13.07.2015 Views

Van Gunst, K.J. 2012. Forest Mortality in Lake Tahoe Basin from ...

Van Gunst, K.J. 2012. Forest Mortality in Lake Tahoe Basin from ...

Van Gunst, K.J. 2012. Forest Mortality in Lake Tahoe Basin from ...

SHOW MORE
SHOW LESS
  • No tags were found...

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

University of Nevada, Reno<strong>Forest</strong> <strong>Mortality</strong> <strong>in</strong> <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong> <strong>from</strong> 1985-2010: Influences of <strong>Forest</strong> Type,Stand Density, Topography and ClimateA thesis submitted <strong>in</strong> partial fulfillment of the requirements for the degree of Master ofScience <strong>in</strong> Natural Resources and Environmental SciencebyKrist<strong>in</strong> Jane <strong>Van</strong> <strong>Gunst</strong>Dr. Peter J. Weisberg/Thesis AdvisorDecember, 2012


Copyright by Krist<strong>in</strong> Jane <strong>Van</strong> <strong>Gunst</strong> 2012All Rights Reserved


iTHESIS ABSTRACTWidespread and synchronous outbreaks of forest mortality dur<strong>in</strong>g the 1990s andcont<strong>in</strong>u<strong>in</strong>g to present day have drastically altered millions of hectares of forestlands <strong>from</strong>Mexico to Alaska (Bentz 2009). Tree mortality is a complex process with a variety ofboth exogenous and endogenous factors <strong>in</strong>fluenc<strong>in</strong>g the extent, pattern, and severity oftree mortality (Frankl<strong>in</strong> et al. 1987; Holdenreider et al. 2004; Powers et al; Raffa et al.2008; Simard et al 2012). Although the vast majority of mortality is mediated by nativebark beetles (Coleoptera: Scolyt<strong>in</strong>ae), more proximal exogenous factors such as drought,and endogenous factors such as stand density and environmental sett<strong>in</strong>g have also beenimplicated (Allen et al. 2010). This study utilizes remote sens<strong>in</strong>g imagery <strong>from</strong> the 1985-2010 Landsat Thematic Mapper archive to explore relationships among forest types,climate, stand density, and environmental gradients and mortality <strong>in</strong> the <strong>Lake</strong> <strong>Tahoe</strong>Bas<strong>in</strong> <strong>in</strong> the central Sierra Nevada.In coniferous forests of the western U.S. and Europe, <strong>in</strong>creased stand densityresult<strong>in</strong>g <strong>from</strong> forest management and land use practices is widely hypothesized to<strong>in</strong>crease forest-wide mortality levels (Dobbert<strong>in</strong> et al. 2007; Guarín and Taylor 2005;Millar et al. 2012; Maloney and Rizzo 2002; Maloney et al. 2011). In my first chapter, Iexam<strong>in</strong>e how the relationship between mortality and stand density has differed amongfive forest types and wet and dry climatic periods. This part of the project uses annualmaps of forest mortality (Plates 1-2) and an <strong>in</strong>dex of stock<strong>in</strong>g level, a proxy for standdensity, derived <strong>from</strong> classification of remotely sensed imagery. Logistic regression isthen used to evaluate how a 20% <strong>in</strong>crease <strong>in</strong> stock<strong>in</strong>g level affects probability ofmortality. Our analysis showed that the strength of density dependence for <strong>in</strong>fluenc<strong>in</strong>g


iimortality is variable: all forest types exhibited periods when density-dependent mortalitywas positive (<strong>in</strong>creased density <strong>in</strong>creases risk of mortality), negative (<strong>in</strong>creased densitydecreases risk of mortality) or when density was not associated with mortality. However,positive density-dependent mortality was found to a greater degree <strong>in</strong> Jeffrey p<strong>in</strong>e andlodgepole p<strong>in</strong>e-dom<strong>in</strong>ated forests and dur<strong>in</strong>g climatic periods characterized by drought.In middle- and upper-elevation forests, density-dependent mortality was more variable byclimate with negative density-dependent mortality apparent over much of the time period.In my second chapter, I explore 1) Variations <strong>in</strong> mortality levels by forest type andclimatic period, and 2) How the probability of mortality is associated with elevation and<strong>in</strong>cident solar radiation. These two topographical variables are important <strong>in</strong> govern<strong>in</strong>gproductivity, resource availability, and temperature and climate regimes <strong>in</strong> thetopographically-complex LTB (Urban et al. 2000). We used generalized l<strong>in</strong>ear modelsand regression tree analysis to elucidate relationships between mortality levels, foresttype and climatic period, and logistic regression to quantify association between mortalityrisk and environmental gradients. We discovered that mortality is not consistentlyassociated with drought. <strong>Mortality</strong> levels were higher for all forest types dur<strong>in</strong>g an earlycold and dry period, but not dur<strong>in</strong>g a later warm and dry period. Although forestmortality was episodic for upper-elevation forests, lower-and mid-elevation forestsexhibited more stable mortality levels. Over the majority of the time period, we foundthat mortality risk was greater on more north-fac<strong>in</strong>g slopes. The relationship betweenmortality and elevation differed among forest types. In lower-elevation forests and dur<strong>in</strong>gthe first dry period, lower elevations were associated with decreased mortality risk. For


iiimiddle and upper-elevation forests, higher elevations were associated with greatermortality risk.Results highlight the need for further study <strong>in</strong>to the relative role of species traits,stand characteristics, <strong>in</strong>sect herbivory, and climate for <strong>in</strong>fluenc<strong>in</strong>g forest mortality at bothstand and landscape scales. Variable response to both climate and stand density meansthat scientists should be cautious about generaliz<strong>in</strong>g these effects to all western forests.Th<strong>in</strong>n<strong>in</strong>g treatments should not be expected to improve forest health <strong>in</strong> all forest types.Consistent relationships of forest mortality with forest type and environmental sett<strong>in</strong>g canbe used to guide landscape-level management of forest health.LITERATURE CITEDAllen, Craig D., Alison K. Macalady , Haroun Chenchouni, Dom<strong>in</strong>ique Bachelet, NateMcDowell,Michel Vennetier, Thomas Kitzberger, Andreas Rigl<strong>in</strong>g , David D.Breshears, E.H. (Ted) Hogg, Patrick Gonzalez , Rod Fensham, Zhen Zhang, JorgeCastro, Natalia Demidova, Jong-Hwan Lim, Gillian Allard, Steven W. Runn<strong>in</strong>g,Akk<strong>in</strong> Semerci, and Neil Cobb. 2010. A global overview of drought and heat<strong>in</strong>ducedtree mortality reveals emerg<strong>in</strong>g climate change risks for forests. <strong>Forest</strong>Ecology and Management 259 (4): 660–684.Bentz Barbara J. ed. 2009. Bark Beetle Outbreaks <strong>in</strong> Western North America: Causes andConsequences. University of Utah Press.Bentz, Barbara, Jacques Régnière, Christopher J. Fettig, E. Matthew Hansen, Jane L.Hayes, Jeffrey A. Hicke, Rick G. Kelsey, Jose F. Negrón, and Steven J. Seybold.


iv2010. Climate change and bark beetles of the western United States and Canada:Direct and <strong>in</strong>direct effects. Bioscience 60(8): 602-613.Dobbert<strong>in</strong>, Matthias, Beat Wermel<strong>in</strong>ger, Christof Bigler, Matthias Bürgi, MatthiasCarron, Beat Forster, Urs Gimmi, and Andreas Rigl<strong>in</strong>g. 2007. L<strong>in</strong>k<strong>in</strong>g <strong>in</strong>creas<strong>in</strong>gdrought stress to Scots p<strong>in</strong>e mortality and bark beetle <strong>in</strong>festations. The ScientificWorld 7(S1): 231–239.Frankl<strong>in</strong>, Jerry F., H.H. Shugart, and M.E. Harmon. 1987. Tree death as an ecologicalprocess. Bioscience 27: 259–288.Guarín, Alejandro and Alan H. Taylor. 2005. Drought triggered tree mortality <strong>in</strong> mixedconifer forests <strong>in</strong> Yosemite National Park, California, USA. <strong>Forest</strong> Ecology andManagement 218: 229–244.Holdenrieder, Ottmar, Marco Pautasso, Peter J. Weisberg, and David Lonsdale. 2004.Tree diseases and landscape processes: the challenge of landscape pathology.Trends <strong>in</strong> Ecology and Evolution 19(8): 446-452.Maloney, Patricia E. and David M. Rizzo. 2002. Pathogens and <strong>in</strong>sects <strong>in</strong> a prist<strong>in</strong>e forestecosystem: the Sierra San Pedro Martir, Baja, Mexico. Canadian Journal of <strong>Forest</strong>Research 32: 448-457.Maloney, Patricia E., Detlev R. Vogler , Andrew J. Eckert, Camille E. Jensen, and DavidB. Neale. 2011. Population biology of sugar p<strong>in</strong>e (P<strong>in</strong>us lambertiana Dougl.) withreference to historical disturbances <strong>in</strong> the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>: Implications forrestoration. <strong>Forest</strong> Ecology and Management 262: 770–779.Millar, Constance I., Robert D. Westfall, Diane L. Delany, Matthew J. Bokach, Alan L.Fl<strong>in</strong>t, and Lorra<strong>in</strong>e E. Fl<strong>in</strong>t. <strong>2012.</strong> <strong>Forest</strong> mortality <strong>in</strong> high-elevation whitebark


vp<strong>in</strong>e (P<strong>in</strong>us albicaulis) forests of eastern California, USA; <strong>in</strong>fluence ofenvironmental context, bark beetles, climatic water deficit, and warm<strong>in</strong>g.Canadian Journal of <strong>Forest</strong> Research 42: 749–765.Powers, Jennifer Sarah, Phillip Soll<strong>in</strong>s, Mark E. Harmon and Julia A. Jones. 1999. Plantpest<strong>in</strong>teractions <strong>in</strong> time and space: A Douglas-fir bark beetle outbreak as a casestudy. Landscape Ecology 14: 105–120.Raffa, Kenneth F., Brian H. Aukema, Barbara J. Bentz, Allan L. Carroll, Jeffrey A.Hicke, Monica G. Turner, and William H. Romme. 2008. Cross-scale drivers ofnatural disturbances prone to anthropogenic amplification: the dynamics of barkbeetle eruptions. BioScience 58(6): 501-517.Simard, Mart<strong>in</strong>, Er<strong>in</strong>n N. Powell, Kenneth F. Raffa, and Monica G. Turner. <strong>2012.</strong> Whatexpla<strong>in</strong>s landscape patterns of tree mortality caused by bark beetle outbreaks <strong>in</strong>Greater Yellowstone? Global Ecology and Biogeography 21: 556–567.Urban, Dean L., Carol Miller, Patrick Halp<strong>in</strong>, and Nathan L. Stephenson. 2000. <strong>Forest</strong>gradient response <strong>in</strong> Sierran landscapes: the physical template. LandscapeEcology 15: 603–620.


viI dedicate this thesis to my family:Marilyn, Roger, Sara & Andrea <strong>Van</strong> <strong>Gunst</strong>


viiACKNOWLEDGMENTS: Primary f<strong>in</strong>ancial support for this project was provided bythe <strong>Lake</strong> <strong>Tahoe</strong> License Plate Program (Project No. LTLP 08-06). I thank my advisor Dr.Peter Weisberg for his guidance and support throughout this project. As a researchassistant <strong>in</strong> the Great Bas<strong>in</strong> Landscape Ecology Lab, my understand<strong>in</strong>g of remotesens<strong>in</strong>g, data analysis, and forest ecology has progressed greatly and I thank Peter for hisrole <strong>in</strong> foster<strong>in</strong>g such a comprehensive graduate school education. This project has alsobenefited <strong>from</strong> <strong>in</strong>put and guidance <strong>from</strong> my committee, Dr. Bob Nowak and Dr. ScottBassett. I thank them for their <strong>in</strong>sights and suggestions <strong>in</strong> improv<strong>in</strong>g and strengthen<strong>in</strong>gthis project. I thank members of the Great Bas<strong>in</strong> Landscape Ecology Lab, especiallySarah Karam and Jian Yang for advice on all th<strong>in</strong>gs R, statistics, and remote sens<strong>in</strong>g. Iespecially thank Yuanchao Fan for his guidance on remote sens<strong>in</strong>g, process<strong>in</strong>g Landsatimagery, and collect<strong>in</strong>g field measurements: this project would have been exponentiallymore difficult without him. I thank Roland Shaw and Elizabeth Harrison at the NevadaDivision of Lands for their comments dur<strong>in</strong>g the progression of this project and BrentObl<strong>in</strong>ger of the USFS <strong>Forest</strong> Health Program <strong>in</strong> decipher<strong>in</strong>g the where and when of forestmortality <strong>in</strong> the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>. I also thank Kurt Fesenmyer, Steve Hanser, andMatthias Leu: their support and guidance <strong>in</strong> mentor<strong>in</strong>g a new scientist is unsurpassed.Lastly, I thank friends and family far and wide, who have provided countless hours ofsupport and comic relief.


viii1986-87 1994-952002-03 2009-10Plate 1: Remote sens<strong>in</strong>g derived maps of annual (Fall Year One: Fall Year Two) canopydieback and healthy pixels <strong>in</strong> coniferous forests of the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>. High mortalityyears, 1986-1987 and 1994-1995, occurred dur<strong>in</strong>g reported drought periods. Lower levelsof mortality occurred <strong>in</strong> a later drought period (2002-2003) with 2009-2010 <strong>in</strong>dicative ofmore average mortality levels <strong>in</strong> addition to patches of tree mortality with<strong>in</strong> densityreduction treatments.


ixPlate 2. Photograph of mounta<strong>in</strong> p<strong>in</strong>e beetle damage <strong>in</strong> High Meadows area <strong>in</strong> lodgepolep<strong>in</strong>e forests <strong>in</strong> southern region of LTB. Figure on right shows 2006-2009 canopymortality derived <strong>from</strong> remote sens<strong>in</strong>g imagery classification.


xTABLE OF CONTENTSAbstract ............................................................................................................................... iLiterature Cited ................................................................................................................ iiiDedication ......................................................................................................................... viAcknowledgments ............................................................................................................ viiPlate 1: Remote sens<strong>in</strong>g derived maps of annual (Fall Year One-Fall Year Two) canopydieback and healthy pixels <strong>in</strong> coniferous forests of the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>. High mortalityyears, 1986-1987 and 1994-1995, occurred around a reported drought period. Lowerlevels of mortality occurred <strong>in</strong> a later drought period (2002-2003) with 2009-2010<strong>in</strong>dicative of more average mortality levels <strong>in</strong> addition to patches of tree mortality with<strong>in</strong>density reduction treatments……………………………………………………………viiiPlate 2. Photograph of mounta<strong>in</strong> p<strong>in</strong>e beetle damage <strong>in</strong> High Meadows area <strong>in</strong> lodgepolep<strong>in</strong>e forests <strong>in</strong> southern region of LTB. Figure on right shows 2006-2009 canopymortality derived <strong>from</strong> remote sens<strong>in</strong>g imagery classification…………………………...ixList of Tables .................................................................................................................. xiiiList of Figures ................................................................................................................. xivChapter 1 – Are Denser <strong>Forest</strong>s Less Healthy: A Remote Sens<strong>in</strong>g Analysis of Density-Dependent <strong>Forest</strong> <strong>Mortality</strong> .................................................................................................1Abstract ...............................................................................................................................2Introduction .........................................................................................................................2Methods .............................................................................................................................10Study Area ..................................................................................................................10


xi<strong>Forest</strong> Types and Ecoregions.......................................................................................12Image Calibration and Normalization .........................................................................13Leaf Area Index and Stock<strong>in</strong>g Inde ..............................................................................14<strong>Forest</strong> <strong>Mortality</strong> ...........................................................................................................15<strong>Mortality</strong> Validation.....................................................................................................18Climate .........................................................................................................................19Data Analysis ...............................................................................................................21Results ...............................................................................................................................22Interaction of Climatic Period and <strong>Forest</strong> <strong>Mortality</strong> ........................................................22Influences on the Density Dependence of <strong>Forest</strong> <strong>Mortality</strong> .....................................…..…23A. By <strong>Forest</strong> Type ……………………………………………………………23B. By Climatic Period………………………………………………….....24C. <strong>Forest</strong> Type and Climatic Period…………………………...…….....25Discussion .........................................................................................................................26Management Implications ............................................................................................30Literature Cited .................................................................................................................32Appendix A. 95 th Percentile of LAI values <strong>in</strong> 17 forest groups .......................................72Appendix B. Relationships among modeled snow deficits, Snow Water Equivalent(SWE) <strong>from</strong> California Cooperative Snow Surveys, and Palmer Drought Severity Index(PDSI) over the study period……………...…………………………………………..…74Appendix C. F<strong>in</strong>al models <strong>in</strong> 5 forest types <strong>in</strong> 25-year study period…............................76


xiiChapter 2 – Spatial and Temporal Patterns of <strong>Forest</strong> <strong>Mortality</strong> <strong>in</strong> <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>,USA: Influence of Climate and Environment ....................................................................80Abstract .............................................................................................................................81Introduction .......................................................................................................................82Topographic <strong>in</strong>fluences on forest mortality……………………...………....82Phytophagous <strong>in</strong>sect population dynamics and forest mortality...................85Climate and temporal variation <strong>in</strong> forest mortality y...................................86Study Objectives............................................................................................87Methods .............................................................................................................................88Study Area...................................................................................................88<strong>Forest</strong> Types and Ecoregions......................................................................89Image Process<strong>in</strong>g and <strong>Mortality</strong> Classification………..............................89Climate……………………………………........………..............................90Environmental Variables.............................................................................91Data Analysis...............................................................................................91Results ...............................................................................................................................92<strong>Forest</strong> <strong>Mortality</strong> across Drought and Wet Periods......................................92Ecoregional Influences on <strong>Forest</strong> <strong>Mortality</strong> ................................................93Topographic effects on forest mortality: Solar Radiation andElevation........................................................................................................94Discussion .........................................................................................................................95Conclusion ......................................................................................................................103Literature Cited ...............................................................................................................104


xiiiAPPENDIX A. 25 years of annual mortality (Fall Year One – Fall Year Two) <strong>from</strong> 1985-2010 <strong>in</strong> the LTB. <strong>Mortality</strong> maps are derived by subtract<strong>in</strong>g year one NDWI <strong>from</strong> yeartwo NDWI to create cont<strong>in</strong>uous maps of dNDWI show<strong>in</strong>g forested areas <strong>in</strong> vary<strong>in</strong>gstages of health, <strong>from</strong> canopy dieback to vegetation recovery to growth ……………...134Thesis Summary...............................................................................................................137LIST OF TABLESTable 1-1. Description of LTB forest types def<strong>in</strong>ed us<strong>in</strong>g the USFS CalVeg vegetationclassification (USDA <strong>Forest</strong> Service. 1981. CALVEG: A Classification of CaliforniaVegetation)……………………………………………………………………..………...49TABLE 1-2. Host-specific native bark beetles of LTB forests and impact of successfulattack on canopy foliage…………………………………………………………………50TABLE 1-3a,b. Validation of remote sens<strong>in</strong>g mortality classification with overallclassification accuracy (oca) and kappa statistics for two values of dNDWI,correspond<strong>in</strong>g to a 5% and 10% loss of green canopy per pixel. The kappa statistic is ameasure of agreement between predicted and actual values <strong>in</strong> a category………………51TABLE 1-4 Annual mortality summary statistics by LTB forest type and climatic period(d1: 1987-1995, d2: 1999-2006, w1: 1985-1987, w2: 1995-1999, w3: 2006-2010)…….52TABLE 1-5. Summary of odds ratios for the stock<strong>in</strong>g-level predictor variable <strong>from</strong>annual models of mortality by forest type over the 25-year time series (SD=StandardDeviation). Odds ratios are derived <strong>from</strong> annual mortality models <strong>in</strong> each forest type and


xivshow the change <strong>in</strong> the probability of mortality for every 0.2 <strong>in</strong>crease <strong>in</strong> stock<strong>in</strong>glevel……………………………………………………………………………...……….53TABLE 1-6. Summary of odds ratios for the stock<strong>in</strong>g-level predictor variable <strong>from</strong>annual models of mortality <strong>in</strong> each climate period occurr<strong>in</strong>g <strong>in</strong> the 25-year time series(SD=Standard Deviation)……………………………………………………………..…54TABLE 2-1. Relationships between current and lagged year modeled snow deficits, asratio of current year snow to 30-year norm (1980-2009) by forest type. Derivation of lagsis shown below, us<strong>in</strong>g 1985 as an example year. Current year is def<strong>in</strong>ed as most currentw<strong>in</strong>tertime precipitation. Positive relationships <strong>in</strong>dicate <strong>in</strong>creased moisture is associatedwith <strong>in</strong>creased mortality. Negative relationships <strong>in</strong>dicate decreased moisture associatedwith <strong>in</strong>creased mortality………………………………………………………………...118TABLE 2-2. L<strong>in</strong>ear regression analysis results for Annual <strong>Mortality</strong> ~ Climatic Period <strong>in</strong>each of five forest types (JP, LP, MF, RF, WF)with significance at the p < 0.05 level………………………………………………….120LIST OF FIGURESFIGURE 1-1. Location of <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong> (LTB) study area…………………………55FIGURE 1-2. <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong> (LTB) forest types per the USFS CalVeg Classification(USDA <strong>Forest</strong> Service. 1981. CALVEG: A Classification of California Vegetation )….56FIGURE 1-3a, b. LAI Validation. Relationship between field-measured leaf area <strong>in</strong>dex(LAI) measured with LICOR LAI-2000 Plant Canopy Analyzer <strong>in</strong> 30 plots <strong>in</strong> the LTBand LAI modeled <strong>from</strong> normalized difference vegetation <strong>in</strong>dex (NDVI), derived <strong>from</strong>remote sens<strong>in</strong>g imagery (a). Model<strong>in</strong>g efficiency of observed vs. modeled LAI. Model<strong>in</strong>g


xvefficiency measures agreement between observed and simulated values aga<strong>in</strong>st the l<strong>in</strong>e ofperfect fit (blue) rather than the regression l<strong>in</strong>e (b, Mayer and Butler 1993)……...…….57FIGURE 1-4. Values of the s<strong>in</strong>gle year normalized difference vegetation <strong>in</strong>dex (NDWI)across 254, 30m x 30m pixels digitized <strong>in</strong> areas rang<strong>in</strong>g <strong>from</strong> severe and extensivemortality across the pixel (High) to pixels without evident canopy damage (None).Results of the ANOVA with follow-up Bonferroni comparison are significant, validat<strong>in</strong>gthe sensitivity of the s<strong>in</strong>gle-year NDWI to LTB mortality (F=33.27, p < 0.006)….……58FIGURE 1-5. Relationship between percent green canopy per plot as measured <strong>in</strong> 24,60m x 60m pixels and the s<strong>in</strong>gle year normalized difference vegetation <strong>in</strong>dex (NDWI)value derived <strong>from</strong> Landsat TM imagery. Percent green canopy is derived <strong>from</strong> percentgreen canopy x average percent green canopy per tree……………………………….…59FIGURE 1-6. Pictorial sequence of remote sens<strong>in</strong>g process<strong>in</strong>g used to derive annualmortality maps. Level 1 terra<strong>in</strong> corrected Landsat TM 5 satellite image is acquired,resized to study area, and normalized to reference imagery (a).Year one (ex.2009) andyear two (ex. 2010) normalized images are transformed <strong>in</strong>to s<strong>in</strong>gle-year NDWI maps (b1,2009: b2, 2010). Cont<strong>in</strong>uous differenced NDWI (dNDWI) map (c) is derived bysubtract<strong>in</strong>g 2009 NDWI (Year 1) <strong>from</strong> 2010 NDWI (Year 2). Map of cont<strong>in</strong>uous dNDWIis classified <strong>in</strong>to maps of healthy and canopy dieback pixels us<strong>in</strong>g validated thresholds(dNDWI < -.053)………………………………………….……………………….……60FIGURE 1-7. Derivation of five climatic periods throughout the 1985-2009 study period,with 1980-1984 for reference. S<strong>in</strong>gle-year snow deficits are derived <strong>from</strong> the currentyear/30-year norm precipitation as snow variable <strong>from</strong> the ClimateWNA dataset (Wanget al. 2012). Four-year lagged deficits (average of current year deficit + 3 previous years)


xvireplicate historical accounts of dry and wet periods <strong>in</strong> the LTB. Climatic periods:w1=1985-1987 (August 1985-July 1986, August1986-July 1987, 2-year duration);d1=1987-1995, 8-year duration: w2= 1995-1999, 4-year duration; d2= 1999-2006, 7-year duration; w3: 2006-2010, 4-year duration.……………………………….………...61FIGURE 1-8. <strong>Mortality</strong> levels and pattern by forest type over the 25-year time series,with smooth local estimator (lowess) curve……………………………………………...62FIGURE 1-9. <strong>Mortality</strong> levels by forest type and climatic period (w1: 1985-1987, d1:1987-1995, w2: 1995-1999, d2: 1999-2006, w3: 2006-2010)…………………………...63FIGURE 1-10. Direction and magnitude of density dependent mortality as odds ratios forevery absolute 20% <strong>in</strong>crease <strong>in</strong> stock<strong>in</strong>g level by forest type over the 25-year time<strong>in</strong>terval. Confidence <strong>in</strong>tervals at the 95 th percentile are shown. Odds ratios symbolized by“x” show years <strong>in</strong> which the stock<strong>in</strong>g variable was not <strong>in</strong>cluded <strong>in</strong> f<strong>in</strong>al model, due tooutcome of likelihood ratio test. Odds ratios symbolized by “0” show years <strong>in</strong> which thestock<strong>in</strong>g variable was <strong>in</strong>cluded <strong>in</strong> f<strong>in</strong>al model……………………………………….….64FIGURE 1-11. Box plot of odds ratios of density-dependent mortality across 25 annualmodels derived for each forest type. Odds ratios illustrate the change <strong>in</strong> the probability ofmortality for every 0.2 (20%) <strong>in</strong>crease <strong>in</strong> stock<strong>in</strong>g level……………………………..…65FIGURE 1-12. Number of models where; 1) Stock<strong>in</strong>g level alone or <strong>in</strong> comb<strong>in</strong>ation withenvironmental variables appeared <strong>in</strong> f<strong>in</strong>al, most parsimonious model, 2) Environmentalvariables only appeared <strong>in</strong> f<strong>in</strong>al, most parsimonious model or, 3) <strong>Mortality</strong> was notexpla<strong>in</strong>ed by stand structure or environmental variables……………………………...…66FIGURE 1-13. Type of density dependent mortality exhibited over percent of 25-yeartime series by direction. Positive DDM: <strong>in</strong>creased density <strong>in</strong>creases probability of


xviimortality; Negative DDM: decreased density <strong>in</strong>creases risk of mortality; and NoAssociation: no association between density and mortality…………………………......67FIGURE 1-14. Ratio of positive to negative density-dependent mortality over five foresttypes. Positive density-dependent mortality <strong>in</strong>dicates an <strong>in</strong>crease <strong>in</strong> the probability ofmortality with <strong>in</strong>creased density. Negative density-dependent mortality <strong>in</strong>dicates adecrease <strong>in</strong> the probability of mortality with <strong>in</strong>creased density………………..……..…68FIGURE 1-15. Type of density dependent mortality exhibited over percent of eachclimatic period <strong>in</strong> all five climatic periods by direction. Positive DDM: <strong>in</strong>creased density<strong>in</strong>creases probability of mortality; Negative DDM: decreased density <strong>in</strong>creases risk ofmortality; and No Association: no association between density and mortality………….69FIGURE 1-16. Average odds ratio for every 0.2 (20%) <strong>in</strong>crease <strong>in</strong> stock<strong>in</strong>g level byclimatic period and forest type. Error bars reflect standard errors……………………....70FIGURE 1-17. Relative frequency of negative (N) or positive (P) density-dependentmortality found <strong>in</strong> each forest type <strong>in</strong> each of four climatic periods (d1, d2, w2, w3)…..71FIGURE 2-1a-d. Ecoregions (a) <strong>in</strong> the LTB with ecoregional differences <strong>in</strong> (b) averageannual precipitation (<strong>in</strong>); (c) average maximum temperature ( o F) ; (d) average m<strong>in</strong>imumtemperature ( o F), derived <strong>from</strong> PRISM 1971-2000 norms.. ………………………..…121FIGURE 2-2a,b. Average w<strong>in</strong>ter m<strong>in</strong>imum temperature (a) and number of frost free days(b) <strong>in</strong> five climatic periods of study time series……………………………………..….122FIGURE 2-3. Solar radiation (WH/m 2 ) calculated for August 1, 2010. Solar radiationcomb<strong>in</strong>es slope and aspect to measure both direct and diffuse radiation over a 30m x 30mpixel. Higher values are associated with south-fac<strong>in</strong>g and lower values with north-fac<strong>in</strong>gaspects…………………………………………………………………………………..123


xviiiFIGURE 2-4. Elevation (m) derived <strong>from</strong> a 30m Digital Elevation Model……………124FIGURE 2-5. Boxplots of annual mortality by forest types <strong>in</strong> five climatic periods ofstudy [d1:1987-1995, d2:1999-2006, w1:1985-1987, w2: 1995-1999, w3:2006-20010]………………………………………………………………………….…...…125FIGURE 2-6. Regression tree of predicted mortality levels by forest type and climaticperiod. Regression trees are comprised of a series of b<strong>in</strong>ary splits created throughrecursive partition<strong>in</strong>g with splitt<strong>in</strong>g criterion based on overall deviance reduction.Predicted values of annual mortality are shown at the term<strong>in</strong>us of each node for thatgroup. For example, the predicted annual mortality value for JP, MF, and WF dur<strong>in</strong>geither d2 or w3 is 9.69……………………………………………………………….….126FIGURE 2-7. Effect plots for Annual <strong>Mortality</strong> ~ Climatic Period (ClimPd: d1, d2, w1,w2, w3) <strong>in</strong> each of the five forest types…………………………………………...……127FIGURE 2-8. Effect plots for regression analysis of Annual <strong>Mortality</strong> ~ EcoRegion *Climatic Period <strong>in</strong> JP, MF, and RF forests. Red l<strong>in</strong>es <strong>in</strong>dicate the upper and lower 95%confidence <strong>in</strong>tervals around regression coefficients……………………………………128FIGURE 2-9a-c. Levels of JP(a), MF (b), and RF(c) forest mortality by ecoregion (eh, ej,ek, el, et) over the 25 year time period…………………………………………………129FIGURE 2-10. Effect of a 1,000 WH/m 2 <strong>in</strong>crease <strong>in</strong> solar radiation on mortality for fiveforest types across 25-year time period as odds ratios with 95% confidence <strong>in</strong>tervals.N=odds ratios with confidence <strong>in</strong>tervals that overlap 1 (CI shown) or years when solarradiation did not appear <strong>in</strong> f<strong>in</strong>al model (CI not shown). U= odds ratios with confidence<strong>in</strong>tervals that do not overlap 1. Results <strong>from</strong> annual models occur<strong>in</strong>g dur<strong>in</strong>g dry periodsare <strong>in</strong> orange. Results <strong>from</strong> annual models occur<strong>in</strong>g <strong>in</strong> dry periods are <strong>in</strong> green. Odds


xixratios greater than one <strong>in</strong>dicate <strong>in</strong>crease <strong>in</strong> probability of mortality associated with<strong>in</strong>creases <strong>in</strong> south-fac<strong>in</strong>g slopes. Odds ratios less than one <strong>in</strong>dicate probability ofmortality associated with north-fac<strong>in</strong>g slopes…………………………………………..130FIGURE 2-11. Location of <strong>in</strong>creased risk of mortality by aspect derived <strong>from</strong> solarradiation for five forest types across 25-year time period (Not Significant: confidence<strong>in</strong>terval of regression coefficient and result<strong>in</strong>g odds ratio overlapped 0 or 1; NoAssociation: no association between probability of mortality and elevation)…………131FIGURE 2-12. Effect of a 500m <strong>in</strong>crease <strong>in</strong> elevation on mortality for five forest typesacross 25-year time period as odds ratios with 95% confidence <strong>in</strong>tervals. N=odds ratioswith confidence <strong>in</strong>tervals that overlap 1 (CI shown) or years when elevation did notappear <strong>in</strong> f<strong>in</strong>al model (CI not shown). U= odds ratios with confidence <strong>in</strong>tervals that donot overlap 1. Results <strong>from</strong> annual models occur<strong>in</strong>g dur<strong>in</strong>g dry periods are <strong>in</strong> orange.Results <strong>from</strong> annual models occur<strong>in</strong>g <strong>in</strong> dry periods are <strong>in</strong> green. Odds ratios greater thanone <strong>in</strong>dicate <strong>in</strong>crease <strong>in</strong> probability of mortality associated with <strong>in</strong>creases <strong>in</strong> elevation.Odds ratios less than one <strong>in</strong>dicate probability of mortality associated with decreases <strong>in</strong>elevation………………………………………………………………………………132FIGURE 2-13. Location of <strong>in</strong>creased risk of mortality by elevation for five forest typesacross 25-year time period (Upper: Upper elevation; Lower: lower elevation; No Effect:no association between probability of mortality and elevation; Not Significant:confidence <strong>in</strong>terval of regression coefficient and result<strong>in</strong>g odds ratio overlapped0 or 1)…………………………………………………………......................................133


