wecompared the mean
cavitationpressurewiththemeanpitareapercavitatingcon-
Scanning Electron Microscopy Observations
duit (A ). Normally, this would be the same A per tracheid
P P
usedtocalculatethetracheidpitarearesistance(r );thiswas Asubsetofspecies(A.trichopoda,T.aralioides,D.granadensis,
P
based on the length of tracheids measured in macerations. P. traversii, and Z. pancherii) were examined in the SEM to
However,asexplainedin‘‘Results,’’theconduitlengthscalcu-
lated from silicone/Uvitex injections were often longer than
the actual tracheid. Many tracheids were connected by pit
membraneswithholeslargeenoughtopassthesilicone,which
is normally stopped by pit membranes without the torus-
margo structure. Any pit membrane that lacks a torus and is
permeabletosiliconeshouldalsohaveaverylowair-seeding
pressure. A series of tracheids linked by such perforated pit
membraneswouldbeexpectedtocavitateasasingleunit.To
testwhetherthecavitationpressureofthismultitracheidunit
islinkedtoitstotalpitarea,thepitareaisbestestimatedfrom
thelengthofthemultitracheidunit.Forthatreason,weused
the silicone/Uvitex-based lengths for calculating the A asso-
P
ciated with cavitation pressure. We designate this parameter
asA *todistinguishitfromtheA pertracheidcalculatedfrom
P P
maceration-basedtracheidlength.
Wood Density Fig.3 Flowresistanceacrosspitsonapitmembranesurfacearea
basis.Boxesmarkthetwenty-fifthtoseventy-fifthpercentilerangeand
To determine whether resistance to cavitation in vesselless
themedianvalue(middlehorizontalbar)forallspeciesmeansforthe
angiospermsincreases withwooddensityasseen forconifers
three xylem types. Vertical lines show tenth to ninetieth percentile
andeudicots(Hackeetal.2001),wemeasuredthedryweight
range,withoutlyingpointsassymbols.Meanssharingthesameletter
per fresh volume of secondary xylem samples whittled from were not significantly different (one-way ANOVA followed by least
the same stems used for the vulnerability curves. Measure- significantdifferencetest).EudicotandconiferdatafromHackeetal.
mentdetailsarereportedbyHackeetal.(2001). (2006)andPittermannetal.(2006a).
1120 INTERNATIONAL JOURNAL OF PLANTSCIENCES
assess the porosity of the intertracheid pit membranes. For versii; D 9:1 mm) to 92% for the species with the widest
¼
comparison, we also include observations of intervessel pit tracheids(Z.pommiferum;D 26mm).Overall,77%62%
¼
membranes in Trimenia neocaledonica (Sperry et al. 2007). ofthexylemresistivitywasintheendwall.Thiscontrastswith
Wood samples were preserved in 50% ethanol and shipped conifers,whereendwallresistivitywasindependentoftracheid
toHokkaidoUniversity,Japan,wheretheywerepreparedfor sizeandaveraged64%64%overall(Pittermannetal.2006a).
microscopy by Y. Sano. Samples were prepared as described Theincreaseinpercentageofendwallresistivitywithdiam-
previously(Sano2005;SanoandJansen2006).Inbrief,sam- eter could not be attributed to tracheid allometry. Figure 2
ples were cut into small cubes (ca. 5 mm3) in wet condition. shows that the across-species allometry of length and diam-
The blocks were air-dried after dehydrationin absolute etha- eter in angiosperm tracheids was very similar to conifer tra-
nol. Wood surfaces to be observed were exposed by splitting cheids. The slope of the reduced major axis regression was
(cracking)witharazorblade.Theywerecoatedwithplatinum not different from 1.5 in either case (P 0:05). This value
(
by vacuum evaporation (JE-4, Jeol, Tokyo) and viewed with maintains a size-independent endwall percentage if the frac-
a field-emission scanning electron microscope (JSM-6301F, tionoftracheidwallthatispitted(F )andpitarearesistance
P
Jeol)atanacceleratingvoltageof2.5kV. (r ) do not change systematically with diameter (Lancashire
P
andEnnos2002).ThiswastrueforF (datanotshown),but
P
Statistics itwasnottrueforrP.
Thevalueofr increasedstronglywithincreasingdiameter
P
Allometricscalingofconduitlengthwithdiameterwasde-
and pit area (fig. 4), the major exception being in Tetracen-
termined from reduced major axis regression (Niklas 1994)
tronsinense(fig.4,TSI).Thetrendforincreasingpitareare-
ratherthanleastsquaresregression(fig.2)becauseneitherdi-
sistance in some of the species with larger tracheids (fig. 4,
mension was independently controlled. Probability threshold
e.g.,ZPO,DGR)madetheendwallresistivityincreasinglylim-
forstandardstatisticaltestsofsignificanceforregressionlines
iting and prevented the overall tracheid area resistivity from
and mean differences was 0.05 unless noted. We used SPSS
decreasingwithdiameter(fig.1B).
softwareforthesetests(SPSS,Chicago).
