UUnniivveerrssiittyy ooff NNeebbrraasskkaa -- LLiinnccoollnn DDiiggiittaallCCoommmmoonnss@@UUnniivveerrssiittyy ooff NNeebbrraasskkaa -- LLiinnccoollnn U.S. Department of Agriculture: Agricultural Publications from USDA-ARS / UNL Faculty Research Service, Lincoln, Nebraska 2015 GGrroowwiinngg tthhee uurrbbaann ffoorreesstt:: ttrreeee ppeerrffoorrmmaannccee iinn rreessppoonnssee ttoo bbiioottiicc aanndd aabbiioottiicc llaanndd mmaannaaggeemmeenntt Emily E. Oldfield Yale University, [email protected] Alexander J. Felson Yale University D.S. Novem Auyeung NYC Department of Parks & Recreation, NYC Urban Field Station, Bayside, NY Thomas W. Crowther Yale University Nancy F. Sonti USDA Forest Service, Northern Research Station, NYC Urban Field Station, Bayside, NY See next page for additional authors Follow this and additional works at: https://digitalcommons.unl.edu/usdaarsfacpub Oldfield, Emily E.; Felson, Alexander J.; Auyeung, D.S. Novem; Crowther, Thomas W.; Sonti, Nancy F.; Harada, Yoshiki; Maynard, Daniel S.; Sokol, Noah W.; Ashton, Mark S.; Warren, Robert J. II; Hallett, Richard A.; and Bradford, Mark A., "Growing the urban forest: tree performance in response to biotic and abiotic land management" (2015). Publications from USDA-ARS / UNL Faculty. 1536. https://digitalcommons.unl.edu/usdaarsfacpub/1536 This Article is brought to you for free and open access by the U.S. Department of Agriculture: Agricultural Research Service, Lincoln, Nebraska at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Publications from USDA-ARS / UNL Faculty by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. AAuutthhoorrss Emily E. Oldfield, Alexander J. Felson, D.S. Novem Auyeung, Thomas W. Crowther, Nancy F. Sonti, Yoshiki Harada, Daniel S. Maynard, Noah W. Sokol, Mark S. Ashton, Robert J. Warren II, Richard A. Hallett, and Mark A. Bradford This article is available at DigitalCommons@University of Nebraska - Lincoln: https://digitalcommons.unl.edu/ usdaarsfacpub/1536 RESEARCH ARTICLE Growing the urban forest: tree performance in response to biotic and abiotic land management Emily E. Oldfield1,2, Alexander J. Felson1,3, D. S. Novem Auyeung4, Thomas W. Crowther1, Nancy F. Sonti5,6, Yoshiki Harada1,7, Daniel S. Maynard1, Noah W. Sokol1, Mark S. Ashton1, Robert J. Warren II8, Richard A. Hallett9, Mark A. Bradford1 Forestsarevitalcomponentsoftheurbanlandscapebecausetheyprovideecosystemservicessuchascarbonsequestration, storm-water mitigation, and air-quality improvement. To enhance these services, cities are investing in programs to create urban forests. A major unknown, however, is whether planted trees will grow into the mature, closed-canopied forest on whichecosystemserviceprovisiondepends.Weassessedtheinfluenceofbioticandabioticlandmanagementonplantedtree performance as part of urban forest restoration in New York City, U.S.A. Biotic treatments were designed to improve tree growth,withtheexpectationthathighertreespeciescomposition(sixvs.two)andgreaterstandcomplexity(withshrubsvs. without)wouldfacilitatetreeperformance.Similarly,theabiotictreatment(compostamendmentvs.without)wasexpected toincreasetreeperformancebyimprovingsoilconditions.Growthandsurvivalwasmeasuredforapproximately1,300native saplingsacrossthreegrowingseasons.Thebioticandabiotictreatmentssignificantlyimprovedtreeperformance,whereshrub presenceincreasedtreeheightforfiveofthesixtreespecies,andcompostincreasedbasalareaandstemvolumeofallspecies. Species-specificresponses,however,highlightedthedifficultyofachievingrapidgrowthwithlimitedmortality.Pioneerspecies hadthehighestgrowthinstemvolumeover3years(upto3,500%),butalsothehighestmortality(upto40%).Mid-successional species had lower mortality (<16%), but also the slowest growth in volume (approximately 500% in volume). Our results suggestthattherewillbetrade-offsbetweenoptimizingtreegrowthversussurvivalwhenimplementingurbantreeplanting initiatives. Keywords:afforestation,compost,ecosystemservices,greeninfrastructure,nativespecies,restoration,urbanforestry heatislandeffect,sequesteringcarbon,increasingstorm-water ImplicationsforPractice infiltration,andpromotingwildlifehabitat(Nowak2006;Pataki • Speciesgenerallyconformtotheirecologicalgrowthrate etal.2011;Gaffinetal.2012).Plantingtreesisoneofthepri- classifications for non-urban systems, highlighting the mary means to achieve these services, and the benefits of the utilityoftheseprinciplesinpredictinggrowthpatternsand urbantreecanopyareincreasinglyrecognized(McPhersonetal. speciesinteractionsinurbanafforestationinitiatives. • Pioneer species had the highest incremental growth in Authorcontributions:AJF,MAB,EEO,RAH,MSAconceivedanddesignedthe basal area, height, and stem volume. They also suffered research;allauthorsperformedfieldwork;EEOanalyzeddata;DSM,RJWprovided thehighestmortality,highlightingthatlandmanagerswill statisticalguidance;EEOwrotethefirstdraftofthemanuscript;allauthors need to balance trade-offs in achieving survival versus contributedtodatainterpretationandpaperwriting. growthgoals. 