1Are Denser <strong>Forest</strong>s Less Healthy: A Remote Sens<strong>in</strong>g Analysis of Density-Dependent<strong>Forest</strong> <strong>Mortality</strong>Krist<strong>in</strong> Jane <strong>Van</strong> <strong>Gunst</strong> ,a,c , Peter J. Weisberg a,ba Department of Natural Resources and Environmental Science, University of NevadaReno, 1664 N. Virg<strong>in</strong>ia Street, Reno, NV, 89557.b Program <strong>in</strong> Ecology, Evolution and Conservation Biology, University of Nevada Reno,Mail Stop 314, Reno, NV, 89557.cCorrespond<strong>in</strong>g Author: Phone: 775.784.4020, Fax: 775.784.4583 Email:kvangunst@cabnr.unr.eduEmail. P.Weisberg: pweisberg@cabnr.unr.edu


2ABSTRACTIncreased stand density is commonly cited as a predispos<strong>in</strong>g factor <strong>in</strong> forest mortality,although evidence over larger landscapes and longer timeframes is lack<strong>in</strong>g. We<strong>in</strong>vestigated how the relationship between stand-level mortality and stand density variesby forest type and climatic period <strong>in</strong> the mixed conifer forests of the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>(LTB) <strong>in</strong> the central Sierra Nevada. Our analyses used Landsat TM data <strong>from</strong> 1985-2010to derive annual mortality maps and an annual stock<strong>in</strong>g <strong>in</strong>dex. We found that the strengthof density-dependent mortality was variable <strong>in</strong> the LTB, with no forest show<strong>in</strong>g aubiquitous relationship between mortality and density. However, <strong>in</strong>creases <strong>in</strong> densitywere weakly associated with <strong>in</strong>creased probability of mortality dur<strong>in</strong>g major dry periodsand <strong>in</strong> lower-elevation and p<strong>in</strong>e-dom<strong>in</strong>ated forests. In middle- and upper-elevationforests, <strong>in</strong>creased density was associated with <strong>in</strong>creased mortality risk <strong>in</strong> dry periods,with this effect reversed <strong>in</strong> wet periods. Years when particular forest types experiencedthe greatest mortality levels did not correspond to years when density-dependentmortality was greatest, suggest<strong>in</strong>g that agents of epidemic mortality may act <strong>in</strong> a density<strong>in</strong>dependentmanner.INTRODUCTIONWidespread <strong>in</strong>creases <strong>in</strong> forest mortality have become prevalent over much of thewestern United States, with mortality rates doubl<strong>in</strong>g every 17-29 years <strong>in</strong> some areas (vanMantgem et al. 2009). Recent <strong>in</strong>creases <strong>in</strong> tree mortality have been evidenced <strong>in</strong>whitebark p<strong>in</strong>e (P<strong>in</strong>us albicaulis) <strong>in</strong> the Greater Yellowstone Ecosystem (Logan et al.2010) and the Sierra Nevada (Millar et al. 2012); limber p<strong>in</strong>e (P<strong>in</strong>us flexilis) and mixedconifer stands of the Sierra Nevada (Egan et al. 2011; Ferrell et al. 1994; Guarín and


3Taylor 2005; Maloney et al. 2011; Millar et al. 2007, 2012; van Mantgem andStephenson 2007; van Mantgem et al. 2009); ponderosa p<strong>in</strong>e (P<strong>in</strong>us ponderosa), mixedconifer, and p<strong>in</strong>yon-juniper woodlands of the Southwestern US (Breshears et al. 2005;Floyd et al. 2009; Negrón et al. 2009; Ganey and Vojta 2011); subalp<strong>in</strong>e and middleelevationforests of Utah and Colorado (DeRose and Long 2012; Hebertson and Jenk<strong>in</strong>s2008), lodgepole p<strong>in</strong>e forests of British Columbia, Canada (Aukema et al. 2006); spruceforests of Alaska (Berg et al. 2006), and temperate coniferous forests of Europe (Allen etal. 2010 and references there<strong>in</strong>; Dobbert<strong>in</strong> et al. 2005, 2007). These dieback events havebeen caused by a diversity of host-specific mortality agents <strong>in</strong>clud<strong>in</strong>g mounta<strong>in</strong> p<strong>in</strong>ebeetle (MPB, Dendroctonus ponderosae Hopk<strong>in</strong>s) <strong>in</strong> lodgepole p<strong>in</strong>e (P<strong>in</strong>us contorta),sugar p<strong>in</strong>e (P<strong>in</strong>us lambertiana), and whitebark p<strong>in</strong>e stands; spruce beetle (Dendroctonusrufipennis Kirby) <strong>in</strong> Englemann spruce stands; p<strong>in</strong>yon ips (Ips confusus Leconte) <strong>in</strong>p<strong>in</strong>yon-juniper woodlands, and several <strong>in</strong>sect species <strong>in</strong>clud<strong>in</strong>g fir engraver (Scolytusventralis) and Jeffrey p<strong>in</strong>e beetle (Dendroctonus jeffreyi) <strong>in</strong> mixed conifer forests (Eganet al 2011; Ferrell et al. 1994; Ferrell 1996). <strong>Forest</strong> dieback has been l<strong>in</strong>ked to climatewarm<strong>in</strong>g and drought (Bentz et al. 2010; Berg et al. 2006; Guarín and Taylor 2005;Hebertsen and Jenk<strong>in</strong>s 2008; Millar et al. 2012; van Mantgem and Stephenson 2007), butalso attributed to denser forests associated with fire exclusion (Dobbert<strong>in</strong> et al. 2005;Egan et al.2011; Ferrell et al. 1994; Galiano et al. 2010; Guarín and Taylor 2005;Maloney et al. 2011; Millar et al. 2012; Negrón et al. 2009).<strong>Forest</strong> mortality can be either patchy or widespread, occurr<strong>in</strong>g at chronic backgroundlevels with small groups of trees affected, or at outbreak levels with widespread tree dieoffacross much of the landscape. Causes of forest mortality and the conditions that give


4rise to outbreaks of widespread mortality are complex and hard to predict (Bentz et al.2010; Berg et al. 2006; Dukes et al. 2009; Frankl<strong>in</strong> et al. 1987 but see Auclair 2006). Inconiferous forests, mortality is mediated by a diverse array of host-specific pests andpathogens that <strong>in</strong>teract with edaphic and climatic factors, as well as <strong>in</strong>dividual treecharacteristics, to suppress growth or cause tree mortality (Castello et al.1995). Manion(1991) framed reductions <strong>in</strong> tree vigor often lead<strong>in</strong>g to mortality as a result of threefactors: “predispos<strong>in</strong>g factors” due to long-term stress, “<strong>in</strong>cit<strong>in</strong>g factors” of shortduration, and “contribut<strong>in</strong>g factors” that ultimately lead to mortality if a tree cannotrecover <strong>from</strong> the <strong>in</strong>cit<strong>in</strong>g stress. While a plethora of both endogenous and exogenousfactors comb<strong>in</strong>e to <strong>in</strong>fluence tree mortality, competition among trees <strong>in</strong> dense foreststands is often considered a predispos<strong>in</strong>g factor lead<strong>in</strong>g to episodes of widespread forestmortality (Guarín and Taylor 2005; Millar et al. 2012; Maloney and Rizzo 2002;Maloney et al. 2011; Raumann and Cablk 2008).The primary role of competition <strong>in</strong> govern<strong>in</strong>g forest dynamics dur<strong>in</strong>g early standdevelopment is widely recognized, but less well understood is the <strong>in</strong>tensity and impact ofcompetitive forces <strong>in</strong> structur<strong>in</strong>g mature forests (Das et al. 2011; Frankl<strong>in</strong> et al 2002; Peetand Christensen 1987; Re<strong>in</strong>eke 1933; Vygodskaya et al. 2002; Yoda et al. 1963). Inmature forests, the “zone of <strong>in</strong>fluence” of competition is typically restricted to a smallarea proximate to the plant, with abiotic stressors more important <strong>in</strong> driv<strong>in</strong>g tree mortality(Frankl<strong>in</strong> et al. 2002; Peet and Christensen 1987). Background mortality rates <strong>in</strong> matureand old-growth forests are hypothesized to be low, with elevated levels caused by largescaleepisodic factors, such as bark beetle outbreaks (Frankl<strong>in</strong> et al. 2002; Peet andChristensen 1987).


5Across the western United States, <strong>in</strong>creased stand density result<strong>in</strong>g <strong>from</strong> over acentury of fire exclusion is often considered a significant cause of heightened forestmortality. The strength of density-dependent mortality can be positive, where <strong>in</strong>creases <strong>in</strong>stand density are associated with <strong>in</strong>creased probability of mortality or negative, wheredecreases <strong>in</strong> density area associated with <strong>in</strong>creased probability of mortality. The l<strong>in</strong>kbetween mortality and density is based on the Optimal Defense Theory <strong>in</strong> whichsusta<strong>in</strong>ed <strong>in</strong>ter-tree competition for shared critical resources limits the ability of trees togenerate the energy-<strong>in</strong>tensive chemical and physical structures necessary to survive barkbeetle attacks (Raffa and Berryman 1983; Showalter and Filip 1993; Stamp 2003).However, a compet<strong>in</strong>g theory of plant defense, the Growth-Differentiation-Balancehypothesis, states that when growth is slowed by environmental factors, the resource poolavailable to allocate to secondary defense <strong>in</strong>creases (Coley et al. 1985; Loomis 1932,1953; Herms and Mattson 1992; Stamp 2003). Although a comprehensive understand<strong>in</strong>gabout the impacts of plant stress on mortality may be lack<strong>in</strong>g (Stamp 2003), many studieshave shown that decreased tree vigor <strong>in</strong>creases risk of successful bark beetle attack.Although some have shown that decl<strong>in</strong><strong>in</strong>g growth rate separates surviv<strong>in</strong>g trees <strong>from</strong>dead trees (Bigler et al. 2007; Das et al. 2007; L<strong>in</strong>ares et al. 2009, 2010; Pedersen 1998;Suarez et al. 2004), others have not found that dist<strong>in</strong>ction (Kane and Kolb 2010; Powerset al. 1999; Reid and Robb 1999). Whether vigorous or weakened trees are preferred notonly depends on the bark beetle species, with some such as MPB preferentially attack<strong>in</strong>ghealthy trees, but also on mortality <strong>in</strong>duced dur<strong>in</strong>g non-epidemic or epidemic outbreakswhen mass attacks can overwhelm tree defenses of even healthy trees (Logan et al. 1998;Powers et al. 1999; Raffa and Berryman 1983; Wallner 1987). In addition, manufacture


6of compounds and structures necessary to mount plant defense systems is likely moreproximal to understand<strong>in</strong>g why some trees die and others do not (Kane and Kolb 2010).Increased stand density can also directly <strong>in</strong>fluence the transmission of pests andpathogens by alter<strong>in</strong>g landscape connectivity. In Western forests, a majority of mortalityis mediated by bark beetles (Dendroctonus, Ips, Scolytus) which attack dur<strong>in</strong>g the lateSpr<strong>in</strong>g to early Fall and, <strong>in</strong> the case of a successful attack, excavate egg galleries <strong>in</strong> the<strong>in</strong>ner bark <strong>from</strong> which new beetles emerge <strong>in</strong> the next Spr<strong>in</strong>g (Christensen et al. 1987;Ferrell 1996). Successful attack by bark beetle <strong>in</strong> one tree <strong>in</strong> a crowded stand can <strong>in</strong>creaseprobability of attack to proximate conspecifics, due to <strong>in</strong>creased numbers of bark beetles<strong>in</strong> the stand as well as proximity to <strong>in</strong>fected trees (Christensen et al. 1987; Das et al.2008; Raffa and Berryman 1983). Proximate and touch<strong>in</strong>g canopies and tree root systemscan facilitate the spread of damage agents such as dwarf mistletoe (Arceuthobium spp.)and root rots which decrease photosynthetic capacity (Aukema et al. 2010; Cruickshanket al. 1997; Maloney and Rizzo 2002). Infection by these damage agents may furtherweaken trees, predispos<strong>in</strong>g them to mortality due to bark beetle attack (Ferrell 1996;Kenaley et al. 2008, but see Maloney and Rizzo 2002). Increased forest homogeneity,decreased structural complexity, decreased species diversity, and <strong>in</strong>creased stand densityresult<strong>in</strong>g <strong>from</strong> fire exclusion can lead to more extensive mortality by <strong>in</strong>creas<strong>in</strong>g supplyand cont<strong>in</strong>uity of food resources for host-specific bark beetles (Moritz et al. 2010; Perryet al. 2011).Several studies, however, have po<strong>in</strong>ted out the limited role that stand density can play<strong>in</strong> forest mortality, with recent forest decl<strong>in</strong>es attributed more to the effects of <strong>in</strong>creas<strong>in</strong>gmoisture stress due to prolonged drought, spread and proliferation of native bark beetle


7populations, edaphic conditions, and topographic gradients (Bentz et al. 2010; Das et al.2011; Ganey and Vojta 2011; L<strong>in</strong>es et al. 2010; Nelson et al. 2007; Sánchez-Martínezand Wagner 2002; van Mantgem and Stephenson 2007; van Mantgem et al. 2009).Drought-<strong>in</strong>duced mortality has been implicated as a primary factor <strong>in</strong> many widespreadtree die-offs both globally and across the western United States (Allen et al. 2010;Breshears et al. 2005; McDowell et al. 2008; Millar et al. 2011; van Mantgem andStephenson 2007). The role of drought versus elevated <strong>in</strong>sect herbivory on tree mortalitycan be confounded, however, as warm<strong>in</strong>g temperatures can also directly <strong>in</strong>fluence barkbeetle populations (Bentz et al. 2010). Numerous damaged and stand<strong>in</strong>g dead treescaused by w<strong>in</strong>dstorms or heavy snow loads have also been shown to lead to widespreadforest mortality (Christiansen et al.1987). Local conditions def<strong>in</strong>ed by environmental andedaphic gradients can also <strong>in</strong>fluence mortality risk (Allen et al. 2010; Nelson et al. 2007)and confound importance of stand density. <strong>Forest</strong> structure is tied to the physicaltemplate, and failure to account for the effects of topography on mortality can stronglyoverestimate the role of density. He and Duncan (2000) found that account<strong>in</strong>g forelevation rendered spurious their <strong>in</strong>itial f<strong>in</strong>d<strong>in</strong>g of strong density-dependent mortalitybetween Douglas fir (Pseudotsuga menziesii) and western hemlock (Tsuga heterophylla).F<strong>in</strong>ally, the epidemic or non-epidemic nature of forest mortality can obscure primarydrivers. Dur<strong>in</strong>g epidemic periods, <strong>in</strong>creased populations of bark beetles capable ofoverwhelm<strong>in</strong>g tree defenses through susta<strong>in</strong>ed mass attacks may overshadow theimportance of tree vigor and stand-level characteristics (Nelson et al. 2006). Differences<strong>in</strong> the importance of tree vigor and spatial aggregation may help expla<strong>in</strong> why hazardrat<strong>in</strong>gs based on stand characteristics have little predictive power when applied to


8landscapes (Logan et al. 1998; Nelson et al. 2006). Logan et al (1998) posit that current<strong>in</strong>terest <strong>in</strong> bark beetle epidemics has biased our understand<strong>in</strong>g of the role of standstructure <strong>in</strong> augment<strong>in</strong>g mortality risk: with more self-focus<strong>in</strong>g damage agents on thelandscape, the role of spatial proximity and landscape configurations may become moreimportant than stand characteristics (Holdenreider et al. 2004; Logan et al. 1998;MacQuarrie and Cooke 2011).The <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong> (LTB) <strong>in</strong> the central Sierra Nevada provides an ideal studyarea for explor<strong>in</strong>g the relative role of stand density for <strong>in</strong>fluenc<strong>in</strong>g forest mortalitypatterns. The LTB is characterized by steep elevational gradients, facilitat<strong>in</strong>g study ofvary<strong>in</strong>g forest types and <strong>in</strong>fluences of environmental gradients on forest mortality.Dur<strong>in</strong>g the study period, the LTB experienced two drought cycles and three wet periodswith bark beetle populations at both epidemic and non-epidemic levels. Extensive forestdecl<strong>in</strong>e occurred dur<strong>in</strong>g the late 1980s to the mid-1990s. By 1991, Elliot-Fisk (1996)estimated 300 million board feet of timber <strong>in</strong> the <strong>Tahoe</strong> Bas<strong>in</strong> was dead or dy<strong>in</strong>g. Us<strong>in</strong>gremotely sensed imagery, Macomber and Woodcock (1994) quantified mortality <strong>in</strong> theconifer and red fir zones of the LTB at a 15% loss of timber volume between 1988 and1992. In addition, as elsewhere <strong>in</strong> the western United States, the LTB experienced major<strong>in</strong>creases <strong>in</strong> stand density follow<strong>in</strong>g 19 th Century timber harvests and subsequent fireexclusion (Barbour et al. 2002; Taylor 2004).Chang<strong>in</strong>g climate regimes, post-settlement logg<strong>in</strong>g and fire suppression policies<strong>in</strong>stituted <strong>in</strong> the early 1900s have altered post-settlement (mid 1800s) forest structure andcomposition, with lower elevation post-settlement forests more dense and dom<strong>in</strong>ated byshade-tolerant white fir compared to pre-settlement forests (Barbour et al. 2002; Millar


9and Woolfenden 1999; North et al. 2005; Taylor 2004). In Carson Range forests, Taylor(2004) estimates a five-fold <strong>in</strong>crease <strong>in</strong> tree density and two-fold <strong>in</strong>crease <strong>in</strong> basal areabetween pre and post-settlement Jeffrey p<strong>in</strong>e-white fir forests. Post-settlement lodgepolep<strong>in</strong>e and red fir (Abies magnifica)-western white p<strong>in</strong>e (P<strong>in</strong>us monticola) forests have also<strong>in</strong>creased <strong>in</strong> density, with species composition <strong>in</strong> red fir-western white p<strong>in</strong>e forests<strong>in</strong>creas<strong>in</strong>gly comprised of lodgepole p<strong>in</strong>e (Taylor 2004).Increases <strong>in</strong> forest mortality rates, projected climatic changes, and concerns aboutforest health and wildfire severity have led forest management agencies <strong>in</strong> the LTB, aselsewhere <strong>in</strong> the western United States, to plan and implement forest-wide densityreductions. The objectives of such density reduction treatments are to mitigate fire riskand restore forest health (Fettig et al. 2007). Although fuel reduction treatments havedemonstrated effectiveness <strong>in</strong> reduc<strong>in</strong>g fire <strong>in</strong>tensity dur<strong>in</strong>g subsequent wildfires <strong>in</strong> theSierra Nevada (Graham et al. 1999; Safford et al. 2009), the effectiveness of suchtreatments for improv<strong>in</strong>g long-term forest health is not well demonstrated (Bradley andTueller 2001; Fettig et al. 2007, 2008, 2012; Schwilk et al. 2006).Although <strong>in</strong>creased density is often implicated <strong>in</strong> studies of forest mortality, fewstudies have directly addressed the question: how are forest density and forest mortalityrelated? Remote sens<strong>in</strong>g analysis of Landsat Thematic Mapper (TM) archival imagery <strong>in</strong>the LTB provides <strong>in</strong>sight <strong>in</strong>to how stand structure affects mortality at the wholewatershedlevel over decadal time scales. The overall objective of this study was toexam<strong>in</strong>e how density dependent mortality has varied with forest type and for wet and dryperiods over the past 25 years <strong>in</strong> the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>, which <strong>in</strong>cludes forests that arerepresentative of p<strong>in</strong>e and mixed-conifer types throughout the western United States. We


11(Sierra Nevada) of the LTB with m<strong>in</strong>or ra<strong>in</strong>fall amounts dur<strong>in</strong>g the summer and fallmonths. Average annual precipitation is approximately 79cm.January and February are the coldest months with average m<strong>in</strong>imum temperaturesjust above 19F (-7.22 o C) (<strong>Tahoe</strong>,CA Western Regional Climate Center (WRCC)http://www.wrcc.dri.edu/cgi-b<strong>in</strong>/cliMAIN.pl?ca8758 ). July and August are the warmestmonths with average maximum temperatures just above 77F (25 o C). The eastern side ofthe LTB, formed by the Carson Range, is much drier. December-March snowfall stillaccounts for the majority of the precipitation <strong>in</strong>puts with an average monthly snowfall of46.23cm (Glenbrook, NV WRCC http://www.wrcc.dri.edu/cgi-b<strong>in</strong>/cliMAIN.pl?nv3205 ).Average annual precipitation is approximately 45cm.LTB forests have been affected by land use practices <strong>from</strong> both pre- and post-EuroAmerican settlement periods (SNEP 1996). Fire, due to both Native American andlightn<strong>in</strong>g ignitions, was common <strong>in</strong> the Sierra Nevada dur<strong>in</strong>g the pre-settlement period(SNEP 1996). The LTB experienced extensive deforestation start<strong>in</strong>g <strong>in</strong> 1860 to providetimber for m<strong>in</strong>es of the Comstock lode, with approximately two-thirds of LTB forests cutby 1930 (SNEP 1996). After the Comstock Era, sheep graz<strong>in</strong>g was widespread withmillions of sheep trailed <strong>from</strong> California to the Great Bas<strong>in</strong> <strong>from</strong> 1865 through the 1890s.In the early to mid 1900s, fire suppression policies were <strong>in</strong>stituted by forest managementagencies across the West, <strong>in</strong>clud<strong>in</strong>g the Sierras and the LTB (SNEP 1996).In many lower and middle-elevation forests of the western United States, frequent fireregulated stand density and favored dom<strong>in</strong>ance by fire-tolerant, shade-<strong>in</strong>tolerant p<strong>in</strong>es(Taylor 2004). Dendrochronological reconstructions for the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong> of theSierra Nevada <strong>in</strong>dicate that historic mixed-severity surface fire occurred with a mean fire


12rotation <strong>in</strong>terval of 11.4 years (Taylor 2004) <strong>in</strong> the Jeffrey p<strong>in</strong>e-white fir forest type. Inupper elevation red fir forests, the fire free <strong>in</strong>terval was much greater. Scholl and Taylor(2006) estimated a mean po<strong>in</strong>t fire <strong>in</strong>terval of 76 years for old-growth stands on the eastside of the LTB, with fires characterized as hav<strong>in</strong>g been of low to moderate severity.Lodgepole p<strong>in</strong>e <strong>in</strong> the Sierra Nevada typically does not produce serot<strong>in</strong>ous cones, likelydid not depend on crown fire for regeneration, and probably experienced crown andsurface fire only <strong>in</strong>frequently (Parker 1986; Taylor 2004).<strong>Forest</strong> Types and EcoregionsWe exam<strong>in</strong>ed forest mortality patterns for five forest types and five ecoregions acrossthe <strong>Tahoe</strong> Bas<strong>in</strong> us<strong>in</strong>g the 2009 USFS CalVeg vegetation classification (Figure 1-2;USDA <strong>Forest</strong> Service. 1981. CALVEG: A Classification of California Vegetation.CalVegZone3. EVeg Exist<strong>in</strong>g Vegetation Tiles 17B, 21A, and 21B). <strong>Forest</strong> types wereidentified accord<strong>in</strong>g to dom<strong>in</strong>ant forest composition as: Jeffrey P<strong>in</strong>e Alliance (JP)(Jeffrey P<strong>in</strong>e or Eastside P<strong>in</strong>e), Red Fir (RF), Lodgepole P<strong>in</strong>e (LP), White Fir (WF), andthe Mixed Conifer Alliance (MF) (Table 1-1). We further divided each forest type by oneof five ecoregions def<strong>in</strong>ed by the CalVeg dataset to def<strong>in</strong>e 15 unique forest type xecoregion areas. We selected only those pixels identified by the coniferous cover typewith the dom<strong>in</strong>ant vegetation as WF, RF, EP (East-side p<strong>in</strong>e, <strong>in</strong>corporated <strong>in</strong>to the JPforest type), JP, LP, and MF. We removed all non-forested areas designated as urban <strong>in</strong>the CalVeg dataset, except for certa<strong>in</strong> misclassified areas that were not near urbandevelopment. We buffered all major roads <strong>in</strong> the LTB by 120 meters (60 meters bothsides of the road) and removed those pixels <strong>from</strong> the analysis. Pixels located <strong>in</strong> historicfires (California Department of <strong>Forest</strong>ry and Fire Protection) and <strong>in</strong> fuel reduction


13treatment areas were also excluded. Pixels dom<strong>in</strong>ated by tall (>2m) chaparral cover wereomitted as these contributed to an overestimation of tree leaf area <strong>in</strong>dex. Areas formerlydom<strong>in</strong>ated by chaparral but classified as tree-dom<strong>in</strong>ated pixels <strong>in</strong> 2009 may have been<strong>in</strong>cluded <strong>in</strong> our analysis. However, us<strong>in</strong>g high-resolution imagery <strong>from</strong> the ArcServer(ArcGIS 10, ESRI), we removed all pixels <strong>in</strong> which chaparral has dom<strong>in</strong>ated <strong>in</strong> recentyears. Other dom<strong>in</strong>ant shrub and herbaceous vegetation had dim<strong>in</strong>ished photosyntheticactivity by the time of image collection <strong>in</strong> late September to mid-October, and so did notbias leaf area <strong>in</strong>dex estimations. We removed all meadows, edges of small water bodies,areas dom<strong>in</strong>ated by aspen (Populus tremuloides), riparian vegetation and developed areassurround<strong>in</strong>g campgrounds, ski areas, or residential areas.Image Calibration and NormalizationWe used Fall Landsat Thematic Mapper (TM) imagery <strong>from</strong> 1985 – 2010, available<strong>from</strong> the USGS Glovis site (http://glovis.usgs.gov/). All pixels used <strong>in</strong> our analyses werecloud- and snow-free. We subset all images to a bound<strong>in</strong>g rectangle that <strong>in</strong>cluded theLTB and performed an absolute radiometric correction of errors associated withatmospheric transmission and reflectance between sensor and earth surface for theSeptember 2010 reference image. Images <strong>from</strong> other years (1985 – 2009) werenormalized to the reference imagery us<strong>in</strong>g the IRMAD method (Canty and Nielsen2008). The IRMAD method uses canonical correlation to isolate “true change” due tovegetation change rather than cloud cover or satellite discrepancies between the referenceand the target imagery. This process allowed images <strong>from</strong> different years, with differentatmospheric conditions, to be truly comparable. Pixels with<strong>in</strong> <strong>Lake</strong> <strong>Tahoe</strong> were not<strong>in</strong>cluded <strong>in</strong> the normalization process.


14Leaf Area Index and Stock<strong>in</strong>g IndexAll images were transformed to Normalized Difference Vegetation Index (NDVI), aratio of red to near-<strong>in</strong>frared bands which conta<strong>in</strong>s <strong>in</strong>formation about the health and extentof green vegetation <strong>in</strong> a pixel (Huete et al. 1997). NDVI is calculated us<strong>in</strong>g[1]where NIR=Near-Infrared wavelength, Landsat TM band 4 and VIS=Visible (Red)wavelength, Landsat TM band 3. We used regression model<strong>in</strong>g aga<strong>in</strong>st field-measureddata to further transform NDVI <strong>in</strong>to leaf area <strong>in</strong>dex (LAI) based on the nonl<strong>in</strong>ear modelof Baret and Guyot (1991), which accounts for NDVI saturation when LAI <strong>in</strong>creases tohigh values. This model uses the follow<strong>in</strong>g equation:[2] –where VI equals NDVI, VI g equals NDVI value of bare soil, VI∞ equals asymptoticvalue of NDVI when LAI tends toward <strong>in</strong>f<strong>in</strong>ity, and K VI equals the ext<strong>in</strong>ction coefficientcontroll<strong>in</strong>g slope of relationship between NDVI and LAI.For model calibration, we used the NDVI and LAI values <strong>from</strong> 30 field-sampled plotssampled with a LI-COR 2000 Plant Canopy Analyzer (LI-COR, L<strong>in</strong>coln, Nebraska). andfound the NDVI value of 0.257 at bare soil (VI g ) <strong>in</strong> 5 areas <strong>in</strong> the study area almostcompletely dom<strong>in</strong>ated by bare soil and solved for the rema<strong>in</strong><strong>in</strong>g two parameters: theasymptotic value of NDVI at 0.760 (VI∞ ) and the ext<strong>in</strong>ction rate between NDVI andLAI of 0.345 (K VI ). In many coniferous systems, the asymptotic value of NDVI isbetween 0.6 - 0.8 (White et al. 1997).


