wculturalsystemand 498 bf/ac/yrunder clearcuttingThis 20% reduction approximates the long-termreduction m growth,and thusfutureharvestlevels, that may resultfrom suchan alternative silviculturalsystem. Reduction in MAI is about 20%, but the reductionin yield for the first rotation may be somewhat greater. While standingvolumesat the end of rotation are comparablefor the two systems(Table3), thisis primarilydue to the large volumecontainedin the overstorytrees(Figure2). If the alternativesystemis continuedbeyondthe first rotation, some of the overstory trees and some of the mature under- storytrees,will needto be retainedon site for coarsewoody debrisand second-rotationgreentree reserves. For our example,PROGNOSISpredictedthat 16 of the original20 overstory trees remainedafter 100 years. Fourof the originalreservetreeseither died or were removed in the commer- SUMMARY We believethat the developmentof alternative silvicultural systems, including those associatedwith "new forestry"canyield considerable benefits. Suchsystemscan combinethe in- herent advantagesof even-agedand uneven-aged management, e.g., greater structural diversity than strictlyeven-agedstands,andreduced management complexity compared with classic uneven-aged systems.Silvicultural prescriptions must, of course,be executedproperly to capture potential benefits and minimize potential risks. Green tree retention carries with it risks and tradeoffs. For example, adverse effects related to windfirmness and future stand health maymakethe systemillustratedin our discussion inappropriate in somestands. The associated reductionsin growth and yield are not trivial.Actualreductionsin growth, yield, and value will dependon suchfactorsas the level of residual growing stock and species mix. In our example,currentand first cialthinning.We assumedthatfour of the 16 remainingwould be killedand left standing as snags and future downedwoody material.Four others, along with 16 mature understory trees,were designatedfor "green re- rotation harvest levels were about one- tention" storycompetitionmay furtherreduce for the next rotation. There- third lower than in the clearcutting system.Smallertree size due to over- fore, about 13,700 bf/ac was retained the value of future harvests. on site,representingnearly33%of the approximation,our analysissuggests that this type of alternativesilvicultural systemmay result in a 20% re- standingvolumeat theendof thefirst rotation (Table 3). As a first Site Index and Height Growth Curves for Ponderosa Pine, Western Larch, LodgepolePine, and Douglas-Fir in ductionanlong-termgrowth, and thus future harvestlevels,comparedto traditionaleven-agedsystems. An importantpoint to emphasizeis that many of the potentialdisadvan- tagesof suchalternativesilvicultural systemsare only negativein relation to some other system, e.g., clearcutting or single-treeselection.As prioritieschangesodo the objectivesof forest management;when management objectives andconstraints do not allow eitherclassic even-agedor uneven-aged systems,a thoughtfullydesignedand carefullyexecutedalternativesilvicultural systemmay be a viableoption. [] LITERATURE CITED DAmœL,T.W., J.A. HœLMS,A•r• F.S. B•CER. 1979. Principlesof silviculture.Ed. 2. McGraw-Hill, New York. 500 p. F•CLIN, J.F.1989.Towarda new forestry.Am. For. 95:37-44. LEa•c, W.B., a•r• J.H. Gorrsacr,•a. 1985. New approachesto uneven-agedmanagementin New England.North. J. Appl. For. 2:28-31. MAttHews, J.D. 1989.Silviculturalsystems.Oxford Science Publications, Clarendon Press, Oxford, 284 p. ORtaNS,G.H. 1990."New Forestry"and the OldGrowth Forests of Northwestern North Amer- ica. A conversationwith Jerry F. Franklin. Northwest Environ. J. 6:445-461. SaLWaSS•R, H. 1990. Gaining perspective:Forestryfor the future. J. For. 88(11):32-38. SMtTH,D.M. 1986.The practiceof silviculture. Ed. 8. Wiley, New York. 527 p. WYCOFF, W.R., N.L. CROOKSTO•, a•r• A.R. STAG•.1982.User'sguideto the standprognosis model. Gen. Tech. Rep. INT-133. 