Wood Density Profiles and Their Corresponding Tissue Fractions in Tropical Angiosperm Trees

14 Pages·2017·2.4 MB·English

Article
Wood Density Profiles and Their Corresponding
Tissue Fractions in Tropical Angiosperm Trees
TomDeMil1,2,*,YegorTarelkin1,2,3,StephanHahn4 ,WannesHubau1,2 ,
VictorDeklerck2 ,OlivierDebeir4 ,JorisVanAcker2,CharlesdeCannière3,
HansBeeckman1 andJanVandenBulcke2
1 RoyalMuseumforCentralAfrica,WoodBiologyService,Leuvensesteenweg13,B-3080Tervuren,Belgium;
[email protected](Y.T.);[email protected](W.H.);
[email protected](H.B.)
2 UGCT-UGent-Woodlab,LaboratoryofWoodTechnology,DepartmentofEnvironment,GhentUniversity,
CoupureLinks653,B-9000Gent,Belgium;[email protected](V.D.);[email protected](J.V.A.);
[email protected](J.V.d.B.)
3 LandscapeEcologyandPlantProductionSystemsUnit,UniversitéLibredeBruxelles,CP264/2,
B-1050Bruxelles,Belgium;[email protected](C.d.C.)
4 LaboratoryofImage,SignalprocessingandAcoustics—BrusselsSchoolofEngineering,Universitélibrede
Bruxelles(ULB),B-1050Brussels,Belgium;[email protected](S.H.);[email protected](O.D.)
* Correspondence:[email protected];Tel.:[phone]
(cid:1)(cid:2)(cid:3)(cid:1)(cid:4)(cid:5)(cid:6)(cid:7)(cid:8)(cid:1)
(cid:1)(cid:2)(cid:3)(cid:4)(cid:5)(cid:6)(cid:7)
Received:12October2018;Accepted:27November2018;Published:7December2018
Abstract:Wooddensityprofilesrevealatree’slifestrategyandgrowth. Densityprofilesare,however,
rarely defined in terms of tissue fractions for wood of tropical angiosperm trees. Here, we aim
atlinkingthesefractionstocorrespondingdensityprofilesoftropicaltreesfromtheCongoBasin.
Coresof8treespecieswerescannedwithX-rayComputedTomographytocalculatedensityprofiles.
Then,coresweresandedandtheoutermost3cmwereusedtosemi-automaticallymeasurevessel
lumen,parenchymaandfibrefractionsusingtheWekasegmentationtoolinImageJ.Fibrewalland
lumenwidthsweremeasuredusinganewlydevelopedsemi-automatedmethod. Anassessment
of density variation in function of growth ring boundary detection is done. A mixed regression
modelestimatedtherelativecontributionofeachtraittothedensity,withaspecieseffectonslope
andinterceptoftheregression. Position-dependentcorrelationsweremadebetweenthefractions
andthecorrespondingwooddensityprofile. Onaverage,densityprofilevariationmostlyreflects
variationsinfibrelumenandwallfractions,butthesearespecies-andposition-dependent: onsome
positions,parenchymaandvesselshaveamorepronouncedeffectondensity. Themodellinking
densitytotraitsexplains92%ofthevariation,with65%ofthedensityprofilevariationattributedto
thethreemeasuredtraits. Theremaining27%isexplainedbyspeciesasarandomeffect. Thereisa
clearvariationbetweentreesandwithintreesthathaveimplicationsforinterpretingdensityprofiles
inangiospermtrees: theexactdrivinganatomicalfractionbehindeverydensityvaluewilldependon
thepositionwithinthecore. Theunderlyingfunctionofdensitywillthusvaryaccordingly.
Keywords: wooddensity;woodspecificgravity;parenchyma;fibres;vessellumen;tissuefractions;
CongoBasin;X-rayCTdensitometry;tropicalforests
1. Introduction
Wood specific gravity or wood density is an integrating variable [1] and a property defined by
chemicalandanatomicaltraits[2].Ingeneral,wooddensityismeasuredintheframeworkofecologyand
carbon dynamics within and across species, taxa, and across a continent [3]. Wood density varies
considerably from pith to bark, especially in tropical trees [4,5]. A large portion of the variation,
Forests2018,9,763;doi:10.3390/f9120763 www.mdpi.com/journal/forests
Forests2018,9,763 2of14
especiallyinsoftwood,occursbetweenringboundariesduetoseasonaldynamicsofwoodformation[6],
whichisofparticularimportanceinclimatereconstructionsandstudiesonimpactofclimatechange
ontrees.
To study these inter and intra-annual patterns of density variation, high-resolution density
profilesareneeded. High-resolutiondensitometrycanbeobtainedwithblueintensity[7]orX-ray
densitometry[8]systemsthatcombinedensitometrywithanatomicalobservations[9,10],andmany
otherdevicesexistaswell[11]. X-rayComputedTomography(CT)microdensitometryisatechnique
that allows obtaining 3D density volumes, that can be converted to large datasets of density
profiles[12,13],withresolutionrangingfromsubmicronlevel[14]tocoarserresolutionstoperform
tree-ringanalysis[15,16]. Maximumlatewooddensityofconifersisnexttotree-ringwidthoneof
theparametersusedto,forinstance,reconstructsummertemperature[17],andearlywoodisbeing
exploredaswell[18]. Therelationbetweendensityandcellfractionsisstraightforwardforconiferous
species: tracheidsizeluminaandthecellwallwidthoftracheidsdeterminethedensityvariationina
ring[18]. Assuch,agrowthringboundaryismarkedatthepositionwherelatewoodtoearlywood
transitionisassociatedwithasharpdecrease. Densitometryisapromisingtechniquetoassistintree
ringanalysisofsometropicaltreespecies[19],butknowledgeofunderlyinganatomicalpatternsthat
determineringboundariesisnecessaryforitsapplication.