TheSEMobservationsindicatedageneraltendencyformore
porouspitmembranestobeassociated withlowerr (fig.5).
P
Of the five species examined in the SEM (labeled in fig. 4),
Results themostconspicuouslyporousmembraneswerefromspecies
withthelowerpitarearesistances:P.traversii(r 11:161:4
P
twiehSsaa(tpRwwCaeo)ohodafdavrepesraseedlrlieecsstisestdainvfigrtoiioemssp(ceRormXnaisf)ewarnetrdreaccchooennidsdiuddietirmaabreelnyasilroeesnsssisttahinvaidn- 5MM:8PPaaMssPmma!!s11;m;fifi!gg.1.;55fiABg),.),T5arCon)cd.hCoAdamernlbqdourroiesntllaa(1rta9rl9iico2hiado)epsho(adrsPan¼¼(orP2te4¼d:4t2h62e:94ex:61-
eudicotpitresistances(fig.1,cf.filledcircleswithcrosses).Con- ceptionallyporousmembranesinT.sinense(hisfig.34),consis-
tributing to this lower-than-expected resistivity was the fact tentwithitsverylowpitresistance(rP 7:661:4MPasm!1).
¼
thatangiospermtracheidstendedtobelongerfortheirdiam-
etersthanconifertracheids,althoughnotbyastatisticallysig-
nificantdegree(fig.2).Longertracheidswouldtendtominimize
theconduitendwallcontribution.Themainreasonforthelow
resistivity,however,wasthattheestimatedpitarearesistance
forangiospermtracheids(fig.3)wasmuchlower(mean6SE:
1662 MPas m!1) than the eudicot average (336681 MPa
s m!1; Hacke et al. 2006) and not significantlydifferent from
thetorus-margovalueforconifers(5:761:3MPasm!1;Pit-
termannetal.2005).Thefraction of tracheid wall pitted was
nodifferentbetweenangiosperm(F 0:07560:008;table1)
P
¼
and conifer tracheids (F 0:08660:008; Pittermann et al.
P
¼
2006a).
Another departure from expectation was that R for ves-
Ca
selless angiosperms did not decrease with increasing tracheid
diameter to the 2 power as was statistically the case in co-
!
nifer tracheids and also eudicot vessels (fig. 1B; Sperry et al.
2006).Remarkably,R forvessellessangiospermsdidnotde-
Ca
crease at all with increasing diameter, and two of the species
withthelargesttracheids(ZygogynumpommiferumandDrimys Fig. 4 Pit resistance standardized for pit membrane area versus
intertracheid pit area per tracheid. Pit areas were based on tracheid
granadensis) had R values high enough to be within the pre-
Ca
length measured in macerations. Species means6SE; abbreviations
dictedrange(fig.1B,ZPO,DGR).
fromtable1.Largertracheidswithlargerpitareas(e.g.,DGR,ZPO)
Concordant with R being independent of tracheid diam-
Ca generally had higher area-specific pit resistances. Area-specific pit
eter in vesselless angiosperms, the percentage of the xylem
resistancealsoincreasedwithtracheiddiameter(datanotshown).A
resistivity in tracheid endwalls increased significantly with notableoutliernotincludedintheregressionisTetracentronsinense
tracheiddiameter (r2 0:38;datanotshown),from64%for (TSI).Otherlabeledspecieswereviewedinthescanningelectronmi-
¼
the species with the narrowest tracheids (Pseudowintera tra- croscope(fig.5).
Fig. 5 Scanning electron microscope images of intertracheid (A–E) and intervessel (F) pit membranes. Insets show magnified sections of
membrane.Scalebar 1mminallcases;micrographstakenatbetween37000(D)and310,000(A)magnification.A,Pseudowinteratraversii.
¼
B,Trochodendronaralioides.C,Amborellatrichopoda.D,Zygogynumpancherii.E,Drimysgranadensis.F,Trimenianeocaledonica.
1122 INTERNATIONAL JOURNAL OF PLANTSCIENCES
In contrast, pit membranes from the species with the highest
pitresistance,Drimysgranadensis,hadrelativelysmallpores
(rP 4269:9 MPa s m!1; fig. 5E). By comparison, pores
¼
are often not visible in the SEM for intervessel pit mem-
branes of many vessel-bearing species such as Trimenia neo-
caledonica (fig. 5F). Not fitting this correspondence between
observedporosityandpitresistancewasZygogynumpancherii,
whichhadrelativelylowpitresistance(r 23:565:6MPas
P
m!1;fig.4)butbarelyvisiblemembranepo¼res(fig.5D).