1SchoolofForestryandEnvironmentalStudies,YaleUniversity,370ProspectStreet, • Plantingshrubsalongsidetreesathighstockingdensities NewHaven,CT06511,U.S.A. 2AddresscorrespondencetoE.Oldfield,[email protected] canleadtogreatergrowthintheheightofplantedtreesto 3SchoolofArchitecture,YaleUniversity,180YorkStreet,NewHaven,CT06511, achievefastercanopyclosure. U.S.A. 4NYCDepartmentofParks&Recreation,NYCUrbanFieldStation,431WalterReed • Theuseofcompostcanincreasetreegrowth;however,it Road,Bayside,NY11359,U.S.A. maytakemultipleyearsforcomposteffectstomanifest. 5USDAForestService,NorthernResearchStation,NYCUrbanFieldStation,431 WalterReedRoad,Bayside,NY11359,U.S.A. 6USDA Forest Service, Northern Research Station, Baltimore Field Station, 5523 ResearchParkDrive,Baltimore,MD21228,U.S.A. 7DepartmentofHorticulture,CornellUniversity,134PlantSciencesBuilding,Ithaca, Introduction NY14853,U.S.A. 8DepartmentofBiology,SUNYBuffaloState,1300ElmwoodAvenue,Buffalo,NY Asurbanpopulationscontinuetogrow,investingingreeninfra- 14222,U.S.A. 9USDAForestService,NorthernResearchStation,Durham,NH03824,U.S.A. structurehasbecomeapolicyimperativeforlargecitiesacross the globe. In the pursuit of sustainability and resilience, cities ©2015SocietyforEcologicalRestoration areinvestinginprojectsintendedtosupportarangeofecosys- doi:10.1111/rec.12230 Supportinginformationat: temservicessuchasimprovingairquality,reducingtheurban http://onlinelibrary.wiley.com/doi/10.1111/rec.12230/suppinfo September2015 RestorationEcology Vol.23,No.5,pp.707–718 707 Urbanafforestationandlandmanagement 1997;Nowaketal.2001;McPhersonetal.2005).Researchers demonstrateimprovedperformancecomparedwithtreeswithin havequantifiedtheeconomicvalueoftheurbanforestcanopy, two species plots. The addition of shrubs tested the impact of finding that trees provide millions of U.S. dollars worth of competitiveandfacilitativeinteractionsbetweenplantedshrubs ecosystem service benefits (Brack 2002; Nowak etal. 2002). andtrees,whichcouldeitherleadtoreducedorenhancedtree To bolster urban tree canopy, cities are now investing equally performance when compared to plots without shrubs (Van- large sums of money in street tree planting programs and the dermeer 1989). The addition of compost tested whether plots creationofnewurbanforests(Patakietal.2011;Nowak2012). receiving amendment promoted vigorous growth and biomass Indeed,anumberofafforestationeffortsincitiessuchasNew accumulationforcarbonstorage,givenpositiveeffectsofcom- York City, London, Los Angeles, and Auckland are investing postonwateravailability(Ryals&Silver2013).Weassessed inthecreationofclosed-canopyurbanforests,plantingnative the performance of approximately 1,300 individuals of six treesaplingswiththeexpectationthattheywilldevelopintoa nativespeciesacross54experimentalplotsin2011,2012,and matureforestcapableofprovidingahostofecosystemservices 2013,followingplantingofthe3-to5-year-oldseedlingsinthe totheurbanlandscape(Oldfieldetal.2013). autumnof2010. Toconstruct urbanforests, land managers generallychoose treesnativetotheregionwithnoknownsusceptibilitytopests andpathogens.Theplantingchoicesforthesenewurbanforests Methods typically contain assemblages of native species growing in an environmentwithoutanaturalanalogue.Assuch,itisdifficult SiteDescriptionandExperimentalDesign toprojectwithcertaintyhowwelltheplantedtreeswillperform The experiment, the New York City Afforestation Project in the urban environment. Yet the success of these programs (NY-CAP), is located in Kissena Corridor Park (40∘44′N, relies on the creation of closed-canopy forests that will even- 73∘49′W; 114cm MAP, 13∘C MAT), a 40-ha urban park in tually attain maturity, as mature trees provide the majority of eastern Queens, New York, U.S.A. The project was devel- environmentalbenefits(Nowaketal.2002;Troxeletal.2013). oped as a “designed experiment,” which incorporates ecolog- Research in non-urban systems shows that