112 p. tendency tounderpredict heights atgreater ages.New siteindexandheightgrowth curves arepresented for eachspecies. West.J. Appl. For. 7(1):9-14. Siteindex anddominant height growth curvesare commonlyused to identify the potential height-growth patternthat individualsof a givenspeciesmightbe expectedto achievein a particularplace.This potentialheight growthcurvemay thenbe usedas the sitequality"driver" in a yield prediction system.Identificationof the appropriate potential curve is essential for reliable predictions of future growthand yield by suchsystems. Traditionalsitecurve systemssumKelsey S. Milner, School of Forestry,Universityof Montana, marize the relationships between Missoula, Montana, 59812. height,age (eitherbreastheightor total), andheightat someindexage(SI), for somespecifiedset of treesbelonging to a defined population. For a ABSTRACT.Heightgrowthpatterns from thatdiffered significantly fromthose in the given species,the summarizationwill patterns several published sitecurvesytems forpon- data.Themagnitude ofthedeviations often showdifferentheight-growth derosa pine(Pinusponderosa),Douglas- variedby levelof siteindex;thisappeared dependingon the populationdefinitions,samplingprocedures,and confir (Pseudotsuga menziesiivarglauca), to berelatedto differences in therangeof western larch (Larix occidentalis) and sitequalitiessampled. Thosecurvescon- structionmethodologyused.Any one lodgepole pine(Pinuscontortavarlatifostructed fromstemanalysis datafromgeolia) werecompared to stemanalysisdata graphically similarpopulations compared 1Fundingandfieldsupportforthisstudy fromwestern Montana.Mostof thepub- mostclosely.Curvesconstructed using were providedby ChampionInternational lishedcurveshadheightgrowthpatterns guidecurvetechniques showed a consistent Corporation,Milltown, MT. Western Montana WJAF7(1)1992 9 of thesefactorscouldcausea parhcular sitecurvesystemto be suboptimal in estimatingpotentialheight-growth patternsfor a particularsite.In recog- wassampledat 31 locations,Douglasfir at 46, western larch at 37, and derestimated by a half-year'sgrowth. The effectof 1 pointin 11 was consid- ered trivial and no correction was lodgepolepine at 39. The meansand rangesof siteand standcharacteristics made to the data. nition of this fact, there has been confor the sampleare givenin Table1. Evaluationof ExistingSiteCurves Site trees, which were selected in a tinual regional activity aimed at asFor each species the observed sessingthe appropriateness of partic- fashionsimilar to that describedby ular site curve systemsto a specific Monserud (1984), had the following height:BHagepatternsfor a givensite indexlevel were comparedwith those area, and at developingnew curves characteristics:(1) dominant crown predicted by various published site where necessary (Monserud 1984, classin the senseof receivinglight curve systems.The publishedcurves from above and from all sides (thus, VanderPloegand Moore 1989,Means and Sabin1989).In additionto providtrees growing in openingsin cutover comparedto the data were as follows ing a closermatchto populations,the or unevenagedstandswere potential ponderosapine--Tesch 1980, Lynch 1958asadjustedby Brickell1970,Sumnew curvesoften utilize methodology candidates), (2) no substantial top and data that provide improvedestidamage,(3) healthycrownsof at least merfield1980,and Barrett1978;Dougmatesof potentialheightgrowthpat40% of totalheight,(4) no substantial las-fir--Monserud1984,Cochran1979, terns. Summerfield 1980,and VanderPloeg periodsof suppression,basedon inDue to its ubiquitousdistribution crementcorestaken at breastheight 1989; westernlarch--Schmidt 1976; and lodgepole pine--Alexander1967. andsubstantial economic value,Rocky (BH), and (5)potentiallyrepeatable on To compareheightgrowth patterns Mountain Douglas-fir (Pseudotsuga the site. That