DensityprofileswereconstructedfortropicaltreespeciesshortlyaftertheestablishmentofX-ray
densitometry[8]forfastgrowingspeciessuchasAucoumeaklaineanaPierre[20]andTerminaliaivorensis
A.Chev.[21],andmorerecentlyinsemi-aridregions[22]inordertodelineateringboundariesorto
assesswoodquality. However, theseprofilesneverrelatedwooddensitytotheunderlyingwood
anatomicalfractions. Densityprofilesforangiospermsareindeedlessclearlydefinedcomparedto
softwoods, as the underlying anatomical signal is more variable due to a combination of vessels,
axialandrayparenchymaandfibresthatvarywithinthegrowthring. Attemptstodisentangledensity
profiles locally intovessels, parenchymaandfibres havebeenreportedfor angiosperm temperate
speciessuchasQuercuspetraea(Matt.) Liebl.[23]andinsometropicalones[24], butaplethoraof
angiosperm tropical species with a wide range of wood anatomical patterns [25] remains largely
understudied. Inthetropics,thereisalsoahighlevelofunclearandlessdistinctringbordersand
intra-andinter-ringvariabilityofanatomicalpatternsduetoweakseasonality. Moreover,densityis
regardedasafunctionaltrait[1],butmanyuncertaintiesexistinwhattheunderlyingcomponents
causedensityvariations,whichforangiospermsisacombinationofwoodanatomicalvariables[26]
andotherchemicalcomponents.
In this study, we define wood density profiles in terms of (i) ring boundaries and (ii) the
underlying wood anatomical fractions. To do so, we compare high-resolution X-ray CT profiles
with vessel, parenchyma and fibre wall fractions of eight common tropical species at high detail
(110µm). Weinvestigatewhichanatomicalfeatureinfluenceswooddensitythemostandexplorethe
effectofspecies. Finally,weassesslocalradialvariationsthathelpininterpretingthedensityprofile.
Wehypothesizethatduetohighanatomicalvariability,thisrelationwillbedifferentfromspeciesto
speciesanddependsontheradialposition.
2. MaterialsandMethods
2.1. StudySiteandSamples
To assess the general variability between density and tissue fractions, tree increment cores
(ø5.15mm)weretakenatbreastheightfromeightspecies,onecoreperspecies(Table1)fromtheLuki
Biospherereserve(KongoCentralprovince;5.39◦S13.4◦E),theBoloboforestsofMalebo(Mai-Ndombe
province; 2.49◦ S, 16.50◦ E) and the Yoko reserve (Tshopo province; 0.33◦ N, 25.31◦ E), all in the
DemocraticRepublicoftheCongo. Thosespecieswereselectedinordertorepresentawiderange
ofwooddensities,lifestrategiesandleafsheddinghabits. AllsamplesreceivedauniqueTervuren
Wood identification number of the Xylarium (Tw) and are stored at the Africamuseum Tervuren
Forests 2018, 9, x FOR PEER REVIEW 3 of 14
(http://www.africamuseum.be/en/research/collections_libraries/biology/collections/xylarium).
These samples were also verified via reference samples where microsections show detailed anatomy
of the selected species (Figure 1).
Forests2018,9,763 3of14
Table 1. Summary of the collected species that were used to link wood density to the tissue fractions.
Diameter at breast height (DBH) of the sampled tree and Tervuren Wood identification number at the
Xylaxryiulamrium(h t(tTpw:/) /isw giwvewn. aafsr wicealml. Luisfee ustmra.tbegey/ eann/d rleeasef aprhcehn/olcooglyle wctaios nexst_rlaicbtreadr fireosm/b [i2o7l]o. .g y/collections/
xylarium). Thesesampleswerealsoverifiedviareferencesampleswheremicrosectionsshowdetailed
Tw DBH Life
anatomTyesotf stpheecsieelse ctedspecies(Figure1)F.amily Site Phenology
label (cm) strategy
Anonidium mannii Shade
Table1.SummaryoftheTcwol6le4c3te6d6 speAciensntohnaatcweaeer euse2d6.t7o linkMwaoloedbod ensitytothetissuefEravcetriognrse.en
(Oliv.) Engl. & Diels bearer
Diameteratbreastheight(DBH)ofthesampledtreeandTervurenWoodidentificationnumberatthe
Canarium schweinfurthii Light
xylarium(Tw)isgivenaswell.Lifestrategyandleafphenologywasextractedfrom[27].
Tw78659 Burseraceae 31.7 Luki Deciduous
Engl. demanding
EntaTnedsrtoSpphercaigemsa TwLabel Family DBH Site Life Phenology
(cm) SLtriagthegt y
angolense (Welw.) Tw78727 Meliaceae 37.2 Malebo Deciduous
demanding
Anonidiummannii(Oliv.)
C.DC. Tw64366 Annonaceae 26.7 Malebo Shadebearer Evergreen
Engl.&Diels
Milicia excelsa (Welw.) Light
Canariumschweinfurthii Tw78904 Moraceae 32 Luki Light Deciduous
C.CE. nBgelr.g Tw78659 Burseraceae 31.7 Luki ddeemmaannddiinngg Deciduous
PEnytcannadnrothphursa gamngaoalnengosliesn se LLiigghhtt
TwT7w87728782 7MyrMisteilciaacceeaaee 4347 .2 MMalaelebboo EDveecridgureoeuns
(W(Welewlw.). )WCa.DrbC.. ddeemmaannddiinngg
Miliciaexcelsa(Welw.) Light
Polyalthia suaveolens Tw78904 Moraceae 32 Luki Shade Deciduous
C.C.Berg Tw68771 Annonaceae 27.5 Yoko demanding Evergreen
Engl. & Diels bearer
Pycnanthusangolensis Light
Tw78728 Myristicaceae 44 Malebo Evergreen
Staud(tWia eklawm.)eWruanrben.sis deSmhaanddei ng
Tw68713 Myristicaceae 23.8 Luki Evergreen
PolyaWlthaiarbsu. aveolens bearer
Tw68771 Annonaceae 27.5 Yoko Shadebearer Evergreen
Engl.&Diels
Tetrorchidium
StaudtiakamerunensisWarb. Tw68713 Myristicaceae 23.8 Luki ShaLdiegbheta rer Evergreen
didymostemon (Baill.) Tw78828 Myristicaceae 33.4 Luki Evergreen
Tetrorchidiumdidymostemon Tw78828 Myristicaceae 33.4 Luki demLaignhdting Evergreen
(BPaailxl. )&P aKx.H&oKff.Hmo.f fm. demanding

Figure1.Cont.