mixed species ical research with urban design (Felson etal. 2013). The park plantings can provide ecological benefits over monocultures, contains 54 research plots (Fig. 1A). The Natural Resources includingincreasedbiodiversity,improvednutrientcyclingand ConservationService(NRCS)classifiedsoilsacrossKissenaas soilfertility,andincreasedcarbonsequestration(Forresteretal. Laguardia-Ebbetscomplex,meaningthesoilsarewelldrained, 2006; Piotto 2008). These benefits, however, depend on the loamy-skeletal to coarse-loamy fill soils with more than 10% speciesplanted,thewaytheyinteractwitheachother,andlocal human-derived artifacts. Our research plots fall in the Ebbets sitecontingencies(Piotto2008). soilseries,characterizedbylessthan35%ofcoarsefragments Usingknownecologicalgrowthrateandshadetoleranceclas- (NRCS2009). sificationscanhelpguideexpectationsforforestgrowth.How- Theresearchplotswereseparatedintoeightdifferenttreat- ever,ahostofstressorsintheurbanenvironmentcanthreaten ments, consisting of a crossed arrangement of tree species theviabilityofnativeplantings.Theseincludedegradedsoils, composition (six species vs. two species), stand complexity drought,andaggressiveexoticsthatcanoutcompetethenatives (with shrubs vs. without), and soil amendment (with compost (Robinson&Handel2000;Rawlinsonetal.2004;Sullivanetal. vs. without). We refer to these treatments as composition, 2009;Overdyck&Clarkson2012).Theseissueshighlightthe shrub,andcompost.Notethatthisexperimentwasnotdesigned fact that there is no prior knowledge that the planted vegeta- totestmechanismsofdiversityonproductivity(e.g.sampling tionissuitabletothesite(Robinson&Handel2000;Kowarik effect or complementarity), but rather to evaluate if species 2011).Whatisneededarestudiesthatredressthepaucityofdata composition in plots had an effect on performance of the assessingthesuccessofnative,urbanplantingefforts(Oldfield planted trees. Replication is uneven and is organized as fol- etal.2013).Withoutsuchdata,urbanafforestationprojectsare lows:twospecies/noshrubs/nocompost,n=9;twospecies/no investmentsthatlacksufficientinformationinformingthelike- shrubs/compost, n=5; two species/shrubs/no compost, n=9; lihood of success. A key question is then: will planted native two species/shrubs/compost, n=5; six species/no shrubs/no treesgrowandcreateforestscapableofprovidingtheintended compost, n=5; six species/no shrubs/compost, n=8; six ecosystem services? Understanding how species will interact species/shrubs/no compost, n=5; six species/shrubs/compost, witheachotherandwiththeirenvironmentiskeytodesigning n=8. Each plot is 15×15m (225m2) and includes 56 trees successful afforestation programs (Piotto 2008), in non-urban planted 2.1m apart from each other. The two species plots andurbansystemsalike. include 28 Tilia americana and 28 Quercus rubra (Fig. 1B). This study—a research component of New York City’s The six species plots are composed of 8 individuals of MillionTreesNYC Initiative—explored the performance of an T. americana and Q. rubra, plus 10 individuals each of assemblage of six native species under various treatments Celtis occidentalis, Carya spp., Prunus serotina, and Quercus known to impact tree growth. Species composition explored alba (Fig. 1C). As the NY-CAP exists within a heavily used the effect of the number of planted species (two vs. six) on urban park, the plot layout was designed as an offset grid tree performance.Given researchhighlighting greater produc- (Fig. 1B & 1C) to both facilitate research and serve as an tivity of trees planted in mixture (Forrester etal. 2006; Piotto esthetic landscape amenity (Felson etal. 2013). Tree species 2008), we expected that trees within six species plots would selected for this project had to meet the following criteria: 708 RestorationEcology September2015 Urbanafforestationandlandmanagement (A) (B) (C) Figure1. (A)AnaerialviewoftheafforestationresearchplotsinKissenaCorridorParkinQueens,NewYork,U.S.A.;(B)plotlayoutoftheplantedtreesin twospeciesplots;and(C)layoutoftheplantedtreesinsixspeciesplots.Measurementsofdiameter,height,andstemvolumeweretakenforthecentral24 treesineachplotlocatedinthesubplotslabeleda–f. September2015 RestorationEcology 709 Urbanafforestationandlandmanagement all species are native to the region, and have no known sus- s ee ee cterereapettei,bmialoinrtytdaltisotlyop.weSstegslreaocnwtdeedrpsatrtdehieofsgfeecrnionsngktaniionnwatnhemtiorixcsahouafsdeerawptioiddlee,srapmnreocaedds- percentage