The silicone injections also provided indirect evidence for
relativelylargeporesinintertracheidpitmembranesofatleast
some of the vesselless species (fig. 6). When the red pigment
oflargeparticlesize(0:6060:11mm)wasusedwithsilicone,
mean tracheid length was similar whether calculated from
the injections or measured directly from macerations (fig. 6,
open circles; regression not different from 1:1 line). How-
ever, when the soluble Uvitex dye was used, mean tracheid
Fig.7 Vulnerability-to-cavitationcurvesforvessellesstaxashow-
length measured by injection exceeded the maceration mea-
ingpercentagelossofconductivityasafunctionofnegativepressure.
surement in several species, especially so in A. trichopoda, Curves are means6SE for six stems per species. Symbols are iden-
Bubbiaqueenslandiana,Zygogynumbaillonii,Tasmanniamem- tifiedfrommostvulnerable(topright,opensymbols)toincreasingly
branea, and Zygogynum bicolor (fig. 6, filled circles, abbre- resistant(left,fromtoptobottom,grayandblacksymbols).Species
viations). On average, the mean tracheid length from the abbreviationsfromtable1.
silicone/Uvitex injectionwas 1:560:1 times longer thanthe
meanfrommacerations,rangingfromjustover1.0inseveral
speciestoahighof2.9inA.trichopoda. Queensland population of Tasmannia insipida is not shown
Vulnerabilitycurvesofthevessellesstaxawereverysimilar but was very similar to the New South Wales population for
in shape: a sigmoidal relationship with a well-defined cavita- the first half of the vulnerability curve (up to 50% loss of
tionthreshold(fig.7).However,therewasconsiderablevaria- conductivity). However, beyond this point, the xylem was
tioninthemeancavitationpressure,rangingfrom 1:560:1 much more resistant. Because of the very different curve for
!
MPainT.sinenseto 5:560:2MPainP.traversii.Thegrand this apparently disjunct Queensland population, we did not
!
mean for all species was 3:460:3 MPa, indicating a sur- use it in further analysis but represented the cavitation resis-
!
prising degree of protection from cavitation, given their gen- tanceofthespecieswiththeNewSouthWalesdata.
erally wet and often shaded habitats (table 1). The north Overall, the threshold cavitation pressure compared favor-
ably with the threshold air-entry pressure for the air-injected
species(fig.8;regressionconfidencelimitsinclude1:1line),
supportingthehypothesisthatairentrythroughpitmembranes
was causing the cavitation. Species furthest from agreement
were P. traversiiand T. aralioides, wherethe average air-entry
pressurewaslessthanthecavitationthreshold,andB.queens-
landiana,whichshowedtheoppositesituation(fig.8).
The vesselless angiosperms did appear to follow the pit
area hypothesis but not in the simple way expected from the
publishedeudicotdata(fig.9;Hackeetal.2006).Thevessel-
less taxa were consistent with the hypothesis because there
wasasignificantrelationshipbetweenincreasinginterconduit
pit area (A * from silicone length data) and increasing vul-
P
nerability to cavitation. The slope of the log-log relationship
( 1:3060:33)wasnodifferentfromthatoftheeudicotdata
!
( 1:9260:20). Using the pit area estimated from macera-
!
tions (A instead of A *) also gave a significant relationship
P P
(r2 0:45;datanotshown),withasimilarslope.Notably,the
¼
vessellesstaxahadsubstantiallylesstotalpitareaperconduit
for the same cavitation pressure than eudicot taxa (fig. 9).
Fig.6 Conduitlengthsmeasuredbythesilicone-injectionmethod This is consistent with intertracheid pits having more porous
versusbythemacerationmethod.Solidlineis1:1agreement.Data
membranesthanintervesselpits(see‘‘Discussion’’).
points are species means6SE, non-log-transformed. In five species,
Therewasevidenceofatrade-offbetweenincreasingresis-
the silicone contained a red pigment whose average particle size of
tance to cavitation and increasing sapwood area resistivity
ca. 0.6 mm inhibited penetration through porous intertracheid pit
membranes (open symbols). All but one of these species, plus the (r2 0:28)andasignificanttrend(r2 0:27;datanotshown)
¼ ¼
remaining ones, were reinjected with silicone plus soluble Uvitex fornarrower tracheids tobe more resistantto cavitation.We
fluorescentdye(filledsymbols).Speciesabbreviationsfromtable1. also found a relationship between greater wood density and