15We evaluated the accuracy of modeled LAI results us<strong>in</strong>g both regression analysis(Figure 1-3a) and the model<strong>in</strong>g efficiency (EF) statistic (Figure 1-3b, Mayer and Butler1993). The EF statistic is similar to the common R 2 statistic, but compares fit betweenpredicted and observed values aga<strong>in</strong>st the l<strong>in</strong>e of true fit, with values close to one<strong>in</strong>dicat<strong>in</strong>g a perfect fit. Results <strong>in</strong>dicate that modeled LAI is strongly proportional tofield-measured LAI, but is overestimated. However, the overestimation is notproblematic for the stock<strong>in</strong>g <strong>in</strong>dex calculation (described below) because this <strong>in</strong>dexnormalizes LAI <strong>in</strong> that it is used <strong>in</strong> both numerator and denom<strong>in</strong>ator.We used modeled leaf area <strong>in</strong>dex to derive an estimate of stock<strong>in</strong>g level (“stock<strong>in</strong>g<strong>in</strong>dex”) where higher values <strong>in</strong>dicate greater conversion of available resources <strong>in</strong>to foliarbiomass. Stock<strong>in</strong>g <strong>in</strong>dex was developed <strong>from</strong> estimated LAI as the ratio of LAI at agiven pixel <strong>in</strong> a given year, to the 95 th percentile of LAI for the ecoregion encompass<strong>in</strong>gthat pixel (Appendix A). Thus, values for the stock<strong>in</strong>g <strong>in</strong>dex ranged between 0 and 1,with values above the 95 th percentile truncated at 1. A stock<strong>in</strong>g <strong>in</strong>dex near 1 <strong>in</strong>dicates thatthe leaf area <strong>in</strong>dex of a forest is near the maximum encountered for a given ecoregion,which represents a particular comb<strong>in</strong>ation of forest composition, soil type, andmicroclimate.<strong>Forest</strong> <strong>Mortality</strong>To estimate reductions <strong>in</strong> canopy vigor, we used the Normalized Difference WetnessIndex (NDWI) and the change <strong>in</strong> NDWI values (differenced NDWI, dNDWI) betweentwo consecutive years (Gao 1996). NDWI is calculated us<strong>in</strong>g[3]


16where NIR=Near-Infrared wavelength, Landsat TM band 4 and SWIR=ShortwaveInfrared wavelength, Landsat TM band 5. NDWI is less sensitive to atmosphericscatter<strong>in</strong>g and more sensitive to water absorption, render<strong>in</strong>g the <strong>in</strong>dex better suited todetection of coniferous forest mortality than other vegetation <strong>in</strong>dices, such as NDVI orthe Tasseled Cap Indices (Gao 1996; Hunt and Rock 1989; J<strong>in</strong> and Sader 2005;Vogelmann 1990; Vogelmann et al. 2009; Wilson and Sader 2002). Index strength <strong>in</strong>captur<strong>in</strong>g gradations of foliage health was measured aga<strong>in</strong>st 254 pixels <strong>in</strong> 80 polygons ofvary<strong>in</strong>g mortality levels, del<strong>in</strong>eated with<strong>in</strong> USFS aerial survey damage polygons for 2008that were reconstructed us<strong>in</strong>g 2009 high-resolution imagery (Figure1-4; USDA <strong>Forest</strong>Service-<strong>Forest</strong> Health Monitor<strong>in</strong>g, 2008 Damage Polygons). Trees <strong>in</strong> high mortality plotswere def<strong>in</strong>ed by USFS as hav<strong>in</strong>g over 75% of crown as red/brown/gray needles with treeswith fad<strong>in</strong>g crown foliage cover<strong>in</strong>g over 75% of the digitized polygon. Areas with lowmortality had less than 25% of fad<strong>in</strong>g trees cover<strong>in</strong>g the polygon, and medium mortalitypolygons had <strong>from</strong> 25-50% of the polygon covered by trees with 25-50% of fad<strong>in</strong>gfoliage. Two way-ANOVA with a follow-up Bonferroni multiple comparison test wasused to compare NDWI values with the USFS forest mortality polygons. NDWI valueswere significantly different among high, medium and low, and no mortality pixels (p


17field data. In early fall of 2010, we measured canopy cover us<strong>in</strong>g the moosehorn canopymeasurement tool and recorded percent green canopy cover for trees over 5cm DBHacross 24, 60m x 60m plots with vary<strong>in</strong>g species composition, density, and amount ofmortality. Average percent green canopy per plot was regressed aga<strong>in</strong>st the s<strong>in</strong>gle-year2010 NDWI (Figure1-5). We used l<strong>in</strong>ear regression to estimate percent green canopycover per pixel, differenced NDWI among consecutive years to obta<strong>in</strong> dNDWI, and thenconsidered three possible thresholds of canopy mortality to convert cont<strong>in</strong>uous annualmortality maps, based on dNDWI, <strong>in</strong>to b<strong>in</strong>ary maps of “healthy” and “unhealthy” pixels.The three dNDWI thresholds considered corresponded to an absolute loss of 5, 10, or20% of green canopy per year. Initial sensitivity analyses revealed that the 20% thresholdwas too strict, but that the 5% loss (correspond<strong>in</strong>g dNDWI= -0.0503) and 10% loss(correspond<strong>in</strong>g dNDWI= -0.106) thresholds exhibited reasonable behavior. Simard et al.(2012) used a similar approach to transform remote-sens<strong>in</strong>g derived mortality maps tobeetle-killed and undisturbed pixels us<strong>in</strong>g a threshold of 10% beetle-killed basal area. Insummary, annual mortality maps were derived <strong>in</strong> the follow<strong>in</strong>g sequence (Figure 1-6): 1)Image acquisition and process<strong>in</strong>g (a), 2) Transformation of image pairs <strong>in</strong>to s<strong>in</strong>gle-yearNDWI (b1 and b2), 3) Calculation of differenced NDWI between image pairs <strong>in</strong> forestedareas of LTB (c), and 4) Transformation of cont<strong>in</strong>uous dNDWI <strong>in</strong>to b<strong>in</strong>ary classificationof healthy v canopy dieback pixels (d).<strong>Mortality</strong> processes <strong>in</strong> tree crowns are def<strong>in</strong>ed by progressive changes <strong>in</strong> spectral<strong>in</strong>dices as canopy foliage dies, with needles turn<strong>in</strong>g <strong>from</strong> green to red to gray andeventually dropp<strong>in</strong>g <strong>from</strong> the tree (Ahern 1988; Dennison et al. 2010; Wulder et al.2006). Pixel NDWI may decl<strong>in</strong>e further follow<strong>in</strong>g the <strong>in</strong>itial onset of mortality due to


18spatial spread of mortality agents among proximate conspecifics, and due to changes <strong>in</strong>spectral reflectance as needles become chlorotic and fall <strong>from</strong> tree.Therefore we used only the first <strong>in</strong>cidence of mortality with<strong>in</strong> a four-year period (asdef<strong>in</strong>ed by the dNDWI threshold), <strong>in</strong> a previously-healthy pixel. We used the temporaltrajectories of canopy fade by major bark beetle types with<strong>in</strong> the Bas<strong>in</strong> (Table 1-2) to seta mortality period at four years to allow for the longest sequence of bark beetle attackexpressed as canopy foliage change. Once a pixel reached the dNDWI threshold of atleast five percent loss of green canopy, that pixel was considered “available” formodel<strong>in</strong>g, provided it had at least 6% green canopy.Us<strong>in</strong>g thresholded values and the four year w<strong>in</strong>dow, we derived annual mortality foreach forest type as a quantification of the amount of new mortality seen on the landscape<strong>in</strong> each year. For the years 1986-1988, we excluded years that had not experiencedmortality <strong>in</strong> the preced<strong>in</strong>g 1-3 years. Thus, “mortality” as used <strong>in</strong> this study equates to theonset of a significant canopy dieback event occurr<strong>in</strong>g with<strong>in</strong> a 30m x 30m pixel.<strong>Mortality</strong> ValidationWe validated the five and ten percent loss of green canopy thresholds <strong>in</strong> 776 pixels,us<strong>in</strong>g 2010 high resolution imagery to select areas where gray and red canopies wereeither present (“mortality” pixels, n=277) or absent (“healthy”, n=499). We used the2010-2009 dNDWI values for the healthy pixels and the 4-year m<strong>in</strong>imum of 2007-2006,2008-2007, 2009-2008, or 2010-2009 dNDWI for the mortality pixels. The majority ofmortality pixels were digitized <strong>in</strong> the High Meadows mounta<strong>in</strong> p<strong>in</strong>e beetle outbreak <strong>in</strong>the southern portion of the LTB (Plate 2). Nearly all pixels were a mix of both red and


19gray canopies, a result of the cont<strong>in</strong>ued presence and spread of MPB damage <strong>from</strong> about2005-2009 (USDA <strong>Forest</strong> Service Report #SS09-03).Agreement (kappa statistic or Κ) between our Landsat classification and USFSmortality mapp<strong>in</strong>g was greater at the lower threshold of green canopy loss (5% loss:Κ=0.78, 10% loss: Κ=0.56) (Table 1-3a,b). The 10% loss threshold was sensitive tooutbreak mortality but was unable to consistently detect more <strong>in</strong>cremental decreases <strong>in</strong>canopy vigor. We set the mortality threshold at a loss of at least 5% of green canopy peryear as most sensitive to the dist<strong>in</strong>ction between “ healthy” and “mortality” pixels with ahigher overall classification accuracy (5% loss: 89.69%, 10% loss: 81.69%) and higherkappa statistic.ClimateOver the 25-year time series we quantified how relationships between stand structureand mortality changed by forest type and climatic period, us<strong>in</strong>g five climatic periodsbased on historical snow levels and w<strong>in</strong>ter m<strong>in</strong>imum temperatures (ClimateWNA dataset;Wang et al. 2012). ClimateWNA downscales PRISM (Parameter-elevation Regressionson Independent Slopes Model; Daly et al. 2002) data to derive estimates for seasonalclimatic values based on elevation and geographical location. Because w<strong>in</strong>ter snowaccounts for over 80% of precipitation <strong>in</strong>puts to the LTB, we used snow departure <strong>from</strong>norm over a four-year w<strong>in</strong>dow to identify years with significant moisture deficits. Wealso used w<strong>in</strong>ter m<strong>in</strong>imum temperatures to account for relative differences <strong>in</strong> snowmeltand also because w<strong>in</strong>ter temperatures are known to regulate the distribution and lifecycles of bark beetles, key mortality agents <strong>in</strong> the <strong>Tahoe</strong> Bas<strong>in</strong> (Logan et al. 1998).


20We sampled approximately 9,000 po<strong>in</strong>ts <strong>from</strong> the ClimateWNA dataset, located on auniform grid with spac<strong>in</strong>g of 300 m. From this data set we derived yearly averages forsnow and temperature, annual departure <strong>from</strong> norm, as well as a four year mov<strong>in</strong>gaverage (current year + 3 previous) for the 1980-2009 period. We correlatedClimateWNA-modeled snow levels with snow water equivalent collected by theCalifornia Cooperative Snow Surveys for the Ward Creek (WR2) and Glenbrook (GL2)areas (Appendix B) to ensure that modeled snow levels agreed with on-the-groundmeasurements. We then identified five climatic periods based on historical accounts ofdry and wet years and periods <strong>in</strong> the LTB and surround<strong>in</strong>g areas as well as lagged snowdeficits (Egan et al. 2011; Ferrell et al. 1994; Guarín and Taylor 2005; Millar et al. 2012;Nevada Department of Water Resources, SNEP 1996; USDA <strong>Forest</strong> Service ReportSS09-12). Because reported dry periods differed <strong>in</strong> beg<strong>in</strong>n<strong>in</strong>g and end dates depend<strong>in</strong>gon location and study timel<strong>in</strong>e, we tested the relationship between mortality and vary<strong>in</strong>glags of snow deficit to establish beg<strong>in</strong>n<strong>in</strong>g and end<strong>in</strong>g dates of the five climatic periods.Here, we accounted not only for antecedent moisture conditions but also the impact ofanomalous wet years that occurred dur<strong>in</strong>g overall dry periods and dry years that occurreddur<strong>in</strong>g overall wet periods. In the Sierra Nevada and elsewhere, tree mortality is not oftencorrelated with s<strong>in</strong>gle year precipitation deficits and overall tree mortality levels areaffected dissimilarly by differences <strong>in</strong> drought tim<strong>in</strong>g, duration, and magnitude (Bigler etal. 2007; Guarín and Taylor 2005; Millar et al. 2007, 2010). Based on historicalprecipitation and temperature records, Truckee River chronology, historical accounts ofdrought periods <strong>in</strong> the LTB, and lagged snow deficits derived <strong>from</strong> the ClimateWNA, we


21established five climatic periods dur<strong>in</strong>g the 1985-2010 study series (Figure 1-7) ofvary<strong>in</strong>g duration (w1=2years, d1=8years, w2=4years, d2=7years, w3=4years).Data AnalysisWe used the Rare Events Logistic Regression (ReLogit) method to developgeneralized l<strong>in</strong>ear models of mortality probability as a function of environmental andstand structure predictors (K<strong>in</strong>g and Zeng 2001). As normal logistic regression cansharply underestimate the probability of occurrence for events that are <strong>in</strong>tr<strong>in</strong>sically rare,the Relogit method adds a pseudo-Bayesian prior to the logistic regression model andadjusts parameter estimates based on the probability of a rare event’s occurrence <strong>in</strong> thepopulation. This leads to narrower standard errors for each parameter and thus a moreconservative estimate of model error. In our analysis, we found very slight differencesbetween parameter estimates derived <strong>in</strong> a typical generalized l<strong>in</strong>ear model and thoseresult<strong>in</strong>g <strong>from</strong> the ReLogit algorithm.Logistic regression us<strong>in</strong>g the Relogit method was used to model the probability of amortality event with<strong>in</strong> a pixel as a statistical function of forest density and environmental<strong>in</strong>fluences. For each forest type <strong>in</strong> each year, we sampled 20% of pixels that had notexperienced mortality <strong>in</strong> the preced<strong>in</strong>g four years. For JP, MF, and RF forests that areseparated <strong>in</strong>to five ecoregions, we sampled 20% of pixels <strong>from</strong> each ecoregion.We used a two-part regression model<strong>in</strong>g approach to test whether stand structure issignificantly associated with forest health. <strong>Mortality</strong> at the 5% threshold was statisticallymodeled as a function of environmental variables (ecoregion, elevation and solarradiation) to create a “covariate model” of best fit. Likelihood ratio tests were then usedto evaluate whether <strong>in</strong>clusion of the stock<strong>in</strong>g variable, <strong>in</strong> turn, improved the regression


22model beyond the covariate model. The f<strong>in</strong>al model was chosen based on highestexplanatory power and model parsimony (Appendix C). Results are <strong>in</strong>terpreted as oddsratios, derived <strong>from</strong> parameter coefficients as effect sizes for every 0.2 (20%) <strong>in</strong>crease <strong>in</strong>stock<strong>in</strong>g level. Odds ratios greater than 1 <strong>in</strong>dicate a positive relationship betweenpredictor and response and less than 1, a negative relationship. In models where thestock<strong>in</strong>g variable was not <strong>in</strong>cluded <strong>in</strong> the f<strong>in</strong>al model, the odds ratio for the stock<strong>in</strong>gvariable implicitly equals 1, show<strong>in</strong>g no relationship between mortality and stock<strong>in</strong>g. Therelative frequency of models with positive or negative odds ratios was used to describedifferences among forest types with respect to density-dependent <strong>in</strong>fluences on forestmortality. Similarly, the frequency of models with positive or negative odds ratios acrosseach wet and dry period was further used to describe the variable effects of climate ondensity dependent mortality.RESULTSInteraction of Climatic Period and <strong>Forest</strong> <strong>Mortality</strong><strong>Forest</strong> mortality has varied by year and forest type throughout the study period, butwas generally greatest prior to 1995 which spanned the first wet period and first drought(Table 1-4; Figure 1-8; Figure 1-9). Upper-elevation forests (RF and LP) showed highmagnitude spikes of mortality at the end of drought periods, follow<strong>in</strong>g successive yearsof drought, or follow<strong>in</strong>g years of abundant snowfall, with low mortality <strong>in</strong> <strong>in</strong>terven<strong>in</strong>gyears. These upper-elevation forest types also experienced the greatest mean levels offorest mortality overall. JP and MF showed more stable mortality levels, with slightlygreater mortality dur<strong>in</strong>g the first dry period. WF forests were <strong>in</strong>termediate, with majormortality events occurr<strong>in</strong>g at the end of prolonged drought periods. <strong>Mortality</strong> levels were


23far greater dur<strong>in</strong>g the first drought period (1987 – 1994) than dur<strong>in</strong>g the second droughtperiod of record (1999 – 2005) (Figure 1-9).Influences on the Density Dependence of <strong>Forest</strong> <strong>Mortality</strong>A. By <strong>Forest</strong> TypeThe strength of density dependent mortality was variable <strong>in</strong> magnitude and directionamong forest types, years, and climatic periods (Figure 1-10). For all forest types, thestrongest positive relationship between stock<strong>in</strong>g level and forest mortality (positiveDDM) occurred between 1993 and 1994, when the risk of mortality was <strong>in</strong>creased by 1.7-2.8 times for every 20% <strong>in</strong>crease <strong>in</strong> stock<strong>in</strong>g level. The strongest negative relationshipbetween stock<strong>in</strong>g level and forest mortality (negative DDM), consistent among foresttypes, occurred <strong>in</strong> 1985. Negative DDM was exhibited to a greater degree <strong>from</strong> 1995-1999 with positive DDM more frequent dur<strong>in</strong>g the 1987-1994 and 2000-2009 periods.Over the time series, JP had the highest average density-dependent mortality with anodds ratio of 1.16 (95% CI = 1.05 to 1.27), followed by LP, with an average odds ratio of1.10 (95% CI = 1.01 to 1.19). Average odds ratios for MF, RF, and WF forests weremuch lower at 1.06 (95% CI = 0.92 to 1.20), 1.05 (95% CI = 0.96 to 1.14), and 1.03(95% CI = 0.87 to 1.19) respectively (Table 1-5; Figure 1-11). Relative to other foresttypes, JP and LP forests exhibited negative density-dependence least frequently andpositive density-dependence most frequently at 44% and 40% of the time seriesrespectively (Figure 1-12; Figure 1-13). <strong>Mortality</strong> <strong>in</strong> RF forests exhibited positive densitydependence over 40% of the time series, but exhibited negative density dependence over36% of the time series. In MF forests, positive DDM occurred half as frequently asnegative DDM, which occurred over 40% of the time period.


24Over at least half of the time period, WF and JP models showed no relationshipbetween density and mortality. <strong>Mortality</strong> and density were not associated <strong>in</strong> 48% of LPmodels. In MF and RF forests, density and mortality were associated over a greatermajority of the time series, at 36% and 24% respectively. Relative frequency of positiveto negative DDM over the time period was highest for JP (11), LP (3.3), and RF (1.1)(Figure 1-14). Ratio of positive to negative density-dependence was less than one and ofsimilar magnitude for WF (0.57) and MF (0.6), <strong>in</strong>dicat<strong>in</strong>g that these forests experiencednegative DDM more frequently than positive DDM over the time series.B. By Climatic PeriodClimate period <strong>in</strong>fluenced the strength and direction of average density-dependentmortality, with both drought periods exhibit<strong>in</strong>g positive DDM and the first two wetperiods exhibit<strong>in</strong>g negative DDM (Table 1-6). Average strength of DDM was highestoverall dur<strong>in</strong>g the first drought period at 1.20 (95% CI = 1.06 to 1.34), followed by thesecond drought at 1.08 (95% CI = 1.02 to 1.14), and the third wet period at 1.03 (95% CI= 0.96 to 1.11). Average DDM was lowest <strong>in</strong> the first wet period (0.89, 95% CI = 0.82 to1.11) and the second wet period at 0.98 (95% CI = 0.89 to 1.07).In wet periods, the frequency of negative DDM equaled or exceeded that of positiveDDM (Figure 1-15). Dur<strong>in</strong>g the first wet period, negative DDM was found <strong>in</strong> 60% of allmodels with lack of DDM <strong>in</strong> the rema<strong>in</strong><strong>in</strong>g models. In the later two wet periods, negativeDDM was evidenced <strong>in</strong> 40% and 30% respectively of all models. Dur<strong>in</strong>g both dryperiods, positive DDM accounted for 40% of the time period and mortality was notassociated with density over roughly 45% of each dry period. Negative DDM was


25slightly more frequent dur<strong>in</strong>g the second dry period compared to the earlier droughtperiod.Model results show<strong>in</strong>g no relationship between density and mortality occurred mostfrequently <strong>in</strong> dry periods. Models show<strong>in</strong>g no relationship between mortality and densityaccounted for approximately 47% and 45% of d1 and d2, respectively. In the first andthird wet periods, mortality was not associated with density <strong>in</strong> over 40% of all modelsoccurr<strong>in</strong>g dur<strong>in</strong>g the climate period. In the second wet period, mortality was notassociated with density <strong>in</strong> over 35% of all models.C. <strong>Forest</strong> Type and Climatic PeriodAverage strength of DDM varied by forest type and climatic period, with positiveDDM evident for JP and LP <strong>in</strong> both wet and dry periods and DDM more variable for MF,RF, and WF (Figure 1-16). Dur<strong>in</strong>g the first wet period, all forests exhibited averageDDM between 0.85 and 0.96. Dur<strong>in</strong>g the first dry period, all forests exhibited positiveDDM, with highest average DDM for JP and MF and lowest for LP. In the second andthird wet period, average DDM was between 0.93 and 0.98 for MF and RF. Dur<strong>in</strong>g thesewet periods, average DDM for JP and LP was approximately 1.1 <strong>in</strong> the second wetperiod, and decl<strong>in</strong>ed to approximately 1.06 <strong>in</strong> the third wet period. For WF, averageDDM was positive <strong>in</strong> the third wet period with an average odds ratio of approximately1.13 and between 0.83 and 0.91 <strong>in</strong> the second wet period and second dry periodrespectively.In the first dry period, all forest types exhibited positive DDM to a much greaterdegree than negative DDM (Figure 1-17). This separation is magnified <strong>in</strong> JP and RFforests. Negative and positive DDM were of more similar frequency <strong>in</strong> LP, MF, and WF


26forests. In the second dry period, both JP and RF ma<strong>in</strong>ta<strong>in</strong>ed D1 disparities betweenpositive and negative DDM, with positive relationships account<strong>in</strong>g for a greaterpercentage of each time period. However, <strong>in</strong> MF and WF forests negative DDM is eitherequal (MF) or of greater (WF) frequency than positive DDM.JP and LP forests ma<strong>in</strong>ta<strong>in</strong>ed highest or near-highest positive DDM over all climaticperiods. For MF and RF forests, however, negative DDM is more frequent than positiveDDM dur<strong>in</strong>g each wet period. WF behaves as other forest types dur<strong>in</strong>g the first drought,but then exhibits a different pattern than the other forest types, with positive DDMdom<strong>in</strong>ant <strong>in</strong> the second wet period and negative DDM dom<strong>in</strong>ant <strong>in</strong> the second dry periodand the follow<strong>in</strong>g wet period.DISCUSSIONWe found evidence for both positive and negative density-dependent mortality acrossall forests of the LTB. No forest type showed a persistent, unidirectional relationshipbetween density and mortality over the entire time series and no forest type showed arelationship between density and mortality <strong>in</strong> every year of the 25-year time period.Rather, relationships between mortality and density were variable over the 25-year timeperiod <strong>in</strong> ways that were somewhat consistent for each forest type.Positive density dependent mortality suggests competition as a causal mechanism,although other explanations are possible. For example, denser stands may facilitatespread of pathogens and <strong>in</strong>sect herbivores (Das et al. 2008), or may simply <strong>in</strong>dicate areasof previously high rates of establishment on currently unfavorable microsites(Greenwood and Weisberg 2008). However, if competition is to be <strong>in</strong>ferred <strong>from</strong> positivedensity dependent mortality, our results show that competition has been a relatively weak


27but persistent factor for forest mortality <strong>in</strong> the LTB, with its importance heightened <strong>in</strong>drier forest types and dur<strong>in</strong>g drought periods. Our results parallel those of other studiesthat have <strong>in</strong>ferred a limited role of competition for lead<strong>in</strong>g to <strong>in</strong>creased forest mortality <strong>in</strong>mature and old-growth forests (Das et al. 2011; Floyd et al. 2009; Ganey and Vojta2011; L<strong>in</strong>es et al. 2010; van Mantgem and Stephenson 2007; van Mantgem et al. 2009;Vygodskaya et al. 2002).Our study agrees with others f<strong>in</strong>d<strong>in</strong>g that extended dry periods and <strong>in</strong>creased moisturestress are more proximal causes of elevated mortality than are stand densities (Das et al.2012; Floyd et al. 2009; L<strong>in</strong>es et al. 2010; van Mantgem and Stephenson 2007; vanMantgem et al. 2009), with one important caveat. While upper-elevation RF and LPforests experienced high mortality levels dur<strong>in</strong>g dry periods, we surmise that this forestdecl<strong>in</strong>e was augmented by the preced<strong>in</strong>g wet and warm period dur<strong>in</strong>g the early 1980s.High numbers of damaged trees, relicts of prolonged w<strong>in</strong>ter storms <strong>in</strong> the early 1980’smay have provided abundant resources and substrate for epidemic bark beetlepopulations (Christiansen et al. 1987). In the <strong>in</strong>terior West, MPB outbreaks often followextremely wet periods, with elevated temperatures important <strong>in</strong> regulat<strong>in</strong>g MPBabundance, distribution, and epidemics (Bentz et al. 2010; Logan et al. 1998). In addition,while extended drought often reduces tree vigor, it does not necessarily predispose treesto susta<strong>in</strong>ed and successful attack as many beetles select high-quality food resources(Christiansen et al. 1987; Logan et al. 1998).Variations <strong>in</strong> the strength of density dependent mortality dur<strong>in</strong>g non-epidemic orepidemic periods have been posited <strong>in</strong> many studies and the preponderance of forestmortality studies dur<strong>in</strong>g or follow<strong>in</strong>g epidemics may have biased our understand<strong>in</strong>g of


28the role of stand density (Logan et al. 1998). In the Sierra Nevada and elsewhere, studiesshow<strong>in</strong>g strong, positive density-dependence have been conducted dur<strong>in</strong>g periods ofepidemic forest decl<strong>in</strong>e (Ferrell et al. 1994, Egan et al. 2011), with little support for suchrelationships dur<strong>in</strong>g non-epidemic periods (Das et al. 2011; Sanchez-Mart<strong>in</strong>ez andWagner 2002; van Mantgem and Stephenson 2007; van Mantgem et al. 2009, but seeFloyd et al. 2009). As mortality becomes more widespread and spatially aggregated,forest structure at the stand level may become less important. The issue becomes one ofscale: mortality dur<strong>in</strong>g non-epidemic periods is expressed at smaller scales and driven byprocesses that operate at that scale. <strong>Mortality</strong> dur<strong>in</strong>g epidemic periods is more of alandscape-scale phenomenon (Aukema et al. 2006; Holdenreider et al. 2004; Lev<strong>in</strong> 1992;Nelson et al 2006). In our study, the strength of DDM was generally highest for all foresttypes dur<strong>in</strong>g a period of reported <strong>in</strong>sect outbreak <strong>in</strong> the LTB. In this case, strength ofdensity dependent mortality at the stand level was imposed by a larger-scale constra<strong>in</strong>t atthe landscape level (beetle population clusters) with low magnitude variations imposedby the characteristics of species assemblages and forest types (Lev<strong>in</strong> 1992).Dur<strong>in</strong>g non-epidemic bark beetle conditions that were more typical of the yearsfollow<strong>in</strong>g 1994, dist<strong>in</strong>ctions between strength of DDM <strong>in</strong> forest types and climates aremore evident. In the mesic and lower-stress environments of WF and MF, negative DDMdom<strong>in</strong>ated or was extremely weak. In more stressful environments at upper and lowerelevations, <strong>in</strong>creased resource scarcity for shared and limited resources may have fostered<strong>in</strong>creased competition <strong>in</strong> less diverse (JP and LP) or more clumped (RF) forests. Das et al(2008) found similar dist<strong>in</strong>ctions <strong>in</strong> the relative strength of DDM between fir and p<strong>in</strong>e.


29For sugar p<strong>in</strong>e, mortality risk <strong>in</strong>creased with <strong>in</strong>creas<strong>in</strong>g density of conspecifics but forwhite fir, mortality risk decreased with <strong>in</strong>creas<strong>in</strong>g density of conspecifics.In JP forests, DDM dur<strong>in</strong>g the second drought period was similar to subsequent andpreced<strong>in</strong>g wet periods. For LP and RF forests, however, positive DDM dur<strong>in</strong>g the seconddry period might be more representative of environmental conditions conducive togrowth <strong>in</strong> subalp<strong>in</strong>e communities: warm and dry weather (Millar et al. 2004). Acceleratedgrowth rates and <strong>in</strong>creased recruitment <strong>in</strong> exist<strong>in</strong>g clumped RF forests and <strong>in</strong> gaps <strong>in</strong> LPstands may have fostered <strong>in</strong>creased competition for space and nutrients.While weak but significant and persistent density-dependent mortality was found <strong>in</strong>our study, we do not f<strong>in</strong>d support for the widely held assumption that <strong>in</strong>creased densitydependentmortality drives more severe and widespread mortality. For example, althoughpositive density dependence is evident <strong>in</strong> JP forests, forest mortality levels for this typeare the most stable and of lowest magnitude of all forest types over the time period. Ourresults further suggest that f<strong>in</strong>d<strong>in</strong>gs developed at the stand level may not translate directlyto understand<strong>in</strong>g forest mortality patterns at the landscape level.Our f<strong>in</strong>d<strong>in</strong>gs suggest that climate may play a greater role than forest structure for<strong>in</strong>fluenc<strong>in</strong>g forest mortality patterns <strong>in</strong> the central Sierra Nevada. Further, the effect offorest structure on forest mortality may be closely l<strong>in</strong>ked to climate variability. For red firand mixed fir forests <strong>in</strong> particular, the direction of DDM is l<strong>in</strong>ked to climate, with dryperiods characterized by positive DDM and wet periods characterized by negative DDM.These results are <strong>in</strong> accord with numerous studies of plant-plant <strong>in</strong>terspecific <strong>in</strong>teractionsthat have found facilitation to be most important <strong>in</strong> environments of high abiotic stress,and competition most important <strong>in</strong> moderate conditions where stress is low (e.g. Bertness


30and Callaway 1994; Callaway 1998). Dense and clumped stands at upper elevations canameliorate harsh environmental extremes, such as high w<strong>in</strong>ds, heavy snowloads, andprolonged extreme m<strong>in</strong>imum temperatures (James et al. 1994; L<strong>in</strong>gua et al. 2008;Tranquill<strong>in</strong>i 1979). In our study, density dependence switched <strong>from</strong> a positive to negativeeffect dur<strong>in</strong>g wet periods characterized by <strong>in</strong>creas<strong>in</strong>g stress <strong>from</strong> heavy snow loads.Stand density may play an important facilitative role <strong>in</strong> favor<strong>in</strong>g establishment,recruitment, and survival dur<strong>in</strong>g such wet w<strong>in</strong>ters.Management ImplicationsOur results provide guidance for forest managers seek<strong>in</strong>g to use density reductiontreatments to improve forest health <strong>in</strong> the Sierra Nevada. First, stand density reductionsmay not consistently lead to forest health benefits <strong>in</strong> middle and upper-elevation forests.Red fir and mixed fir stands with greater densities experienced less mortality, especiallydur<strong>in</strong>g wet periods. In such forests, any beneficial impacts of th<strong>in</strong>n<strong>in</strong>g treatments to foresthealth are likely to be short-lived, with reversal <strong>in</strong> subsequent wet periods. Second,density reduction treatments may improve forest health for low-elevation forests where<strong>in</strong>creased tree density was associated with <strong>in</strong>creased forest mortality. However, suchforests did not experience widespread mortality over the period of record. For Jeffreyp<strong>in</strong>e forests, where positive DDM is most frequent of all forest types, annual forestmortality levels were lowest and most stable, suggest<strong>in</strong>g limited forest-wide impacts ofstand-level risk. Positive relationships between forest density and subsequent forestmortality were apparent <strong>in</strong> less than half of years, even for the Jeffrey p<strong>in</strong>e forests. <strong>Forest</strong>managers of the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong> should expect the forest health benefits of th<strong>in</strong>n<strong>in</strong>g


31treatments to be variable, <strong>in</strong>consistent, limited to periods of non-epidemic mortality, andto be greatest for forests dom<strong>in</strong>ated by shade-<strong>in</strong>tolerant p<strong>in</strong>e species.