is, wolf trees or trees in the data to thosein the published menziesii var. glauca)hasreceivedmost growingon microsites obviouslydifof the attention in such research in the ferent from the general conditions curves, a site index had to be obtained for eachsampletree that was consisInland Empire.Thereis little informa- were avoided. tionon the appropriateness of existing Three trees on roughly a half-acre tent with the variousindexagesused in thepublishedsitecurves.The study site curve systemsfor westernlarch area were selected and felled for meadata were appropriatefor the BH-age surement of "actual" site index (de(Larixoccidentalis), ponderosapine (Pi50-yrbasecurvesof Monserud(1984), nus ponderosa), or lodgepolepine (Pifined to be the total height correCochran (1979), and Summerfield nuscontorta var. latifolia).Thepurposes spondingto a ring countof 50 at 4.5 (1980).Forthosecurvesindexedby toof this paper are to compareseveral ft). Becauseof time and moneycontal age, 50-yrbase(Tesch1980,Lynch sitecurvesystemsfor thesespeciesto straintsit was assumedthat the pat1958,and Schmidt1976),a compatible data collected in western Montana, tern of height-growthfor the tree of value was estimated from the data via median site index, of the three felled, and to provide a new set of height growthand siteindexcurvesfor each would representthe averagepotential linear interpolation between the for that location.Thisassumption ap- species. peared reasonableat six sites where METHODS stem analysiswas performedon all three site trees. Data Collection The tree of median site index was Thesedatawere collectedas part of a generalinvestigation into sourcesof variation in dominant tree height growth patterns (Milner 1987). The study area was westernMontana, definedastheareaboundedby Missoula in the south,Libby to the north, the Montana/Idaho border on the west, and Helena to the east. Sites were se- lectedaccordingto levelsof precipitation, temperature, solar insolation, andthe presenceof suitablesitetrees. Since values for the stratification variables were not available at the time of field sampling,slope, aspect,elevation, and geographiclocation were usedasinitialproxies.Ponderosapine sectioned at ground level, breast height,and at intervalsof one-tenthof treeheightabove4.5 ft, for a totalof 12 sectionsper tree. Annual rings were countedat the top of eachsection.The ring count at each sectionwas subtractedfrom that at breastheight to obtain the BH age of the tree correspondingto the heightof that section above4.5 ft. Assumingthat, onthe average, sectioncuts (includingthat at BH) were made at internode mid- points (see discussionby Dyer and Bailey1987),the resultingheightdata are unbiasedwith respectto "true" BH age (ring count--0.5 yr), for all pointsexceptthe tip, whichwill be un- Table1. Meansand rangesof standand sitecharacteristics, by species,for the westernMontana site index studysites. Species Ponderosa pine Number of sites Site Index (ft) Elevation (100 ft) Slope(degrees) Precipitation(in.) WJAF 7(1)1992 Lodgepole pine 31 46 37 39 59 (27-91) 46 (24-66) 71 (50-97) 42 (24-59) 60 (31-86) 48 (20-64) 14 (0-350) 27 (15-50) 10 Western larch 60 (41-84) 39 (28-57) (0-32) Aspect(degrees) Douglas-fir 20 17 (1-36) (1-33) (0-31) (0-352) (0-360) (36-350) 30 (16-55 ) 35 (15-60) 14 37 (15-60) height:totalagepairsbracketingtotal age50. However,for thosecurvesthat were total age-base100yr (Alexander 1967),no observationswere available Therefore,a regressionequationwas first developedfrom the published height growth curvesthat converted heightat totalage50 to heightat total age 100.This equationwas then used to convertthe SI base50-totalage observations in the data to a SI base 100- total age basis.For thosecurvesthat were base100-BHage (Barrett1978)a similarregressionequationwas developed from the publishedcurvesthat convertedthe SI base50-BH age valuesto SI base100-BHagevalues.Use