ESTIMATING TREE TRANSPIRATION ACCURATELY DEPENDS ONWOOD TYPE AND SPECIES:

A STUDY OF FOUR SOUTHERN APPALACHIAN TREE SPECIES

Ava Hoffman

The measurement of sap movement in trees is a valuable way to estimate forest evapotranspi-ration, or the return of water vapor to the atmosphere. The accuracy of these measurements isessential to determining forest evapotranspiration. Importantly, forest evapotranspiration is amajor component of water budget, and subsequent water lost through this process affects otherfluxes such as streamflow and surface water supply. Using a dual thermal dissipation probesystem inserted into sapwood, the difference in temperature (AT) between the two probescan be related to sap flux density using an empirically derived equation. The universality ofthe power function coefficients proposed by Granier in this empirical equation has recentlycome into question regarding different tree species. Notably, Granier's coefficients may not beuniversal among different wood types and among trees with varying sapwood to heartwoodratios. It has been suggested that each specie* should be validated before using the universalcoefficients. Using four tree species indigenous to the southern Appalachians (Betula lenta,Liriodendron tulipifera, Nyssa sylvatica, Rhododendron maximum), this study estimates sapflux through excised woody stems using two independent methods: gravimetric and thermaldissipation. This study evaluates whether the species differ significantly among each otherby comparing aforementioned gravimetric measurements. Replicates of excised stems wereselected from trees growing in the Coweeta basin. Our results indicated that three of thefour species require coefficients that differ from Granier's (Betula, Nyssa, Rhododendron).Furthermore, there are significant differences between all but two of the species (Lirioden-dron and Rhododendron). Future studies will focus on testing other major tree species in thesouthern Appalachians to improve the accuracy of stand-level transpiration measurements.

BACKGROUND

Sap flux density is the movement of water throughthe vascular system of a tree per unit area per

unit time (/s, g H,O rn~2 sapwood s"1). Sap flux mea-surements at the tree-level allow scalar calculationof stand-level canopy transpiration (Ef) and stomatalconductance [Tang et al. 2006], allowing estimates ofthe forest water budget. Other variables in the waterbudget include stream runoff, interception, changein groundwater storage, and soil evaporation. Anunderstanding of forest water budgets and £( areparticularly important to mountainous areas with ahistory of erosion caused by prevalent logging andconstruction [Mattson and Swank 1989; Meyer and Tate1983].

Ava Hoffman is a fourth-year distinguished major in Biol-ogy. Following graduation, she hopes to pursue a gradu-ate career, applying her knowledge afforest processes todifficult questions regarding natural resource manage-ment, conservation, and climate change. This study wasconducted during the summer of'2011 through the U.S.Forest Service at Coweeta Hydrologic Laboratory, part ofthe Long-Term Ecological Research Network. She is pro-foundly grateful to her advisors, Drs. Steven Brantley andChelcy Ford, for their luisdom and guidance.

30 The Oculus: The Virginia Journal of Undergraduate Research

Sap flux density may be measured using a dualthermal dissipation probe system (Figure 1), which,although invasive, causes minimal damage to thetree. Thermal dissipation probes are commonly usedwithin the United States, though other methods suchas the heat pulse velocity technique [Cermdk et al.2004; Smith and Allen 1996; Tognetti et al. 2004] maybe used as well. In the thermal dissipation system,a heated probe is placed at a fixed distance down-stream ("higher" on a living trunk) from a non-heatedreference probe. The heated probe measures the tem-perature of the sapwood as well as the temperatureof the water moving around the probe. Water move-ment cools the probe as heat is removed downstream(hence "thermal dissipation" probe). The non-heatedreference probe measures only the sapwood temper-ature. The difference in temperature (AT) between thetwo probes provides a useful numerical value; it indi-cates the heat conducted away from the probe due tosap flow, where AT decreases as flow increases. In otherwords, values of AT and sap flux (JJ vary inversely;this can be observed in a diurnal time series of ATand resulting Js values (Figure 2).