32LITERATURE CITEDAhern, F.J. 1988. The effects of bark beetle stress on the foliar spectral reflectance oflodgepole p<strong>in</strong>e. International Journal of Remote Sens<strong>in</strong>g 9: 1451–1468.Allen, Craig D., Alison K. Macalady , Haroun Chenchouni, Dom<strong>in</strong>ique Bachelet, NateMcDowell,Michel Vennetier, Thomas Kitzberger, Andreas Rigl<strong>in</strong>g , David D.Breshears, E.H. (Ted) Hogg, Patrick Gonzalez , Rod Fensham, Zhen Zhang, JorgeCastro, Natalia Demidova, Jong-Hwan Lim, Gillian Allard, Steven W. Runn<strong>in</strong>g,Akk<strong>in</strong> Semerci, and Neil Cobb. 2010. A global overview of drought and heat<strong>in</strong>ducedtree mortality reveals emerg<strong>in</strong>g climate change risks for forests. <strong>Forest</strong>Ecology and Management 259 (4): 660–684.Auclair, Allan N.D. 2005. Patterns and general characteristics of severe forest dieback<strong>from</strong> 1950 to 1995 <strong>in</strong> the northeastern United States. Canadian Journal of <strong>Forest</strong>Research 35: 1342–1355.Aukema, B. H., A.L. Carroll, J.Zhu, K.F. Raffa, T.A. Sickley, and S.W. Taylor. 2006.Landscape level analysis of mounta<strong>in</strong> p<strong>in</strong>e beetle <strong>in</strong> British Columbia, Canada:spatiotemporal developments and spatial synchrony with<strong>in</strong> the present outbreak.Ecography 29: 427-441.Aukema, Brian H., Jun Zhu, Jesper Møller, Jakob G. Rasmussen , Kenneth F. Raffa.2010. Predisposition to bark beetle attack by root herbivores and associatedpathogens: Roles <strong>in</strong> forest decl<strong>in</strong>e, gap formation, and persistence of nonepidemicbark beetle populations. <strong>Forest</strong> Ecology and Management 259: 374–382.


33Baret, F. and G. Guyot. 1991. Potentials and limits of vegetation <strong>in</strong>dexes for LAI andAPAR assessment. Remote Sens<strong>in</strong>g of Environment 35:161-173.Barbour, M, E. Kelley, P. Maloney, D.Rizzo, E. Royce and J. Fites-Kaufmann. 2002.Present and past old-growth forest of the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>, Sierra Nevada, US.Journal of Vegetation Science 13: 461-472.Bentz, Barbara, Jacques Régnière, Christopher J. Fettig, E. Matthew Hansen, Jane L.Hayes, Jeffrey A. Hicke, Rick G. Kelsey, Jose F. Negrón, and Steven J. Seybold.2010. Climate change and bark beetles of the Western United States and Canada:Direct and <strong>in</strong>direct effects. Bioscience. 60(8): 602-613.Berg, Edward E, J. David Henry, Christopher L. Fastie, Andrew D. De Volder, Steven M.Matsuoka. 2006. Spruce beetle outbreaks on the Kenai Pen<strong>in</strong>sula, Alaska, andKluane National Park and Reserve, Yukon Territory: Relationship to summertemperatures and regional differences <strong>in</strong> disturbance regimes. <strong>Forest</strong> Ecology andManagement 227: 219–232.Bertness,M.D. and R.M. Callaway.1994. Positive <strong>in</strong>teractions <strong>in</strong> communities. Trends <strong>in</strong>Ecology and Evolution 9: 191–193.Bigler, C., D.G. Gav<strong>in</strong>, C. Gunn<strong>in</strong>g, and T. T.Veblen. 2007. Drought <strong>in</strong>duces lagged treemortality <strong>in</strong> a subalp<strong>in</strong>e forest <strong>in</strong> the Rocky Mounta<strong>in</strong>s. Oikos 116: 1983–1994.Bradley, Tim and Paul Tueller. 2001. Effects of fire on bark beetle presence on Jeffreyp<strong>in</strong>e <strong>in</strong> the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>. <strong>Forest</strong> Ecology and Management 142: 205-214.Breshears, D.D.,N.S. Cobb, P.M. Rich, K.P Price, C.D. Allen, R.G. Balice, W.H.Romme, J.H. Kastens, M.L. Floyd, J. Belnap, J.J. Anderson, O.B. Myers, andC.W. Meyer. 2005. Regional vegetation die-off <strong>in</strong> response to global-change-type


34drought. Proceed<strong>in</strong>gs of the National Academy of Sciences of the United States ofAmerica 102 (42): 15144–15148.California Department of <strong>Forest</strong>ry and Fire Protection. Fire and Resource AssessmentProgram. Fire Perimeters Dataset: http://frap.cdf.ca.gov/data/frapgisdata/select.aspCallaway, R.M. 1998. Competition and facilitation on elevation gradients <strong>in</strong> subalp<strong>in</strong>eforests of the northern Rocky Mounta<strong>in</strong>s, USA. Oikos 82: 561-573.Canty, Morton J. and Allan A. Nielsen. 2008. Automatic radiometric normalization ofmultitemporal satellite imagery with the iteratively re-weighted MADtransformation. Remote Sens<strong>in</strong>g of Environment 112: 1025–1036.Castello, John D., Donald J. Leopold, and Peter J. Smallidge. 1995. Pathogens, Patterns,and Processes <strong>in</strong> <strong>Forest</strong> Ecosystems. BioScience 45(1): 16-24.Coley, Phyllis D., John P. Bryant, and F. Stuart Chap<strong>in</strong> III. 1985. Resource Availabilityand Plant Antiherbivore Defense. Science 230: 895-899.Cruickshank, M.G. D.J. Morrison, and Z.K. Punja. 1997. Incidence of Armillaria ostoyae<strong>in</strong> precommercial th<strong>in</strong>n<strong>in</strong>g stumps and spread of Armillaria ostoyae to adjacentDouglas-fir trees. Canadian Journal of <strong>Forest</strong> Research 27: 481-490.Daly, Christopher, Wayne P. Gibson, George H. Taylor, Gregory L. Johnson, andPhillip Pasteris. 2002. A knowledge-based approach to the statistical mapp<strong>in</strong>g ofclimate. Climate Research 22: 99–113.Das, Adrian, John Battles, Nathan L. Stephenson, Phillip J. van Mantgem. 2007. Therelationship between tree growth patterns and likelihood of mortality: a study oftwo tree species <strong>in</strong> the Sierra Nevada. Canadian Journal of <strong>Forest</strong> Research 38:580-597.


35Das, Adrian, John Battles, Nathan L. Stephenson, Phillip J. van Mantgem. 2008. Spatialelements of mortality risk <strong>in</strong> old-growth forests. Ecology 89 (6): 1744-1756.Das, Adrian, John Battles, Nathan L. Stephenson, Phillip J. van Mantgem. 2011. Thecontribution of competition to tree mortality <strong>in</strong> old-growth coniferous forests.<strong>Forest</strong> Ecology and Management 261: 1203–1213.Dennison, Philip E., Andrea R. Brunelle, andVachel A. Carter. 2010. Assess<strong>in</strong>g canopymortality dur<strong>in</strong>g a mounta<strong>in</strong> p<strong>in</strong>e beetle outbreak us<strong>in</strong>g GeoEye-1 high spatialresolution satellite data. Remote Sens<strong>in</strong>g of Environment 114: 2431–2435.DeRose, R. Just<strong>in</strong> and James N. Long. <strong>2012.</strong> Factors <strong>in</strong>fluenc<strong>in</strong>g the spatial and temporaldynamics of Engelmann spruce mortality dur<strong>in</strong>g a spruce beetle outbreak on theMarkagunt Plateau, Utah. <strong>Forest</strong> Science 58(1): 1-14.Dobbert<strong>in</strong>, M, P. Mayer, T. Wohlgemuth, E. Feldmeyer-Christe, U. Graf, N.E.Zimmermann, and A. Rigl<strong>in</strong>g. 2005. The decl<strong>in</strong>e of P<strong>in</strong>us sylvestris L. forests <strong>in</strong>the swiss Rhone Valley-a result of drought stress? Phyton-Annales Rei Botanicae45: 153-156.Dobbert<strong>in</strong>, Matthias, Beat Wermel<strong>in</strong>ger, Christof Bigler, Matthias Bürgi, MatthiasCarron, Beat Forster, Urs Gimmi, and Andreas Rigl<strong>in</strong>g. 2007. L<strong>in</strong>k<strong>in</strong>g <strong>in</strong>creas<strong>in</strong>gdrought stress to Scots p<strong>in</strong>e mortality and bark beetle <strong>in</strong>festations. The ScientificWorld 7(S1): 231–239.Dukes, Jeffrey S., Jennifer Pontius, David Orwig, Jeffrey R. Garnas, Vikki L. Rodgers,Nicholas Brazee, Barry Cooke, Kathleen A. Theoharides, Erik E. Strange, RobenHarr<strong>in</strong>gton, Joan Ehrenfeld, Jessica Gurevitch, Manuel Lerdau, Krist<strong>in</strong>a St<strong>in</strong>son,Robert Wick, and Matthew Ayres. 2009. Responses of <strong>in</strong>sect pests, pathogens,


36and <strong>in</strong>vasive plant species to climate change <strong>in</strong> the forests of northeastern NorthAmerica: What can we predict? Canadian Journal of <strong>Forest</strong> Research 39: 231-248.Egan, Joel, Dave Fournier, Hugh Safford, J.McLean Sloughter, Tamre Cardoso, PatrickTra<strong>in</strong>or, and John Wenz. 2011. Assessment of a Jeffrey P<strong>in</strong>e Beetle Outbreak<strong>from</strong> 1991-1996 near Spooner <strong>Lake</strong> Junction, <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>. FHP Report#3311-09. U.S. Department of Agriculture, <strong>Forest</strong> Service, <strong>Forest</strong> HealthProtection, Sonora, CA. 24pp.Elliott-Fisk, D.L. 1996. In: <strong>Lake</strong> <strong>Tahoe</strong> case study. Addendum to the Sierra NevadaEcosystem Project; f<strong>in</strong>al report to Congress. University of California, Centers forWater and Wildland Resources. Wildland Resources Center Report No. 36.Ferrell, G.T., W. J. Otros<strong>in</strong>a, W.J., and C.J. Demars. 1994. Predict<strong>in</strong>g susceptibility ofwhite fir dur<strong>in</strong>g a drought associated outbreak of the fir engraver, Scolytusventralis, <strong>in</strong> California. Canadian Journal of <strong>Forest</strong> Research 24: 302–305.Ferrell, G.T. 1996. The <strong>in</strong>fluence of <strong>in</strong>sect pests and pathogens on Sierra <strong>Forest</strong>s. In:Sierra Nevada Ecosystem Project: F<strong>in</strong>al Report to Congress, vol. II, Assessmentsand Scientific Basis for Management Options. Univ. of California, Davis, WaterResources Center Report No. 37, pp. 1177–1192.Fettig, Christopher J., Christopher J. Fettig, Kier D. Klepzig, Ronald F. Bill<strong>in</strong>gs, A.Steven Munson, T. Evan Nebeker, Jose F. Negrón, and John T. Nowak. 2007. Theeffectiveness of vegetation management practices for prevention and control ofbark beetle <strong>in</strong>festations <strong>in</strong> coniferous forests of the western and southern UnitedStates. <strong>Forest</strong> Ecology and Management 238: 24–53.


37Fettig, Christopher J., Robert R. Borys, Stephen R. McKelvey, and Christopher P.Dabney. 2008. Blacks Mounta<strong>in</strong> Experimental <strong>Forest</strong>: bark beetle responses todifferences <strong>in</strong> forest structure and the application of prescribed fire <strong>in</strong> <strong>in</strong>teriorponderosa p<strong>in</strong>e. Canadian Journal of <strong>Forest</strong> Research 38: 924-935.Fettig, Christopher J, Christopher J. Hayes, Karen J. Jones, Stephen R. McKelvey, SylviaL. Mori and Sheri L. Smith. <strong>2012.</strong> Th<strong>in</strong>n<strong>in</strong>g Jeffrey p<strong>in</strong>e stands to reducesusceptibility to bark beetle <strong>in</strong>festations <strong>in</strong> California, U.S.A. Agricultural and<strong>Forest</strong> Entomology 14: 111–117.Floyd, M. Lisa, Michael Clifford, Neil S. Cobb, Dust<strong>in</strong> Hanna, Robert Delph, PauletteFord, and Dave Turner. 2009. Relationship of stand characteristics to drought<strong>in</strong>ducedmortality <strong>in</strong> three Southwestern piñon–juniper woodlands. EcologicalApplications 19(5): 1223–1230.Frankl<strong>in</strong>, Jerry F., H.H. Shugart, and M.E. Harmon. 1987. Tree death as an ecologicalprocess. Bioscience 27: 259–288.Frankl<strong>in</strong>, Jerry F., Thomas A. Spies, Robert <strong>Van</strong> Pelt, Andrew B. Carey, Dale A.Thornburgh, Dean Rae Berg, David B. L<strong>in</strong>denmayer, Mark E. Harmon, WilliamS. Keeton, David C. Shaw, Ken Bible and Jiquan Chen. 2002. Disturbances andstructural development of natural forest ecosystems with silviculturalimplications, us<strong>in</strong>g Douglas-fir forests as an example. <strong>Forest</strong> Ecology andManagement 155: 399–423.Galiano, L., J.Martínez-Vilalta, and F. Lloret. 2010. Drought-<strong>in</strong>duced multifactor decl<strong>in</strong>eof Scots p<strong>in</strong>e <strong>in</strong> the Pyrenees and potential vegetation change by the expansion ofco-occurr<strong>in</strong>g oak species. Ecosystems 13: 978–991.


38Ganey, Joseph L. and Scott C. Vojta. 2011. Tree mortality <strong>in</strong> drought-stressed mixedconiferand ponderosa p<strong>in</strong>e forests, Arizona, USA. <strong>Forest</strong> Ecology andManagement 261: 162–168.Gao, Bo-Cai. 1996. NDWI: A normalized difference water <strong>in</strong>dex for remote sens<strong>in</strong>g ofvegetation liquid water <strong>from</strong> space. Remote Sens<strong>in</strong>g of Environment 58: 257-266.Greenwood, David L. and Peter J. Weisberg. 2008. Density-dependent tree mortality <strong>in</strong>p<strong>in</strong>yon-juniper woodlands. <strong>Forest</strong> Ecology and Management 255(7): 2129–2137.Graham, R., A. Harvey, T. Ja<strong>in</strong>, and J. Tonn. 1999. Effects of th<strong>in</strong>n<strong>in</strong>g and similar standtreatments on fire behavior <strong>in</strong> western forests. USDA <strong>Forest</strong> Service, PacificNorthwest Research Station, General Technical Report PNW-GTR-463.Guarín, Alejandro and Alan H. Taylor. 2005. Drought triggered tree mortality <strong>in</strong> mixedconifer forests <strong>in</strong> Yosemite National Park, California, USA. <strong>Forest</strong> Ecology andManagement 218: 229–244.He, Fangliang and Richard P. Duncan. 2000. Density-dependent effects on tree survival<strong>in</strong> an old-growth Douglas fir forest. Journal of Ecology 88: 676-688.Hebertson, Elizabeth G. and Michael J. Jenk<strong>in</strong>s. 2008. Climate factors associated withhistoric spruce beetle (Coleoptera: Curculionidae) outbreaks <strong>in</strong> Utah andColorado. Environmental Entomology 37(2): 281-292.Herben, Tomas, Frantisek Krahulec, Vera Had<strong>in</strong>cova, Sylvie Pechackova, and RadkaWildova. 2003. Year-to-year variation <strong>in</strong> plant competition <strong>in</strong> a mounta<strong>in</strong>grassland. Journal of Ecology 91: 103-113.Herms, Daniel A. and William J. Mattson. 1992. The dilemma of plants: To grow ordefend. The Quarterly Review of Biology 67(3): 283-335.


39Huete, A.R., H.Q. Liu, K. Batchily, and W. vanLeeuwen. 1997. A comparison ofvegetation <strong>in</strong>dices over a global set of TM Images for EOS-MODIS. RemoteSens<strong>in</strong>g of Environment 59: 440-451.Hunt, E. Raymond Jr. and Barrett N. Rock. 1989. Detection of changes <strong>in</strong> leaf watercontent us<strong>in</strong>g near- and middle-<strong>in</strong>frared reflectances. Remote Sens<strong>in</strong>g ofEnvironment 30: 43-54.James, J.C., J. Grace, J. and S.P. Hoad. 1994. Growth and photosynthesis of P<strong>in</strong>ussylvestris at its altitud<strong>in</strong>al limit <strong>in</strong> Scotland. Journal of Ecology 82: 297-306.J<strong>in</strong>, Sum<strong>in</strong>g and Steven A. Sader. 2005. Comparison of time series tasseled cap wetnessand the normalized difference moisture <strong>in</strong>dex <strong>in</strong> detect<strong>in</strong>g forest disturbances.Remote Sens<strong>in</strong>g of Environment 94: 364–372.Kane, Jeffrey M. and Thomas E. Kolb. 2010. Importance of res<strong>in</strong> ducts <strong>in</strong> reduc<strong>in</strong>gponderosa p<strong>in</strong>e mortality <strong>from</strong> bark beetle attack. Oecologia 164: 601-609.Kenaley, Shawn, Robert Mathiasen, and James E. Harner. 2008. <strong>Mortality</strong> associatedwith a bark beetle outbreak <strong>in</strong> dwarf mistletoe-Infested Ponderosa P<strong>in</strong>e stands <strong>in</strong>Arizona. Western Journal of Applied <strong>Forest</strong>ry 23(2): 113-120.K<strong>in</strong>g, Gary and Langche Zeng. 2001. Logistic regression <strong>in</strong> rare events data. PoliticalAnalysis 9: 137-163.LAI 2000-Plant Canopy Analyzer Instruction Manual. LI-COR.ftp://ftp.licor.com/perm/env/LAI-2000/Manual/LAI-2000_Manual.pdfLev<strong>in</strong> , Simon A. 1992. The problem of pattern and scale <strong>in</strong> ecology: The Robert H.MacArthur award lecture. Ecology 73(6): 1943-1967.


40L<strong>in</strong>ares, Juan-Carlos Antonio Delgado-Huertas, J. Julio Camarero, José Mer<strong>in</strong>o and JoséA. Carreira. 2009. Competition and drought limit the response of water-useefficiency to ris<strong>in</strong>g atmospheric carbon dioxide <strong>in</strong> the Mediterranean fir Abiesp<strong>in</strong>sapo. Oecologia 161: 611–624.L<strong>in</strong>ares, Juan Carlos, Jesús Julio Camarero, and José A. Carreira. 2010. Competitionmodulates the adaptation capacity of forests to climatic stress: <strong>in</strong>sights <strong>from</strong>recent growth decl<strong>in</strong>e and death <strong>in</strong> relict stands of the Mediterranean fir Abiesp<strong>in</strong>sapo. Journal of Ecology 98: 592-603.L<strong>in</strong>es, Emily R., David A. Coomes, and Drew W. Purves. 2010. Influences of foreststructure, climate and species composition on tree mortality across the EasternU.S. PLoS ONE 5(10): e13212. doi:10.1371/journal.pone.0013212L<strong>in</strong>gua, Emanuele, Paolo Cherub<strong>in</strong>i, Renzo Motta, and Paola Nola. 2008. Spatialstructure along an altitud<strong>in</strong>al gradient <strong>in</strong> the Italian central Alps suggestscompetition and facilitation among coniferous species. Journal of VegetationScience. Published onl<strong>in</strong>e 13 March 2008.Lloyd, Andrea H. and Lisa J. Graumlich. 1997. Holocene dynamics of treel<strong>in</strong>e forests <strong>in</strong>the Sierra Nevada. Ecology 78(4): 1199-1210.Logan, Jesse A., William W. MacFarlane, and Louisa Wilcox. 2010. Whitebark p<strong>in</strong>evulnerability to climate-driven mounta<strong>in</strong> p<strong>in</strong>e beetle disturbance <strong>in</strong> the GreaterYellowstone Ecosystem. Ecological Applications 20(4): 895–902.Logan, Jesse A., Peter White, Barbara J. Bentz and James A. Powell. 1998. Modelanalysis of spatial patterns <strong>in</strong> mounta<strong>in</strong> p<strong>in</strong>e beetle outbreaks. TheoreticalPopulation Biology 53: 236-255.


41Loomis W E. 1932. Growth-differentiation balance vs carbohydrate-nitrogen ratio.Proceed<strong>in</strong>gs of the American Society for Horticultural Science 29: 240-245.Loomis W E. 1953. Growth and differentiation--an <strong>in</strong>troduction and summary. Pages 1-17 <strong>in</strong> Growth and Differentiation <strong>in</strong> Plants, edited by W E Loomis. Ames (IA):Iowa State College Press.MacQuarrie, Chris J.K., and Barry J. Cooke. 2011. Density-dependent populationdynamics of mounta<strong>in</strong> p<strong>in</strong>e beetle <strong>in</strong> th<strong>in</strong>ned and unth<strong>in</strong>ned stands. CanadianJournal of <strong>Forest</strong> Research 41: 1031–1046.Macomber, S.A., Woodcock, C.E., 1994. Mapp<strong>in</strong>g and monitor<strong>in</strong>g conifer mortalityus<strong>in</strong>g remote sens<strong>in</strong>g <strong>in</strong> the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>. Remote Sens<strong>in</strong>g of Environment50: 255–266.Maloney, Patricia E. and David M. Rizzo. 2002. Pathogens and <strong>in</strong>sects <strong>in</strong> a prist<strong>in</strong>e forestecosystem: the Sierra San Pedro Martir, Baja, Mexico. Canadian Journal of <strong>Forest</strong>Research 32: 448-457.Maloney, Patricia E., Detlev R. Vogler , Andrew J. Eckert, Camille E. Jensen, and DavidB. Neale. 2011. Population biology of sugar p<strong>in</strong>e (P<strong>in</strong>us lambertiana Dougl.) withreference to historical disturbances <strong>in</strong> the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>: Implications forrestoration. <strong>Forest</strong> Ecology and Management 262: 770–779.Manion, P.D., 1991. Tree Disease Concepts, 2nd ed. Prentice-Hall Inc., Upper SaddleRiver, NJ. 416 pp.Mayer, D.G. and D.G. Butler. 1993. Statistical validation. Ecological Model<strong>in</strong>g 68: 21–32.


42McDowell, Nate, William T. Pockman, Craig D. Allen, David D. Breshears, Neil Cobb,Thomas Kolb, Jennifer Plaut, John Sperry, Adam West, David G. Williams, andEnrico A. Yepez. 2008. Mechanisms of plant survival and mortality dur<strong>in</strong>gdrought: why do some plants survive while others succumb to drought? NewPhytologist 178: 719–739.Miao, ShiLi, Chris B. Zou, and David D. Breshears. 2009. Vegetation responses toextreme hydrological events: Sequence Matters. The American Naturalist 173 (1):113-118.Millar, Constance. I., and W. B. Woolfenden. 1999. The role of climate change <strong>in</strong><strong>in</strong>terpret<strong>in</strong>g historical variability. Ecological Applications 9: 1207–1216.Millar, Constance I., Robert D. Westfall, Diane L. Delany, John C. K<strong>in</strong>g, and LisaJ.Graumlich. 2004. Response of subalp<strong>in</strong>e conifers <strong>in</strong> the Sierra Nevada,California, U.S.A, to 20 th -Century warm<strong>in</strong>g and decadal climate variability.Arctic, Antarctic, and Alp<strong>in</strong>e Research 36(2): 181-200.Millar, Constance.I., Robert D. Westfall, and Diane L. Delany. 2007. Response of highelevationlimber p<strong>in</strong>e (P<strong>in</strong>us flexilis) to multiyear droughts and 20th-centurywarm<strong>in</strong>g, Sierra Nevada, California, USA. Canadian Journal of <strong>Forest</strong> Research37: 2508–2520.Millar, Constance I., Robert D. Westfall, Diane L. Delany, Matthew J. Bokach, Alan L.Fl<strong>in</strong>t, and Lorra<strong>in</strong>e E. Fl<strong>in</strong>t. <strong>2012.</strong> <strong>Forest</strong> mortality <strong>in</strong> high-elevation whitebarkp<strong>in</strong>e(P<strong>in</strong>us albicaulis) forests of eastern California, USA; <strong>in</strong>fluence ofenvironmental context, bark beetles, climatic water deficit, and warm<strong>in</strong>g.Canadian Journal of <strong>Forest</strong> Research 42: 749–765.


43Moritz, Max A., Paul F. Hessburg, and Nicholas A. Povak. 2010. Native Fire Regimesand Landscape Resilience. In: McKenzie, D., Miller, C., Falk, D.A. (Eds.). TheLandscape Ecology of Fire. Verlag, Spr<strong>in</strong>ger, pp. 51–86, Vol. 21.Negrón, José.F., Joel D. McMill<strong>in</strong>, John A. Anhold, and Dave Coulson. 2009. Barkbeetle-caused mortality <strong>in</strong> a drought affected ponderosa p<strong>in</strong>e landscape <strong>in</strong>Arizona, USA. <strong>Forest</strong> Ecology and Management 257: 1353–1362.Nelson, T., B. Boots, M.A. Wulder, T. Shore, L. Safranyik, and T. Ebata. 2006. Rat<strong>in</strong>gthe susceptibility of forests to mounta<strong>in</strong> p<strong>in</strong>e beetle <strong>in</strong>festations: the impact ofdata. Canadian Journal of <strong>Forest</strong> Research 36: 2815-2825.Nelson, T.A., B. Boots, M.A. Wulder, and A.L. Carroll. 2007. Environmentalcharacteristics of mounta<strong>in</strong> p<strong>in</strong>e beetle <strong>in</strong>festation hot spots. BC Journal ofEcosystems and Management 8(1): 91–108.North, Malcolm, Matthew Hurteau, Robert Fiegener, and Michael Barbour. 2005.Influence of fire and El Niño on tree recruitment varies by species <strong>in</strong> Sierranmixed conifer. <strong>Forest</strong> Science 51(3): 187-197.Parker, Albert J. 1986. Persistence of lodgepole p<strong>in</strong>e forests <strong>in</strong> the central Sierra Nevada.Ecology 67(6): 1560-1567.Pedersen, Brian S. 1998. The role of stress <strong>in</strong> the mortality of Midwestern oaks as<strong>in</strong>dicated by growth prior to death. Ecology 79(1): 79-93.Peet, Robert K. and Norman L. Christensen. 1987. Competition and tree death.BioScience 37(8): 586-595.Perry, David A., Paul F. Hessburg , Carl N. Sk<strong>in</strong>ner, Thomas A. Spies, Scott L. Stephens,Alan Henry Taylor, Jerry F. Frankl<strong>in</strong>, Brenda McComb, and Greg Riegel. 2011.


44The ecology of mixed severity fire regimes <strong>in</strong> Wash<strong>in</strong>gton, Oregon, and NorthernCalifornia. <strong>Forest</strong> Ecology and Management 262: 703–717.Powers, Jennifer Sarah, Phillip Soll<strong>in</strong>s, Mark E. Harmon and Julia A. Jones. 1999. Plantpest<strong>in</strong>teractions <strong>in</strong> time and space: A Douglas-fir bark beetle outbreak as a casestudy. Landscape Ecology 14: 105–120.Raffa, K.F. and A. A. Berryman. 1983. The Role of host plant resistance <strong>in</strong> thecolonization behavior and ecology of bark beetles (Coleoptera: Scolytidae).Ecological Monographs 53(1): 27-49.Raumann, Christian G. and Mary E. Cablk. 2008. Change <strong>in</strong> the forested and developedlandscape of the <strong>Lake</strong> <strong>Tahoe</strong> bas<strong>in</strong>, California and Nevada, USA, 1940-2002.<strong>Forest</strong> Ecology and Management 255: 3424-3439.Reid, M. L. and T. Robb. 1999. Death of vigorous trees benefits bark beetles. Oecologia120(4): 555-562.Re<strong>in</strong>eke, L. H. 1933. Perfect<strong>in</strong>g a stand-density <strong>in</strong>dex for even-aged forests. Journal ofAgriculture Research 46 : 627-638.Safford, Hugh D., David A. Schmidt, and Chris H. Carlson. 2009. Effects of fueltreatments on fire severity <strong>in</strong> an area of wildland–urban <strong>in</strong>terface, Angora Fire,<strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>, California. <strong>Forest</strong> Ecology and Management 258: 773–787.Sánchez-Martínez, Guillermo and Michael R. Wagner. 2002. Bark beetle communitystructure under four ponderosa p<strong>in</strong>e forest stand conditions <strong>in</strong> northern Arizona.<strong>Forest</strong> Ecology and Management 170: 145–160.