Forests2018,9,763 4of14
Forests 2018, 9, x FOR PEER REVIEW 4 of 14

FFigiguurree 11. .MMicircoroscsocoppici csescetcitoinons soof fTTeervrvuurerenn WWoooodd rreefeferreenncceess ssaammppleles soof fththee ccoorrreressppoonnddiningg spspeeciceies,s ,
sshhoowwiningg wwoooodd aannaattoommyy,,a nadnd,w, whehneanv aaivlaabillaeb,lteh,e tghreo wgrtohwritnhg rbinougn bdoaurinedsa(wriehsi te(wtrhiaitneg ltersia).nTghleess).p eTchiees
sapreecimeso astrley mdoifsftulyse dpifofruosues p,owriotuhsv, ewsistehl svseoslsietalsr ysoolirtasrliyg hotrl ysliggrhotulyp egdr.ou(pa)edA.n (oan) idAinuomnimdiaunmn imi,aTnwni7i2, ,
T(wb)7C2,a (nba)r iCuamnasrcihuwme isncfhuwrtehiini,fuTrwth6ii1, 1T5w1,6(1c1)5E1n, t(acn) dErnoptahnradgrompahraanggmolae nasnegTolwen1s4e3 T2,w(d14)3M2,i l(idci)a Mesicleiclsiaa eTswce9ls4a3 ,
T(we)94P3y,c (nea)n Pthyucnsaannthguolse nansigsolTewns1is5 1T3w,1(5f)13P, o(lfy) aPltoilayaslutiaav seuolaevnesolEennsg El.ng&l. &D Dieiles.ls. TTww3322661188,, ((gg)) SSttaauuddtitaia
kkaammereurunneennssisis TTww6622770077,, aanndd ((hh)) TTeettrroorrcchhiiddiiuumm ddiiddyymmoosstteemmoonn TTww22004499.. SSccaallee bbaarr:: 220000 µµmm..
2.2. X-RayCTDensitometry
2.2. X-Ray CT Densitometry
Woodcoreswereinsertedinpaperstrawsandovendriedfor24hoursbeforebeingscannedat
Wood cores were inserted in paper straws and oven dried for 24 hours before being scanned at
110µmusingtheNanoWoodCTfacility[28],developedincollaborationwithXRE(X-rayEngineering,
110 µm using the NanoWood CT facility [28], developed in collaboration with XRE (X-ray
www.XRE.be).Thescannedimageswerereconstructed(GPUGeForceGPX7704GB)withtheOctopus
Engineering, www.XRE.be). The scanned images were reconstructed (GPU GeForce GPX 770 4 GB)
reconstruction software package ([29,30], X-ray Engineering, www.XRE.be) and further processed
with the Octopus reconstruction software package ([29,30], X-ray Engineering, www.XRE.be) and
(extractionofvolumes,tiltcorrection,tangentialcorrection)throughatailoredtoolchain[13]. X-rayCT
further processed (extraction of volumes, tilt correction, tangential correction) through a tailored
wooddensityprofilesresultfromcorrectingstructuredirectionoffibreandringdeviations,calibrating
toolchain [13]. X-ray CT wood density profiles result from correcting structure direction of fibre and
withareferencematerial,andaveragingintangentialandaxialdirection[15],basedon30–40voxelsat
ring deviations, calibrating with a reference material, and averaging in tangential and axial direction
agivenradialplane. Theobtainedwooddensitymeasurementsaredefinedasovendrywoodweight,
[15], based on 30–40 voxels at a given radial plane. The obtained wood density measurements are
dividedbyovendryvolume,aftercalibratingwithareferencematerialofknowndensity[31].
defined as ovendry wood weight, divided by ovendry volume, after calibrating with a reference
material of known density [31].
Forests2018,9,763 5of14
2.3. WoodAnatomicalMeasurements
Coreswerethengluedinawoodensampleholderandtheirtransversalsurfacewassandedwith
increasinglyfinersandpaper(grit80–1200). Thelastthreecmoftheouterwood(untilthecambial
zone)wasimaged(StreamMotion,Olympus,Japan)onascanningstage(SCAN100×100,Märzhäuser
Wetzlar,Germany)withacamera(3.2MP,2048×1536pixels;UC30,Olympus(Tokyo,Japan))mounted
Forests 2018, 9, x FOR PEER REVIEW 5 of 14
onareflectedlightmicroscope(BX60,Olympus,magnification×10). Thefinalimagesconsistedof
2.3. Wood Anatomical Measurements
overlapping images stitched together with the Multiple Image Alignment algorithm (Olympus).
TheseimagesCwoereres wthereen thuesne gdluteod imn ae awsouordeenth saemfrpalec htiooldnesr oanfdd tihfefeirr ternantsavnerastaol smurifcaacle fwraasc staiondnesd( wveitshs ellumina,
increasingly finer sandpaper (grit 80–1200). The last three cm of the outer wood (until the cambial
axialandrayparenchyma,fibres). Forthefibrefraction,thelocalfibrewallfractions(theratiofibre
zone) was imaged (StreamMotion, Olympus, Japan) on a scanning stage (SCAN 100 × 100,
wall/totalcellsize)weremeasuredaswell.
Märzhäuser Wetzlar, Germany) with a camera (3.2 MP, 2048 × 1536 pixels; UC30, Olympus (Tokyo,
Japan)) mounted on a reflected light microscope (BX60, Olympus, magnification ×10). The final
2.3.1. Pareinmcahgyesm caonasnisdtedV eosf soevlserlapping images stitched together with the Multiple Image Alignment
algorithm (Olympus). These images were then used to measure the fractions of different anatomical
Delineatingvessellumina(VL)andparenchymafractions(PRM)isatime-consumingprocess,
fractions (vessel lumina, axial and ray parenchyma, fibres). For the fibre fraction, the local fibre wall
and thereffroarcetiotnhse (sthee frraaticot fiiobnres wwalel/rtoetasle cgelml seiznet) ewderue smineagsutrhede atsr awienlla. b le Weka plugin [32] in ImageJ [33].