The relationship between AT and Js is representedby a power function. Granier [1985, 1987] expressedAT as a function of /s using the thermal dissipationsystem, resulting in fitted coefficients to the powerfunction which were deemed ubiquitous, applyingto not only all species but all materials (mesh, wood,sawdust, etc.). However, recent studies suggest thatprobe placement, sapwood anatomy, and wound-ing may affect the accuracy of Granier's coefficients[Bush et al. 2010; Clearwater et al. 1999; Taneda andSperry 2008; Wullschleger et al. 2011], often yielding anunderestimate of Js [Bush et al. 2010; Wullschleger et al.2011]. As a result, it has been suggested that each treespecies be independently validated, and calibrated ifneeded to account for anatomical differences.

Sapwood varies widely among different species,based on the arrangement of vessels within an an-nual ring (ring-porous or diffuse-porous), proportionof sapwood to heartwood, and variable Js gradientswithin the sapwood [Ford et al. 2004; Phillips et al.1996]. Ring-porous species have one or occasionallytwo years growth of active sapwood (shallower sap-wood) whereas diffuse-porous species have multipleyears of active sapwood and therefore a greater depthfrom the outer surface (deeper sapwood) (Figure 3a).The difference is based on the diameter of the indi-vidual conducting elements in the sapwood — xylemvessels. / gradients occur within these classificationsof sapwood according to Ford et al. [2004]. Variationwithin the sapwood may drastically influence mea-surements of AT because the probe measures an aver-age AT along the probe. Ideally, probe length shouldcover the entire sapwood radius. This avoids acci-dently contacting non-conductive heartwood (probeis too long) as well as measuring only faster movingsap in the outermost vessels (probe is too short).

The purpose of this study is to measure sap fluxdensity using two independent methods, gravimetricsap flux density (Js) and thermal dissipation, of fourwoody species in the southern Appalachians andcompare their derived power function coefficients tothose proposed by Granier. Species selected for thestudy were Betula lenta (BELE), Liriodendron tulipifera(LITU), Nyssa sylvatica (NYSY), and Rhododendronmaximum (RHMA). We hypothesized that Granier'scoefficients would be inadequate for providing esti-mates of Js (often underestimating sap flux), and thatthe fitted coefficients among species would vary.

METHODSStem samples were gathered in low-elevation

sites in the Coweeta basin in Otto, NC. Sites werechosen based ongoing sap flow studies in the basin.Terrain in the area is considered variable: mountain-ous and often semi-riparian. Five stems of 4.5-5.5 cmdiameter were selected from each species. These fourspecies represent >60% of the forest community leafarea, and >70% of stand level Etin these sites. Thesespecies show variation in amount of heartwood, al-though all may be classified as diffuse-porous (Fig-ure 3b). LITU and NYSY generally contain lower sap-wood to heartwood ratios.

Stems were cut at a length twice the average ves-sel length for each species to prevent embolism andto ensure that water would flow through vessel endwalls. The stem was secured vertically with clamps,and a positive pressure system consisting of a bot-tomless graduated cylinder was affixed over the barkonto the top of the stem with paraffin wax. A largeErlenmeyer vacuum flask maintained constant -waterpressure within the graduated cylinder, supplyingwater through the open end. This system maintainedconstant hydraulic pressures to force water downthrough the stem's vascular system. This system alsoallowed changes in the hydraulic pressure to achieve

' faster or slower sap flow though the stem. Three tar-get hydraulic pressure ranges were used: 2-5, 8-12,and 14+ cm. Typically, 20-30 minutes were allowedfor flow to equilibrate to each pressure. The positivepressure applied simulated the effect of transpiration(maintained via tension within living trees). Greaterhydraulic pressure within the graduated cylinderrepresented a greater tension via transpiration. Ther-mal dissipation probes were installed in each stem bydrilling holes of appropriate length into the sapwood.The signal output from the probes, AT , was logged

1 every 5 minutes using a Campbell CR10X data logger(Campbell Scientific, Logan, UT, USA). At the sameintervals, water flowing out of the stem was inde-pendently recorded and weighed with a balance (i.e.,gravimetric values). Cross-sectional sapwood areawas measured on each stem; conductive sapwoodwas determined by staining with Congo Red dye.Gravimetric values were then converted to volumet-ric units based on water temperature (g nr2 s'1).

Spring 2012: Volume 11, Issue 1

DATA ANALYSISValues of AT collected from the data logger were

first converted to the dimensionless quantity k, whereAT is the maximum AT under zero flow.

m

Jc = (AT m -AT)/AT

The quantity k was then regressed onto Js using apower function and two coefficients:

Coefficients were fit using numerical maximumlikelihood solutions with the PROCNLMIXED func-tion in SAS (SAS Institute, Gary, NC, USA) describedby Peek et al. (2002). We also used Granger's coeffi-cients with k values and compared them to our fittedvalues on the same stems.