of theseconversionequationsmeans that the published height growth curves are accessed according to heightat 50 yearsof age rather than 100.Thus, conclusions regardingade- quacyof thepublishedcurvesareonly relevant to applications consistent with this modification. Had observa- tionsof heightat age 100 been available,differentresultsmighthavebeen evident. The regression equations used were: Alexander(LPP) SI100 = - 2.99666 + 1.6129 * SI50 Barrett(PP) SI100 = -2.64174 + 0.6853 * SI50 For each(SI, Age, Height)tripletcontainedin the stemanalysisdata, a predicted height was obtainedfrom the •20- .•-5 o 10 20 30 40 50 60 70 80 90 AgeOotal orbreast height) on the site curve). To aid in detection of age-relatedbiasthe residualswere grouped by 5-yr age classesand the means, standard deviations, and standard errors calculated for each class. The residuals were also examined at variouslevelsof site index, sinceagerelated trends could vary by level of site index. 60 80 100 120 -'-Schmid{-WL +Alexander ] -•-J/•erud-.DF +Mo•serud• +Monserud-SAF Figure3. Mean heightresidualsfor Douglas-fir height growth curvesby Monserud (by habitat type series). appropriatepublishedH = f (SI, Age) equation. The residual, (observedpredictedheight),wasformedfor each SI and Age pair and plotted against age (total or breastheight depending 40 Total Age Breast Height Age -mLynch +Bo'reit -z-.S•erfield -•-Tesch ] Figure 1. Mean height residualsfor ponderosa pine height growth curves by Lynch,Barrett,Summerfield,and Tesch. 20 lO 20 30 40 50 60 70 80 90 Site Curve Construction Siteindex and heightgrowthequations were developedfrom the data usingproceduresdescribedby Heger (1968), Dahms (1973), and many others. The proceduresinvolve first fitting simple,linear modelsto the decadal height and site index data and then predictingthe regressioncoefficientsof theseequationsas functions of age. Final modelsare obtainedby substitutingthe latter equationsfor the appropriate coefficientsin the former. Though the resulting equations are over-parameterized, they fit the dataquitewell andwerejudgedto Figure4. Mean height residualsfor lodgepole pine and westernlatch height growth curvesby Alexanderand Schmidt,respectively. satisfactorily representthe patternsof height-growthin the sample.No attempt was made to refine or advance site curve constructionmethodology in this study. RESULTS ResidualsAnalyses Figures 1-4 show the mean height residuals(observed-predicted) by age class. Positive values indicate under- predictions;negative values indicate overpredictions.The associatedsite indexcurveswill giveoverestimates of site index where the height growth curve underestimates, and underesti- Table 2. Site index and heightgrowth equationsfor ponderosapine (PP), Douglas-fir(DF), westernlarch (WL), and lodgepolepine (LPP). Site Index Equations: PP $1 = 59.6 + (4.787 + 0.012544* A - 1.141 * LnA + 11.44/A2) * (H-4.5 - 121.4 * [1-EXP(-0.01756 * A)] ** 1.483) DF SI = 57.3 + (7.06 + 0.02275* A - 1.858 * LnA + 5.496/A2) * (H-4.5 - 114.6 * [1-EXP(-0.01462 * A)] ** 1.179) WL SI = 69.0 + (-0.8019 + 17.06/A + 0.4268 * LnA - .00009635* A2) * (H-4.5 - 127.8 * [1-EXP(-0.01655 * A)] ** 1.196) LP SI = 59.6 + (1.055 - 0.006344* A + 14.82/A - 5.212/A2) * (H-4.5 - 96.93 * [1-EXP(-0.01955 * A)] ** 1.216) Height Growth Equations: o 10 20 30 40 50 60 70 80 90 100 Breast Height Age -•-Cochran +SL•merfi•l • VarKler Ploeg J Figure2. Mean height residualsfor Douglas-fir height growth curvesby Cochran, Summerfield,and Vander Ploeg. PP H = 4.5 + + DF H = 4.5 + + WL H = 4.5 + + LP H = 4.5 + + 121.4 * [1-EXP(-0.01756 (1.189 * [1-EXP(-0.05799 114.6 * [1-EXP(-0.01462 (1.703 * [1-EXP(-0.02214 127.8 * [1-EXP(-0.01655 (1.289 * [1-EXP(-0.03211 96.93 * [1-EXP(-0.01955 (1.410 * [1-EXP(-0.02656 * A)] ** 1.483 * A)] ** 2.63) * (SI * A)] ** 1.179 * A)] ** 1.321) * (SI * A)] ** 1.196 * A)] ** 1.047) * (SI * A)] ** 1.216 * A)] ** 1.297) * (SI 