TheWekaplugincombinesasetofmachinelearningalgorithms,toperformpixel-basedsegmentations
2.3.1. Parenchyma and Vessels
of images. In a graphical user interface, a sub-region was selected as training set and all vessels,
parenchymaaDndelifinebarteinsgw veesrseelm luamniunaa l(lVyLl)a abnedl lpeadreinnchtyhmata sfruabct-iroengs i(oPnR.MT) hise at rtiamine-ecdonmsuomdinegl wpraocsetshs,e napplied
and therefore these fractions were segmented using the trainable Weka plugin [32] in ImageJ [33].
ontheentireimage. Ablockprocessingapproachwasimplementedforhandlingtheratherlarge
The Weka plugin combines a set of machine learning algorithms, to perform pixel-based
imagesansdegrmedenutactiinognsp orf oimceasgseisn. Ign sap gereapdhoicnal ausdere sinktteorfpacce,o am supbu-rteegrio(nIn wteals® seCleocrteed™ ais5 t-r4ai5n7in0gC sePt Uan@d 3.20GHz
3.20GHz,a1ll6 vGesBselRs, ApaMre)n.cThyhmeai manadg fiebrwesa wsetrhe emraenfouarellys luabbedllievdi dine tdhaut ssuinb-greagiuons.e Trh-de etrfiainneedd mgordidel, weaasc hgridcell
wasclassitfiheend aspeppliaerda otne ltyhea enndtirael ilmgargied. Ace bllloscwk perroecesstsiitncgh aepdpraofatcehr wparso cimespsleimngenrteeds ufoltr ihnagndinlinag cthlae ssification
rather large images and reducing processing speed on a desktop computer (Intel®Core™i5-4570 CPU
of the entire image. The results are displayed as a single probability map per feature, and each
@ 3.20 GHz 3.20 GHz, 16 GB RAM). The image was therefore subdivided using a user-defined grid,
probabilityeamch agpridh caesll twhaes scalamsseifisediz seepaasrattheely oarnidg ainll aglriidm caelglse waenred steitvcehreyd apftiexre plrhocaesssainvga rleusueltbinegt win eae n0and1,
indicatingcltahsseifpicraotibona bofi ltihtye eonftitrhe aimtafgeea.t Tuhree .reTsuhletss aerep droisbplaabyeildi tays ma sainpgslew perorbeavbiilsituy amllayp cpoern fveeatrutreed, tobinary
mapsviatahnrde esahcohl pdrionbga,bifloitry tmhaepfi hnasa tlhde esalimnee asitzieo ans othfet ohreigainnaal timomagiec aanldf reavcetriyo pnixse.lV heass sae vlaslufiel lbeedtwweeinth sawdust,
0 and 1, indicating the probability of that feature. These probability maps were visually converted to
notrecognizedbythealgorithm,wereindicatedmanuallyinImageJ(Figure2).
binary maps via thresholding, for the final delineation of the anatomical fractions. Vessels filled with
Binarsyawmduaspt,s nootf rveceosgsneilzseda nbyd thpea arlegnorcihthymm, waewre einrdeiccaotendv meratneudalltyo inp IrmofiagleeJs (Fbiygutrae n2g). entialaveragingin
ImageJ.TheimBinaagrey smwapesr oef avlessoselus saendd paasremncahsykmsa ownerteh ceonovreirgteind atol ipmroafiglees sbyp trainogretnotiaml aevaesraugrining gint helumen
ImageJ. The images were also used as masks on the original images prior to measuring the lumen
diametersandfibrewallthicknesses(Figure2a). Fibresweredelineatedassubtractoftotalfraction
diameters and fibre wall thicknesses (Figure 2a). Fibres were delineated as subtract of total fraction
(i.e.,1)minusthesumofparenchymafractionandvessellumenfractionatagivenradialposition.
(i.e. 1) minus the sum of parenchyma fraction and vessel lumen fraction at a given radial position.

Figure2. FFiglouwre c2h. aFrlotwfcohrardt efolirn deealitnienagtinvge vsesseslesls aanndd ppaarernecnhcyhmyam ona ao nsegamseengt mof ean stanodfead ssuarnfadcee dof surfaceof
Polyalthia suaveolens with the trainable Weka plugin. The plugin assesses the probability for different
PolyalthiasuaveolenswiththetrainableWekaplugin.Thepluginassessestheprobabilityfordifferent
portions of the image to be classified as vessel or parenchyma before being manually corrected. (a)
portions of the image to be classified as vessel or parenchyma before being manually corrected.
Original sanded surface is imaged, (b) probability map for parenchyma with mismatches for vessel
(a)Originalulmsainnad, e(cd) csourrrefcatcioeni sthirmouagghe md,an(bua)l pinrodibcaatbioinli toyf mveasspelsf otro pbeacroemnec hbyinmaray wimiatghesm oifs (md)a vtcehsseesl forvessel
lumina,(c)correctionthroughmanualindicationofvesselstobecomebinaryimagesof(d)vessel

luminaand(e)parenchymafractions.(f)Fibrefraction,asthesubtractofthetotalimagewithvessel
andparenchymafraction.Scale500µm.
Forests 2018, 9, x FOR PEER REVIEW 6 of 14
lumina and (e) parenchyma fractions. (f) Fibre fraction, as the subtract of the total image with vessel
and parenchyma fraction. Scale 500 µm.
2.3.2. Fibres
Fibre fraction (F) is then subdivided in fibre walls and fibre lumina. Fibre lumina diameter and
wall thickness were measured using pattern recognition based on self-developed software. First, the
original images were binarized in order to clearly separate fibre walls from lumina for further steps
(Figure 3b): light-coloured pixels (corresponding to lumina) were selected with the colour
thresholding tool in ImageJ. Adjacent selected pixels formed different shapes that were filtered based
on their circularity and size to ensure that the retained shapes correspond to the fibre lumina.