Figure 1. Model of thermal dissipation probes. Note theheated probe (red) is upstream of the non-heated probe (sil-ver).

10

9

8

< 7

RESULTS AND DISCUSSIONVisual comparison of k to Js indicates that use of

Granier's coefficients will likely yield an underesti-mate of sap flux density depending on the species(Figure 4). Our analyses indicate that coefficients forBELE, NYSY, and RHMA differed from predictionsusing Granier's coefficients (all p < 0.05); however,coefficients for the power function for LITU did notsignificantly differ from Granier's coefficients (Fig-ure 5).

Because sap flux density though RHMA stemswas slow, we show those values magnified for clari-fication in Figure 6. Analyses also indicate that mostspecies differ from each other. BELE/LITU, BELE/NYSY, BELE/RHMA, LITU/NYSY, and NYSY/RHMA all showed significant differences (p < 0.05)in their respective coefficients. Coefficients for LITUand RHMA did not significantly differ from one an-other.

Errors in sap flux measurement are likely dueto misplacement of the thermal dissipation probesor differences in sapwood anatomy among species.Wide variation in coefficients among species andfrom Granier's coefficients necessitates a cautiousapproach if very accurate evapotranspirative mea-surements are needed. Ideally, all species must bevalidated before use to determine if Granier's coef-ficients will yield a suitable estimate.

A Web of Science search on scientific literatureciting Granier's original work [1985] shows thatthis method is quite popular and is increasing inrelevance; indeed, it has been cited over 300 times.Granier's coefficients have been used with little criti-cism in many species in other environments. In spe-cies occupying the southern Appalachians, thesecoefficients provide a more accurate £( estimate forLITU than for the other species we evaluated. Ourdata suggest that Granier's coefficients may yieldunderestimates for many species. Based on our workfor southern Appalachian woody species, species-specific coefficients will yield more accurate esti-mates of sap flux density than those published byGranier.

0.004

121 122 123 H24 125 126

Day of Year

.0000

121 122 123 124 125

Day of Year

Figure 2. Values of AT (left) and gravimetric sap flux (Js) (right) vary inversely, diurnally. Figure credit to Steven T. Brant-ley, Coweeta Hydrologic Laboratory.

32 The Oculus: The Virginia Journal of Undergraduate Research

Future plans include validating and calibrating,if needed, other southern Appalachian woody spe-cies. Our methods may also be optimized to includesources of error: minimizing probe contact withheartwood, minimizing wounding/trauma to theprobe insertion site, and increasing understanding ofsapwood properties and gradients of sap flux den-sity at different sapwood depths. Minimizing probecontact with heartwood yields more accurate mea-surements, should the Granier coefficients be used.Minimizing wounding/trauma may be the most

pressing challenge, as cavitation may cause underes-timates of flux. Embolized vessels often close aroundthe wound site.

Effectiveness of sap flux probes on ring-porousspecies must also be maximized. Ring-porous spe-cies, such as Quercus alba, can often reach old-growthstatus in the southern Appalachians and, as such, cancontribute substantially to stand-level Et (Figure 3a).Differences between individuals of the same speciesshould also be examined, as they may vary amongdifferent environments or growing conditions (e.g.,

(a) (b

Figure 3. Staining study of stems used in sap flow measurements; hydroactive sapwood is stained red, nonfunctionalheartwood is not stained, (a) Top left, Liriodendron tulipifera, a diffuse-porous species, and bottom left Quercus alba, aring-porous species. Yellow boxes indicate sapwood radius. Note the absence of dye where the probe was inserted, (b)BELE (top left), LITU (top right), NYSY (bottom left) and RHMA (bottom right) sapwood, stained red. Note the unstainedheartwood in LITU and NYSY. LITU and NYSY generally contain lower sapwood to heartwood ratios.

Granier Flux = 0.0119*1-231

0.020 I LITU

, Rux-0.017*0661

oo,5 j BELE

Flux-O.OSU"""ooto[ -^.