45Scholl, Andrew E. and Alan H. Taylor. 2006. Regeneration patterns <strong>in</strong> old-growth redfir–western white p<strong>in</strong>e forests <strong>in</strong> the northern Sierra Nevada, <strong>Lake</strong> <strong>Tahoe</strong>, USA.<strong>Forest</strong> Ecology and Management 235: 143-154.Schowalter, T.D. and G.M. Filip. 1993. Beetle-pathogen <strong>in</strong>teractions <strong>in</strong> conifer forests.Academic Press, San Diego.Schwilk, Dylan W., Eric E. Knapp , Scott M. Ferrenberg, Jon E. Keeley , Anthony C.Caprio. 2006. Tree mortality <strong>from</strong> fire and bark beetles follow<strong>in</strong>g early and lateseason prescribed fires <strong>in</strong> a Sierra Nevada mixed-conifer forest. <strong>Forest</strong> Ecologyand Management 232: 36–45.Simard, Mart<strong>in</strong>, Er<strong>in</strong>n N. Powell, Kenneth F. Raffa, and Monica G. Turner. <strong>2012.</strong> Whatexpla<strong>in</strong>s landscape patterns of tree mortality caused by bark beetle outbreaks <strong>in</strong>Greater Yellowstone? Global Ecology and Biogeography 21: 556–567.SNEP [Sierra Nevada Ecosystem Project]. 1996. <strong>Lake</strong> <strong>Tahoe</strong> Case Study. Pages 217–276<strong>in</strong> Sierra Nevada Ecosystem Project: F<strong>in</strong>al report to Congress, Addendum. Davis:University of California, Centers for Water and Wildland Resources, 1996.Stamp, Nancy. 2003. Out of the quagmire of plant defense hypotheses. The QuarterlyReview of Biology 78(1): 23-55.State of Nevada. Division of Water Resources. Truckee River Chronology:http://water.nv.gov/mapp<strong>in</strong>g/chronologies/truckee/part1.cfmSuarez, Maria Laura, Luciana Ghermandi, and Thomas Kitzberger. 2004. Factorspredispos<strong>in</strong>g episodic drought-<strong>in</strong>duced tree mortality <strong>in</strong> Nothofagus– site, climaticsensitivity andgrowth trends. Journal of Ecology 92: 954–966.


46Taylor, Alan H. 2004. Identify<strong>in</strong>g forest reference conditions on early cut-over lands,<strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>, USA. Ecological Applications 14(6): 1903-1920.Tranquill<strong>in</strong>i, W. 1979. Physiological Ecology of the alp<strong>in</strong>e timberl<strong>in</strong>e: tree existence athigh altitudes with special reference to the European Alps. Ecological Studies 31.Berl<strong>in</strong> and New York:S pr<strong>in</strong>ger-Verlag 137 pp.Urban, Dean, Carol Miller, Patrick N. Halp<strong>in</strong>, and Nathan L. Stephenson. 2000. <strong>Forest</strong>gradient response <strong>in</strong> Sierran landscapes: the physical template. LandscapeEcology 15: 603–620.USDA <strong>Forest</strong> Service. <strong>Forest</strong> Health Monitor<strong>in</strong>g. Aerial Detection Monitor<strong>in</strong>g. PacificSouthwest Region.1979-2011 Insect Damage Polygons.http://www.fs.usda.gov/detail/r5/forest-grasslandhealth/?cid=fsbdev3_046696USDA <strong>Forest</strong> Service. 1981. CALVEG: A Classification of California Vegetation.Pacific Southwest Region, Regional Ecology Group, San Francisco CA. 168 pp.2009 Update.http://www.fs.usda.gov/detail/r5/landmanagement/resourcemanagement/?cid=stelprdb5365219http://www.fs.usda.gov/detail/r5/landmanagement/resourcemanagement/?cid=stelprdb5347175USDA <strong>Forest</strong> Service. <strong>Forest</strong> Health Protection. South Sierra Shared Service Area.Report no. SS09-12. Evaluation of Jeffrey p<strong>in</strong>e mortality near Fallen Leaf <strong>Lake</strong>Campground. May 29 th , 2009.USDA <strong>Forest</strong> Service. <strong>Forest</strong> Health Protection. Pacific Southwest Region. Report No.SS09-03. File No. 3420. March 2, 2009. Current situation of mounta<strong>in</strong> p<strong>in</strong>e beetleat High Meadows, <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong> Management Unit.


47van Mantgem, P.J., Stephenson, N.L., 2007. Apparent climatically <strong>in</strong>duced <strong>in</strong>crease oftree mortality rates <strong>in</strong> a temperate forest. Ecology Letters 10: 909–916.van Mantgem, P.J., N.L. Stephenson, J.C.Byrne, L.D. Daniels, J.F. Frankl<strong>in</strong>, P.Z. Fulé,M.E. Harmon, A.J. Larson, J.M. Smith, A.H. Taylor, T.T. Veblen. 2009.Widespread <strong>in</strong>crease of tree mortality rates <strong>in</strong> the western United States. Science323: 521–524.Vogelmann, J. E. 1990. Comparison between two vegetation <strong>in</strong>dices for measur<strong>in</strong>gdifferent types of forest damage <strong>in</strong> the northeastern United States. InternationalJournal of Remote Sens<strong>in</strong>g 11: 2281−2297.Vogelmann, James E., Brian Tolk, and Zhiliang Zhu. 2009. Monitor<strong>in</strong>g forest changes <strong>in</strong>the southwestern United States us<strong>in</strong>g multitemporal Landsat data. RemoteSens<strong>in</strong>g of Environment 113: 1739–1748.Vygodskaya, N.N, E.D. Schulze, N. M. Tchebakova. O. Karpachevski, D. Kozlov, K. N.Sidorov, M. I. Panfyorov, M. A. Abrazko, E. S. Shaposhnikov, O. N. Solnzeva, T.Y. M<strong>in</strong>aeva, A. S. Jeltuch<strong>in</strong>, C. Wirth and A. V. Pugachevskii. 2002. Climaticcontrol of stand th<strong>in</strong>n<strong>in</strong>g <strong>in</strong> unmanaged spruce forests of the southern taiga <strong>in</strong>European Russia. Tellus 54: 443–461.Wallner, W. E. 1987. Factors affect<strong>in</strong>g <strong>in</strong>sect population dynamics: Differences betweenoutbreak and non-outbreak species. Annual Review of Entomology 32: 317-340.Wang, T., A. Hamann, D.Spittlehouse, and T.N.Murdock. <strong>2012.</strong> ClimateWNA - Highresolutionspatial climate data for western North America. Journal of AppliedMeteorology and Climatology 61: 16-29.


48Westoby, Mark. 1984. The self-th<strong>in</strong>n<strong>in</strong>g rule. Advances <strong>in</strong> Ecological Research. 14: 167-225.White, Joseph D., Steven W. Runn<strong>in</strong>g, Ramakrishna Nemani, Robert E. Keane, andKev<strong>in</strong> C. Ryan. 1997. Measurement and remote sens<strong>in</strong>g of LAI <strong>in</strong> RockyMounta<strong>in</strong> montane ecosystems. Canadian Journal of <strong>Forest</strong> Research 27: 1714—1727.Wilson, E. H., and Steven A. Sader. 2002. Detection of forest harvest type us<strong>in</strong>g multipledates of Landsat TM imagery. Remote Sens<strong>in</strong>g of Environment 80: 385– 396.Wulder , Michael A., Caren C. Dymond, Joanne C. White, Donald G. Leckie, and AllanL. Carroll. 2006. Survey<strong>in</strong>g mounta<strong>in</strong> p<strong>in</strong>e beetle damage of forests: A review ofremote sens<strong>in</strong>g opportunities. <strong>Forest</strong> Ecology and Management 221: 27–41.Yoda, K., T. Kira, H. Ogawa, and H. Hozumi. 1963. Self th<strong>in</strong>n<strong>in</strong>g <strong>in</strong> overcrowded purestands under natural and cultivated conditions. J. Inst. Polytech. Osaka City Univ.,Ser.D 14: 107-129.


49Table 1-1. Description of LTB forest types def<strong>in</strong>ed us<strong>in</strong>g the USFS CalVeg vegetationclassification (USDA <strong>Forest</strong> Service. 1981. CALVEG: A Classification of CaliforniaVegetation).<strong>Forest</strong> Type Species Composition Elevation Number of Number ofRange(m) Ecoregions PixelsJeffrey P<strong>in</strong>eAlliance (JP)Lodgepole (LP)Mixed Conifer –Fir Alliance(MF)Red Fir Alliance(RF)White FirAlliance (WF)Jeffrey P<strong>in</strong>e (P<strong>in</strong>us Jeffreyi Balf.),Ponderosa P<strong>in</strong>e (P<strong>in</strong>us ponderosae Dougl.ex. Loud.), White Fir (Abies concolor A.Murray), Washoe P<strong>in</strong>e (P<strong>in</strong>us washoensis)1890-2746 5 32,024Lodgepole p<strong>in</strong>e (P<strong>in</strong>us contorta var. 1897-3031 1 25,449murrayana Grev.& Balf.)White Fir, Jeffrey P<strong>in</strong>e, Lodgepole P<strong>in</strong>e, 1889-2717 5 142,878Sugar P<strong>in</strong>e (P<strong>in</strong>us lambertiana A.Murray), Incense Cedar (Calocedrusdecurrens)Lower Elevations: White FirUpper Elevations: Red Fir (Abiesmagnifica A. Murr.)Red Fir, White Fir, Western White P<strong>in</strong>e 1987-2878 5 82,521(P<strong>in</strong>us monticola Dougl.), Lodgepole P<strong>in</strong>e,isolated Mounta<strong>in</strong> Hemlock (Tsugamertensiana)Pure white fir <strong>in</strong> cool, moist, shady 1900-2638 1 14,522environments


50TABLE 1-2. Host-specific native bark beetles of LTB forests and impact of successfulattack on canopy foliage.Mounta<strong>in</strong> P<strong>in</strong>eBeetle(Dendroctonusponderosae)Common Bark Beetles of Sierra Nevada <strong>Forest</strong>sCanopy FadeSpectral SignatureCanopy Fade TemporalSignatureFlightSeasonGreen –Yellow- Generally 8-10June toReddish/Brown- months after attack. In AugustBrownunusually dry years,Progresses upward foliage fade may beg<strong>in</strong>or downward with<strong>in</strong> a fewthroughout crown weeksHostLP,PP,SP,WWPNo. Trees KilledWestern P<strong>in</strong>eBeetle(Dendroctonusbrevicomis)Pale green –Yellow-Red-Reddish/BrownEarly Season Attack:yellow to red <strong>in</strong> sameseason,Late Season Attack: yellowto red by Spr<strong>in</strong>g offollow<strong>in</strong>g seasonPP Low population #:S<strong>in</strong>gle tree, smallgroupsHigh pop. #:groups of 10-20treesRedTurpent<strong>in</strong>eBeetle(Dendroctonusvalens)Jeffrey P<strong>in</strong>eBeetle(Dendroctonusjeffreyi)Green –Yellow-Reddish/Brown-BrownFades evenlythroughout crownYellow to red <strong>in</strong> yearfollow<strong>in</strong>g <strong>in</strong>festationSpr<strong>in</strong>g toearly FallJune toOctoberJP,LP,PP, SPJPTree mortalitytends to occur <strong>in</strong>groups,rang<strong>in</strong>g <strong>from</strong> 10 to50 to severalhundredFir Engraver(Scolytusventralis)P<strong>in</strong>e Engraver(Ips p<strong>in</strong>i)CaliforniaFive-Sp<strong>in</strong>nedIps (Ipsparaconfusus)Yellow to redGreen –Lime green- Yellow-Reddish/Brown-BrownGreen –Lime green- Yellow-Reddish/Brown-BrownYellow to red <strong>in</strong> yearfollow<strong>in</strong>g <strong>in</strong>festation.USFS: Can kill upperportion of crown or entiretree <strong>in</strong> a s<strong>in</strong>gle summer.Beg<strong>in</strong>s fade with<strong>in</strong> fewmonths after attack.Depend<strong>in</strong>g on tree speciesand weather, may fade bylate summer/early fall orSpr<strong>in</strong>g of follow<strong>in</strong>g season.June toSeptemberRF,WFJP,LP, PPJP,LP,PP, SPUsually


51TABLE 1-3a,b. Validation of remote sens<strong>in</strong>g mortality classification with overallclassification accuracy (oca) and kappa statistics for two values of dNDWI,correspond<strong>in</strong>g to a 5% and 10% loss of green canopy per pixel. The kappa statistic is ameasure of agreement between predicted and actual values <strong>in</strong> a category.a. Classification matrix of healthy v mortality pixels classified at the >=5% loss of greencanopy threshold (dNDWI= -0.053)Predicted (5%)Actual(5%)healthymortalityhealthy mortality N458(91.78%) 41 (8.22%) 49939 238(14.08%) (85.92%) 277oca=89.69%, kappa=.78b. Classification matrix of healthy v mortality pixels classified at the >=10% loss of greencanopy threshold (dNDWI= -0.106)Predicted (10%)healthymortalityNActual(10%)healthy 498(99.8%) 1 (.2%) 499mortality 139 (50.18%)138(49.82%) 277oca=81.69%, kappa=.56


52TABLE 1-4. Annual mortality summary statistics by LTB forest type and climatic period(d1: 1987-1995, d2: 1999-2006, w1: 1985-1987, w2: 1995-1999, w3: 2006-2010).<strong>Forest</strong> TypeMean SD M<strong>in</strong> Max nJP 12.73 5.00 2.65 21.13 25LP 15.88 9.60 5.56 44.02 25MF 12.61 6.19 1.92 24.41 25RF 17.39 9.41 3.98 42.88 25WF 12.74 7.73 2.52 40.96 25Climatic Periodd1 17.60 7.53 6.64 42.88 40d2 10.71 5.65 2.01 30.87 35w1 18.87 13.67 5.25 44.02 10w2 14.87 6.64 6.60 32.52 20w3 10.96 5.85 1.92 25.13 20


53TABLE 1-5. Summary of odds-ratios for the stock<strong>in</strong>g-level predictor variable <strong>from</strong>annual models of mortality by forest type over the 25-year time series (SD=StandardDeviation). Odds ratios are derived <strong>from</strong> annual mortality models <strong>in</strong> each forest type andshow the change <strong>in</strong> the probability of mortality for every 0.2 <strong>in</strong>crease <strong>in</strong> stock<strong>in</strong>g level.<strong>Forest</strong> Type Mean SD M<strong>in</strong> Max nJP 1.16 0.28 0.92 2.23 25LP 1.10 0.23 0.64 1.77 25MF 1.06 0.36 0.76 2.62 25RF 1.05 0.23 0.71 1.92 25WF 1.03 0.42 0.60 2.70 25


54TABLE 1-6. Summary of odds-ratios for the stock<strong>in</strong>g-level predictor variable <strong>from</strong>annual models of mortality <strong>in</strong> each climate period occurr<strong>in</strong>g <strong>in</strong> the 25-year time series(SD=Standard Deviation).Climatic Mean SD M<strong>in</strong> Max nPeriodd1 1.20 0.46 0.60 2.70 40d2 1.08 0.19 0.66 1.53 35w1 0.89 0.11 0.71 1.00 10w2 0.98 0.21 0.64 1.51 20w3 1.03 0.19 0.67 1.65 20


FIGURE 1-1. Location of <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong> (LTB) study area.55


56FIGURE 1-2. <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong> (LTB) forest types per the USFS CalVeg Classification(USDA <strong>Forest</strong> Service. 1981. CALVEG: A Classification of California Vegetation ).


57a.b.FIGURE 1-3a, b. LAI Validation. Relationship between field-measured leaf area <strong>in</strong>dex(LAI) measured with LICOR LAI-2000 Plant Canopy Analyzer <strong>in</strong> 30 plots <strong>in</strong> the LTBand LAI modeled <strong>from</strong> normalized difference vegetation <strong>in</strong>dex (NDVI), derived <strong>from</strong>remote sens<strong>in</strong>g imagery (a). Model<strong>in</strong>g efficiency of observed vs. modeled LAI. Model<strong>in</strong>gefficiency measures agreement between observed and simulated values aga<strong>in</strong>st the l<strong>in</strong>e ofperfect fit (blue) rather than the regression l<strong>in</strong>e (b, Mayer and Butler 1993).


58FIGURE 1-4. Values of the s<strong>in</strong>gle year normalized difference vegetation <strong>in</strong>dex (NDWI)across 254, 30m x 30m pixels digitized <strong>in</strong> areas rang<strong>in</strong>g <strong>from</strong> severe and extensivemortality across the pixel (High) to pixels without evident canopy damage (None).Results of the ANOVA with follow-up Bonferroni comparison are significant, validat<strong>in</strong>gthe sensitivity of the s<strong>in</strong>gle-year NDWI to LTB mortality (F=33.27, p < 0.006).


60b1.b2.d. c.FIGURE 1-6. Pictorial sequence of remote sens<strong>in</strong>g process<strong>in</strong>g used to derive annualmortality maps. Level 1 terra<strong>in</strong> corrected Landsat TM 5 satellite image is acquired,resized to study area, and normalized to reference imagery (a).Year one (ex.2009) andyear two (ex. 2010) normalized images are transformed <strong>in</strong>to s<strong>in</strong>gle-year NDWI maps (b1,2009: b2, 2010). Cont<strong>in</strong>uous differenced NDWI (dNDWI) map (c) is derived bysubtract<strong>in</strong>g 2009 NDWI (Year 1) <strong>from</strong> 2010 NDWI (Year 2). Map of cont<strong>in</strong>uous dNDWIis classified <strong>in</strong>to maps of healthy and canopy dieback pixels us<strong>in</strong>g validated thresholds(dNDWI


Average W<strong>in</strong>ter M<strong>in</strong>imumTemperature ( o C)Snow: Depature <strong>from</strong> 30 Year Norm610-2-41980 1985 1990 1995 2000 2005w1 d1 w2 d2 w31.51.41.31.2SnowDeficit-6-8-10-121.110.90.80.70.6AverageW<strong>in</strong>terM<strong>in</strong>imumTemp.FIGURE 1-7. Derivation of five climatic periods throughout the 1985-2009 study period,with 1980-1984 for reference. S<strong>in</strong>gle-year snow deficits are derived <strong>from</strong> the currentyear/30-year norm precipitation as snow variable <strong>from</strong> the ClimateWNA dataset (Wanget al. 2012). Four-year lagged deficits (average of current year deficit + 3 previous years)replicate historical accounts of dry and wet periods <strong>in</strong> the LTB. Average w<strong>in</strong>ter m<strong>in</strong>imumtemperature is current-year temperature. Climatic periods: w1=1985-1987 (August 1985-July 1986, August1986-July 1987, 2-year duration); d1=1987-1995, 8-year duration:w2= 1995-1999, 4-year duration; d2= 1999-2006, 7-year duration; w3: 2006-2010, 4-year duration.


62FIGURE 1-8. <strong>Mortality</strong> levels and pattern by forest type over the 25-year time series(with smooth local estimator (lowess) curve).


63FIGURE 1-9. <strong>Mortality</strong> levels by forest type and climatic period (w1: 1985-1987, d1:1987-1995, w2: 1995-1999, d2: 1999-2006, w3: 2006-2010).


64FIGURE 1-10. Direction and magnitude of density dependent mortality as odds ratios forevery absolute 20% <strong>in</strong>crease <strong>in</strong> stock<strong>in</strong>g level by forest type over the 25-year time<strong>in</strong>terval. Confidence <strong>in</strong>tervals at the 95 th percentile are shown. Odds ratios symbolized by“x” show years <strong>in</strong> which stock<strong>in</strong>g variable was not <strong>in</strong>cluded <strong>in</strong> f<strong>in</strong>al model, due tooutcome of likelihood ratio test. Odds ratios symbolized by “0” show years <strong>in</strong> which thestock<strong>in</strong>g variable was <strong>in</strong>cluded <strong>in</strong> f<strong>in</strong>al model.


65FIGURE 1-11. Box plot of odds ratios of density-dependent mortality across 25 annualmodels derived for each forest type. Odds ratios illustrate the change <strong>in</strong> the probability ofmortality for every 0.2 (20%) <strong>in</strong>crease <strong>in</strong> stock<strong>in</strong>g level.


66FIGURE 1-12. Number of models where; 1) Stock<strong>in</strong>g level alone or <strong>in</strong> comb<strong>in</strong>ation withenvironmental variables appeared <strong>in</strong> f<strong>in</strong>al, most parsimonious model, 2) Environmentalvariables only appeared <strong>in</strong> f<strong>in</strong>al, most parsimonious model or, 3) <strong>Mortality</strong> was notexpla<strong>in</strong>ed by stand structure or environmental variables.


Percent of 25-Year Time Period671009080706050403020100JP LP MF RF WF<strong>Forest</strong> TypePositiveDDMNoAssociationNegativeDDMFIGURE 1-13. Type of density dependent mortality exhibited over percent of 25-yeartime series by direction. Positive DDM: <strong>in</strong>creased density <strong>in</strong>creases probability ofmortality; Negative DDM: decreased density <strong>in</strong>creases risk of mortality, and NoAssociation: no association between density and mortality.


Ratio: Positive DDM:Negative DDM6811109876543210JP LP MF RF WF<strong>Forest</strong>TypeFIGURE 1-14. Ratio of positive to negative density-dependent mortality over five foresttypes. Positive density-dependent mortality <strong>in</strong>dicates an <strong>in</strong>crease <strong>in</strong> the probability ofmortality with <strong>in</strong>creased density. Negative density-dependent mortality <strong>in</strong>dicates adecrease <strong>in</strong> the probability of mortality with <strong>in</strong>creased density.


Percent of 25-Year Time Period691009080706050403020100d1 d2 w1 w2 w3Climatic PeriodPositive DDMNoAssociationNegative DDMFIGURE 1-15. Type of density dependent mortality exhibited over percent of eachclimatic period <strong>in</strong> all five climatic periods by direction. Positive DDM: <strong>in</strong>creased density<strong>in</strong>creases probability of mortality; Negative DDM: decreased density <strong>in</strong>creases risk ofmortality, and No Association: no association between density and mortality.


70FIGURE 1-16. Average odds ratio for every 0.2 (20%) <strong>in</strong>crease <strong>in</strong> stock<strong>in</strong>g level byclimatic period and forest type. Error bars reflect standard errors.


71FIGURE 1-17. Relative frequency of negative (N) or positive (P) density-dependentmortality found <strong>in</strong> each forest type <strong>in</strong> each of four climatic periods (d1, d2, w2, w3).


72APPENDIX A. 95 th Percentile of Leaf Area Index ( LAI) values <strong>in</strong> 17 forest groups (Five <strong>Forest</strong> Types (LP, MF, JP, RF, WF)with MF, JP, and RF forests further separated by five ecoregions (EH, EJ, EK, EL, ET) per the 2009 update to the USFS CalVegvegetation classification. Year: Value of LAI at 95 th percentile by year.<strong>Forest</strong>Type EcoRegion Group 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997LP LP LP 3.31 3.42 3.15 3.50 3.44 3.48 3.50 3.25 3.46 3.07 4.16 3.19 3.16MF EH MFEH 6.30 6.32 5.93 5.81 5.29 6.14 5.22 5.21 4.96 3.89 5.35 4.63 4.88MF EJ MFEJ 5.04 5.02 4.63 4.66 4.41 4.83 4.49 4.52 4.36 3.51 4.82 4.36 4.47MF EK MFEK 4.06 4.26 4.06 4.38 4.11 4.08 4.04 3.79 4.14 3.43 4.56 3.97 4.03MF El MFEl 5.53 5.68 5.37 5.46 5.03 5.43 4.61 4.55 4.64 3.87 5.06 4.77 4.81MF ET MFET 4.06 4.17 3.91 3.98 3.76 3.70 3.40 3.23 3.13 2.97 3.55 3.37 3.49JP EH PEH 4.84 5.00 4.59 4.39 4.05 4.34 3.95 4.03 3.90 4.03 4.39 3.65 3.98JP EJ PEJ 4.25 4.58 4.18 4.48 4.11 4.41 4.54 4.12 4.55 3.35 5.15 3.96 4.11JP EK PEK 4.53 4.79 4.46 4.71 4.40 4.18 4.04 3.89 4.14 3.67 4.19 4.16 4.21JP EL PEL 4.64 4.59 4.51 4.46 4.18 4.08 3.91 3.83 3.90 3.46 4.19 3.86 3.88JP ET PET 3.69 3.62 3.33 3.21 3.31 3.23 2.98 2.86 2.69 2.87 2.97 3.03 2.90RF EH RFEH 3.52 3.48 3.32 3.32 3.53 3.66 3.61 3.51 3.49 2.97 3.83 3.25 3.33RF EJ RFEJ 4.16 3.88 3.76 4.10 4.06 4.28 4.22 4.29 3.93 3.37 4.39 4.08 3.95RF EK RFEK 3.32 3.23 2.99 3.38 3.36 3.35 3.38 3.24 3.29 3.08 3.66 3.22 3.16RF EL RFEL 4.34 4.26 3.97 4.27 4.28 4.38 3.95 3.67 3.85 3.35 4.58 3.89 3.90RF ET RFET 3.52 3.55 3.15 3.38 3.42 3.26 3.19 3.08 2.84 2.94 3.28 3.04 3.01WF WF WF 5.49 5.46 5.14 5.27 4.99 5.60 5.21 5.01 5.08 3.96 5.47 4.98 5.00


<strong>Forest</strong>Type EcoRegion Group 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010LP LP LP 3.12 3.33 3.21 3.30 3.65 3.46 3.48 3.12 3.40 3.23 3.48 3.37 3.56MF EH MFEH 4.77 5.16 5.13 4.95 5.21 5.20 5.32 4.96 5.11 4.92 5.03 4.74 5.32MF EJ MFEJ 4.32 4.71 4.50 4.35 4.71 4.58 4.63 4.49 4.70 4.46 4.28 4.33 4.79MF EK MFEK 3.93 4.16 4.20 4.27 4.37 4.20 4.39 3.90 4.19 4.04 4.35 4.08 4.38MF El MFEl 4.60 4.98 4.78 4.72 4.85 4.88 5.14 4.60 4.84 4.46 4.69 4.55 4.96MF ET MFET 3.45 3.60 3.48 3.55 3.54 3.49 3.65 3.42 3.68 3.52 3.59 3.46 3.68JP EH PEH 4.01 4.13 4.14 4.04 4.30 4.16 4.25 3.83 4.05 4.14 4.21 3.87 4.28JP EJ PEJ 4.11 4.50 4.22 4.46 4.67 4.67 5.06 4.46 4.65 4.32 4.86 4.59 4.56JP EK PEK 4.11 4.41 4.30 4.55 4.50 4.22 4.66 4.00 4.47 4.32 4.46 4.16 4.38JP EL PEL 3.87 4.05 3.98 3.94 3.94 3.98 4.02 3.91 4.07 4.06 3.98 3.99 4.19JP ET PET 3.03 3.05 2.82 2.90 2.82 2.89 2.91 2.96 3.09 2.97 2.91 2.79 2.99RF EH RFEH 3.25 3.52 3.31 3.30 3.54 3.47 3.47 3.42 3.42 3.28 3.53 3.35 3.61RF EJ RFEJ 3.86 4.29 4.05 3.99 4.30 4.08 4.26 4.24 4.30 4.32 4.35 4.16 4.57RF EK RFEK 3.12 3.33 3.18 3.30 3.33 3.30 3.32 3.22 3.29 3.11 3.28 3.22 3.36RF EL RFEL 3.73 4.13 3.93 3.93 4.03 4.06 4.25 4.00 3.95 3.76 3.88 3.84 4.15RF ET RFET 3.11 3.22 3.07 3.01 3.08 3.02 3.15 3.10 3.18 2.99 3.09 2.98 3.17WF WF WF 4.88 5.50 4.99 4.98 5.21 5.25 5.35 5.18 5.44 4.91 4.86 4.59 5.2973


198019821984198619881990199219941996199820002002200420062008Departure <strong>from</strong> 30-yr Norm (as ratio)74APPENDIX B. Relationships among modeled snow deficits, Snow Water Equivalent (SWE)<strong>from</strong> California Cooperative Snow Surveys, and Palmer Drought Severity Index (PDSI) over thestudy period. Appendix C.1: Snow deficit is 4-year averaged snow deficit (Current Year/30 YearNorm (1980-2009)) derived <strong>from</strong> the ClimateWNA dataset. Appendix C.2 Monthly April PDSIfor California Region 3 compared aga<strong>in</strong>st Snow Survey SWE and Snow Deficits.1.61.51.41.31.21.110.90.80.70.6April SnowWaterEquivalent: 4-Year AverageSnow Deficit: 4-Year AverageAppendix C.1


198019821984198619881990199219941996199820002002200420062008Snow Departure (Current Year/30-Year Norm)PDSI751.7101.51.31.10.90.70.586420-2-4-6April PDSI4-YearAveragedSnow Deficit4-Yr AverageSWE: CAApril 1 SnowSurveysAppendix C.2


76APPENDIX C. F<strong>in</strong>al models <strong>in</strong> five forest types <strong>in</strong> 25-year study period. LRT =Likelihood Ratio Test<strong>Forest</strong>Type Year Model LRT ModelTypePr(>Chisq)JP 1985 <strong>Mortality</strong>~Group+sr+stk 2.98E-03 environment + stkLP 1985 <strong>Mortality</strong>~elevation+sr 3.01E-01 environmentMF 1985 <strong>Mortality</strong>~Group+sr+stk


77<strong>Forest</strong>Type Year Model LRT ModelTypePr(>Chisq)LP 1992 <strong>Mortality</strong>~1 7.45E-01 nullMF 1992 <strong>Mortality</strong>~Group+elevation+sr+stk 1.47E-02 environment + stkRF 1992 <strong>Mortality</strong>~Group+sr+stk 1.04E-04 environment + stkWF 1992 <strong>Mortality</strong>~sr 3.24E-01 environmentJP 1993 <strong>Mortality</strong>~Group+sr+elevation+stk


78<strong>Forest</strong>Type Year Model LRT ModelTypePr(>Chisq)LP 2000 <strong>Mortality</strong>~sr+stk 3.23E-05 environment + stkMF 2000 <strong>Mortality</strong>~Group+elevation+sr+stk 3.93E-03 environment + stkRF 2000 <strong>Mortality</strong>~Group+elevation+stk 6.64E-11 environment + stkWF 2000 <strong>Mortality</strong>~elevation+sr 1.69E-01 environmentJP 2001 <strong>Mortality</strong>~Group+stk 7.82E-05 environment + stkLP 2001 <strong>Mortality</strong>~stk 4.85E-03 stkMF 2001 <strong>Mortality</strong>~Group+elevation+sr 1.68E-01 environmentRF 2001 <strong>Mortality</strong>~Group+sr+stk 6.23E-08 environment + stkWF 2001 <strong>Mortality</strong>~sr 2.47E-01 environmentJP 2002 <strong>Mortality</strong>~Group+elevation+sr 6.37E-01 environmentLP 2002 <strong>Mortality</strong>~elevation+sr


79<strong>Forest</strong>Type Year Model LRT ModelTypePr(>Chisq)WF 2007 <strong>Mortality</strong>~elevation+sr+stk 1.62E-03 environment + stkJP 2008 <strong>Mortality</strong>~Group+elevation 5.41E-01 environmentLP 2008 <strong>Mortality</strong>~elevation+stk 9.86E-03 environment + stkMF 2008 <strong>Mortality</strong>~Group+sr 5.34E-02 environmentRF 2008 <strong>Mortality</strong>~Group+stk 1.65E-03 environment + stkWF 2008 <strong>Mortality</strong>~elevation 6.87E-01 environmentJP 2009 <strong>Mortality</strong>~Group+elevation+stk 1.40E-02 environment + stkLP 2009 <strong>Mortality</strong>~sr 7.34E-01 environmentMF 2009 <strong>Mortality</strong>~Group+sr+stk 2.13E-02 environment + stkRF 2009 <strong>Mortality</strong>~Group+sr+stk 7.67E-04 environment + stkWF 2009 <strong>Mortality</strong>~sr+stk 1.96E-02 environment + stk


80Chapter 2 - Spatial and Temporal Patterns of <strong>Forest</strong> <strong>Mortality</strong> <strong>in</strong> <strong>Lake</strong> <strong>Tahoe</strong>Bas<strong>in</strong>, USA: Influence of Climate and Environment.Krist<strong>in</strong> Jane <strong>Van</strong> <strong>Gunst</strong> a , Peter J. Weisberg a,b .a Department of Natural Resources and Environmental Science, University of NevadaReno, 1664 N. Virg<strong>in</strong>ia Street, Reno, NV, 89557.b Program <strong>in</strong> Ecology, Evolution and Conservation Biology, University of Nevada Reno,Mail Stop 314, Reno, NV, 89557.cCorrespond<strong>in</strong>g Author: Phone: 775.784.4020, Fax: 775.784.4583 Email:kvangunst@cabnr.unr.eduEmail. P. Weisberg: pweisberg@cabnr.unr.edu.