For the detection of individual fibre lumina and wall thickness, we followed a protocol [34]
consisting of a segmentation, tracking and measurement sequence. Segmentation detects and labels
individual cells based on criteria such as the size, the shape and the surrounding pixel intensity
variation ([35,36]; Figure 3c). Tracking allows connecting different fibre’s centroids to form the
Forests2018,9,763 6of14
longest chains in the radial direction by using a nearest neighbours (KNN) algorithm: from the nth
centroid, the nth+1 was searched within a specified distance and with a maximum allowed angle of
2.3.2. Fibres
deviation from the radial direction (Figure 3d). Pixel intensity was then measured along numerous
pathsF cirberaetefrda cbtyio tnhe(F t)raiscktihnegn, wsuibthd ifvibidreesd luinmfiinbare/wwalallsl sshaonwdifinbgr epilxuemls inwai.thF hibigreh/lluomwi nvaaludeiasm. Tehteer
waniddthw oafl flibthreicsk lnuemsisnwa aenred mfiberaes uwraeldls uwsains gthpenat mteernasurerceodg bnyit cioonunbtainsegd thoen nusemlfb-dere voef laodpjeadcensot fptwixealrse .
wFiitrhst ,hitghhe/loorwig iinnatelnimsitayg veasluweesr erebspineacrtiizveedly.i nA oursdeerr intotecrlfeaacerl ywases pdaervaetelofipebdre tow ianlslspefrcot mthel urmesiunlats faotr
efaucrhth perroscteespss s(tFeipg uarned3 tbo): aldigahptt- ctohleo usergedmpenixtealtsio(cno prraersapmonetdeirnsg ift orelquumirinead). wImerpelesmeleecntteadtiownit ohft thheisc oulsoeurr
itnhtreersfahcoel dwinags tboaosleidn oImn athgee JS.cAikdijta icmenatgsee l[e3c7t]e adnpdi xSecliskfiot rlmeaerdn d[3i8ff]e rliebnrtasrhieasp iens Pthyathtowne.r Tehfiel tceoredde bisa soend
aonn atvhaeiilracbilrec urelaproitsyitoarnyd (shitztepsto://egnitshuureb.tchoamtt/hsteeprehtaahinne/dWsohoadpSeescctoiornre).s p ondtothefibrelumina.

FFiigguurree 33. .FFlolowwcchhaarrt tooff tthhee pprreeppaarraattioionn ooff aa sseeggmmeenntt ooff aa ssaannddeedd ssuurrffaaccee oonn PPyyccnnaanntthhuuss aannggoolelennssisis ffoorr
mmeeaassuurreemmeennt toof ffifibbrree wwaalll ltthhiicckknneessss aanndd lluummininaa ddiaiammeeteterr. .(a(a) )OOrrigigininaal lssaannddeedd aanndd ssccaannnneedd imimaaggee, ,
wwhheerree ppaarreenncchhyymmaa aanndd vveesssseel llluummininaa wweerree mmaasskkeedd ((oobbtatainineedd wwitihth WWeekkaa sseeggmmeennttaattiioonn pplluugginin iinn
imimaaggeeJJ)). .((bb)) BBininaarriizzaattiioonn ooff tthhee imimaaggee inin oorrddeerr ttoo sseeppaarraattee ffiibbrree lluummiinnaa aanndd wwaallllss.. ((cc)) RReeccooggnnititioionn aanndd
lalabbeelllilningg ooff iinnddiivvididuuaal lcceelllsls aanndd ((dd)) ttrraacckkiinngg aanndd lliinnkkiinngg ooff iinnddiivviidduuaall cceellllss.. PPixixeel liinntteennssitiyty iiss tthheenn
aannaalylysseedd aalloonngg tthhee nnuummeerroouuss ccrreeaatteedd ppaatthhss aanndd tthhee wwiiddtthhss ooff lluummiinnaa aanndd ffiibbrree wwaallllss eexxttrraacctteedd.. SSccaalele
110000 µµmm..
For the detection of individual fibre lumina and wall thickness, we followed a protocol [34]
All measurements of the anatomical variables were interpolated to the coarser X-ray CT
consistingofasegmentation,trackingandmeasurementsequence. Segmentationdetectsandlabels
resolution of 110 µm so that each wood density at a given position corresponds to a value of fibre,
individual cells based on criteria such as the size, the shape and the surrounding pixel intensity
vessel lumen and parenchyma fraction, as well as to the ratio of fibre wall/total fibre diameter (further
variation([35,36];Figure3c). Trackingallowsconnectingdifferentfibre’scentroidstoformthelongest
referred to as fibre wall fraction (FW)). Fibre fraction is inversely related to the sum of the vessel
chainsintheradialdirectionbyusinganearestneighbours(KNN)algorithm: fromthenthcentroid,
lumen and parenchyma fraction and is thus a linear combination. Therefore, it was not taken in
thenth+1wassearchedwithinaspecifieddistanceandwithamaximumallowedangleofdeviation
account for further analysis and model construction.
fromtheradialdirection(Figure3d). Pixelintensitywasthenmeasuredalongnumerouspathscreated
bythetracking,withfibreslumina/wallsshowingpixelswithhigh/lowvalues. Thewidthoffibres
2.4. Model and Local Correlation
luminaandfibrewallswasthenmeasuredbycountingthenumberofadjacentpixelswithhigh/low
For the overall relationship between wood density and anatomy, we used a Gaussian linear
intensityvaluesrespectively. Auserinterfacewasdevelopedtoinspecttheresultsateachprocessstep
mixed framework with random intercepts and slopes to investigate the potential of predicting wood
andtoadaptthesegmentationparametersifrequired. Implementationofthisuserinterfacewasbased
density using the vessel lumina, parenchyma and fibre wall fraction as fixed effects and species as
ontheScikitimage[37]andScikitlearn[38]librariesinPython. Thecodeisonanavailablerepository
random effects. This model assesses the importance of a species effect in the relationship between
(https://github.com/stephahn/WoodSection).
wood density and wood anatomy, and was performed in R software (version 3.4.1, R Foundation for
AllmeasurementsoftheanatomicalvariableswereinterpolatedtothecoarserX-rayCTresolution
Statistical Computing, Vienna, Austria).
of110µmsothateachwooddensityatagivenpositioncorrespondstoavalueoffibre,vessellumen
andparenchymafraction,aswellastotheratiooffibrewall/totalfibrediameter(furtherreferred
toasfibrewallfraction(FW)).Fibrefractionisinverselyrelatedtothesumofthevessellumenand
parenchymafractionandisthusalinearcombination.Therefore,itwasnottakeninaccountforfurther
analysisandmodelconstruction.