ShadeTolerance ntermediatntermediatIntolerantntermediatntermediatTolerant h II II (Table1). wt bn cTurerTeerdseefwrsoewmreePr3ei-npteollaan5nt-deysdeNairnu-orOsledcrtyosabipnelriCn2og0ls1u0mm.bePaulssau,nrNtienedgwtarJepeeprssreowyx,eiUmre.aSpt.erAoly-. mentalgro GrowthRateassificatio RapidSlowRapidSlowModerateModerate e Cl 0.6–1.2minheight,withrootmassescontainedineither1or2 cr gTgaerelrleoisnnwh(oaelpreepsrpoolaxfniamtpeapdtrewolpyitrhi3a.at7eh9saionzrde7-th.o5e8lhdoLmu,sereecsthhpaeenctitrzieevedelrpyoo)ostctso-h(naotpaleipnrdeoirxgs--. vals,andin Mortality(%) 40122016128 iostmhhferatuhtcbeeolsym5(542p5ospspcltemocatsimdersieea,ncm4de1emivtepeernldatanscn.tosdSmphapeprorupsbprtosloxaatni)nmddianhtheaaellyfrcbrw2os5sewsrceemedrpedldaeepnseltiapegn)dn.tewwHdiiattalhhft confidenceinter 3me(cm) MeanStemVolumeaIncrementYears1–3)(%) 355654611894575321022 the same time as the trees, and were planted in holes large % u ( easAinnpsdo,ouHchgyeahnrmubtomsamawcceicatlohinsminnvamiebroagincdinuhaimtapen,lroaoAt,osiatcnnmlcedlpauVsidasieebssuss.rhyTnrruhuiabemcsmadS,oeasnamttnaabdbtuuucPmnaad;snaaicncnatudnsmahhdrevueribbnrss-- edin2013,95 StemVol 95%ConfidenceInterval 370–994346–4931762–2707358–500518–6201980–2321 gHyaeeatrurbmaoc.feApoluafsnuptlillnasgnp,teiancngiedssssuluifsbftseeriseqduinesncigltuncdiofielcdoanniintzadFtiieeolbsnaocnokfeattfhtaeelr.pt(lh2oe0ts1fi3rbs)yt. suresrecord MeanSE± 682±159419±372234±241429±36569±262150±87 pocspSFfisSiaaTA2twmtwWdsitwwtarnrnhhimrhpmofl0iioueeevokhpheetageaceeriprl1aeileeaoeaepenepenmoedallrrdts1frlsnrdntdstrtmemPtosolehMiee)sebeodi.ddefyifx,rtcnrwdettdifelfexetoeeeeeeinAotereoprwrsntmierddeparosangarmrmofeetptalrttaledaesmoiod.lraohei.ecrwhlpenaheruletftsesechaharl(aetdctnero1ilwpyoiseet2hrteteoslehrnrnhfo5oiahdylsnr0uibiseeooyagiewuseorrbiwdnc1rsaeansldasca1toeeemutlmdcue4elt.ohorr1s5lttiaiecnrdet)mceewoneom0arert.cotdciivhahiidicnedinaueogumemntenpBhaademgmelalnhsglpiomtltlpr.onnsnpitiatlaesaiaetv-vtptnodTuelrotsehtprnrnaeoelafitetflielimhde.dldths,agtcctnrrtaeyedevitsohahhefddTnala,srroarasdtteeatteemtcihpaonttrtebaevmabrc“ohoter-stcmfsdiroaeiideseritmeuotanooeoptieohei,pmtsacelrstvginvoarptrleov,palnooroeeoausitotrfeldlmlnmrwaddtcluhfbotaaieysrtlidohsenot”fvntmhemmeatitoiaslehotfeooodedenlt2bifiiemiaelrfsnnnylesggaz.es;ter1ar5dsg-rnntuheccealnr,iphciewecea2dairtgamtntdoaelea,-afidamamus9eageictldalrn3drucacasei6nr2oeenoiipdlm-eiernafcct4apnmeatsduteeebfiendoshrsaitletedrbgdeetcepclpdmanosrtasrhpeumlereliath,btihatasetftsp1lntrlhscetoesooyor.t.n.cgro0ttetrsuerttueroeeWthraIts0mtobsHreabsoroiennrkelttvvsdloeamesueadsewblp.ee(splicyimeanirr5wdllAlnd2sdgeoa(e,easadi,psAlerechnrfnuescnsnooaetesolapmitrtsoracwtuacnnlsexsereiitOlnrhtnougansuh)ctwoothdcoetitttttnchuoglrwooneeeheheeanoodueheisrsdddnddogeesss---f-f-tt.,. Table1.Growthmeasuresandgeneralspeciescharacterizationsforplantedtreespecies.Means±SEforgrowthmeafrom2011to2013. 2BasalArea(cm)Height(m) MeanMeanBasalArea95%Heightaa95%ConfidenceIncrementConfidenceIncrementSpeciesMeanSEInterval(Years1–3)(%)MeanSEInterval(Years1–3)(%)±± Celtisoccidentalis7.66±1.145.44–9.898061.72±0.131.46–1.98240Caryaspp.6.49±0.425.67–7.312351.73±0.061.60–1.8536Prunusserotina21.60±1.8218.03–25.175522.55±0.112.34–2.7772Quercusalba8.16±0.517.16–9.162871.46±0.051.37–1.5541Quercusrubra7.98±0.277.45–8.523111.88±0.031.81–1.9437Tiliaamericana23.05±0.7121.65–24.453702.54±0.042.46–2.6291 aCalculatedastheproportionincreaseintreesizeduringyears1–3relativizedtoinitialmeasurestakeninyear1.bGrowthrateclassificationsandshadetolerancearealsoincluded,whichwereobtainedfromtheUSDAPlantdatabase(USDANRCS2015). 710 RestorationEcology September2015 Urbanafforestationandlandmanagement on some species prevented taking diameter measures 10cm Results above ground level. Given the differing morphologies of the TreeSurvival trees,rootcollardiameterprovidedaconsistentmeasureacross speciestotestfordifferencesingrowth,anditisoftenusedas After 3years, a total of 180 of the 1,296 assessed trees were ameasuretotestforstatisticaldifferencesinthegrowthoftree scored as either dead or missing. By year 3, more trees were seedlings (Halter etal. 1993; Page-Dumroese etal. 1998). We recordedasmissingversusdead(109vs.71).Whentreeswere repeated these assessments of height and diameter in August visiblydead,wescoredthemassuch.Incaseswhenwecould 2012andAugust2013.Weusedrootcollardiametertocalculate notlocatetrees,wescoredthemasmissing.Missingtreeslikely basalareaonanindividualtreebasis,whichwasthensummed resultedfrommultiplefactors,includingvandalism,theft,fire, acrosstreestodeterminetotalbasalareaperplot.Toestimate andsevereherbivory.Sixspeciesplotshad1.6-timesmoredead stembiomassvolume,weusedtheformulaforacone,usingthe andmissingtreesthantwospeciesplots(p<0.001).Composted heightandrootcollardiametermeasures(Magnussen&Reed plots also had a higher degree of mortality (17±1%) than 2004).Likebasalarea,stemvolumewasrecordedattheindivid- non-compostedplots(11±1%)(p<0.01).Mortalityincreased ualtreelevelandthensummedtodeterminetotalstemvolume acrossthe3yearsfrom1%inyear1,to6%inyear2,toalmost attheplotlevel. 