59.6) - 57.3) - 69.0) - 59.6) where SI = site index, breastheightage50 yr base H = total height Ln = naturallogarithm A = breastheightage EXP= exponentialfunction,basee WJAF7(1)1992 11 Table3. Mean heightresiduals(observed-predicted), R, and associated standarderrors,SE,for new heightgrowth curves. Species PP Age n R 10 20 30 40 50 60 70 80 26 29 30 36 31 22 9 4 2.1 0.8 1.1 0.7 0.4 0.7 0.1 1.7 - - WL DF SE n R SE n R 0.41 0.66 0.59 0.43 0.22 0.44 0.75 4.46 51 44 46 35 33 23 14 8 - 0.3 - 0.2 0.1 0.3 - 0.1 0.3 - 0.5 1.1 0.29 0.39 0.44 0.30 0.23 0.49 1.1 6 1.92 31 32 29 29 24 18 14 8 - 0.3 - 0.3 - 0.2 -0.7 0.1 - 0.5 - 0.7 1.4 matesof site index where the height growthcurvesoverestimate. Trendsof residuals by ageclassfor levelsof site agreewell with the testdata at young agesbut there is a cleartrend toward underpredictionat older ages. This pattern is stronglypresentat site indexesbelow55 ft but virtuallydisappearsat siteindexesabove60 ft. The results for Schmidt's (1976) index are discussed but not shown. Figure 1 showsthe resultsfor the ponderosapine heightgrowthcurves by Tesch(1980),Lynch (1958),Summerfield (1980), and Barrett (1978). Tesch's,Lynch's,and Barrett'scurves showsimilartrends:an overprediction of heightbetween20 and50 yearsand an underprediction of heightat older ages.Thispatternis evidentacrossall western larch curves and Alexander's (1967)curvesfor lodgepole pine(asreindexed for use with these data) are givenin Figure4. Neithersetof curves appearsto fit the data particularly well. levels of site index but becomes more Schmidt's curves underestimate heightsat ageslessthan40 yearsand at agesgreaterthan 80. The pattern pronouncedas site index decreases. Tesch'sand Lynch'scurvesalsoshow a tendencyto underpredictheightat intensifies as site index increases. Al- exander'scurvesoverestimateheights at young ages,but may be adequate very young ages. Summerfield's curves fit the data quite well; they for agesaboveindexage.The pattern couldbe appliedto the studypopula- of residuals is similar across all levels tion without incurringany substantial of site index. bias. The resultsfor the Douglas-fir sitecurvesby Cochran,Summerfield, Site Curves andVanderPloegare shownin Figure 2; those for Monserud's curves are Table 2 gives the site index and given in Figure 3. Cochran'scurves height-growthequations,developed from the study data, for ponderosa and Monserud'scurvesfor the grand pine, Douglas-fir,westernlarch and fir habitattype seriesfit the datawell, but Summerfield's curves and Monselodgepolepine. Graphsof the height growth equationsare shown in Figrud's curvesfor the Douglas-firand ures 7-10. subalpinefir habitat type series, all As a basisfor assessing reliabilityof show various degreesof departure the new curves, the residuals (obfromthe heightgrowthpatternsin the data. All three curves tend to underserved-predicted)of SI and height were obtainedby using the present predict height for young ages and stemanalysisdataasinputsto the site overpredictheightat older ages.This index and height growth equations. patternof residualsis mostevidentat LP SE n R 0.48 0.66 0.65 0.40 0.30 0.49 1.05 1.67 37 33 30 27 22 17 12 10 0.2 0.1 0.4 1.0 0.1 1.5 0.6 2.3 - - SE 0.41 0.55 0.67 0.51 0.41 0.45 0.94 1.77 greeof biasin the new equationsmay be assessedin Tables 3 and 4, which give the mean residuals,along with the associated standard errors, for the site index and height growth equations by age class for each set of curves.There does not appear to be any significant bias. Figures5 and 6 show the standard deviationsof the height and SI residuals,respectively,plottedagainstage. The standard deviation is a measure of the variationremainingafterthe relationshipsbetweenheight,SI, and age have been modelled. This variation consistsof two orthogonal components (Monserud 1984); within plot variabilityassociated with the sample basedmean;and variabilityof the plot means