2.4. ModelandLocalCorrelation
Fortheoverallrelationshipbetweenwooddensityandanatomy,weusedaGaussianlinearmixed
frameworkwithrandominterceptsandslopestoinvestigatethepotentialofpredictingwooddensity
usingthevessellumina,parenchymaandfibrewallfractionasfixedeffectsandspeciesasrandom
effects. Thismodelassessestheimportanceofaspecieseffectintherelationshipbetweenwooddensity
and wood anatomy, and was performed in R software (version 3.4.1, R Foundation for Statistical
Computing,Vienna,Austria).
WD = β +β x +α +α ε +ε (1)
i,j 0,k 1,k i,j,k 0,j,k 1,j,k i,j,k i,j
Forests2018,9,763 7of14
whereWD(i,j)arei-thmeasureofwooddensityofspeciesj, β(0,k)istheinterceptand β(1,k)isthe
slopeforthek-thvariableregardlessofspecies,α(0,j,k)isthedeviationfromtheglobalinterceptand
α(1,j,k)isthedeviationfromtheglobalslopeforthek-thvariableofspeciesj.
Toanalyzethespatiallyexplicitlocalrelationshipbetweenwooddensityandwoodanatomy,
wesubdividedthedensityandanatomicalmeasurementsprofilesintosectionsof30datapointsand
constructed a linear model with parenchyma, vessel lumen and fibre wall fractions as explaining
variables. Allanatomicaldatawasstandardizedperspecies(subtractedthevalueswiththemeanand
dividedbythestandarddeviation)inordertoassesstherelativeimportanceofthethreevariables.
Furthermore,wecalculatedPearsoncorrelationbetweenwooddensity,parenchyma,vessellumen
fraction, total fibre fraction and fibre wall fraction, with a moving window of 30 points (trade-off
betweensufficientdatapointsandthe110µmresolution),whichwasperformedinMatlabR2016b
(Mathworks,Natick,MA,USA).
3. Results
3.1. GeneralWoodDensity—AnatomicalFractionsRelationship
Averagewooddensityvaluesrangefrom510kg·m−3to794kg·m−3. Onaverage,fibresformthe
highestfraction,followedbyparenchymaandvessellumenfractions. Thefibrewallfractionofthe
totalfibrefractionpartvariedbetween0.22and0.49(Table2).
Table2. Measuredvaluesofthe8testedsamples. Wooddensity(WD,kg/m3),parenchymafraction
(PRM),vesselluminafraction(VL),aswellasfibre(F)andfibrewallfraction(FW,definedasthefibrewall
widthdividedbythesumoffibrewallandlumendiameter).Standarddeviationisgivenbetweenbrackets.
Species WD(kg·m−3) PRM(-) VL(-) F(-) FW(-)
A.mannii 513.15(48.77) 0.34(0.07) 0.04(0.03) 0.62(0.09) 0.23(0.06)
C.schweinfurthii 500.91(60.75) 0.13(0.03) 0.09(0.04) 0.78(0.04) 0.22(0.04)
E.angolense 651.89(47.61) 0.23(0.07) 0.08(0.04) 0.69(0.08) 0.35(0.06)
M.excelsa 571.17(57.71) 0.28(0.14) 0.06(0.04) 0.66(0.16) 0.25(0.06)
P.angolensis 570.79(36.30) 0.23(0.03) 0.07(0.04) 0.7(0.04) 0.33(0.03)
P.suaveolens 794.36(25.82) 0.42(0.03) 0.12(0.04) 0.46(0.04) 0.49(0.06)
S.kamerunensis 768.67(26.51) 0.31(0.03) 0.05(0.02) 0.64(0.03) 0.35(0.05)
T.didymostemon 510.30(30.72) 0.19(0.08) 0.13(0.05) 0.67(0.09) 0.25(0.05)
Overall, fibre wall fraction is positively correlated with wood density while vessel lumina
and parenchyma slightly decrease density (Table 3). Fibre wall fraction impacts density the most,
followedrespectivelybyvesselandparenchymafractions. Themodelalsoshowstheimportanceof
thespecieseffectontherelationshipbetweenwooddensityandwoodanatomy: morethan25%of
wooddensityvariabilityisinfluencedbyspecies.
Table3.CoefficientsfromaGaussianlinearmixedmodel(randomintercept+slope)predictingwood
densityfromtheanatomicalvariablesfor8tropicalspeciesinvestigated. Coefficientestimatesare
providedforthefixedeffects(at95%ofconfidenceinterval).
Fixedeffects
Intercept 490.92
Parenchymafraction −24.39
Vesselluminafraction −94.22
Fibrewallfraction 453.67
%ofvariance 65%
Randomeffects Contributiontototalvariance(%)
Species 4
Species|parenchyma 4
Species|vessels 6
Species|fibrewalls 13
Residuals 8
Forests 2018, 9, x FOR PEER REVIEW 8 of 14
Overall, fibre wall fraction is positively correlated with wood density while vessel lumina and
parenchyma slightly decrease density (Table 3). Fibre wall fraction impacts density the most,
followed respectively by vessel and parenchyma fractions. The model also shows the importance of
Forests2018,9,763 8of14
the species effect on the relationship between wood density and wood anatomy: more than 25% of
wood density variability is influenced by species.
FFoorr aallll ssppeecciieess,, tthhee fifibbrree wwaallll ffrraaccttiioonn aappppeeaarrss ttoo bbee tthhee mmoosstt iimmppoorrttaanntt ddrriivveerr ooff wwoooodd ddeennssiittyy
vvaarriiaabbiilliittyy,, aass hhiigghheerr fifibbrree wwaallll ffrraaccttiioonnss iinnccrreeaassee wwoooodd ddeennssiittyy ((FFiigguurree 44aa)).. TThhee rreellaattiioonn ooff vveesssseell
lluummeenn ffrraaccttiioonn wwiitthh wwoooodd ddeennssiittyy iiss,, iinn mmoosstt ccaasseess,, tthhee lloowweesstt ooff aallll aannaattoommiiccaall ffrraaccttiioonnss..