81ABSTRACTGlobal <strong>in</strong>creases <strong>in</strong> forest mortality have been widely attributed to chang<strong>in</strong>g climate andaltered disturbance regimes, yet there is limited understand<strong>in</strong>g of how forest mortalitytrends and spatial patterns vary with forest type, climate and topographic variability.Us<strong>in</strong>g the Landsat TM archival database, we <strong>in</strong>vestigated how patterns of forest mortalitychanged annually over five forest types and five climatic periods <strong>from</strong> 1985-2010 <strong>in</strong> theconiferous forests of the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>, <strong>in</strong> the central Sierra Nevada. In addition, wequantified the relationships between mortality and environmental gradients def<strong>in</strong>ed byelevation and solar radiation. Our results show that mortality levels and patterns aredissimilar across forest types and climatic periods, suggest<strong>in</strong>g that <strong>in</strong>creases <strong>in</strong> forestmortality are driven more by unique sequences of extreme events rather than variability<strong>in</strong> normative conditions. Throughout the time series for almost all forests, mortality riskwas greater on north-fac<strong>in</strong>g slopes. Relationships between mortality and elevation arepredictable for lower-elevation forests, with decreases <strong>in</strong> elevation consistently associatedwith greater mortality risk. For middle- and upper-elevation forest types, <strong>in</strong>creasedmortality risk is associated with <strong>in</strong>creases <strong>in</strong> elevation dur<strong>in</strong>g wet periods and decreases<strong>in</strong> elevation dur<strong>in</strong>g dry periods. Results of our study corroborate those of other studies <strong>in</strong>xeric forests that have found <strong>in</strong>creased mortality risk on north-fac<strong>in</strong>g slopes and withdecreas<strong>in</strong>g elevation dur<strong>in</strong>g droughty periods. In the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>, widespreadforest mortality has occurred dur<strong>in</strong>g both wet and dry periods but with differentcharacteristic patterns.


83mounta<strong>in</strong>ous regions of snow-dom<strong>in</strong>ated hydrology, lower elevation forests are subject toearlier draw-down of snowpack runoff <strong>in</strong>puts to soil moisture, experienc<strong>in</strong>g a seasonaldrought last<strong>in</strong>g <strong>from</strong> mid-Summer to early-Fall depend<strong>in</strong>g on previous w<strong>in</strong>ter snowpack.Increased droughty conditions have been correlated with <strong>in</strong>creased mortality, likely dueto moisture stress that predisposes trees to successful beetle attack (Allen et al.2010).Studies of both short-term and long-term <strong>in</strong>creases <strong>in</strong> coniferous forest mortalityassociated with drought conditions have found <strong>in</strong>creas<strong>in</strong>g mortality at lower elevations(Ferrell et al. 1994; Millar et al. 2007, 2012; Negrón et al. 2009; van Mantgem andStephenson 2007) although others have found no relationship between elevation andmortality (Ganey and Vojta 2011).Upper-most elevations receive highest snowfall <strong>in</strong>puts and experience a shortenedgrow<strong>in</strong>g season bracketed by low m<strong>in</strong>imum temperatures not conducive to tree growth.Studies of treel<strong>in</strong>e dynamics have illustrated the role of climate <strong>in</strong> foster<strong>in</strong>g differentialgrowth and mortality patterns. Establishment above treel<strong>in</strong>e or <strong>in</strong>to subalp<strong>in</strong>e meadows isgreatest dur<strong>in</strong>g warm and dry periods; dur<strong>in</strong>g cold and wet periods mortality levels often<strong>in</strong>crease (Lloyd and Graumlich 1997; Millar et al. 2004; Villalba et al. 1994). Lloyd andGraumlich (1997) also found that mortality of subalp<strong>in</strong>e conifers <strong>in</strong>creased with<strong>in</strong>creas<strong>in</strong>g elevation and posited that, although mortality severity is unlikely to be greatestat upper elevations, higher transpiration rates at upper elevations or <strong>in</strong>creasedvulnerability of trees established <strong>in</strong> marg<strong>in</strong>al sites might lead to mortality due to waterstress. Villalba et al. (1994) also showed that upper-elevation xeric habitats can beaffected by moisture stress.


84Many studies have attributed elevated levels of forest mortality on south-fac<strong>in</strong>gslopes or drier sites due to <strong>in</strong>creased moisture stress (Dobbert<strong>in</strong> et al. 2005; Oberhuber2001; Vilà-Cabrera et al. 2011; Worrall et al. 2008). However, other studies haveobserved the opposite relationship where <strong>in</strong>creased mortality was associated with northfac<strong>in</strong>gslopes (Guarín and Taylor 2005; Millar et al. 2012). Differences between northfac<strong>in</strong>gand south-fac<strong>in</strong>g slopes exert a major <strong>in</strong>fluence on soil moisture. South-fac<strong>in</strong>g,gentle slopes experience heightened and prolonged solar radiation while north-fac<strong>in</strong>gslopes are more mesic and characterized by lower heatloads. Stands on gentle, northfac<strong>in</strong>gslopes are often more productive and of higher tree density while stands onsouthern aspects are more sparse. Studies of the <strong>in</strong>fluence of aspect or solar radiation onmortality susceptibility engender two hypotheses: 1)<strong>Mortality</strong> is elevated on southernslopes because soil moisture is reduced below critical levels dur<strong>in</strong>g dry periods, and2)<strong>Mortality</strong> is elevated on north-fac<strong>in</strong>g slopes dur<strong>in</strong>g dry periods because trees havelower drought tolerance compared to trees on south-fac<strong>in</strong>g slopes.Studies that have associated <strong>in</strong>creased mortality with north-fac<strong>in</strong>g slopes suggest thatgreater drought tolerance on south-fac<strong>in</strong>g slopes might mitigate mortality risk (Kubiskeand Abrams 1994). In a study of 19 tree species <strong>in</strong> xeric (xeric barrens), mesic (valleyfloor), and “wet-mesic” (floodpla<strong>in</strong>) ecosystems, Kubiske and Abrams (1994)found thatspecies <strong>in</strong> mesic and wet-mesic sites were most affected by decreas<strong>in</strong>g soil moisture andsuggested this was due to lack of drought tolerance mechanisms. Many studies havefound that more drought-tolerant trees or trees <strong>in</strong> more moisture-stressed areas have<strong>in</strong>creased resistance to xylem cavitation, thereby ma<strong>in</strong>ta<strong>in</strong><strong>in</strong>g water potentials dur<strong>in</strong>g soilmoisture scarcity (Alder et al. 1996; Hacke et al. 2001; Maherali and DeLucia 2000;


85McDowell et al. 2008; Rice et al. 2004). Maherali (1999) suggest that <strong>in</strong>creased xylemefficiency is largely phenotypic and a plastic response to environmental conditions. Inblack cottonwood (Populus trichocarpa), Sparks and Black (1999) found that trees <strong>in</strong>more mesic sites showed poor stomatal control <strong>in</strong> ma<strong>in</strong>ta<strong>in</strong><strong>in</strong>g <strong>in</strong>ternal water balance,lowered cavitation resistance, and higher stem mortality, likely due to a trade-off betweencavitation resistance and maximiz<strong>in</strong>g plant conductance.Phytophagous <strong>in</strong>sect population dynamics and forest mortalityThe environmental sett<strong>in</strong>g and tree species composition of a forest stand, oftensummarized as “forest type”, <strong>in</strong>fluence population processes of phytophagous <strong>in</strong>sectssuch as bark beetles. Bark beetles are the primary mortality agents <strong>in</strong> coniferous forests,exhibit<strong>in</strong>g both non-epidemic populations with tree-kills conf<strong>in</strong>ed to small and patchilydistributed groups of trees, and epidemic populations, caus<strong>in</strong>g wide-spread mortalityacross entire forests (Bentz et al. 2010; Christansen et al. 1987). Beetles arepoikilotherms, with developmental processes and overw<strong>in</strong>ter<strong>in</strong>g survival tied totemperature (Bentz et al. 2010; Logan and Powell 2001). For example, fir engraverbeetles (Scolytus ventralis), key herbivores of red fir (Abies magnifica) and white fir(Abies concolor), can produce one generation per year at lower elevations but require 2-3years to produce one generation at upper elevations (Ferrell and Hall 1975). In lowerelevationforests, <strong>in</strong>sects are generally more abundant, with <strong>in</strong>creas<strong>in</strong>gly harsh conditionsand cold temperatures at upper elevations regulat<strong>in</strong>g population size and activity (Loganand Powell 2001; Millar et al. 2007). Shifts <strong>in</strong> generation duration associated withwarmer temperatures have been found for spruce beetle (Dendroctonus rufipennis)populations <strong>in</strong> Alaska, Utah, Colorado and mounta<strong>in</strong> p<strong>in</strong>e beetle (MPB, Dendroctonus


86ponderosae) <strong>in</strong> higher-elevation forests (Bentz and Schen-Langenheim 2007; Hansen etal. 2001; Werner and Hosten 1985). Warmer temperatures and earlier and prolongedwarmer temperatures lead to <strong>in</strong>creased bark beetle overw<strong>in</strong>ter survival, earlier andprolonged fly<strong>in</strong>g seasons, and <strong>in</strong>creased numbers of life cycles per year. Such conditionslikely augment abundance and <strong>in</strong>crease exposure to beetle attacks at lower- and middleelevationforests, which may overwhelm tree defense systems (Christensen et al. 1987).Upper elevation forests may have reduced populations of bark beetles and many fungalpathogens because of cold temperatures, widely-spaced trees, rocky substrates, and lowrelative humidity (Millar et al. 2007). However, mechanical damage susta<strong>in</strong>ed <strong>from</strong>heavy snowpack and high w<strong>in</strong>ds dur<strong>in</strong>g a prolonged w<strong>in</strong>ter period can provide food andshelter resources for bark beetles. W<strong>in</strong>ter conditions not only augment <strong>in</strong>creased spatialrisk of death or damage <strong>from</strong> neighbor<strong>in</strong>g tree-falls but <strong>in</strong>creased densities of dead anddamaged trees can lead to <strong>in</strong>creases <strong>in</strong> the abundance of bark beetle populations,especially dur<strong>in</strong>g warm periods (Bouget and Duelli 2004; Gilbert et al. 2005; Peltonen1999; Schroeder and L<strong>in</strong>delöw 2002).Climate and temporal variation <strong>in</strong> forest mortalityStand characteristics comb<strong>in</strong>e <strong>in</strong> unique ways to <strong>in</strong>fluence differential levels andtim<strong>in</strong>g of mortality across forested landscapes. Further regulation of mortality is imposedby top-down control <strong>from</strong> climatic variation and bottom-up control as altered standstructure, species and age class composition, and forest spatial pattern <strong>from</strong> forestresponse to past historic climates or disturbance events (DeRose and Long 2007;Vygodskaya et al. 2002) Not all drought events are associated with forest healthproblems. Yet few studies have explored how and why differences between drought


87periods <strong>in</strong> the same forest may result <strong>in</strong> differ<strong>in</strong>g levels of forest mortality. Many recentstudies of forest mortality have centered on large-scale and severe mortality outbreaks,with much focus on the l<strong>in</strong>k between persistent drought and forest mortality (Allen et al.2010; Breshears et al. 2005; Dobbert<strong>in</strong> et al. 2007; Floyd et al. 2009; Ganey and Vojta2011; Guarín and Taylor 2005; Hebertson and Jenk<strong>in</strong>s 2008; Millar et al. 2012; vanMantgem et al. 2009). While a strong association between drought and mortality seemsreasonable, few studies have more comprehensively addressed variations <strong>in</strong> mortalitylevels over multiple dry and wet periods.In a study of three tree species <strong>in</strong> Florida, subjected to differ<strong>in</strong>g tim<strong>in</strong>g andmagnitude of drought and flood events, Miao et al. (2009) found that plantgrowth/mortality responded <strong>in</strong> unique ways to disturbance events. The sequence ofdrought and flood events more strongly <strong>in</strong>fluenced plant mortality response than didspecies-specific life history traits. Plant mortality was heightened <strong>in</strong> droughts follow<strong>in</strong>gflood, but not necessarily <strong>in</strong> floods follow<strong>in</strong>g drought. Auclair (1993) found this samepattern when study<strong>in</strong>g tree mortality <strong>in</strong> the Pacific Rim. Canopy dieback <strong>in</strong> dry periodswas correlated not just with the occurrence of dry periods, but with extreme climaticfluctuation between wet and dry periods.Study ObjectivesAvailability of a cont<strong>in</strong>uous, 25-year remote sens<strong>in</strong>g image archive <strong>from</strong> the Landsatsatellite allowed us to quantify and statistically model forest mortality patterns across a623-km 2 watershed, the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong> (LTB), spann<strong>in</strong>g the border of California andNevada, USA. We studied how the magnitude and spatial distribution of forest mortalitychanged over environmental gradients and over wet and dry periods that occurred dur<strong>in</strong>g


88the 1985-2010 timeframe. Our study goals were: 1) To quantify how mortality levelsvaried by forest type and climatic period, 2) To quantify how mortality levels variedaccord<strong>in</strong>g to elevation and solar radiation <strong>in</strong> <strong>in</strong>dividual forests, and 3) To quantify howassociations among topography and mortality levels varied among forest types andclimatic periods.METHODSStudy Area<strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong> (LTB) is <strong>in</strong> the central Sierra Nevada, comprised of approximately62,300 hectares (ha) of forestland rang<strong>in</strong>g <strong>from</strong> 1880 meters (m) to over 3000 m <strong>in</strong>elevation (Figure 1-1). The LTB has a typical Mediterranean climate characterized bycold, wet w<strong>in</strong>ters and warm, dry summers. Most precipitation <strong>in</strong> the LTB arrives assnowfall <strong>from</strong> December-March with small ra<strong>in</strong>fall <strong>in</strong>puts dur<strong>in</strong>g the summer and fallmonths. The Sierra Nevada range, which forms the western side of the LTB, exert a largera<strong>in</strong>-shadow effect on the eastern side (Carson Range) of the LTB, result<strong>in</strong>g <strong>in</strong> decreasedannual precipitation and more regional xeric conditions. Snow amounts decrease along awest-east gradient <strong>in</strong> the LTB, with lower snow <strong>in</strong>put <strong>in</strong> the Carson Range. Soils at thelowest elevations, e.g. alluvial flats and flood pla<strong>in</strong>s, <strong>in</strong> the LTB are primarily derived<strong>from</strong> igneous rocks (mostly granodiorite) with some soils derived <strong>from</strong> igneous extrusiverocks (mostly andesitic lahar) <strong>from</strong> the watershed above. Middle and upper-elevationmounta<strong>in</strong>s <strong>in</strong> the LTB are formed by colluvium <strong>from</strong> either granodiorite or volcanicmaterials. Glaciation and glacial deposition strongly <strong>in</strong>fluenced the LTB as well, withglacial outwash leav<strong>in</strong>g large quantities of sediment at the lower elevations (USDANRCS 2007).


89<strong>Forest</strong> Types and EcoregionsWe exam<strong>in</strong>ed mortality patterns <strong>in</strong> five forest types across the LTB. <strong>Forest</strong> types weredef<strong>in</strong>ed us<strong>in</strong>g the 2009 USFS CalVeg vegetation classification (USDA <strong>Forest</strong> Service.1981. CALVEG; CalVeg Zone3: EVEG (Exist<strong>in</strong>g Vegetation) Tiles 17B, 21A, and21B). <strong>Forest</strong> types were identified accord<strong>in</strong>g to dom<strong>in</strong>ant forest composition as: P<strong>in</strong>es(JP) (Jeffrey P<strong>in</strong>e or Eastside P<strong>in</strong>e), Red Fir (RF), Lodgepole P<strong>in</strong>e (LP), White Fir (WF),and the Mixed Conifer Alliance (MF) (Table 1-2). Because of their widespreaddistribution around the LTB, JP, MF, and RF forests were further separated <strong>in</strong>to fiveecoregions def<strong>in</strong>ed by the CalVeg classification (Figure 2-1).We removed all urban areas, <strong>in</strong>clud<strong>in</strong>g urban forests, as well as all meadows, edges ofsmall water bodies, areas dom<strong>in</strong>ated by aspen (Populus tremuloides), riparian vegetation,chaparral fields, restoration projects and historical fires. Developed areas surround<strong>in</strong>gcampgrounds, ski areas, or residential areas were also removed. We buffered all majorroads <strong>in</strong> the LTB by 120 meters (60 meters both sides of the road) and removed thoseareas <strong>from</strong> the analysis.Image Process<strong>in</strong>g and <strong>Mortality</strong> ClassificationWe used fall Landsat Thematic Mapper (TM) imagery <strong>from</strong> 1985 – 2010 available<strong>from</strong> the USGS Glovis site (see Chapter One for further detail) and normalized all imagesto the September 2010 image us<strong>in</strong>g the IRMAD change detection algorithims (Canty andNielsen 2008). To estimate reductions <strong>in</strong> canopy vigor, we used the NormalizedDifference Wetness Index (NDWI) and the change <strong>in</strong> NDWI (Differenced NDWI,dNDWI) values between two subsequent years (Gao 1996). NDWI uses Landsat spectralbands 4 and 5, which are sensitive to leaf water content, and is more sensitive to detection


90of coniferous forest mortality than other vegetation <strong>in</strong>dices, such as NDVI (Gao 1996;Vogelmann et al. 2009). To estimate annual mortality, we used dNDWI to quantifysignificant deterioration of tree canopy that exceeded pre-def<strong>in</strong>ed thresholds calibratedwith field data. <strong>Mortality</strong> classification and validation procedures are further expla<strong>in</strong>ed <strong>in</strong>Chapter One.ClimateWe quantified how relationships between stand structure and mortality changedby forest type and climatic period with<strong>in</strong> our 25-year time series (see Chapter One forfurther description). We designated five climatic periods based on historical accounts ofclimatic variation <strong>in</strong> the LTB and proximate areas as well as historical hydrologicaldatasets, such as the California Cooperative Snow Survey Program. To verify dates <strong>from</strong>these accounts, we used the “annual precipitation as snow” (PAS) variable derived <strong>from</strong>the ClimateWNA dataset to determ<strong>in</strong>e annual snow deficits as departure <strong>from</strong> the 30-yearnorm (Wang et al. 2012). We then tested accuracy of historical accounts aga<strong>in</strong>st variouslagged modeled snow deficits (<strong>from</strong> 2-5 years) as well as annual w<strong>in</strong>ter m<strong>in</strong>imumtemperature to ensure correct representation of climatic periods. We found the four yearaveraged snow deficit most accurately portrayed historical accounts and established fiveclimatic periods dur<strong>in</strong>g the 1985-2010 study series of vary<strong>in</strong>g duration (w1=2years,d1=8years, w2=4years, d2=7years, w3=4years) (Figure 1-3). We additionally used thenumber of frost free days and w<strong>in</strong>ter m<strong>in</strong>imum temperature <strong>in</strong> our analyses of variabilityacross the five climatic periods (Figure 2-2a,b).


91Environmental VariablesIn addition to ecoregional variables, we used solar radiation (WH/m 2 , Figure 2-3).and elevation (m, Figure 2-4) <strong>from</strong> a 30-meter digital elevation model (DEM) <strong>in</strong>ArcMap10 (ESRI). Solar radiation was calculated us<strong>in</strong>g the Spatial Analyst feature <strong>in</strong>ArcMap and uses aspect and slope derived <strong>from</strong> a 30m DEM to derive a measure of totalclear-sky radiation, both diffuse and direct, over a s<strong>in</strong>gle-day period (Fu and Rich 1999).We used the total radiation received per pixel over the length of daylight on August 1 asrepresentative of the <strong>in</strong>fluences of diffuse and direct sunlight on soil moisture dur<strong>in</strong>gsummer conditions.Data AnalysisWe used the ReLogit (Rare Events Logistic Regression) method to test forrelationships between mortality and environmental and stand structure predictors (K<strong>in</strong>gand Zeng 2001). Logistic regression was used to model the probability of a mortalityevent with<strong>in</strong> a pixel as a statistical function of forest density and environmental<strong>in</strong>fluences, with f<strong>in</strong>al model selection based on highest explanatory power and modelparsimony. Results are <strong>in</strong>terpreted as odds ratios, derived <strong>from</strong> parameter coefficients forevery 500m <strong>in</strong>crease <strong>in</strong> elevation or 1000 WH/m 2 <strong>in</strong>crease <strong>in</strong> solar radiation. Odds ratiosgreater than 1 <strong>in</strong>dicate a positive relationship between predictor and response and oddsratios less than 1 <strong>in</strong>dicate a negative relationship. Odds ratios that equal 1 show norelationship between predictor and response variable.To exam<strong>in</strong>e differences <strong>in</strong> the levels of mortality among forest types across the timeperiod, we regressed annual mortality for an <strong>in</strong>dividual forest type by climatic period. Toexam<strong>in</strong>e differences <strong>in</strong> the levels of mortality among ecoregions across the time period,


92we regressed annual mortality for an <strong>in</strong>dividual forest type with the <strong>in</strong>teraction ofecoregion and climatic period. Results are <strong>in</strong>terpreted as percent of ecoregionexperienc<strong>in</strong>g canopy dieback.To exam<strong>in</strong>e differences <strong>in</strong> mortality levels among forest types and climatic periods,we used a regression tree with forest type and climatic period as the categoricaldependent predictor variables with annual mortality as the response variable. Regressiontrees use b<strong>in</strong>ary recursive partition<strong>in</strong>g to split the dataset <strong>in</strong>to mean<strong>in</strong>gful groups of thepredictor variables that are most similar <strong>in</strong> terms of the value of the response variable(Breiman et al. 1984; De'ath and Fabricius 2000). We did not prune or snip the tree assplits <strong>in</strong> the <strong>in</strong>itial tree were reasonable and of acceptable parsimony. Because antecedentmoisture conditions have been important <strong>in</strong> tree mortality <strong>in</strong> other studies <strong>in</strong> the westernU.S., we exam<strong>in</strong>ed relationships between annual mortality and differ<strong>in</strong>g lag periods ofthe snow deficit variable (Table 1-2; Bigler et al. 2007; Guarín and Taylor 2005; Millar etal. 2012). We <strong>in</strong>vestigated relationships with lagged snow deficits rang<strong>in</strong>g <strong>from</strong> 0-5years, where lag 0 equals most recent w<strong>in</strong>ter snowfall.RESULTS<strong>Forest</strong> <strong>Mortality</strong> Levels across Drought and Wet PeriodsUs<strong>in</strong>g the mortality classification process described <strong>in</strong> Chapter One, we derivedannual mortality levels for each forest type that reflect the amount of new mortality seenon the landscape <strong>in</strong> each year (Appendix A, Figure 2-5). Regression results showed twodist<strong>in</strong>ct periods of differ<strong>in</strong>g forest mortality that were consistent across forest types: anearly to mid-period <strong>from</strong> 1985 to1995 (w1, d1) characterized by heightened mortality,and a later period <strong>from</strong> 2000 to 2009 (w2, d2, w3) characterized by lower mortality levels


93(Figure 2-6). <strong>Forest</strong> mortality magnitude and temporal trend also varied by forest type,with upper-elevation forests (LP and RF) exhibit<strong>in</strong>g higher mortality levels than lowerandmiddle-elevation JP, MF, and WF forests. JP and MF showed higher mortality levelsdur<strong>in</strong>g the first drought than for preced<strong>in</strong>g or subsequent wet periods. For LP forest,mortality levels were highest <strong>in</strong> W1 and W2 and lower <strong>in</strong> the first drought (Figure 2-7;Table 2-2). In LP and RF forests, mortality was characterized by years of heightenedmortality with <strong>in</strong>terven<strong>in</strong>g years characterized by lower mortality levels. In JP, MF, andWF forests mortality was less episodic, with annual mortality levels that were similarover all climatic periods. All forests showed <strong>in</strong>creased mortality <strong>in</strong> the first droughtperiod compared to the second. Although mortality was stable at low levels over the timeperiod, WF forests experienced approximately three years (1988, 1994, and 2008) ofepisodic high mortality similar to LP and RF forests. Dur<strong>in</strong>g the first and second wetperiod, mortality <strong>in</strong> these forests was elevated with respect to mortality <strong>in</strong>curred dur<strong>in</strong>gthe second drought and third wet period.Ecoregional Influences on <strong>Forest</strong> <strong>Mortality</strong>Regression analysis of ecoregion x climatic period <strong>in</strong>teractions did not revealsignificant differences among annual mortality levels <strong>in</strong> ecoregions of similar forests, butshowed dissimilar mortality patterns among ecoregions <strong>in</strong> some forest types (Figure 2-8;Figure 2-9). In JP and MF forests, all ecoregions were characterized by much highermortality levels dur<strong>in</strong>g the first drought, with lower mortality levels throughout the rest ofthe time series. In RF forests, mortality levels were greatest <strong>in</strong> west side ecoregions andleast <strong>in</strong> the upper-elevation southern ecoregion dur<strong>in</strong>g the first wet period. East-side redfir forests did not respond similarly with slight mortality <strong>in</strong>creases seen dur<strong>in</strong>g the first


94drought and subsequent and preced<strong>in</strong>g wet periods. Over all climatic periods andecoregions, mortality levels were typically highest <strong>in</strong> RF ecoregions and lower for JP andMF ecoregions. In the east-side Carson Range forests and upper-elevation southernecoregions, JP forests showed lowest mortality rates over nearly all climatic periods.Topographic effects on forest mortality: Solar Radiation and ElevationAll forests exhibited negative relationships between solar radiation and mortality riskover at least 30% of the 25-year time period, with positive relationships account<strong>in</strong>g foronly 0-9% of the time period with the exception of WF forests (Figure 2-10; Figure 2-11). In WF forests, south-fac<strong>in</strong>g aspects were positively associated with mortality morefrequently than for other forest types, with this effect heightened <strong>in</strong> the last wet period ofthe time series. <strong>Mortality</strong> risk <strong>in</strong> JP forests shows the greatest <strong>in</strong>sensitivity to sitevariations <strong>in</strong> solar radiation followed by LP, RF, WF, and MF forests. In JP, LP, and MFforests, at least 1-2 models showed <strong>in</strong>creased mortality risk on southern or steeperaspects. In MF forests, when solar radiation was associated with mortality risk, morenorth-fac<strong>in</strong>g, gentle aspects were always predisposed to mortality.Throughout the time series, relationships among mortality and elevation were mostvariable <strong>in</strong> direction for LP and RF forests and less so for JP, MF, and WF forests (Figure2-12). In LP and RF forests, positive relationships between mortality and elevation weremore frequent <strong>in</strong> the second half of the time series, and generally negative <strong>in</strong> the first partof the time series. In MF and WF forests, mortality was generally positively associatedwith elevation. When associated with elevation, the probability of mortality <strong>in</strong>creasedwith <strong>in</strong>creas<strong>in</strong>g elevation for upper-elevation RF and LP forests dur<strong>in</strong>g all three wetperiods.


95In JP forests, decreases <strong>in</strong> elevation <strong>in</strong>creased mortality risk over 60% of the timeperiod and decreased mortality risk less than 10% of the time (Figure 2-13). In MFforests <strong>in</strong>creases <strong>in</strong> elevation <strong>in</strong>creased mortality risk over 35% of the time series, themost of any forest type, and decreased mortality risk around 11% of the time period. InRF and WF forests, <strong>in</strong>creases <strong>in</strong> elevation <strong>in</strong>creased mortality risk over approximately20% of the time period, and decreased mortality risk over approximately 22% and 9%respectively. For LP forests, <strong>in</strong>creases <strong>in</strong> elevation <strong>in</strong>creased mortality risk over 20% ofthe time period, and decreased mortality risk <strong>in</strong> only 15% of the time period. Elevationdid not show a relationship with mortality <strong>in</strong> at least 30% of the time period for all foresttypes, with the greatest lack of effect shown for WF over approximately 70% of the timeperiod. WF, however, showed the strongest magnitude spikes where <strong>in</strong>creased elevation<strong>in</strong>creased the risk of mortality by 7 to 8 times for every 500 m <strong>in</strong>crease <strong>in</strong> elevation. RF,WF, LP, and MF showed high magnitude positive relationships between elevation andmortality <strong>in</strong> the first two years of the time period, with smaller magnitude effects seenprimarily <strong>in</strong> the first and second dry periods.DISCUSSIONIn our study, mortality response was variable and <strong>in</strong>fluenced by forest type andclimatic period. While recent studies on forest mortality suggest that extensive dry ordrought-like periods almost always create widespread forest mortality events (reviewed<strong>in</strong> Allen et al. 2010), we found evidence to the contrary. The disparity between mortalitylevels <strong>in</strong> the first drought and the second drought is surpris<strong>in</strong>g, consider<strong>in</strong>g that the laterdrought period, where forest mortality was relatively limited, was characterized byelevated temperatures as well as low precipitation amounts.