«*~ '.•£... •*•

I _ f **** , __ , _

NYSY

Flux-0.037J<ul5*

-

RHMA

Flux-0.<X)5IP'M1

Figure 4: Summary of all species indicated by data points,where k is a unitless quantity based on AT, and sap fluxdensity represents independently measured, gravimetricvalues (J). The solid line represents estimated sap flowbased on Granier's equation and Granger's coefficients.Note the apparent underestimate of the solid line.

Figure 5. Breakdown of species data points (circles) andGranier estimates (crosses), where k is a unitless quantitycalculated from AT, and sap flux density represents inde-pendently measured, gravimetric values (Js). Equationsshown in each panel are based on the best-fit coefficients tothe observed data. The Granier equation and coefficients areshown above. Note the differences in coefficients. RHMA isalso shown with different scale in Figure 6.

Spring 2012: Volume 11, Issue 1 33

RHMAC$P O

<D

O

o CDO

o o

Figure 6: A closer examination of RHMA indicates that Granier estimates are, in fact, an underestimate of sap flux.

soils, elevation, slope, or aspect). Influence of growthrate and yearly biomass accumulation might haveunforeseen effects on the species' coefficients.

The results of this study suggest that the coeffi-cients proposed in Granier [1985] are not universallyapplicable, and that tree species should be indepen-dently validated and calibrated, if needed, to avoiderrors in E( estimates. Differences in xylem anatomynecessitate this tailored approach. Placement of ther-mal dissipation probes must also be examined to en-sure the most accurate coverage of sapwood.

REFERENCES1. Bush, S. E., K. R. Hultine, J. S. Sperry, J. R.

Ehleringer, N. Phillips. 2010. Calibration of ther-mal dissipation sap flow probes for ring- and dif-fuse-porous trees. Tree Physiology. 30:1545-1554.

2. Cermak, J., J. Kucera, N. Nadezhdina. 2004. Sapflow measurements with some thermodynamicmethods, flow integration within trees and scal-ing up from sample trees to entire forest stand.Trees. 18:529-546.

3. Clearwater, M.J., F. C. Meinzer, J. L. Andrade, G.Goldstein, N. M. Holbrook. 1999. Potential errorsin measurement of nonuniform sap flow usingheat dissipation probes. Tree Physiology. 19:681-687.

4. Ford, C. R., M. A. McGuire, R. J. Mitchell, R. O.Teskey. 2004. Assessing variation in the radialprofile of sap flux density in Pinus species and itseffect on daily water use. Tree Physiology. 24:241-249.

5. Granier, A. 1985. Une nouvelle methode pour lamesure du flux de seve brute dans le tronc desarbre. Annals of Forest Science. 42:81-88.

6. Granier, A. 1987. Evaluation of transpiration in aDouglas-fir stand by means of sap flow measure-ment. Tree Physiology. 3:309-320.

7. Mattson, K. G., W. T. Swank. 1989. Soil and detri-tal carbon dynamics following forest cutting inthe southern Appalachians. Earth and Environ-mental Science. 7:247-253.

8.

9.

Meyer, J. L., C. L. Tate. 1983. The effects of wa-tershed disturbance on dissolved organic carbondynamics of a stream. Ecology. 64:33-44.Peek, M. S., E. Russek-Cohen, D. A. Wait, I. N.Forseth. 2002. Physiological response curve anal-ysis using nonlinear mixed model. Oecologia.132:175-180.

10. Phillips, N., R. Oren, R. Zimmermann. 1996. Ra-dial patterns of xylem sap flow in non-, diffuse-and ring-porous tree species. Plant, Cell and En-vironment. 19:983-990.

11. Smith, D. M., S. J. Allen. 1996. Measurement ofsap flow in plant stems. Journal of ExperimentalBotany. 47:1833-1844.

12. Taneda, H., ]. S. Sperry. 2008. A case-study ofwater transport in co-occurring ring- versus dif-fuse-porous trees: contrasts in water-status, con-ducting capacity, cavitation and vessel refilling.Tree Physiology. 28: 1641-1651.

13. Tang, J., P. V. Bolstad, B. E. Ewers, A. R. Desai,K. J. Davis, E. V. Carey. 2006. Sap flux-upscaledcanopy transpiration, stomatal conductance, andwater use efficiency in an old growth forest in theGreat Lakes region of the United States. Journalof Geophysical Research. 111:1-12.

14. Tognetti, R., R. d'Andria, G. Morelli, D. Calan-drelli, F. Fragnito. 2004. Irrigation effects on dai-ly and seasonal variations of trunk sap flow andleaf water relations in olive trees. Plant and Soil.263:249-264.