about the model. The error due to the within-plotvariabilitycanbe reducedby increasing the samplesizeof sitetreeson a plot. The "plot aboutthe model" error cannot be reduced. It is temptingto usethesestandard deviations of residuals to calculate the number of site trees of a given age needed to obtain estimates of site in- dex and/or height to some statistical standard (Monserud 1984). However, due to the samplingschemeused in the study (one tree per plot), these components cannot be separated. rThus,precisioncurvesare not possible. Moreover, the tree selected was the one with median site index from a low siteindex levels(<60 ft) and tends to becomelesspronouncedas site in- Means, standard deviations, and standard errors for the residuals were then sampleof three.Thus,it is likelythat the within plot componentof the varianceis seriouslyunderrepresented. If dex increases.Vander Ploeg'scurves calculatedfor 5-yr ageclasses. The de- this is the case, then the standard de- Table4. Mean site index residuals(observed-predicted), R, and associated standarderrors,SE,for newsiteindex curves. Species PP Age n R 10 20 30 40 50 60 70 80 26 38 31 36 31 22 9 6 -1.25 1.01 0.54 0.77 -0.36 -0.72 -0.50 -0.47 12 WJAF7(1)1992 DF SE n R 2.28 1.09 0.71 0.50 0.23 0.39 0.69 2.15 46 44 46 31 33 23 14 8 1.90 0.81 0.00 0.17 0.22 -0.90 0.30 -0.78 WL SE n R 1.33 0.92 0.67 0.57 0.27 0.45 0.91 1.40 31 32 24 29 18 18 14 6 1.29 1.05 0.02 0.78 -0.03 0.57 0.21 -0.19 LP SE n R 1.21 1.04 0.87 0.44 0.35 0.46 0.83 1.61 37 33 30 27 22 17 12 10 0.28 -0.68 0.77 1.35 -0.05 -1.36 -0.64 1.31 SE 1.44 1.15 1.04 0.62 0.40 0.41 0.81 1.57 10- 160 150 140 130 160 150 140 130 SI • •120 _•,•.-• .. / 80 70 •1oo 90 60 50 40 30 20 10 0 10 20 30 40 50 60 70 80 90 Breost Hei9hl A9e lO 1020 30 4050 60 70 80 90100110120 o lO 2o 5o 4o5o 6o 7o 8o 90100110120 viation of residualsapproximatesthe between-plotvariationonly, and representsthe bestprecisionpossible.Indeed, the plots in Figures5 and 6 resemblethosepresentedby Monserud (1984)for his Douglas-fircurves,both m shapeandmagnitude,forlarge(n > 10) within-plot samplesizes.Assuming similaramountsof variability,approximately10 treesper sitewouldbe requiredto achievethe levelsof preci- •iguze 9. Ne• hei•t fled in the samplingdesign.Depending on the underlyingdistributionof tree heights and ages, greater samplingintensityon a fixedareacansubstantiallyreducethe estimateof SI. For the presentsitecurves,plot sizeis not specified,and considerableresponsibility is placedwith the userto select individualtreesthat (1) exhibitheight growth patternsat or near the potential for the site,and (2) are potentially repeatableacrossthe site. If care is takento ensurethat all sitetreesfully meet the criteria specified,a larger tray the height growth patternscontainedin the studydata. Theseresults were not unexpected;they reflectthe differences in populations, definitions,samplingprocedures, andcurve constructionmethodologiesused by sion indicated. number of trees would be taken on a Of course,it is alsotrue that by increasingthe within-site sample size one may be significantlychangingthe populationof site trees used to estimate siteindex, relativeto that speci- largerarea, the samplingintensityrelative to the underlying distribution would remain constant, and no bias would result. DISCUSSION With a few exceptions,the publishedcurvesdid not accuratelypor- 12 gzow• •e• the various authors, relative to the sampleto which they were compared. Generally, the results agree with Monserud's(1985)conclusions regarding similarityin shapeof Douglas-fir siteindexand heightgrowthcurvesin the Pacific Northwest. Curves con- structed from stem analysis based methodologies applied to data from populationsreasonablycloseto the study area agreedmost closelywith the test data. These included Monse- rud's and Cochran's Douglas-fir curvesand Summerfield'sponderosa pine curves. Site curves that utilized