96Results parallel those of Millar et al (2007) who observed elevated levels of limberp<strong>in</strong>e mortality (P<strong>in</strong>us flexilis) <strong>in</strong> the 1987-1992 drought but not the subsequent 1999-2004drought, for the eastern escarpment of the central-southern Sierra Nevada. The authorshypothesized that earlier drought-<strong>in</strong>duced mortality promoted short-term <strong>in</strong>creases <strong>in</strong>forest health and forest resilience. Support for this hypothesis is found <strong>in</strong> our study <strong>in</strong> theubiquitous decrease <strong>in</strong> mortality observed <strong>in</strong> the latter drought period (d2) compared tomore widespread mortality associated with the first earlier period. Early, <strong>in</strong>tense drought<strong>in</strong> the early part of this time series may have selectively reduced populations of lessdrought-tolerant tree <strong>in</strong>dividuals or species, adjust<strong>in</strong>g forest composition tolerance suchthat the forest experienced less mortality dur<strong>in</strong>g the later droughts of the early 2000s.The tim<strong>in</strong>g of high-mortality events <strong>in</strong> the LTB confirms that droughts, as def<strong>in</strong>ed bya duration of precipitation deficits, do not always impact forests <strong>in</strong> the same way, and thatsequence matters (sensu Miao et al. 2009). Miao et al. (2009) measured forest mortality<strong>in</strong> two disturbance sequences determ<strong>in</strong>ed by either 4-month transitions between floodfollowed by <strong>in</strong>termediate conditions and then drought, or by 4-month transitions betweendrought followed by <strong>in</strong>termediate conditions and then flood. In our study, fluctuationsbetween abnormally high snowloads and drought proved analogous to the drought-wetdroughtsequence of Miao et al. (2009). In the LTB, <strong>in</strong>creases <strong>in</strong> tree mortality are likelydue to the abrupt transition between two climatic extremes as well as the impact of thefirst disturbance on the outcome of the subsequent disturbance.In the search for predictive relationships between tree growth rates and mortality,many researchers have found that variable growth is of equal or greater importance thanslow growth <strong>in</strong> separat<strong>in</strong>g live and dead trees (Ogle et al. 2000; Suarez et al. 2004) and


97that abrupt decl<strong>in</strong>es <strong>in</strong> growth are key <strong>in</strong>dicators of eventual mortality (Das et al. 2007;Pedersen 1998). Active tree growth, fostered dur<strong>in</strong>g periods of abundant moisture andwarm temperatures, such as those that occurred <strong>in</strong> early-mid 1980s, may have decreasedrelative availability of carbon needed for manufacture of secondary defense compounds.Limited and variable supplies of water and nutrients engender different resourcepartition<strong>in</strong>g and allocation strategies with<strong>in</strong> plants that impact trade-offs between growthand defense aga<strong>in</strong>st herbivores and pathogens (Herms and Mattson 1992; Stamp 2003).A landscape with ample numbers of vigorous trees, as required by MPB and likely otherDendroctonus beetles, with lowered carbon stores <strong>in</strong> addition to temperature-mediatedbark beetle <strong>in</strong>creases and refugia for bark beetles <strong>in</strong> snow-damaged trees may have set thestage for outbreak mortality dur<strong>in</strong>g the first drought. For trees <strong>in</strong> which moisture stressoccurs on a seasonal and annual basis, it may be that climate variability and sharpfluctuations <strong>in</strong> moisture stress <strong>in</strong>fluence the magnitude of forest mortality more than theduration of drought-like conditions.<strong>Mortality</strong> patterns were similar across ecoregions <strong>in</strong> all forest types, <strong>in</strong>dicat<strong>in</strong>g thatmore f<strong>in</strong>e-scale, regional conditions, as def<strong>in</strong>ed by the ecoregion variable, may havelimited <strong>in</strong>fluence on overall forest health. Most dist<strong>in</strong>ct mortality patterns amongecoregions occurred <strong>in</strong> RF forest types, where forests <strong>in</strong> the southern and eastern (CarsonRange) portions of the LTB had lower mortality levels than those of western and northernRF forests. This dist<strong>in</strong>ction could be due to lower snow loads on the eastern side of theLTB, <strong>in</strong>creased drought tolerance of these forests, or less <strong>in</strong>sect herbivory. However,Ferrell et al. (1994) reported on widespread fir engraver activity <strong>in</strong> the upper-elevationCarson Range and JP forests <strong>in</strong> the Carson Range experienced heightened mortality <strong>in</strong> the


98first wet and dry period. Similar patterns of forest mortality across all ecoregions couldbe due to the widespread nature of the mid 1980s-mid 1990s beetle epidemic and thesubsequent non-epidemic nature of beetle populations. The conceptual model suggestedby our results <strong>in</strong>cludes top-down <strong>in</strong>fluences (<strong>in</strong>clud<strong>in</strong>g climate, bark beetle populations,and weather) driv<strong>in</strong>g overall forest mortality patterns, with local variations due to edaphicand microclimate conditions important <strong>in</strong> regulat<strong>in</strong>g mortality levels, particularly dur<strong>in</strong>gnon-epidemic periods.We f<strong>in</strong>d that <strong>in</strong>creas<strong>in</strong>gly productive systems often experience highest mortality, butfor different reasons. In contrast to more xeric Carson Range forests, RF forests on thewest side of the LTB experienced heightened mortality levels likely due to extremew<strong>in</strong>ter storms and result<strong>in</strong>g <strong>in</strong>creases <strong>in</strong> bark beetle populations. In JP forests, moremesic areas may experience <strong>in</strong>creased pressure <strong>from</strong> damage agents. For MF forests,percent mortality <strong>in</strong> more mesic west-side ecoregions (EH and EJ) rema<strong>in</strong>s stable dur<strong>in</strong>gthe second and third wet period when mortality decl<strong>in</strong>es <strong>in</strong> eastern and southern stands.The positive relationship between net primary productivity (NPP) and forest mortalityhas been identified at global scales, though common drivers rema<strong>in</strong> elusive (Frankl<strong>in</strong> etal. 1987; Stephenson et al. 2011). <strong>Mortality</strong> patterns <strong>in</strong> the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong> suggest thatthis relationship may hold at f<strong>in</strong>er spatial scales associated with topographic variation.In all forests and across the time period, mortality risk is generally greater on northfac<strong>in</strong>gaspects and steeper slopes, similar to other studies of mortality <strong>in</strong> the SierraNevada (Guarín and Taylor 2005; Millar et al. 2012). This near ubiquitous relationshipcould be <strong>in</strong>dicative of the <strong>in</strong>crease <strong>in</strong> moisture stress and the limited capability of standson north-fac<strong>in</strong>g slopes to adapt to <strong>in</strong>creased moisture stress dur<strong>in</strong>g droughty periods (d1


99and d2) and <strong>in</strong>creased temperatures (w2 and w3) (Maherali and DeLucia 2000; Zwiazekand Blake 1989).The relationship between aspect and mortality risk can also be a result of bark beetlepreferences. Similar to elevation, aspect moderates both over-w<strong>in</strong>ter temperatures andgeneration period, with warmer slopes associated with outbreak conditions <strong>in</strong> MPB(Safranyik et al. 1974; Safranyik et al. 1989). In their study of environmental conditionsassociated with MPB damage <strong>in</strong> lodgepole p<strong>in</strong>e forests, Nelson et al. (2007) found thatsouthern and western aspects were tied to heightened beetle activity. In our study, it maybe that <strong>in</strong>creased stand density on north-fac<strong>in</strong>g slopes promotes warmer w<strong>in</strong>terconditions, as w<strong>in</strong>d speeds are decreased and ambient temperatures <strong>in</strong> the stand aremoderated. In addition, many of the w<strong>in</strong>ter storms <strong>in</strong> the Sierra Nevada arrive <strong>from</strong> awesterly-southwesterly direction, with severity likely heightened on sparsely forested,southern slopes (Dett<strong>in</strong>ger et al. 2004).Increased pathogen pressure <strong>from</strong> root pathogens (Armillaria root disease (Armillariaspp.), Annosus root disease (Heterobasidion annosum), lam<strong>in</strong>ated root rot (Phell<strong>in</strong>usweirii), black sta<strong>in</strong> root disease (Leptographium wageneri) may also help expla<strong>in</strong>elevated mortality risk on more north-fac<strong>in</strong>g slopes. A concurrent exploration ofmortality drivers <strong>in</strong> the LTB found elevated levels of tree root disease <strong>in</strong> stands on morenorth-fac<strong>in</strong>g aspects (I. Munck & R. Nowak, pers. comm., University of Nevada, Reno).Villalba and Veblen (1998) found <strong>in</strong>creased <strong>in</strong>cidence of root disease <strong>in</strong> more mesic areasof their study site and <strong>in</strong>creased mortality <strong>from</strong> lam<strong>in</strong>ated root rot and Annosus rootdisease <strong>in</strong> true firs (Goheen and Goheen 1989; Nelson and Sturrock 1993; Slaughter andParmeter 1989), common <strong>in</strong> more mesic sites with deeper soil (North et al. 2005).


100Our study results parallel those of other studies of mortality <strong>in</strong> coniferous forests thatfound <strong>in</strong>creased mortality at lower elevations <strong>in</strong> the western U.S. dur<strong>in</strong>g drought periodscharacterized by outbreak populations, both <strong>in</strong> the western United States (Ferrell et al.1994; Ferrell 1996; van Mantgem and Stephenson 2007; Negrón et al 2009; Millar et al.2007; Millar et al. 2012) and <strong>in</strong> European temperate forests (Schutt and Cowl<strong>in</strong>g 1985).However, our f<strong>in</strong>d<strong>in</strong>gs also show that the relationship between mortality and elevation isvariable by forest type, with less support for a common relationship between mortalityand elevation <strong>in</strong> either time or place.Increas<strong>in</strong>g average w<strong>in</strong>ter m<strong>in</strong>imum temperatures and number of frost free days wereevidenced after the first drought with similar <strong>in</strong>creases <strong>in</strong> climatic water deficit found <strong>in</strong>other studies of forest mortality <strong>in</strong> the Sierra Nevada (Millar et al. 2012; van Mantgemand Stephenson 2007). Increases <strong>in</strong> temperature <strong>in</strong> addition to lower precipitation dur<strong>in</strong>gthe second drought may have comb<strong>in</strong>ed to cause higher moisture stress for both JP forestsas has been found <strong>in</strong> other studies of forest mortality (Allen et al. 2010) and for relativelyxeric areas with<strong>in</strong> upper-elevation RF, LP and MF forests <strong>in</strong> California (Lloyd andGraumlich 2007). However, beetle activity is also associated with elevation, due to l<strong>in</strong>ksbetween temperature and terra<strong>in</strong>, with elevation critical to host selection for mounta<strong>in</strong>p<strong>in</strong>e beetle (Safranyik and Carroll 2006). Reduced generation time is necessary forelevated numbers of beetles across the landscape and Amman (1973) and Safranyik andCarroll (2006) showed that limitations on a 1-year life cycle are l<strong>in</strong>early related toelevation and latitude. In a study of MPB hot-spots across lodgepole p<strong>in</strong>e forests <strong>in</strong>British Columbia, Canada, Nelson et al. (2007) showed that MPB <strong>in</strong>festations were


101concentrated at lower elevations, and <strong>in</strong>creased <strong>from</strong> northern to southern portions of thestudy area.Increased mortality risk at lower elevations <strong>in</strong> JP forests could be due to <strong>in</strong>creasedbeetle populations and earlier emergence <strong>from</strong> <strong>in</strong>fected trees result<strong>in</strong>g <strong>in</strong> a prolongedflight season. In almost all wet years <strong>in</strong> upper-elevation forests, mortality probability wasgreater at upper elevations, likely <strong>from</strong> mechanical damage and mortality on more recentestablishment on marg<strong>in</strong>al sites dur<strong>in</strong>g preced<strong>in</strong>g dry periods (Lloyd and Graumlich2007; Millar et al. 2004; M<strong>in</strong>nich 1984; Villalba and Veblen 1998). Natural th<strong>in</strong>n<strong>in</strong>g<strong>in</strong>curred dur<strong>in</strong>g the first drought may have <strong>in</strong>creased mortality risk to sparse upperelevation stands, where stands of greater density were observed to experience lowermortality risk (see Chapter One). The same could apply for MF and WF forests dur<strong>in</strong>gthe latter part of the time series. Increased moisture stress due to higher temperaturescompounded by <strong>in</strong>creased susceptibility <strong>in</strong> stands that were formerly sheltered <strong>from</strong>persistent moisture stress may be more proximal explanations (Kubiske and Abrams1994; Maherali et al. 2004; Zwiazek and Blake 1989).Inter-annual variability of forest mortality <strong>in</strong> the LTB has been most similar for lowerand middle-elevation forests Jeffrey p<strong>in</strong>e and mixed-fir forests, which differed <strong>from</strong>upper-elevation lodgeple p<strong>in</strong>e and red fir forests. Elevated mortality <strong>in</strong> JP and MF forestsoccurred dur<strong>in</strong>g the 1987-1994 drought when <strong>in</strong>sect outbreaks, primarily fir engraver andJeffrey p<strong>in</strong>e beetle, were widespread across the LTB (Egan et al. 2012; Elliot-Fisk 1996;Ferrell et al.1994; Ferrell 1996; SNEP 1996). For JP forests <strong>in</strong> the east-side CarsonRange, our study shows that 1985-1986 mortality levels <strong>in</strong>creased before the onset of thefirst drought. Either regional drought began earlier or mortality was dom<strong>in</strong>ated by spread


102of fir engraver mortality that was extensive <strong>in</strong> upper-elevation forests (Ferrell et. al1994). Our results agree with other studies that have found dim<strong>in</strong>ished response <strong>in</strong>mortality levels of p<strong>in</strong>e-dom<strong>in</strong>ated forests to drought-like conditions (Bigler et al. 2007;Ganey and Vojta 2011). Hurteau et al. (2007), work<strong>in</strong>g <strong>in</strong> the southern Sierra Nevada,documented <strong>in</strong>creased climatic sensitivity <strong>in</strong> white fir compared to sugar p<strong>in</strong>e and foundlittle support for a relationship between annual growth rate and climate for Jeffrey p<strong>in</strong>e.Lower elevation forest types may experience less drought-related mortality due tostrong drought-tolerant characteristics of p<strong>in</strong>e dom<strong>in</strong>ants. In the LTB, JP forests aredom<strong>in</strong>ated by larger Jeffrey p<strong>in</strong>e and sugar p<strong>in</strong>e, both species that are characterized byhigh drought tolerance, likely tied to better stomatal control of water loss and cavitationresistance as has been shown for ponderosa p<strong>in</strong>e <strong>in</strong> more xeric environments (Maheraliand DeLucia 2000). In lower-elevation forests which often undergo prolonged seasonaldrought, extended duration of drought <strong>in</strong> addition to widespread <strong>in</strong>sect abundance may benecessary to elevate mortality levelsUpper-elevation forests, however, experienced a more elevated and episodic forestmortality pattern. In RF and LP forests, <strong>in</strong>dividual years experienced severe mortalitylevels up to 4 times higher than that of <strong>in</strong>terven<strong>in</strong>g stable years. Episodic mortalityoccurred throughout the time series and seemed a response to two divergentenvironmental conditions: wet periods characterized by heavy snowfall or drought. Thehighest mortality levels for both forest types were observed <strong>in</strong> 1995 and 1996, possiblydue to damage <strong>from</strong> heavy snowfall followed by extreme drought. Of the two dom<strong>in</strong>antspecies <strong>in</strong> affected forests, LP exhibited greater response to environmental stress dur<strong>in</strong>gwet years and relatively little response to stress dur<strong>in</strong>g dry years. These differential


103responses may be due to the greater drought-tolerance of lodgepole p<strong>in</strong>e (DeClerck et al.2005).CONCLUSIONResults of this study evidence the importance of both endogenous and exogenousfactors and their <strong>in</strong>teraction <strong>in</strong> shap<strong>in</strong>g heterogeneous forest mortality patterns acrossclimatic periods and forest types. <strong>Mortality</strong>, rather than be<strong>in</strong>g an additive comb<strong>in</strong>ation ofspecies traits, environmental position, and climate, is a non-l<strong>in</strong>ear response to factorsoperat<strong>in</strong>g at multiple scales. Needed are more mechanistic models that provide theconceptual, cross-scale l<strong>in</strong>kage among detailed ecophysiological studies, forest pathologysurveys of stand-level patterns and broad-scale remote sens<strong>in</strong>g studies. <strong>Forest</strong> mortalitypatterns are <strong>in</strong>herently multi-causal and unresolvable through observational studies alone.However, observational studies at sufficiently broad scales of space and time are usefulfor generat<strong>in</strong>g alternative hypotheses that can be addressed at f<strong>in</strong>er scales. For example,are apparent effects of environmental gradients on forest mortality ecophysiological <strong>in</strong>nature, or due to <strong>in</strong>direct effects mediated by bark beetle population dynamics? Furtherstudy <strong>in</strong>to separat<strong>in</strong>g causative mechanisms <strong>in</strong> forest mortality is needed.


104LITERATURE CITEDAlder, N.N., J.S. Sperry, and W.T. Pockman. 1996. Root and stem xylem embolism,stomatal conductance, and leaf turgor <strong>in</strong> Acer grandidentatum populations along asoil moisture gradient. Oecologia 105: 293-301.Allen, Craig D., Alison K. Macalady , Haroun Chenchouni, Dom<strong>in</strong>ique Bachelet, NateMcDowell,Michel Vennetier, Thomas Kitzberger, Andreas Rigl<strong>in</strong>g , David D.Breshears, E.H. (Ted) Hogg, Patrick Gonzalez , Rod Fensham, Zhen Zhang, JorgeCastro, Natalia Demidova, Jong-Hwan Lim, Gillian Allard, Steven W. Runn<strong>in</strong>g,Akk<strong>in</strong> Semerci, and Neil Cobb. 2010. A global overview of drought and heat<strong>in</strong>ducedtree mortality reveals emerg<strong>in</strong>g climate change risks for forests. <strong>Forest</strong>Ecology and Management 259 (4): 660–684.Amman, G.D. 1973. Population changes of the mounta<strong>in</strong> p<strong>in</strong>e beetle <strong>in</strong> relation toelevation. Environmental Entomology 2:541–547.Auclair, Allan N.D. 1993. Extreme climatic fluctuations as a cause of forest dieback <strong>in</strong>the Pacific Rim. Water, Air, and Soil Pollution 66: 207-229.Benz, Barbara J. and Greta Schen-Langenheim. 2007. The mounta<strong>in</strong> p<strong>in</strong>e beetle andwhitebark p<strong>in</strong>e waltz: has the music changed? In Proceed<strong>in</strong>gs of the ConferenceWhitebark P<strong>in</strong>e: A Pacific Coast Perspective. August 27-31, 2006, Ashland,OregonBentz, Barbara, Jacques Régnière, Christopher J. Fettig, E. Matthew Hansen, Jane L.Hayes, Jeffrey A. Hicke, Rick G. Kelsey, Jose F. Negrón, and Steven J. Seybold.2010. Climate Change and bark beetles of the western United States and Canada:Direct and <strong>in</strong>direct effects. Bioscience. 60(8): 602-613.


105Berg, Edward E, J. David Henry, Christopher L. Fastie, Andrew D. De Volder, Steven M.Matsuoka. 2006. Spruce beetle outbreaks on the Kenai Pen<strong>in</strong>sula, Alaska, andKluane National Park and Reserve, Yukon Territory: Relationship to summertemperatures and regional differences <strong>in</strong> disturbance regimes. <strong>Forest</strong> Ecology andManagement 227: 219–232.Bigler, C., D.G. Gav<strong>in</strong>, C. Gunn<strong>in</strong>g, and T. T.Veblen. 2007. Drought <strong>in</strong>duces lagged treemortality <strong>in</strong> a subalp<strong>in</strong>e forest <strong>in</strong> the Rocky Mounta<strong>in</strong>s. Oikos 116: 1983–1994.Bouget, Christophe and Peter Duelli. 2004. The effects of w<strong>in</strong>dthrow on forest <strong>in</strong>sectcommunities: a literature review. Biological Conservation 118(3): 281-299.Breshears, D.D., Cobb, N.S., Rich, P.M., Price, K.P., Allen, C.D., Balice, R.G., Romme,W.H., Kastens, J.H., Floyd, M.L., Belnap, J., Anderson, J.J., Myers, O.B., Meyer,C.W., 2005. Regional vegetation die-off <strong>in</strong> response to global-change-typedrought. Proceed<strong>in</strong>gs of the National Academy of Sciences of the United States ofAmerica 102 (42): 15144–15148.Breiman, L., J. H. Friedman, R.A. Olshen, and C. G. Stone. 1984. Classification andregression trees. Wadsworth International Group, Belmont, California, USA.Das, Adrian, John Battles, Nathan L. Stephenson, Phillip J. van Mantgem. 2007. Therelationship between tree growth patterns and likelihood of mortality: a study oftwo tree species <strong>in</strong> the Sierra Nevada. Canadian Journal of <strong>Forest</strong> Research 38:580-597.Das, Adrian, John Battles, Nathan L. Stephenson, Phillip J. van Mantgem. 2008. Spatialelements of mortality risk <strong>in</strong> old-growth forests. Ecology 89 (6): 1744-1756.


106Das, Adrian, John Battles, Nathan L. Stephenson, Phillip J. van Mantgem. 2011. Thecontribution of competition to tree mortality <strong>in</strong> old-growth coniferous forests.<strong>Forest</strong> Ecology and Management 261: 1203–1213.De'ath, Glenn and Kathar<strong>in</strong>a E. Fabricius. 2000. Classification and regression trees: Apowerful yet simple technique for ecological data analysis. Ecology 81(11):3178-3192.DeClerck, Fabrice A.J., Michael G. Barbour, and John O. Sawyer. 2005. Resource UseEfficiency as a Function of Species Richness and Stand Composition <strong>in</strong> UpperMontane Conifer <strong>Forest</strong>s of the Sierra Nevada. Journal of Vegetation Science,16(4): 443-452.DeRose, R. Just<strong>in</strong> and James N. Long. <strong>2012.</strong> Factors <strong>in</strong>fluenc<strong>in</strong>g the spatial and temporaldynamics of Engelmann spruce mortality dur<strong>in</strong>g a spruce beetle outbreak on theMarkagunt Plateau, Utah. <strong>Forest</strong> Science 58(1): 1-14.Dobbert<strong>in</strong>, M, P. Mayer, T. Wohlgemuth, E. Feldmeyer-Christe, U. Graf, N.E.Zimmermann, and A. Rigl<strong>in</strong>g. 2005. The decl<strong>in</strong>e of P<strong>in</strong>us sylvestris L. forests <strong>in</strong>the swiss Rhone Valley-a result of drought stress? Phyton-Annales Rei Botanicae45: 153-156.Dobbert<strong>in</strong>, Matthias, Beat Wermel<strong>in</strong>ger, Christof Bigler, Matthias Bürgi, MatthiasCarron, Beat Forster, Urs Gimmi, and Andreas Rigl<strong>in</strong>g. 2007. L<strong>in</strong>k<strong>in</strong>g <strong>in</strong>creas<strong>in</strong>gdrought stress to Scots p<strong>in</strong>e mortality and bark beetle <strong>in</strong>festations. The ScientificWorld 7(S1): 231–239.Egan, Joel, Dave Fournier, Hugh Safford, J.McLean Sloughter, Tamre Cardoso, PatrickTra<strong>in</strong>or, and John Wenz. 2011. Assessment of a Jeffrey P<strong>in</strong>e Beetle Outbreak


107<strong>from</strong> 1991-1996 near Spooner <strong>Lake</strong> Junction, <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong> . FHP Report #SS11-09.U.S. Department of Agriculture ,<strong>Forest</strong> Service. <strong>Forest</strong> HealthProtection, Sonora, CA. 24pp.Elliott-Fisk, D.L. 1996. In: <strong>Lake</strong> <strong>Tahoe</strong> case study. Addendum to the Sierra NevadaEcosystem Project; f<strong>in</strong>al report to Congress. University of California, Centers forWater and Wildland Resources. Wildland Resources Center Report No. 36.Ferrell, George T. and Ralph C. Hall. 1975. Weather and tree growth associated withwhite fire mortality caused by fir engraver and roundheaded fir borer. USDA<strong>Forest</strong> Service Research Paper PSW-109.Ferrell, G.T., W.J. Otros<strong>in</strong>a, and C.J. Demars.1994. Predict<strong>in</strong>g susceptibility of white firdur<strong>in</strong>g a drought associated outbreak of the fir engraver, Scolytus ventralis, <strong>in</strong>California. Canadian Journal of <strong>Forest</strong> Research 24: 302–305.Ferrell, G.T. 1996. The <strong>in</strong>fluence of <strong>in</strong>sect pests and pathogens on Sierra <strong>Forest</strong>s. In:Sierra Nevada Ecosystem Project: F<strong>in</strong>al Report to Congress, vol. II, Assessmentsand Scientific Basis for Management Options. Univ. of California, Davis, WaterResources Center Report No. 37, pp. 1177–1192.Floyd, M. Lisa, Michael Clifford, Neil S. Cobb, Dust<strong>in</strong> Hanna, Robert Delph, PauletteFord, and Dave Turner. 2009. Relationship of stand characteristics to drought<strong>in</strong>ducedmortality <strong>in</strong> three Southwestern piñon–juniper woodlands. EcologicalApplications 19(5): 1223–1230.Frankl<strong>in</strong>, Jerry F., H.H. Shugart, and M.E. Harmon. 1987. Tree death as an ecologicalprocess. Bioscience 27: 259–288.


108Ganey, Joseph L. and Scott C. Vojta. 2011. Tree mortality <strong>in</strong> drought-stressed mixedconiferand ponderosa p<strong>in</strong>e forests, Arizona, USA. <strong>Forest</strong> Ecology andManagement 261: 162–168.Gao, Bo-Cai. 1996. NDWI: A normalized difference water <strong>in</strong>dex for remote sens<strong>in</strong>g ofvegetation liquid water <strong>from</strong> space. Remote Sens<strong>in</strong>g of Environment 58: 257-266.Gilbert, M., L. –M. Nageleisen, A. Frankl<strong>in</strong>, and J.-C. Grégoire. 2005. Post-stormsurveys reveal large-scale spatial patterns and <strong>in</strong>fluences of site factors, foreststructure and diversity <strong>in</strong> non-epidemic bark-beetle populations. LandscapeEcology 20: 35–49.Goheen, Ellen Michaels and Donald J. Goheen. 1989. Losses caused by Annosus rootdisease <strong>in</strong> Pacific Northwest forests. In Proceed<strong>in</strong>gs of the SymposiumonResearch and Managementof Annosus Root Disease (Heterobasidion annosum) <strong>in</strong>Western North America. April 18-21, 1989, Monterey, California. United StatesDepartment of Agriculture <strong>Forest</strong> Service, Pacific Southwest <strong>Forest</strong> and RangeExperiment Station. General Technical Report PSW-116. 177pp.Guarín, Alejandro and Alan H. Taylor. 2005. Drought triggered tree mortality <strong>in</strong> mixedconifer forests <strong>in</strong> Yosemite National Park, California, USA. <strong>Forest</strong> Ecology andManagement 218: 229–244.Hacke, Uwe G., John S. Sperry, William T. Pockman, Stephen D. Davis, and Kather<strong>in</strong>eA. McCulloh. 2001. Trends <strong>in</strong> wood density and structure are l<strong>in</strong>ked to preventionof xylem implosion by negative pressure. Oecologia 126: 457–461.


109Hansen, E., Bentz, B. and Turner, D. 2001. Temperature-based model for predict<strong>in</strong>gunivolt<strong>in</strong>e brood proportions <strong>in</strong> spruce beetle (Coleoptera : Scolytidae). CanadianEntomologist 133(6): 827-841.Hebertson, Elizabeth G. and Michael J. Jenk<strong>in</strong>s. 2008. Climate factors associated withhistoric Spruce Beetle (Coleoptera: Curculionidae) outbreaks <strong>in</strong> Utah andColorado. Environmental Entomology 37(2): 281-292.Herms, Daniel A. and William J. Mattson. 1992. The dilemma of plants: To grow ordefend. The Quarterly Review of Biology 67(3): 283-335.Hurteau, Matthew, Harold Zald, and Malcom North. 2007. Species-specific response toclimate reconstruction <strong>in</strong> upper-elevation mixed-conifer forests of the westernSierra Nevada, California. Canadian Journal of <strong>Forest</strong> Research 37: 1681-1691.James, J.C., J. Grace, and S.P. Hoad. 1994. Growth and photosynthesis of P<strong>in</strong>ussylvestris at its altitud<strong>in</strong>al limit <strong>in</strong> Scotland. Journal of Ecology 82: 297-306.K<strong>in</strong>g, Gary and Langche Zeng. 2001. Logistic regression <strong>in</strong> rare events data. PoliticalAnalysis 9: 137-163.Kubiske, Mark E. and Marc D. Abrams. 1994. Ecophysiological analysis of woodyspecies <strong>in</strong> contrast<strong>in</strong>g temperate communities dur<strong>in</strong>g wet and dry years.Oecologia 98: 303-312.L<strong>in</strong>es, Emily R., David A. Coomes, Drew W. Purves. 2010. Influences of forest structure,climate and species composition on tree mortality across the Eastern US. PLoSONE 5(10): e13212. doi:10.1371/journal.pone.0013212Lloyd, Andrea H. and Lisa J. Graumlich. 1997. Holocene dynamics of treel<strong>in</strong>e forests <strong>in</strong>the Sierra Nevada. Ecology 78(4): 1199-1210.