14% in year 3. Mortality varied among species, with Celtis occidentalis having the highest percent mortality after 3years at40%andTiliaamericanahavingthelowestat8%(Table1). As tree size varied between species at the time of planting, StatisticalAnalyses wetestedtoseewhetherourfirstheightmeasurementtakenin Weanalyzedhowplottreatments(composition,shrub,compost) 2011 was a significant predictor of tree mortality in 2012 and affectedsurvivalandgrowthintreediameter,height,andstem 2013.Treeheightin2011didhaveasignificanteffecton2012 volume, both across research plots as a whole and also by survival (p<0.05) for C. occidentalis, Carya spp., Quercus species. For the mortality analyses, we used a generalized rubra,andT.americana;andasignificanteffect(p<0.05)on linearmodelwithabinomialerrordistribution.Forthegrowth 2013 mortality for C. occidentalis, Carya spp., and Quercus analyses, we used a linear mixed model (LMM) approach to alba.Largertreessufferedlessmortalitythansmallertreesfor account for the unbalanced replication (see Site description each of those species; however, it is difficult to determine the andexperimentaldesign)andthenon-independentspatialand ultimatecauseofmortality. temporal associations in our data. By specifying individual tree nested in plot as a random effect, our LMMs account for the possibility that plots paired by the shrub treatment Plot-LevelResponse (see Fig. 1B & 1C) are more similar to each other than to For each research plot, we calculated total plot basal area, other plots, and also account for the fact we had repeated the mean height, and total plot stem volume to determine measurements on the same units (i.e. the same tree in 2011, howtreatmentsaffectedtreeperformanceattheplotlevel.All 2012,and2013).Ourapproachisthenconceptuallyequivalent metricsrespondeddifferentlytoplottreatments,althoughthey tousingnestingandrepeatedmeasuresinANOVA,andisnow consistentlyincreasedfromyear1toyear3,demonstratingthat, widely adopted in environmental science (Bolker etal. 2009; overall,thetreesweregrowingasopposedtodyingorindecline Bradfordetal.2012;Krameretal.2012),giventheadvantages (Table2). itprovidesinmodelpowerandevaluation(seenextparagraph). By year 3, total tree basal area per plot was 23% higher in Plottreatments(composition,shrub,compost)andyear(1,2,3) composted versus non-composted plots (p=0.031). Total tree wereidentifiedasfixedeffects.Wealsotestedforinteractions basal area was also 13% higher in plots planted with shrubs betweentreatmentsandbetweentreatmentsandyear.Wetested (p=0.01)and7%higherintwospeciesplotscomparedwithsix for treatment effects on the relative growth in basal area and species plots (p=0.05; see Fig. 2A). Relative growth in basal stem volume (calculated as the proportion increase in growth areaincrement(BAI)wascalculatedasthepercentincreasein during years 1–3 relativized to initial measures in year 1) at growthfromyear1toyear3.Compostwastheonlytreatment the species and plot level. We specified relative growth from tohaveasignificanteffectonBAI(p=0.003),withcomposted year 1 to year 3 as the response variable with plot treatments plots growing 69% faster compared with non-composted identified as fixed effects and tree nested within plot as a plots. randomeffect. Species composition had a significant effect on tree height Toselectthebestmodelsfortreegrowth,weusedthemodel (Table 2). As with basal area, height was greater in two withthelowestAkaikeinformationcriterion(AIC)score(Burn- species plots with trees 10% taller than those in six species ham etal. 2011). Variance inflation factors less than 5 indi- plots (Fig. 2B). The best-fit statistical model retained both a cated that collinearity was sufficiently low among predictor shrub-by-year interaction and a compost-by-year interaction, variables. The LMMs were all fit assuming a Gaussian error