15. Wullschleger, S. D., K. W. Childs, A. W. King,P. J. Hanson, N. Phillips. 2011. A model of heattransfer in sapwood and implications for sap fluxdensity measurements using thermal dissipationprobes. Tree Physiology. 31:669-679.

Water is thecommoditytions. Yet wmaintenanceplasticity prcritical quesal economicsought to c(accordanceservices gertimately traing privatethese polician internatiorder to eluframeworkcifically, thidiverse ecoiThe articleand global

On Febru;Bolivia,

gear enteredseries of maba's rural aichiefly citizewere protestin the Cochain monthlybambinos aster, which thincrease in \s at

Cochabambirescission ofnari, a BolivCorporation

Thomas Colersity of Virginpursuing a dcHis areas of atguage, studiacoming yearsto work in thethank Professsearch on thisof his researd

The Oculus: The Virginia Journal of Undergraduate Research

THE

OCULUS

THE CENTER FOR UNDERGRADUATE EXCELLENCEThe Center for Undergraduate Excellence advises students regarding undergradu-ate research opportunities and national scholarships and fellowships. Students areencouraged to visit the Center throughout their undergraduate careers. For moreinformation about the Center, please visit its website at www.virginia.edu/cue or itsoffice at 305 Harrison Institute/Small Special Collections Library.

THE UNDERGRADUATE RESEARCH NETWORKThe Undergraduate Research Network (URN) was formed in 2001 to foster an un-dergraduate research community at the University of Virginia. URN encourages stu-dents to initiate research projects and also offers guidance and mentorship to thoseinterested in research. To achieve this goal, URN presents information about currentresearch opportunities and publicizes funding availabilities and research-relatedevents. Each semester, URN puts on a series of workshops on topics ranging fromfinding funding for research to giving an effective research presentation, as well asan undergraduate research symposium where students who have completed sig-nificant research projects can present their findings. If you are interested in learningmore about URN or becoming a part of the organization, please visit www.Virginia,edu/cue/urn.

The Oculus is online! Visit http://www.virginia.edu/cue/urn/oculus.html to readthe online version of the journal and learn more about The Oculus.

The OculusVOLUME 11, ISSUE 1

Editors' Note

Dear U.Va. Community,

It is with great pleasure that we present to you the Spring 2012 issue of 7776 Oculus: The Vir-ginia Journal of Undergraduate Research. This semester's edition contains an amazing breadthof research — from plant biology to linguistics to philosophy — and is characteristic of the di-versity of undergraduate research conducted by the University's students and their facultymembers. Within, you will discover research of the highest caliber that demonstrates thefine quality of work that U.Va. students achieve when guided by their intellectual curiosity.It is our hope that through the process of reading this issue, you may also be sparked by thesame thirst for knowledge that drove these student researchers.

As our first semester as Editors-in-Chief, we have brought several fresh aspects to the jour-nal. By bringing together the academic communities of the entire university, we have em-phasized our goal of multidisciplinarity. For the first time in the history of the journal, oureditorial board selected the cover design from a diverse array of competitive submissionsfrom the University's art community, with fourth year architecture student Diana Fang asour winner. We have also added "Lab Notes," a section to showcase the latest researchconducted by students of U.Va.'s scientific community. In addition, we are highlighting thewinners of this year's Undergraduate Research Network Symposium to display an evengreater depth of the research projects conducted at the university. All the while, our edito-rial board has continued the journal's tradition of maintaining the highest standard in pa-per evaluation and selection.

This issue could not have been possible without the support of many. The members of oureditorial staff devoted several hours each week to evaluate and discuss the paper submis-sions. We would like to thank them for all of their efforts, and we would like to give a spe-cial thank you to the editors who were involved in the layout of the journal and to the mem-bers of the Undergraduate Research Network who helped publicize the journal. We wouldalso like to thank the Center for Undergraduate Excellence for their continued support andguidance. Finally, we would like to thank our authors as well as the faculty advisors whoguided them through their research, and we would like to thank you for supporting thissemester's issue of The Oculus.

Thank you, and we hope you enjoy the journal!

Sincerely,

Michelle ChoiCo-Editor-in-Chief

Ko Eun "Janet" ShinCo-Editor-in-Chief