110Logan, Jesse A., Peter White, Barbara J. Bentz and James A. Powell. 1998. Modelanalysis of spatial patterns <strong>in</strong> Mounta<strong>in</strong> P<strong>in</strong>e Beetle outbreaks. TheoreticalPopulation Biology 53: 236-255.Logan, Jesse A. and Jame A. Powell. 2001. Ghost forests, global warm<strong>in</strong>g, and theMounta<strong>in</strong> P<strong>in</strong>e Beetle (Coleoptera: Scolytidae). American Entomologist 47(3):160-173.Macomber, S.A. and C.E. Woodcock. 1994. Mapp<strong>in</strong>g and monitor<strong>in</strong>g conifer mortalityus<strong>in</strong>g remote sens<strong>in</strong>g <strong>in</strong> the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>. Remote Sens<strong>in</strong>g of Environment50: 255–266.Maherali, H. 1999. Water relations of ponderosa p<strong>in</strong>e <strong>in</strong> contrast<strong>in</strong>g environments:implications for global climate change. Ph.D. Thesis, Univ. Ill<strong>in</strong>ois, Urbana, IL,146 p.Maherali, H. and E.H. DeLucia. 2000. Interactive effects of elevated CO 2 andtemperature on water transport <strong>in</strong> ponderosa p<strong>in</strong>e. American Journal of Botany87: 243-249.Maherali, H., W.T. Pockman, and R.B. Jackson. 2004. Adaptive variation <strong>in</strong> thevulnerability of woody plants to xylem cavitation. Ecology 85: 2184–2199.Maloney, Patricia E. and David M. Rizzo. 2002. Pathogens and <strong>in</strong>sects <strong>in</strong> a prist<strong>in</strong>e forestecosystem: the Sierra San Pedro Martir, Baja, Mexico. Canadian Journal of <strong>Forest</strong>Research 32: 448-457.Maloney, Patricia E., Detlev R. Vogler , Andrew J. Eckert, Camille E. Jensen, and DavidB. Neale. 2011. Population biology of sugar p<strong>in</strong>e (P<strong>in</strong>us lambertiana Dougl.) with


111reference to historical disturbances <strong>in</strong> the <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong>: Implications forrestoration. <strong>Forest</strong> Ecology and Management 262: 770–779.Manion, P.D., 1991. Tree Disease Concepts, 2nd ed.Prentice-Hall Inc., Upper SaddleRiver, NJ, 416 pp.McDowell, Nate, William T. Pockman, Craig D. Allen, David D. Breshears, Neil Cobb,Thomas Kolb, Jennifer Plaut, John Sperry, Adam West, David G. Williams, andEnrico A. Yepez. 2008. Mechanisms of plant survival and mortality dur<strong>in</strong>gdrought: why do some plants survive while others succumb to drought? NewPhytologist 178: 719–739.Miao, ShiLi, Chris B. Zou, and David D. Breshears. 2009. Vegetation responses toextreme hydrological events: Sequence matters. The American Naturalist 173 (1):113-118.Millar, Constance. I., and W. B. Woolfenden. 1999. The role of climate change <strong>in</strong><strong>in</strong>terpret<strong>in</strong>g historical variability. Ecological Applications 9: 1207–1216.Millar, Constance I., Robert D. Westfall, Diane L. Delany, John C. K<strong>in</strong>g, and LisaJ.Graumlich. 2004. Response of subalp<strong>in</strong>e conifers <strong>in</strong> the Sierra Nevada,California, U.S.A, to 20 th -Century warm<strong>in</strong>g and decadal climate variability.Arctic, Antarctic, and Alp<strong>in</strong>e Research 36(2): 181-200.Millar, Constance.I., Robert D. Westfall, and Diane L. Delany. 2007. Response of highelevationlimber p<strong>in</strong>e (P<strong>in</strong>us flexilis) to multiyear droughts and 20th-centurywarm<strong>in</strong>g, Sierra Nevada, California, USA. Canadian Journal of <strong>Forest</strong> Research37: 2508–2520.


112Millar, Constance I., Robert D. Westfall, Diane L. Delany, Matthew J. Bokach, Alan L.Fl<strong>in</strong>t, and Lorra<strong>in</strong>e E. Fl<strong>in</strong>t. <strong>2012.</strong> <strong>Forest</strong> mortality <strong>in</strong> high-elevation whitebarkp<strong>in</strong>e (P<strong>in</strong>us albicaulis) forests of eastern California, USA; <strong>in</strong>fluence ofenvironmental context, bark beetles, climatic water deficit, and warm<strong>in</strong>g.Canadian Journal of <strong>Forest</strong> Research 42: 749–765.M<strong>in</strong>nich, Richard A. 1984. Snow drift<strong>in</strong>g and timberl<strong>in</strong>e dynamics on Mount SanGorgonio, California. Arctic and Alp<strong>in</strong>e Research 16(4): 395-412.Negrón, José.F., Joel D. McMill<strong>in</strong>, John A. Anhold, and Dave Coulson. 2009. Barkbeetle-caused mortality <strong>in</strong> a drought affected ponderosa p<strong>in</strong>e landscape <strong>in</strong>Arizona, USA. <strong>Forest</strong> Ecology and Management 257: 1353–1362.Nelson, E.E. and Rona N. Sturrock, <strong>Forest</strong>ry Canada. 1993. Susceptibility of Westernconifers to Lam<strong>in</strong>ated Root Rot (Phell<strong>in</strong>us weirii) <strong>in</strong> Oregon and BritishColumbia field tests. Western Journal of Applied <strong>Forest</strong>ry 8(2): 67-70.Nelson, T.A., B. Boots, M.A. Wulder, and A.L. Carroll. 2007. Environmentalcharacteristics of mounta<strong>in</strong> p<strong>in</strong>e beetle <strong>in</strong>festation hot spots. BC Journal ofEcosystems and Management 8(1): 91–108.North, Malcolm, Matthew Hurteau, Robert Fiegener, and Michael Barbour. 2005.Influence of fire and El Niño on tree recruitment varies by species <strong>in</strong> Sierranmixed conifer. <strong>Forest</strong> Science 51(3): 187-197.Oberhuber, W. 2001. The role of climate <strong>in</strong> the mortality of Scots p<strong>in</strong>e (P<strong>in</strong>us sylvestrisL.) exposed to soil dryness. Dendrochronologia 19: 45-55.


113Ogle, Kiona, Thomas G. Whitham, and Neil S. Cobb. 2000. Tree-r<strong>in</strong>g variation <strong>in</strong> p<strong>in</strong>yonpredicts likelihood of death follow<strong>in</strong>g severe drought. Ecology 81(11): 3237–3243.Pedersen, Brian S. 1998. The role of stress <strong>in</strong> the mortality of Midwestern oaks as<strong>in</strong>dicated by growth prior to death. Ecology 79(1): 79-93.Peet, Robert K. and Norman L. Christensen. 1987. Competition and tree death.BioScience 37(8): 586-595.Peltonen, M. 1999. W<strong>in</strong>dthrows and dead-stand<strong>in</strong>g trees as bark beetle breed<strong>in</strong>g materialat forest-clearcut edge. Scand<strong>in</strong>avian Journal of <strong>Forest</strong> Research 14: 505-511.Powers, Jennifer Sarah, Phillip Soll<strong>in</strong>s, Mark E. Harmon and Julia A. Jones. 1999. Plantpest<strong>in</strong>teractions <strong>in</strong> time and space: A Douglas-fir bark beetle outbreak as a casestudy. Landscape Ecology 14: 105–120.Raffa, K.F. and A. A. Berryman. 1983. The role of host plant resistance <strong>in</strong> thecolonization behavior and ecology of bark beetles (Coleoptera: Scolytidae).Ecological Monographs 53(1): 27-49.Rice, K.J., S.L. Matzner, W. Byer, and J.R. Brown. 2004. Patterns of tree dieback <strong>in</strong>Queensland, Australia: the importance of drought stress and the role of resistanceto cavitation. Oecologia 139: 190–198.Safranyik, L., D.M. Shrimpton, and H.S. Whitney. 1974. Management of lodgepole p<strong>in</strong>eto reduce losses <strong>from</strong> the mounta<strong>in</strong> p<strong>in</strong>e beetle. Natural Resources Canada,Canadian <strong>Forest</strong> Service, Victoria, B.C. <strong>Forest</strong>ry Technical Report No. 1.Safranyik, L., R. Silversides, L. McMullen, and D. L<strong>in</strong>ton. 1989. An empirical approachto model<strong>in</strong>g the local dispersal of the mounta<strong>in</strong> p<strong>in</strong>e beetle (Dendroctonus


114ponderosae Hopk.) (Col., Scolytidae) <strong>in</strong> relation to sources of attraction, w<strong>in</strong>ddirection and speed. Journal of Applied Entomology 108:498–511.Safranyik, L. and A.L. Carroll. 2006. The biology and epidemiology of the mounta<strong>in</strong> p<strong>in</strong>ebeetle <strong>in</strong> lodgepole p<strong>in</strong>e forests. Pages 3-66 <strong>in</strong> L. Safranyik and W.R. Wilson,editors. The mounta<strong>in</strong> p<strong>in</strong>e beetle: a synthesis of biology, management, and impactson lodgepole p<strong>in</strong>e. Natural Resources Canada, Canadian <strong>Forest</strong> Service, Pacific<strong>Forest</strong>ry Centre, Victoria, British Columbia. 304p.Schowalter, T.D. and G.M. Filip. 1993. Beetle-pathogen <strong>in</strong>teractions <strong>in</strong> conifer forests.Academic Press, San Diego.Schroeder, L. Mart<strong>in</strong> and Åke L<strong>in</strong>delöw. 2002. Attacks on liv<strong>in</strong>g spruce trees by the barkbeetle Ips typographus (Col. Scolytidae) follow<strong>in</strong>g a storm-fell<strong>in</strong>g: a comparisonbetween stands with and without removal of w<strong>in</strong>d-felled trees. Agricultural and<strong>Forest</strong> Entomology 4(1): 47–56.Schütt, Peter and Ellis B. Cowl<strong>in</strong>g. 1985. Waldsterben, a General Decl<strong>in</strong>e of <strong>Forest</strong>s <strong>in</strong>Central Europe: Symptoms, Development, and Possible Causes. Plant Disease 69:548-558.Slaughter, G.W. and J.R. Parmeter. 1989. Annosus root disease <strong>in</strong> true firs <strong>in</strong> northernand central California national forests. In Proceed<strong>in</strong>gs of the SymposiumonResearch and Managementof Annosus Root Disease (Heterobasidion annosum) <strong>in</strong>Western North America. April 18-21, 1989, Monterey, California. United StatesDepartment of Agriculture <strong>Forest</strong> Service, Pacific Southwest <strong>Forest</strong> and RangeExperiment Station. General Technical Report PSW-116.177 pp.


115SNEP [Sierra Nevada Ecosystem Project]. 1996. <strong>Lake</strong> <strong>Tahoe</strong> case study. Pages 217–276<strong>in</strong> Sierra Nevada Ecosystem Project: F<strong>in</strong>al report to Congress, Addendum. Davis:University of California, Centers for Water and Wildland Resources, 1996.Sparks, Jed P. and R. 1999. Alan Black. Regulation of water loss <strong>in</strong> populations ofPopulus trichocarpa: the role of stomatal control <strong>in</strong> prevent<strong>in</strong>g xylem cavitation.Tree Physiology 19: 453-459.Stamp, Nancy. 2003. Out of the quagmire of plant defense hypotheses. The QuarterlyReview of Biology 78(1): 23-55.Stephenson, Nathan L., Phillip J. van Mantgem, Andrew G. Bunn, Howard Bruner, MarkE. Harmon, Kari B. O’Connell, Dean L. Urban, and Jerry E. Frankl<strong>in</strong>. 2011.Causes and implications of the correlation between forest productivity and treemortality rates. Ecological Monographs 81(4): 527–555.Suarez, Maria Laura, Luciana Ghermandi, and Thomas Kitzberger. 2004. Factorspredispos<strong>in</strong>g episodic drought-<strong>in</strong>duced tree mortality <strong>in</strong> Nothofagus– site, climaticsensitivity and growth trends. Journal of Ecology 92: 954–966.Tranquill<strong>in</strong>i,W. 1979. Physiological ecology of the alp<strong>in</strong>e timberl<strong>in</strong>e: Tree existence athigh altitudes with special reference to the European Alps. Ecological Studies 31.Berl<strong>in</strong> and New York:S pr<strong>in</strong>ger-Verlag 137 pp.Urban, Dean, Carol Miller, Patrick N. Halp<strong>in</strong>, and Nathan L. Stephenson. 2000. <strong>Forest</strong>gradient response <strong>in</strong> Sierran landscapes: the physical template. LandscapeEcology 15: 603–620.


116USDA <strong>Forest</strong> Service. 1981. CALVEG: A Classification of California Vegetation.Pacific Southwest Region, Regional Ecology Group, San Francisco CA. 168 pp.2009 Update.USDA Natural Resources Conservation Service. 2007. Soil Survey of the <strong>Tahoe</strong> Bas<strong>in</strong>Area, California and Nevada.van Mantgem, P.J., and Nathan L. Stephenson. 2007. Apparent climatically <strong>in</strong>duced<strong>in</strong>crease of tree mortality rates <strong>in</strong> a temperate forest. Ecology Letters 10: 909–916.van Mantgem, P.J., N.L. Stephenson, J.C.Byrne, L.D. Daniels, J.F. Frankl<strong>in</strong>, P.Z. Fulé,M.E. Harmon, A.J. Larson, J.M. Smith, A.H. Taylor, T.T. Veblen. 2009.Widespread <strong>in</strong>crease of tree mortality rates <strong>in</strong> the western United States. Science323: 521–524.Vilà-Cabrera, Albert, Jordi Martínez-Vilalta, Jordi Vayreda, and Javier Retana. 2011.Structural and climatic determ<strong>in</strong>ants of demographic rates of Scots p<strong>in</strong>e forestsacross the Iberian Pen<strong>in</strong>sula. Ecological Applications 21: 1162–1172.Villalba, Ricardo, Thomas T. Veblen and John Ogden. 1994. Climatic Influences on theGrowth of Subalp<strong>in</strong>e Trees <strong>in</strong> the Colorado Front Range. Ecology 75(5): 1450-1462.Villalba, Ricardo and Thomas T. Veblen. 1998. Influences of large-scale climaticvariability on episodic tree mortality <strong>in</strong> northern Patagonia. Ecology 79(8): 2624-2640Vogelmann , James E., Brian Tolk, and Zhiliang Zhu. 2009. Monitor<strong>in</strong>g forest changes <strong>in</strong>the southwestern United States us<strong>in</strong>g multitemporal Landsat data. RemoteSens<strong>in</strong>g of Environment 113: 1739–1748.


117Vygodskaya, N.N., E.D. Schulze, N. M. Tchebakova. O. Karpachevski, D. Kozlov, K. N.Sidorov, M. I. Panfyorov, M. A. Abrazko, E. S. Shaposhnikov, O. N. Solnzeva, T.Y. M<strong>in</strong>aeva, A. S. Jeltuch<strong>in</strong>, C. Wirth and A. V. Pugachevskii.2002. Climaticcontrol of stand th<strong>in</strong>n<strong>in</strong>g <strong>in</strong> unmanaged spruce forests of the southern taiga <strong>in</strong>European Russia. Tellus 54: 443–461.Wallner, W. E. 1987. Factors affect<strong>in</strong>g <strong>in</strong>sect population dynamics: Differences betweenoutbreak and non-outbreak species. Annual Review of Entomology 32: 317-340.Wang, T., Hamann, A., Spittlehouse, D., and Murdock, T. N. <strong>2012.</strong> ClimateWNA - High-Resolution Spatial Climate Data for Western North America. Journal of AppliedMeteorology and Climatology 61: 16-29.Werner, R.A. and E.H. Hosten. 1985. Factors <strong>in</strong>fluenc<strong>in</strong>g generation times of sprucebeetles <strong>in</strong> Alaska. Canadian Journal of <strong>Forest</strong> Research 15: 438-443.Worrall, J. J., L. Egeland, T. Eager, R.A. Mask, E.W. Johnson, P.A. Kemp, and W.D.Shepperd. 2008. Rapid mortality of Populus tremuloides <strong>in</strong> southwesternColorado, USA. <strong>Forest</strong> Ecology and Management 255: 686-696.Zwiazek, J. J., and T. J. Blake. 1989. Effects of precondition<strong>in</strong>g on subsequent waterrelations, stomatal sensitivity, and photosynthesis <strong>in</strong> osmotically stressed blackspruce. Canadian Journal of <strong>Forest</strong> Research 67: 2240–2244.


118TABLE 2-1. Relationships between current and lagged year modeled snow deficits [asratio of current year snow/30-year norm (1980-2009) by forest type. Derivation of lags isshown below, us<strong>in</strong>g 1985 as an example year. Current year is def<strong>in</strong>ed a most currentw<strong>in</strong>tertime precipitation. Positive relationships <strong>in</strong>dicate <strong>in</strong>creased moisture is associatedwith <strong>in</strong>creased mortality. Negative relationships <strong>in</strong>dicate decreased moisture associatedwith <strong>in</strong>creased mortality.Derivation of Lag Timel<strong>in</strong>es (example=1985)Lag Snow Accumulation Timel<strong>in</strong>e Snow Deficit Variable YearCurrent Aug 1984-July 1985 1985Lag 1 Aug1983-July1984 1984Lag 2 Aug 1982-July1983 1983Lag3 Aug 1981-July 1982 1982Lag 4 Aug1980 - July1981 1981Lag 5 Aug1979-July1980 1980Regression coefficients <strong>from</strong> Annual <strong>Mortality</strong> ~ Lagged Snow Deficits<strong>in</strong> Five <strong>Forest</strong> Types<strong>Forest</strong>Type Coefficients Standard Error t-value Pr(>|t| SignificanceJP (Intercept) 4.788 5.873 0.815 0.426Current -4.160 2.431 -1.711 0.104Lag1 6.164 2.603 2.368 0.029 *Lag2 5.651 2.107 2.682 0.015 *Lag3 -0.461 2.001 -0.231 0.820Lag4 -0.043 1.929 -0.022 0.983Lag5 1.064 1.963 0.542 0.594


119<strong>Forest</strong>Type Coefficients Standard Error t-value Pr(>|t| SignificanceLP (Intercept) -11.857 10.806 -1.097 0.287Current 7.058 4.474 1.578 0.132Lag1 10.463 4.790 2.184 0.042 *Lag2 4.742 3.877 1.223 0.237Lag3 0.291 3.681 0.079 0.938Lag4 10.542 3.548 2.971 0.008 **Lag5 -4.159 3.611 -1.152 0.264MF (Intercept) 12.444 9.577 1.299 0.210Current 0.293 3.965 0.074 0.942Lag1 0.051 4.246 0.012 0.991Lag2 2.063 3.436 0.600 0.556Lag3 -1.791 3.263 -0.549 0.590Lag4 -2.915 3.145 -0.927 0.366Lag5 2.578 3.200 0.805 0.431RF (Intercept) -5.087 11.577 -0.439 0.666Current 3.234 4.793 0.675 0.508Lag1 13.059 5.132 2.545 0.020 *Lag2 1.764 4.154 0.425 0.676Lag3 1.045 3.944 0.265 0.794Lag4 7.748 3.802 2.038 0.057 .Lag5 -3.392 3.869 -0.877 0.392WF (Intercept) 6.209 11.738 0.529 0.603Current 5.275 4.860 1.086 0.292Lag1 1.618 5.203 0.311 0.759Lag2 -0.337 4.212 -0.080 0.937Lag3 -4.052 3.999 -1.013 0.324Lag4 3.005 3.855 0.780 0.446Lag5 1.570 3.922 0.400 0.694


120TABLE 2-2. L<strong>in</strong>ear regression analysis results for Annual <strong>Mortality</strong> ~ Climatic Period <strong>in</strong>each of five forest types (JP, LP, MF, RF, WF) with significance at the p < 0.05 level.<strong>Forest</strong>TypeClimaticPeriodStandardErrort-valuePr(>|t| SignificanceJP d1(<strong>in</strong>tercept) 15.490 1.696 9.136 0.000 ***JP d2 -6.008 2.590 -2.320 0.031 *JP w1 -1.189 3.791 -0.314 0.757JP w2 -2.408 2.937 -0.820 0.422JP w3 -4.168 2.734 -1.524 0.143LP d1(<strong>in</strong>tercept) 16.336 3.162 5.166 0.000 ***LP d2 -4.324 4.830 -0.895 0.381LP w1 11.090 7.071 1.569 0.132LP w2 5.103 5.477 0.932 0.363LP w3 -5.600 5.099 -1.098 0.285MF d1(<strong>in</strong>tercept) 18.903 1.547 12.221 0.000 ***MF d2 -10.365 2.363 -4.387 0.000 ***MF w1 -7.180 3.459 -2.076 0.051 .MF w2 -6.235 2.679 -2.327 0.031 *MF w3 -11.167 2.494 -4.477 0.000 ***RF d1(<strong>in</strong>tercept) 19.883 3.206 6.202 0.000 ***RF d2 -6.478 4.897 -1.323 0.201RF w1 9.059 7.168 1.264 0.221RF w2 -2.548 5.552 -0.459 0.651RF w3 -6.253 5.169 -1.210 0.241WF d1(<strong>in</strong>tercept) 17.370 2.709 6.413 0.000 ***WF d2 -7.119 4.137 -1.721 0.101WF w1 -5.434 6.057 -0.897 0.380WF w2 -7.558 4.691 -1.611 0.123WF w3 -6.398 4.367 -1.465 0.158


121a.b.c.d.FIGURE 2-1. Ecoregions (a) <strong>in</strong> the LTB with ecoregional differences <strong>in</strong> (b) averageannual precipitation (<strong>in</strong>); (c) average maximum temperature ( o F) ; (d) average m<strong>in</strong>imumtemperature ( o F), derived <strong>from</strong> PRISM 1971-2000 norms.


122abFIGURE 2-2a,b. Average w<strong>in</strong>ter m<strong>in</strong>imum temperature (a) and number of frost free days(b) <strong>in</strong> five climatic periods of study time series


123FIGURE 2-3. Solar radiation (WH/m 2 ) calculated for August 1, 2010. Solar radiationcomb<strong>in</strong>es slope and aspect to measure both direct and diffuse radiation over a 30m x 30mpixel. Higher values are associated with south-fac<strong>in</strong>g and lower values with north-fac<strong>in</strong>gaspects.


FIGURE 2-4. Elevation (m) derived <strong>from</strong> a 30m Digital Elevation Model.124


125FIGURE 2-5. Boxplots of annual mortality by forest types <strong>in</strong> five climatic periods ofstudy [d1:1987-1995, d2:1999-2006, w1:1985-1987, w2: 1995-1999, w3:2006-20010].


126FIGURE 2-6. Regression tree of predicted mortality levels by forest type and climaticperiod. Regression trees are comprised of a series of b<strong>in</strong>ary splits created throughrecursive partition<strong>in</strong>g with splitt<strong>in</strong>g criterion based on overall deviance reduction.Predicted values of annual mortality are shown at the term<strong>in</strong>us of each node for thatgroup. For example, the predicted annual mortality value for JP, MF, and WF dur<strong>in</strong>geither d2 or w3 is 9.69.


127FIGURE 2-7. Effect plots for Annual <strong>Mortality</strong> ~ Climatic Period (ClimPd: d1, d2, w1,w2, w3) <strong>in</strong> each of the five forest types.


128<strong>Forest</strong> Type Multiple R 2 Adjusted R 2 F DF P-valueJP 0.272 0.097 1.554 24 0.068MF 0.258 0.0799 1.448 24 0.105RF 0.185 -0.010 0.948 24 0.539FIGURE 2-8. Effect plots for regression analysis of Annual <strong>Mortality</strong> ~ EcoRegion *Climatic Period <strong>in</strong> JP, MF, and RF forests. Red l<strong>in</strong>es <strong>in</strong>dicate the upper and lower 95 th %confidence <strong>in</strong>tervals around regression coefficients.


129a: JPb: MFc: RFFIGURE 2-9a-c. Levels and patterns of JP(a), MF (b), and RF(c) forest mortality byecoregion (eh, ej, ek, el, et) over the 25 year time period.


130FIGURE 2-10. Effect of a 1,000 WH/m 2 <strong>in</strong>crease <strong>in</strong> solar radiation on mortality for fiveforest types across 25-year time period as odds ratios with 95% confidence <strong>in</strong>tervals.N=odds ratios with confidence <strong>in</strong>tervals that overlap 1 (CI shown) or years when solarradiation did not appear <strong>in</strong> f<strong>in</strong>al model (CI not shown). U= odds ratios with confidence<strong>in</strong>tervals that do not overlap 1. Results <strong>from</strong> annual models occur<strong>in</strong>g dur<strong>in</strong>g dry periodsare <strong>in</strong> orange. Results <strong>from</strong> annual models occur<strong>in</strong>g <strong>in</strong> dry periods are <strong>in</strong> green. Oddsratios greater than one <strong>in</strong>dicate <strong>in</strong>crease <strong>in</strong> probability of mortality associated with<strong>in</strong>creases <strong>in</strong> south-fac<strong>in</strong>g slopes. Odds ratios less than one <strong>in</strong>dicate probability ofmortality associated with north-fac<strong>in</strong>g slopes.


131FIGURE 2-11. Location of <strong>in</strong>creased risk of mortality by aspect derived <strong>from</strong> solarradiation for five forest types across 25-year time period (Not Significant: confidence<strong>in</strong>terval of regression coefficient and result<strong>in</strong>g odds ratio overlapped 0 or 1; NoAssociation: no association between probability of mortality and elevation).


132FIGURE 2-12. Effect of a 500m <strong>in</strong>crease <strong>in</strong> elevation on mortality for five forest typesacross 25-year time period as odds ratios with 95% confidence <strong>in</strong>tervals. N=odds ratioswith confidence <strong>in</strong>tervals that overlap 1 (CI shown) or years when elevation did notappear <strong>in</strong> f<strong>in</strong>al model (CI not shown). U= odds ratios with confidence <strong>in</strong>tervals that donot overlap 1. Results <strong>from</strong> annual models occur<strong>in</strong>g dur<strong>in</strong>g dry periods are <strong>in</strong> orange.Results <strong>from</strong> annual models occur<strong>in</strong>g <strong>in</strong> wet periods are <strong>in</strong> green. Odds ratios over one<strong>in</strong>dicate <strong>in</strong>crease <strong>in</strong> probability of mortality associated with <strong>in</strong>creases <strong>in</strong> elevation. Oddsratios less than one <strong>in</strong>dicate probability of mortality associated with decreases <strong>in</strong>elevation.


133FIGURE 2-13. Location of <strong>in</strong>creased risk of mortality by elevation for five forest typesacross 25-year time period (Upper: Upper elevation, Lower: Lower elevation, No Effect:no association between probability of mortality and elevation, Not Significant:confidence <strong>in</strong>terval of regression coefficient and result<strong>in</strong>g odds ratio overlapped 0 or 1).


134APPENDIX A. 25 years of annual mortality (Fall Year One – Fall Year Two) <strong>from</strong> 1985-2010 <strong>in</strong> the LTB. <strong>Mortality</strong> maps are derived by subtract<strong>in</strong>g year one NDWI <strong>from</strong> yeartwo NDWI to create cont<strong>in</strong>uous maps of dNDWI show<strong>in</strong>g forested areas <strong>in</strong> vary<strong>in</strong>gstages of health, <strong>from</strong> canopy dieback to vegetation recovery.1986-871985-86 1987-88 1988-891989-901990-91 1991-92 1992-93


1351993-94 1994-95 1995-96 1996-971997-981998-991999-20002000-01


1362001-02 2002-03 2003-04 2004-052005-06 2006-07 2007-08 2008-092009-10


137THESIS SUMMARY<strong>Forest</strong> mortality is an issue of global concern, with outbreaks of bark beetle-mediatedmortality more severe, widespread, and synchronized than evidenced previously. Whilethe association between <strong>in</strong>creas<strong>in</strong>g moisture stress, bark beetle dynamics, and forestmortality has been found <strong>in</strong> numerous studies, it is widely reported that <strong>in</strong>creased standdensity, due to land-use and land-management policies, is as a predispos<strong>in</strong>g factor <strong>in</strong>forest mortality. Global <strong>in</strong>creases <strong>in</strong> forest mortality have also been widely attributed tochang<strong>in</strong>g climate and altered disturbance regimes, yet there is limited understand<strong>in</strong>g ofhow forest mortality trends and spatial patterns vary with forest type, climate andtopographic variability. Though the <strong>in</strong>fluence of density and environmental variables onforest mortality dur<strong>in</strong>g s<strong>in</strong>gle, and often epidemic, episodes of mortality has beenelucidated (though with conflict<strong>in</strong>g results), a more comprehensive understand<strong>in</strong>gencompass<strong>in</strong>g various climatic periods and forest types is lack<strong>in</strong>g.We <strong>in</strong>vestigated how the relationship between stand-level mortality and stand densityhas varied over five forest types and five climatic periods <strong>in</strong> the mixed conifer forests ofthe <strong>Lake</strong> <strong>Tahoe</strong> Bas<strong>in</strong> (LTB) <strong>in</strong> the central Sierra Nevada. Our analyses used LandsatTM data <strong>from</strong> 1985-2010 to derive annual mortality maps and an annual stock<strong>in</strong>g <strong>in</strong>dex.We found that the strength of density-dependent mortality was variable <strong>in</strong> the LTB, withno forest show<strong>in</strong>g a ubiquitous relationship between mortality and density. However, <strong>in</strong>more droughty climatic periods and <strong>in</strong> p<strong>in</strong>e-dom<strong>in</strong>ated forests, <strong>in</strong>creases <strong>in</strong> density wereweakly associated with <strong>in</strong>creased probability of mortality. In middle- and upper-elevationforests, the association between mortality and density was variable. Increased density


138was associated with <strong>in</strong>creased mortality risk <strong>in</strong> dry periods, with this relationshipreversed <strong>in</strong> wet periods.Us<strong>in</strong>g annual mortality maps derived <strong>from</strong> remote sens<strong>in</strong>g analysis, we quantifiedthe relationships between mortality and environmental gradients def<strong>in</strong>ed by elevation andsolar radiation. Our results suggest that <strong>in</strong>creases <strong>in</strong> forest mortality are driven more byunique sequences of extreme climatic events, rather than associated solely with droughtperiods. <strong>Mortality</strong> is highest <strong>in</strong> upper-elevation forests and specifically when droughtfollows periods of heavy and prolonged snowfall. P<strong>in</strong>e-dom<strong>in</strong>ated forests were lesssensitive to drought. Throughout the time series for almost all forests, mortality risk wasgreater on north-fac<strong>in</strong>g slopes. Relationships between mortality and elevation werepredictable for lower-elevation forests: with decreases <strong>in</strong> elevation were consistentlyassociated with elevated mortality risk. For middle-and upper elevation forest types,<strong>in</strong>creased mortality risk is associated with <strong>in</strong>creases <strong>in</strong> elevation dur<strong>in</strong>g wet periods anddecreases <strong>in</strong> elevation dur<strong>in</strong>g dry periods.Our results provide boundary conditions for the widely-accepted hypothesis thatdensity-dependent mortality leads to heightened mortality levels. Patterns revealed <strong>in</strong> ouranalysis between mortality and environmental gradients illustrate that mortality, ratherthan an additive comb<strong>in</strong>ation of species traits, environmental position, and climate, is anon-l<strong>in</strong>ear response to factors operat<strong>in</strong>g at multiple scales. More cross-scale studiesl<strong>in</strong>k<strong>in</strong>g ecophysiological mechanisms of tree death, forest pathology, and landscapedynamics of mortality spread are needed to improve our understand<strong>in</strong>g of the relative roleof endogenous and exogenous factors for <strong>in</strong>fluenc<strong>in</strong>g variable mortality patterns acrossthe landscape.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!