withthosetreatmentshavingpronouncedeffectsinyear3but distribution. We used the “nlme” package in the “R” statis- not years 1 and 2 (Tables 2 & S1, Supporting Information). tical program for our LMMs, and the summary function in In year 3, the measured height of trees was 8% higher in the “LMERtest” package to estimate model coefficients and composted plots and 12% higher in plots planted with shrubs pvalues. (Fig.2B). September2015 RestorationEcology 711 Urbanafforestationandlandmanagement Table 2. Coefficientsandpvaluesforplottreatmenteffectsonplanted (A) Two Species Six Species treegrowthasmeasuredastotalbasalareaperplot,meanheight,andtotal stemvolumeperplotofindividualtreesacross2011–2013.Thenegative coefficientsforthemaineffectofcompostfortotalbasalarea,height,and 400 No Shrub total stem volume are due to their interaction with year and cannot be Shrub ionvteerrptirmetee.dCaoloenffiec,ihenigtshlaignhdtipngvatlhuaetstahreepporessiteinvetedefffoerctesaochf cpoamrapmoestterg.rAowll )−2m tsrteaatitsmtiecnaltlsyrseitganinifiedcawntit(hpin<t0h.e05b)etset-rfimtssataretissthicoawlnminodbeolldarteexptr.esented,and (2*otm 300 pl Coefficient pValue per (A)Totalbasalarea(cm2/m2) al Area 200 Intercept 16.61 0.44 as B Composition −33.55 0.053 al ot 0 Shrub 24.16 0.010 T 0 1 Compost −61.68 0.031 Year 82.76 <0.001 Compost×Year 44.43 <0.001 (B)Height(m) 0 Intercept 1.18 <0.001 No Compost Compost No Compost Compost Composition −0.21 0.01 Shrub −0.16 0.081 Two Species Six Species Compost −0.19 0.12 (B) 3.0 Year 0.27 <0.001 No Shrub Shrub Shrub×Compost 0.13 0.06 5 Shrub×Year 0.11 <0.01 2. Compost×Year 0.11 <0.01 (C)Totalstemvolume(cm3/m2) 0 2. Intercept −3748.3 0.13 Shrub −4412.6 0.098 m) CYeoamrpost −76790339..41 <00..002061 eight ( 1.5 H Shrub×Compost 4069.6 0.032 0 Shrub×Year 2653 0.023 1. Compost×Year 4183.4 <0.001 5 0. Total tree stem volume per plot and growth rate were sig- 0 nificantly affected by compost with an interaction with year 0. No Compost Compost No Compost Compost (p<0.001). Stem volume growth rate was 73% greater in composted plots over the course of 3years. Consequently, Figure2. Treatmenteffectsonaperplotbasisfor(A)basalareaand(B) total stem volume on composted plots was 27% higher than heightofmainstem.Wecalculatedthetotalbasalareaperplotandthe non-composted plots. The best-fit statistical model for total meantreeheightacrosstheresearchplotstoassesswhichplottreatments weredrivingdifferencesinplot-levelgrowth.Errorbarsrepresent±SE. stem volume per plot also retained shrubs as a main effect, Thenumberofreplicatesareasfollows:twospecies/noshrubs/no a compost-by-shrub interaction, and a shrub-by-year interac- compost,n=9;twospecies/noshrubs/compost,n=5;two tion. The presence of shrubs led to significantly more tree species/shrubs/nocompost,n=9;twospecies/shrubs/compost,n=5;six stem volume per plot (p=0.02), with shrub treatment plots species/noshrubs/nocompost,n=5;sixspecies/noshrubs/compost,n=8; containing24%morestemvolume(ofthetrees)thanplotswith- sixspecies/shrubs/nocompost,n=5;sixspecies/shrubs/compost,n=8. out(Fig.3). Thegraphsincludemeasuresonlyfromyear3,andincludeallplot treatmentsaseachwasretainedinthebest-fitstatisticalmodels.See Table2fortreatmentcoefficientsandpvalues. SpeciesResponse Allsixoftheplantedtreespeciesgrewmarkedlyinallrecorded the species) for trees in composted plots. Models for all six measuresacrosseachsuccessivegrowingseason(seeTable3& speciesretainedasignificantcompost-by-yearinteraction,with Tables S2 & S4). We could assess the influence of only two diametersinyear3foreachspeciesmarkedlygreaterincompost treatments, shrubs and compost, for four of the tree species, versus no-compost plots, with minimal differences in years 1 whereasfortheothertwospecies(Q.rubraandT.americana), and2(seeTablesS2&S4).TiliaamericanabutnotQ.rubra we could also assess the effects of the species composition retained a significant species composition by year interaction, treatmentbecausetheywerepresentinboththetwoandthesix withT.americanainsixspeciesplotsshowinggreatergirthin speciesplots(seeFig.1B&1C). rootcollardiameterthanthoseintwospeciesplots,particularly Compost had a positive effect on root collar growth with inyear3(6.11±0.17cmvs.4.86±0.085cm,respectively;see measures of diameter being 15–30% greater (depending on Fig.4A). 