WWhhiillee tthhee ssllooppeess ooff tthhee rreeggrreessssiioonn lliinneess bbeettwweeeenn wwoooodd ddeennssiittyy aanndd tthhee ffiibbrree wwaallll ffrraaccttiioonn aarree
vvaarriiaabbllee bbeettwweeeenn ssppeecciieess bbuutt rreemmaaiinn ppoossiittiivvee ((FFiigguurree 44aa)),, tthhee eeffffeeccttss ooff ppaarreenncchhyymmaa oorr vveesssseellss ccaann
vvaarryy ssttrroonnggllyy bbeettwweeeenn spspeceiceise s(F(iFgiugruer e4b4,cb),.c P).arPeanrcehnycmhyam haash aa gseanegreanl eproasliptiovsei teifvfeecet fofenc ptroonfilperso (fie.lge.s,
(Pe.. gsu.,aPv.esouleanvse oalnends Aa.n mdanAn.iim).a Fnonri it)h.reFeo sratmhpreleess (aEm. apnlgeosle(nEs.ea, nTg. odliednysme,osTt.emdiodny,m Po. satnemgoolnen,sPis.)a, nthgeo lvenesssise)l,
ftrhaectvieosns edlofreasc tnieognadtioveeslyn eingfaltuivenelcye inthfleu ednecnesitthye pdreonfisliet y(Fpirgoufirlee (4Fbi)g. uWreh4ible) .tWheh iinlefltuheenicnefl oufe nficbereosf
rfiebmreasinrse mpaoisnistivpeo stihtirvoeutghhroouutg htohue tpthroefiplero, fitlhee, thinefliuneflnuceen coef opfapreanrecnhcyhmyam aanadn dvvesesseslesl sccaann vvaarryy
ssuubbssttaannttiiaallllyy ffrroomm oonnee ssuubb--sseeccttiioonn ooff tthhee ccoorree ttoo tthhee ootthheerr:: ppaarreenncchhyymmaa ccaann bbee bbootthh ppoossiittiivveellyy aanndd
nneeggaattiivveellyy ccoorrrreellaatteedd ttoo wwoooodd ddeennssiittyy wwiitthhiinn oonnee iinnddiivviidduuaall aanndd vveesssseellss,, aalltthhoouugghh nnoott ssiiggnniifificcaanntt
oovveerraallll,, ccaann ppllaayy aann iimmppoorrttaanntt rroollee llooccaallllyy..
(a) Anonidiummannii (b) Canariumschweinfurthii
800
850
700
750
m-³) 650 600
g. 500
k 550
D ( 400
W 450
300
350 8.5 9.5 10.5 11.5 12.5
7.75 8.75 9.75
(c) Entandrophragmaangolense (d) Miliciaexcelsa
900
1000 800
m-³) 900 700
kg. 800 600
D ( 700
W 600 500
500 400
12.5 13.5 14.5 15.5 12.15 13.15 14.15 15.15
(e) Pycnanthusangolensis (f) Polyalthiasuaveolens
900 1150
800 1050
m-³) 700 950
g. 600 850
k
D ( 500 750
W
400 650
15.25 16.25 17.25 18.25 7.9 8.9 9.9 10.9
(g) (h)
Staudtiakamerunensis Tetrorchidiumdidymostemon
1150 900
g.m-³)1095500 780000
D (k 850 600
W 750
500
650
400
7.3 8.3 9.3 10.3
11.2 12.2 13.2 14.2
Radial distance (cm) Radial distance (cm)

Figure4.Visualinterpretationofthedensityprofile.Sandedwoodcoresectionsinupperpanel,where
Figure 4. Visual interpretation of the density profile. Sanded wood core sections in upper panel,
ringwidthswereindicated. Theoutermostsectionofdensityprofilesof8selectedspeciesfromthe
where ring widths were indicated. The outermost section of density profiles of 8 selected species from
CongoBasin,withazoomshowingtowhatextenttheringboundariesvisuallyindicatedwithwhite
the Congo Basin, with a zoom showing to what extent the ring boundaries visually indicated with
trianglesonthewoodsurface,concurwiththedensityprofileattheringboundary.Wooddensityis
white triangles on the wood surface, concur with the density profile at the ring boundary. Wood
definedastheovendrywooddensity(kg·m−3).Speciesshownare(a)Anonidiummannii,(b)Canarium
density is defined as the ovendry wood density (kg·m−3). Species shown are (a) Anonidium mannii, (b)
schweinfurthii,(c)Entandrophragmaangolense,(d)Miliciaexcelsa,(e)Pycnanthusangolensis,(f)Polyalthia

suaveolens,(g)Staudtiakamerunensis,and(h)Tetrorchidiumdidymostemom.Scalebar500µm.
Forests2018,9,763 9of14
3.2. LocalDefinitionoftheDensityProfile: GrowthRingBoundaryCriterion
Allspecieshavevaryingradialpatternsofwooddensity. Somespecies,suchasP.suaveolensand
S.kamerunensis,showsmall,high-frequencydensityvariationsalongtheprofile(Figure4f,g),whileother
speciesshowlarge,low-frequencyvariations,seenasnarrowandwidegrowthringsrespectively.
Anonidiummanniiboundariesarecharacterizedbyalternatingfibreandparenchymabandsthat
narrowdowntowardsthegrowthringboundary(Figure1a). Thiscausesadensityincrease, seen
intheprofile(Figure4a). Canariumschweinfurthiihasaclearvariationinfibrelumenandwallsize,
causing a density increase (Figure 4b). Entandrophragma angolense has a fibre lumen and wall size
variationcombinedwithterminalparenchyma,butalsowithcleardensityincreases. InMiliciaexcelsa,
there is a fibre variation and in some cases a terminal parenchyma band (Figure 1d). The density
profilesshowclearvariationsindensity,buttheseare,insomecases,maskedbyparenchymabands
(Figure 4d). Ring boundaries of Pyncanthus angolensis are difficult to detect, despite the variation
in density, as the ring boundary is indistinct: fibre lumen decrease is gradual (Figures 1e and 4e).
AlternatingfibreandparenchymabandsdeterminethegrowthringboundaryinPolyalthiasuaveolens,
aswellasdistendedrays(Figure1f),butonthedensityprofilethesesubtleanatomicalchangesare
hardertodetect(Figure4f). Staudtiakamerunensisshowsflattenedfibrestowardstheendofthegrowth
ring,onlyinsomecasesaccompaniedwithaterminalparenchymaband(Figure1g),andthisisclearly
seeninthedensityprofileaspeaks(Figure4g). Tetrorchidiumdidymostemonischaracterisedbyflattened
fibresaswell(Figure4hbutnotseeninFigure1h),butthisissubtle,andalthoughclearvariationis
seen,theringlimitisnotsosharpanddifficulttoobserveinthedensityprofile.