cross-sectional data and guide curve techniques(Tesch1980, Lynch 1958, 160 150 140 130 •120 160 150 140 130 •,120 '; Breost Hei9ht A9e Figure 7. New height growth curves for ponderosapine. -•-ponderosa +Douglas-fir -xbch-•lodgepole curvesø /',/•'"'•'"•' "•60 30 Breast Height Age Figure S. Standarddevition of height residuals vs. age for new height growth 90 •'110 •'100 • 9o E 90 8o 7O 6o 5o 60 o 50 o • 40 4o 20 10 0 50 2O 10 0 0 10 20 30 40 50 60 70 80 90100110120 10 20 ,50 40 50 60 70 80 9O Breast Height Age --•ponderosa +•glas-fir --xlatch -•-Iodge• Figure6. Standarddeviation of site index residualsvs.agefor new siteindexcurves. 0 10 20 50 40 50 60 70 80 90100110120 Breast Height Age Figure 8. New height growth curves for Douglas-fir. Bredst Hei9ht A9e Figure 10. New height growth curvesfor lodgepolepine. WJAF 7(1)1992 13 Schmxdt 1976), on the other hand, change For the Summerfieldand Bar- showed rett curves, it is unclear to what extent a distinct and consistent un- derestimationof height at older ages, suggestingthat the assumption of equalsitequalityrepresentationacross all levels of age was not met. Alexander'scurvesforlodgepolepinewere derived from a sample that encompassed the Rocky Mountain, Inter- the causesof the differencesare primarily genetic, methodological or samplebasedin origin. The new sitecurvesprovidean unbiased summarization of the relation- shipsbetween"true" age (as distinct from ring count) at BH, total tree mountain, and Pacific Northwest reheightcorresponding to that age, and gions. It is not surprising that their SI (totalheightat a BH ring countof comparisonto a small, geographically 50). This configurationis mostapproand geneticallycompactpopulation priate for simulation models, which should show differences. It was also use true age (trees are assumed to notedthat the patternof over and unreach4.5 ft at the end of a growing derpredictions in heightrelativeto the season).Sincering countat BH overstudy data was completelyconsistent estimatestrue BH age by a half year, field estimates of SI will be biased with the age-relatedpattern of predictedsite index noted by Alexander downward slightlyunless0.5 is subin the presentationof his final curves. tractedfromtheringcountpriorto usThus, the differencesmight be both ing the siteindexequation.It is intermethodological and geneticin nature. estingto notethatweretheraw height Stem analysisbased curves from data adjustedto the ring count data reasonablyclose geographicregions (by adding a half years growth), the bias in field estimation of SI would disdid not alwayscorrespondwell with the test data. The study area for appear;only to reappearas an overesVander Ploeg'sDouglas-fircurvesfor timationof heightfor the true agesaswestern Montana had considerable sumedin modelapplications. overlap with the current study area The sitecurvesystemdevelopedin and yet substantialunderestimates of this studyprovidesthe user with the heightwerenotedat olderages.Also, means to identify potential height Barrett'sponderosapine curvesand growth patternsfor ponderosapine, Summerfield's Douglas-fir curves, Douglas-fir,westernlarch,and lodgeboth from east of the Cascade crest pole pine in WesternMontana. Any (similarto Cochran'sDouglas-firand polymorphismin curve shape is reSummerfield's ponderosa pine flectedonly to the extentthat it is associated with levels of site index. [] curves), showed different height growth patterns from those in the LITERATURE CITED study data. The causeof the differences for Vander Ploeg's curves is ALEXANDER,R.R., D. TACKLE,AND w.G. DAH•4S. likely relatedto the absenceof plots 1967.Siteindexesfor lodgepolepine,with corwith site indexes below 55 ft in his sample.In the currentstudy, the daa showa distinctchangein curveshape at low site indexes;Vander Ploeg's samplewould not have containedthis rectionsfor stand density. USDA For. Serv. Res.Pap. RM-29. 