712 RestorationEcology September2015 Urbanafforestationandlandmanagement Table3. Means±SEforeachmeasureofgrowthforspeciesinsixspeciescompositionplotsfor2013(for2011and2012values,seeTableS4).Valuesfor 2013capturetheeffectthatplottreatmentsarehavingonindividualspecies’growthforthelatestmeasurementyear.Significanttreatmenteffects(p<0.05)are showninboldtextforthedifferencesbetweenplotswithandwithouteithershrubsorcompost.SeeFigures4–6andTableS2forsizemeasuresforQuercus rubraandTiliaamericana. SixSpeciesCompositionPlots Shrubs Compost Difference Difference Variable (−) (+) (With–Without) (−) (+) (With–Without) Celtisoccidentalis Diameter(cm) 2.81±0.24 2.58±0.33 −0.23 2.28±0.20 3.03±0.31 0.75 Height(m) 1.74±0.15 1.68±0.23 −0.06 1.44±0.15 1.93±0.15 0.49 Volume(cm3) 594±130 781±309 187 309±75 969±268 660 Caryaspp. Diameter(cm) 2.72±0.13 2.78±0.12 0.06 2.52±0.11 2.92±0.13 0.4 Height(m) 1.61±0.08 1.84±0.09 0.23 1.57±0.10 1.84±0.08 0.27 Volume(cm3) 379±49 456±56 77 316±45 497±54 181 Prunusserotina Diameter(cm) 4.78±0.35 4.86±0.30 0.08 4.33±0.32 5.18±0.31 0.85 Height(m) 2.29±0.15 2.70±0.16 0.41 2.42±0.17 2.65±0.14 0.23 Volume(cm3) 2148±363 2315±323 167 1754±335 2585±331 831 Quercusalba Diameter(cm) 3.16±0.15 3.03±0.12 −0.13 2.77±0.13 3.30±0.12 0.53 Height(m) 1.39±0.06 1.54±0.07 0.15 1.29±0.06 1.56±0.06 0.27 Volume(cm3) 440±61 417±39 −23 285±31 515±52 230 0 Heightresponsestoplottreatmentswerealsofairlyconsis- 0 00 tent in the sign of the effect across species, with significant 4 No Shrub ormarginallysignificantshrub-by-yearinteractionsretainedin )2 Shrub modelsforfiveofthesixspecies(Caryaspp.,Prunusserotina, − m 0 Q.alba,Q.rubra,andT.americana).Allofthesespeciesexhib- * 0 3 0 m 0 ited greater height in plots planted with shrubs versus plots c 3 ( without, with differences most pronounced in year 3 (Tables ot pl S2 & S4). For instance, the height of Q. rubra was 14% me per 20000 gTrileiaatearminersicharunba vaelrssousrentaoinsehdruba pcloomtspobsyt-ybeya-rsh3ru(bFiign.te5rAac)-. u ol tion, indicating that trees were tallest in compost-amended V m 0 plots planted with shrubs and that this response would not al Ste 1000 hmaevnetsbaeleonnepr(eFdigic.t4edB)f.rNomotaebitlhy,ercotmhepocsotmepffoescttsoornshtrreuebhteriegaht-t Tot were less consistent across species than the effects of com- post on root collar diameter and basal area, where there were 0 always compost-by-year interactions. For example, for three No Compost Compost species(C.occidentalis,Q.rubra,andP.serotina),therewere no significant compost or compost-by-year effects on height Figure3. Treatmenteffectsperplotfortotalstemvolume.Theeffectsof (TablesS3&S5). theinteractionbetweentheshrubandcomposttreatmentareclearly Stem volume, like basal area, responded fairly consistently apparent,withcompostedplotscontainingshrubshavingsignificantly morestemvolumecomparedwiththeothertreatments.Errorbars and positively to compost among species, with five of the six represent±SE.Thenumberofreplicatesareasfollows:noshrub/no speciesshowingasignificantcompost-by-yearinteraction(for compost,n=14;noshrub/compost,n=13;shrub/nocompost,n=14; allfivespecies,p<0.05).Quercusrubrawastheonlyspecies shrub/compost,n=13.SeeTable2fortreatmentcoefficientsandpvalues. that did not retain a compost-by-year interaction in its best-fit statistical model for growth in stem volume. However, the Basalarearesponsesmappedontothoseforrootcollardiam- bestmodeldidretainacompost-by-shrubinteraction,withthe eter growth (as it was calculated from each tree’s root collar positive effect of shrubs on stem volume manifesting only in diameter).YetonlycomposteffectswereapparentforBAIfor composted plots (Fig. 5B). Tilia americana retained a signif- eachspeciesfromyear1toyear3.Compostedplotshadhigher icant year interaction for each of the three treatments (Table growthinbasalareathannon-compostedplotsforallspecies, S3), where treatment effects on stem volume were most pro- withsignificantdifferences(p<.05)forC.occidentalis,Carya nouncedinyear3(TableS2),withsixspecies,compostamend- spp., and T. americana; compost had a marginally significant ments, and shrub presence all equating to greater stem vol- effectonQ.alba(p=0.06)(Tables4&5). ume in 2013 (Fig. 6). The best-fit models for relative growth September2015 RestorationEcology 713 Urbanafforestationandlandmanagement (A) Two Species Six Species survivalandgrowtharesparse.Weshowthatsixnativespecies 7 respondrelativelyconsistentlytoplottreatments,yetindividual No Shrub speciesperformancevariesmarkedly.Treespeciescomposition 6 Shrub explaineddifferencesinthebasalareaandheightoftreesatthe plot level, with two species plots having