3.3. LocalRadialVariationintheRelationbetweenWoodDensityandWoodAnatomicalFractions
Whenassessingtheexactlocalcorrelationbetweeneachofthevariablesandthewooddensity
valuesatagivenposition,thesevalueschangeconsiderably. Themovingcorrelations(Figure5)show
thateachofthecomponentscanprevailatacertainradialposition.OnspeciessuchasEntandrophragma
angolense (Figure 5c), the vessel lumina fraction does influence the density profile locally at most,
especiallybetweengrowthringboundaries. ForTetrorchidiumdidymostemon(Figure5h)andCanarium
schweinfurthii(Figure5b),parenchymaandtotalfibrefractionactinverselytoahighextentincertain
regionsnegativelyandpositivelyincertainregions. ForPolyalthiasuaveolens(Figure5f),parenchyma
positively influences the density profile. In species such as Staudtia kamerunensis (Figure 5g) and
Anonidium manii, the fibre wall fractions determine wood density variations the most (Figure 5a),
whichcanalsobeseenduetothegrowthringstructureofflattenedfibres(Figure1a,g). Miliciaexcelsa
showshighcorrelationwithfibrewallfractionatthegrowthringboundaries(Figure1d),whereasin
between, parenchymabandsdeterminethedensitypattern(Figure5d). ForPycnanthusangolensis,
thevariationbetweenprevailingfractionsislarge(Figure5e).
Forests2018,9,763 10of14
Forests 2018, 9, x FOR PEER REVIEW 10 of 14

Figure5.Movingsignificant(p<0.05)Pearsoncorrelationplots(windowwidth=30densityvalues)in
Figure 5. Moving significant (p < 0.05) Pearson correlation plots (window width = 30 density values)
upperpanelwithfibrewallfraction(red),fibrelumina(pink),parenchyma(orange),andvessels(blue).
in upper panel with fibre wall fraction (red), fibre lumina (pink), parenchyma (orange), and vessels
Bottompanelshowsfractionswithfibrewalls(red),fibrelumina(pink),parenchyma(orange),andvessels
(blue). Bottom panel shows fractions with fibre walls (red), fibre lumina (pink), parenchyma (orange),
(blue).Speciesshownare(a)Anonidiummannii,(b)Canariumschweinfurthii,(c)Entandrophragmaangolense,
and vessels (blue). Species shown are (a) Anonidium mannii, (b) Canarium schweinfurthii, (c)
(d)Miliciaexcelsa, (e)Pycnanthusangolensis, (f)Polyalthiasuaveolens, (g)Staudtiakamerunensis, and(h)
Entandrophragma angolense, (d) Milicia excelsa, (e) Pycnanthus angolensis, (f) Polyalthia suaveolens, (g)
Tetrorchidiumdidymostemon.
Staudtia kamerunensis, and (h) Tetrorchidium didymostemon.
4. Discussion
4. Discussion
Ageneralpositiveinfluenceoffibrewallfractiononwooddensitywasdescribedbyprevious
studieAs [g2e4n,3e9ra,4l 0p].oFsiitbivree firnafclutieonncsea onfd fifibbrree wwaallll ffrraaccttiioonn owni twhionodth disefinbsirtey fwraacsti odnesdcreitbeermd ibnye pthreevairoeuas
osctcuudpieiesd [2b4y,3w9,a4l0ls].a Fnidbrbey flruamctiionnasw ahnidch fiabfrfee cwtsawll oforadcdtieonns wityitthhien mthoiss tf(iTbareb lfera3c).tiDonif fdeeretenrcmesinine athvee raargeea
wococoudpdieedn sbityy wbeatlwlse eannds pbeyc ielusmarientah uwshdicrhiv eanffebcytst hweofroadc tdioennssiotfy ththeeir manoastto m(Tya.ble 3). Differences in
averAaglteh wouogohd sdeevnesriatyl tbreotpwiceaelns pspeeciceiesss haroew thruinsg dbriovuennd bayr itehsed ferfiacnteiodnbsy ofd tehnesiirt yanflautcotmuayt.i o nsdueto
fibrewAaltlhlothuigckhn seesvsevraalr itartoiopnicsa[l3 s1p,4e1c]ie,sw sehsohwo wrinthga btoouthnedratryiepse sdoeffintiessdu beya ldseonisnifltyu eflnuccetuwaotioodnsd denusei ttyo
afnibdret hwaatlol tthheicrkdneensssi vtyarviaatriioantiso [n3s1a,4r1e],d wuee tsohofrwa ctthioatn ostohtehre trytpheasn ofif btirsessu.eS aolsfoa ri,niftluaepnpceea wrsodoidffi dceunltsittoy
daentde ctthraitn ogtshdere ldimenisteitdy bvyarpiaatrieonncsh ayrme daubea ntod fsraocrtivoensss eoltsh,ewr hthearena fsibforersfl. Satot efnare,d itfi abprpese,arths idsiaffpicpuelat rtso
edveidteecnt t,riansgsse edneliinmFiitgedur ebsy 1paanrden4c.hTyhmisaa bdadnitdios noarl vvaersisaeblisl,i twyhoefrtehaesd feonrs iftlyatpternoefidle fmibarkese,s tfihbisre a-dprpiveaerns
deevnidsietnyt,v aasr isaeteino nins Fdiigffiucrue l1t atondd Feitgecutrea m4. oTnhgisa aldldthiteioontahle vrsar[i4a2b]i.litDye onfs tihtye dperonfisilteys pcraonfisleh mowakterso pfiibcrael-
tdrereiv-reinn gdbeonusnitdya vriaersiaintiosonms edicfafsiceus,lta ntod dweeteschto wamedonthga atlwl othoed odtehnesristy [i4s2o].n Dlyecnlosisteyl yprreolfaitleeds tcoansp sehcioews
tropical tree-ring boundaries in some cases, and we showed that wood density is only closely related
to species with clear fibre wall variation. The visibility of macroscopically visible growth ring