18 p. BARRETT, J.W. 1978.Heightgrowthand siteindex curvesfor managed,even-agedstandsof ponderosapine in the PacificNorthwest.USDA For. Serv. Res.Pap. PNw-232. 14 p. Establishmentof Jeffrey Pine Seedlings from Animal Caches COCHRAN, P H 1979 Site index and height growthcurvesfor managed,even-agedstands of Douglas-fireastof the Cascades in Oregon and Washington.USDA For. Serv. Res. Pap PNW-251.16 p. DAHI•IS, W.G. 1973.Grossyieldof centralOregon lodgepole pine.Wash.StateUniv., Proc.symp Managementof lodgepolepine ecosystems. DYER, M.E., AND R.L. BAILEY. 1987. A test of six methodsfor estimatingtree heightsfrom stem analysisdata. For. Sci.33(1):3-13. HEaER,L. 1968.A methodof constructingsite indexcurvesfrom stemanalysis.For. Chron 44:11-15. LYNCH,D.W. 1958. Effectsof stockingon site measurement and yieldof second-growth ponderosapine in the InlandEmpire.USDA For Serv.Res.Pap.INT-56.36 p. MEANS,J.E., AND T.E. SAmIN.1989. Height growthand siteindexcurvesfor Douglas-firin the SiuslawNationalForest,Oregon.West.J App1.For. 4(4):136-142. Mn,NER,K.S. 1987.The developmentof sitespecificheightgrowthcurvesfor four conifersin westernMontana. Ph.D diss., University of Montana, Missoula. MONSER•SD, R.A. 1984.Height growth and site index curvesfor inland Douglas-firbasedon stemanalysis dataandforesthabitattype.For Sci. 30:943-965. MONSER•SD, R.A. 1985.Comparisonof Douglasfir siteindexand heightgrowthcurvesin the northwest.Can. J. For. Res. 15(4):673-679. SCH•4IDT, W.C., R.C. SHEA•R, AND A.L. ROE 1976.Ecologyand silvicultureof westernlarch forests. USDA For. Serv. Tech. Bul. INT-1520 96 p. SVM•4ER•IELD, E.R. 1980a.Siteindexand height growthof Douglas-firin easternWashington Oper.Res.Intern.Rep.No. 80-12.Oper.Res Sec., Div. of Manage. Serv, DNR, State of Washington.34 p. SV•4MER•IELD, E.R. 1980b.Siteindexand height growthof ponderosa pinein easternWashington. Oper. Res.Intern. Rep. No. 80-14.Oper Res. Sect.,Div. of Manage. Serv., DNR, State of Washington.34 p. TESCH,S.D., H.R. ZUURING,AND J.L. FAUROT 1980.Ponderosapine sitecurvesfor the BlackfootRiverdrainage,Montana.Res.noteno. 15, Mont. For. & Cons. Exp. Stn., Schoolof For., Univ. of Mont., Missoula, MT. VANDERPLOEG,J.L., AND J.A. MOORE. 1989 Comparison and development of height growthand siteindexcurvesfor Douglas-firin the Inland Northwest. West. J. Appl. For 4(3):85-88. cienttoaccount for mostoftheJeffrey pine seedlingestablishment observed. Cached seeds maygainimportant advantages over uncached seeds in the semiarid habitat oc- cupiedby thispine. West.J. Appl. For. 7(1):14-20 Food-hoarding animals are known to StephenB. Vander Wall, Department of Biology,University of Nevado-Reno, Reno,Nevada89557. dispersethe seedsof many plants These animals cache seeds or nuts un- der a layer of soil or plant litter, and the propagulesgerminatein springif not recovered. ABSTRACT. Jeffrey pine(Pinusjeffreyi) seeds havelargewingsandare effectively dispersed bywind,but37%oftheseedling emergence sitesin spring1989 in theSierraNevadaofwestern Nevadaconsisted of i The author thanks David Charlet, Steve Jenkins,Hal Klieforth,JosephMcAuliffe, Don Minore, and Sue Wilson for their assistance. 14 WJAF7(1)1992 Most attention has tight clumpsof seedlings. By fall 1989, beendirectedtowardcorvids(jaysand 41%ofall emergence siteswhereseedlings nutcrackers) and rodents that store survived hadresulted fromwhatwereorig- large nuts. Relativelylittle is known inallyclumps ofseedlings. AsJeffrey pines concerningthe influenceof rodentsas aged,clumps became smaller,andthefredispersersof the seeds of conifers, quency ofclumps decreased. These seedling shrubs,and grasses(seePriceandJenclumps andmanysingleseedlings emerged kins1986for a review).Amongpines, fromthescattered caches of seed-hoardinginteresthas focusedon the winglessrodents andcorvids,and theactivities of seededspeciesof the softpines(subthese animalsin thestudyareaweresuffi- genusStrobus),which are dispersed