taller trees on aver- 5 ageandgreaterbasalarea.Thisresultappearscounter-intuitive given expected positive effects of plant diversity on ecosys- m) temfunctioning(Isbelletal.2011),butinourstudythiseffect c 4 er ( appears to be driven by a single high-performing species. et m Specifically,halfofthetreesintwospeciesplotswereT.amer- a 3 Di icana. This species’ mean diameter and height in year 3 were substantially greater than those of the other planted species, 2 except Prunus serotina, which is of similar size (Table 1). Differences in individual species performance also appear to 1 account for the fact that six species plots had greater rates of mortality, owing to the high incidence of mortality (40% by 0 No Compost Compost No Compost Compost year3)ofCeltisoccidentalis(Table1).Therefore,treespecies compositioneffects(i.e.2vs.6speciesplots)attheplotlevel (B) 5 appeardrivenmorebythecharacteristicsofindividualspeciesin 3. No Shrub thedifferenttreatmentlevels,ratherthanthecompositiontreat- Shrub 3.0 mentinfluencingthestrengthofpositiveornegativeinteractions amongtreeindividuals. 5 2. There was evidence of species composition treatment 0 effects on the strength of interactions when individual species 2. responses were considered. For example, despite explaining m) ht ( 1.5 the greater growth of trees within the two species treatment g Hei attheplotlevel,T.americanahadgreaterindividualdiameter 0 1. and volume in the six species plots by year 3. In contrast, 5 Q. rubra, the second species in the two species plots, had 0. greater volume in the two versus six species plots, but this 0.0 effect was not statistically significant despite being retained NoCompost Compost in the best-fit model. The effects of the species composition Figure4. TreatmenteffectsforTiliaamericanaon(A)diameterand(B) treatmentonthesetwospeciespointtoinfluencesofcompeti- height.Forbothrootcollardiameterandheight,thebest-fitstatistical tionand/orfacilitationbetweenthesetwospecies.Itisdifficult modelsretainedtreatment-by-yearinteractions,andsotheresultsforyear to disentangle these effects to determine direct causation, but 3onlyarepresentedinthesegraphs,asthatistheonlyyearthatexhibits such interactions are considered key in shaping how mixtures statisticallydifferentresultsinmeasuredparametersbetweentreatments. perform(Forresteretal.2006;Piotto2008).Forexample,itis Errorbarsrepresent±SE.For(A)thenumberofreplicatesareasfollows: possible that T. americana grew less in the two species plots twospecies/noshrubs/nocompost,n=108;twospecies/no shrubs/compost,n=60;twospecies/shrubs/nocompost,n=108;two because the large individual size of this species generated species/shrubs/compost,n=60;sixspecies/noshrubs/nocompost,n=20; intra-specific competition that was more pronounced than the sixspecies/noshrubs/compost,n=32sixspecies/shrubs/nocompost, inter-specificcompetitionwiththesmallerstaturespeciesinthe n=20;sixspecies/shrubs/compost,n=32.For(B)thenumberof sixspeciesplots.Notmutuallyexclusiveisthepossibilitythat replicatesareasfollows:noshrub/nocompost,n=128;noshrub/compost, Q. rubra could have a more suppressive (or less facilitative) n=92;shrub/nocompost,n=128;shrub/compost,n=92.SeeTableS3 effect on the growth of T. americana than other species in fortreatmentcoefficientsandpvalues. the experiment, although the mechanism for this is unclear. Ourexperimentisnotdesignedtoteaseoutthesemechanisms in stem volume from year 1 to year 3 all retained compost, as we lack an experimental control that would help elucidate withithavingasignificanteffect(p≤0.05)onC.occidentalis, these possible effects. Regardless, our results suggest that Caryaspp.,andQ.alba.Composthadamarginallysignificant planting T. americana in more diverse mixtures improves its effect (p=0.06) on T. americana’s relative growth in volume performance. (Table4). Thepresenceofshrubsinfluencedtheperformanceofthetree species,againsuggestingthatthetreatmentsalteredinteractions amongplantedindividuals.Shrubpresencewasassociatedwith Discussion greaterheightoftreespecies,exceptforC.occidentalis.Forthe Growth performance of juvenile trees is a critical determinant otherfivespecies,ashrub-by-yearinteractionwasretainedinall oftheviabilityandsustainabilityofurbanafforestationinitia- ofthebest-fitstatisticalmodels(seeTablesS3&S5),reflecting tives (Ruiz-Jaen & Aide 2006), yet data on planted juvenile the fact that the influence of shrubs was most pronounced by 714 RestorationEcology September2015
Description: