Vegetation of the Great Smoky Mountains Author(S): RH Whittaker Source

Vegetation of the Great Smoky Mountains Author(s): R. H. Whittaker Source: Ecological Monographs, Vol. 26, No. 1 (Jan., 1956), pp. 1-80 Published by: Ecological Society of America Stable URL: http://www.jstor.org/stable/1943577 Accessed: 03-09-2015 19:05 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/ info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecological Monographs.

http://www.jstor.org

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions VEGETATION OF THE GREAT SMOKY MOUNTAINS1

R. H. WHITTAKER Biology Department,Brooklyn College, Brooklyn10, N. Y.

TABLE OF CONTENTS

PAGE PAGE I. GRADIENT ANALYSIS ...... 1 Zonation ...... 34 36 Introduction ...... 1 Ecotones ...... Nature of the Study...... 1 Climax Patterns and Their Comparison...... 37 Literature on Area...... 2 Considerationsof Logic and Method...... 40 Acknowledgments ...... 2 Geology and Climate...... 2 Conclusion ...... 43 Methods ...... 4...... 4 III. VEGETATION TYPES AND THEIR DISTRIBUTIONAL Field Transects and Tree Classes...... 4 RELATIONS ...... 43 Site-Samples and Composite Transects...... 6 Bases of Recognizing and Describing Types ...... 43 Distributions of Species along the Moisture Vegetation Types of the Great Smoky Mountains ... 45 Gradient ...... 8 1. Cove Hardwoods Forest...... 45 in the and Trends in Relation to the Moisture Gradient...... 10 Mixed Mesophytic Smokies Growth-Forms ...... 10 Cumberlands ...... 46 2. Eastern Hemlock Forest...... 48 Coverages ...... 10 48 Diversities ...... 10 3. Gray Beech Forest...... Forest...... 49 Sizes and Numbers of Stems of Trees...... 11 4. Red Oak-Pignut Hickory 5. Chestnut Forest 49 Self-Maintenanceof Stands...... 11 Oak-Chestnut ...... 6. Chestnut Oak-Chestnut Heath ...... 50 High-Elevation Deciduous Forests...... 13 7. Red Oak-ChestnutForest ...... 51 Distributionsof Species in Relation to Elevation... 14 8. White Oak-ChestnutForest ...... 51 Mesic Sites ...... 14 Pine Stands and Their Maintenance . . 51 Submesic Sites ...... 15 9. Virginia Pine Forest...... 52 Subxeric Sites ...... 15 10. Pitch Pine Heath...... 52 Xeric Sites .15 11. Table Mountain Pine Heath . . 53 Trends in Relation to Elevation.16 12. Grassy Bald ...... 53 Growth-Forms.16 The Southern Appalachian Subalpine Tree Statures and Stratal Coverages...... 17 Forest Center ...... 53 Diversity and Environmental Favorableness... 18 13. Red Spruce Forest . . 54 14. Fraser Fir Forest . . 54 Forests .18 Spruce-Fir 15. Heath Bald . 55 Stratal Distributions.19 ...... Trends .20 The Balds as Topographic Climaxes? ... 56 Relations of Species to Unions and Associations 21 Distributional Relations ...... 57 Continuityof Vegetation Types...... 21 The Mosaic Chart ...... 57 Nature of Species Groupings...... 21 Distributions of Types...... 59 Dominance in Relation to CommunityComposition 23 Distribution of Subalpine Forests...... 60 Relation of the Vegetation Pattern to Those of Distributional ...... 24 Summary Groupings of Other Mountain Ranges...... 61 II. DISCUSSION: AN INTERPRETATION OF SUMMARY ...... 62 VEGETATION PATTERNING ...... 25 LITERATURE CITED ...... 63 Distributionsof Species and the Study of Genecology 25 APPENDIXES ...... 69 The Association-UnitTheory and Individualistic Hypothesis ...... 30 A. Population Charts for Major Tree Species 69 The DistributionalBasis of Community-Types...... 32 Note on Supplementary Publication of Appendixes Gradation and the Grouping of Species...... 32 B and C ...... 69

I. GRADIENT ANALYSIS for a study of this vegetation. The work was originally intendedto provide informationon the vege- INTRODUCTION tation for the sake of its own interestand as a basis NATUREOF THE STUDY for studies in animal ecology (Whittaker1952). A major purpose of both this and the precedingstudy, The Great Smoky Mountains of Tennessee and however,was use of the complex pattern of natural North Carolina support vegetation which is particu- communitiesin the Great Smoky Mountainsfor re- larly rich in species and varied in community types. search into the theoryof communityunits or asso- In the summer of 1947 field work was carried out ciations. For this purpose, the approach to vegeta- IBased on a thesis (Whittaker 1948); a contributionfrom tation was based on samplingwithout regard to ap- the Department of Zoology, University of Illinois, Urbana, and the Biology Department, Brooklyn College. parent associations and analysis of the samples in

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 2 R. H. WHITTAKER Ecological R. II. WHITTAKER ~~~~~~~~~~Vol.Monographs26, No. 1 relationto environmentalgradients. It was felt that ACKNOWLEDGMENTS relative validity of vegetationtypes should emerge While at the Universityof Illinois, the authorwas from data impartiallyobtained, and that the rela- aided by the suggestionsand criticismsof S. C. Ken- tions of types to one anothershould be revealed in deigh and A. G. Vestal. The park naturalistof the the studyof theirpopulations in relationto environ- Great Smokies,Arthur Stupka, gave the author the mental gradients. The work thus departs from the cooperationand benefitof broad knowledgeof the traditional approach of studyingintuitively recog- mountainswhich he extendsto studentsin the area. nized types or associations; it is an experimentin Help with the identificationof plant specimenswas population analysis of a whole vegetationpattern. given by Stupka and by A. J. Sharp and R. E. The first part of the monographdescribes results Shanks, who checked all determinations.Responsi- of the analysisin termsof relationsof species popu- bility for statementsof distribution,based on field lations to one anotherand environmentalgradients, determinations,remains with the author. A number and trendsin communitycomposition and structure of people have read part or all of the manuscript along environmentalgradients. The second part in- and offeredcomments on it: E. L. Braun, H. E. terpretsthe vegetationas a complexpattern, within Brewer,H. K. Buechner,W. H. Camp, A. Cronquist, which vegetationtypes may be understoodthrough R. Daubenmire,F. E. Egler, H. A. Gleason, A. R. the distributionalrelations of species populations. Kruckeberg,H. L. Mason, R. E. Shanks,A. F. Sharp, The third part presents a more conventionalde- and A. Stupka. The authoris especiallyindebted to scriptionof vegetationtypes and considersthe re- W. H. Camp for his suggestionsand for information lationsof theseto topography. The studyas a whole which permittedinterpretation of genetic and dis- thus seeks to analyze, interpret,and describe the tributionalphenomena. Cost of publication of the complex vegetational mantle of the Great Smoky tables and chartshas been met in part by a grant-in- Mountains. aid fromthe Society of the Sigma Xi. LITERATURE ON AREA GEOLOGY AND CLIMATE A series of studiesby Cain deal withvegetation of The Great SmokyMountains are part of the Blue the Smokies-the heath balds (1930b), subalpine Ridge Province,a systemof mountainsof great an- forests (1935), and cove hardwoods (1943), floristic tiquity. This part of the Southern Appalachians affinities(1930a), soil reaction (1931), Raunkiaer comprisestwo major ranges,the Blue Ridge proper life-forms(1945), and bryophyteunions (Cain & and the Unaka Mountains, along with their con- Sharp 1938). A number of vegetationtypes were necting cross-ranges (Fenneman 1938). The two describedby Cain et al. (1937). The grassy balds main ranges lie parallel,from northeast to southwest, were reportedon by Camp (1931, 1936) and Wells with the IJnakas,of which the Smokies are part, to (1936a, 1936b, 1937); the subalpine forests were the north. The drainage, north from the divide of recentlydescribed by Oosting& Billings (1951) and the Blue Ridge, is northwestinto the Great Valley of the beech gaps by Russell (1953). These papers and the Tennessee River; and rivers flowingfrom the the descriptionin Braun (1950) are the extentof the Blue Ridge to the Great Valley throughthe Unakas literaturedealing specificallywith the vegetationof cut the latter into a series of segmentsdivided by the Smokies. Other studies include descriptionsof deep gorges. The Great Smoky Mountains are the SouthernAppalachian vegetationtypes, among them largest and highest of these segments,between the Harshberger'sreport (1903) and book (1911), Wells Little Tennesseeand Big Pigeon Rivers. (1924) and the forestryreports of the Message from The crest of the Great Smoky Mountains forms the President (Ayres & Ashe 1902), Reed (1905), the state borderof Tennesseeand NorthCarolina, 25- Holmes (1911), Ashe (1922), and Frothinghamet 50 miles southwestof the city of Knoxville in the al. (1926). Of studiesin nearbyareas Braun's papers Great Valley. The range has the appearance of a on the Cumberland Mountains-Pine Mountain long, sinuous ridge connectingirregularly spaced (1935b), Black Mountain (1940a), and the Cumber- domes,with secondaryridges and hills spreading on lands (1942, 1940b)-and material in the book on each side (Fig. 1). The mountainsare stream-eroded the easternforests (1950) were mostvaluable for the to physiographicmaturity; in form they are sub- relatedvegetation of the Smokies.Other Appalachian dued (Fenneman 1938) thoughrugged. Many of the and easternstudies-Harshberger (1905) and Heim- summitsand ridges are rounded,and almost all the burger (1934) on the Adirondacks,Core (1929) on mountainsurface is coveredby a mantleof soil and Spruce Mountain,Davis (1930) on the Black Moun- vegetation. The resistantrocks have maintainedhigh tains, Conard (1935) on Long Island, Raup (1938) relief in spite of age. Sixteen peaks have elevations on the Black Rock Forest, Oosting& J3illings(1939) above 6000 ft (1830 m); and the highestsummits rise on Ravenel's Woods, Oosting (1942) on the Pied- more than 5000 ft above the vallevs a few miles to mont, and, particularly,Brown (1941) on Roan the north. Valleys are cut deep into the mountain Mountain-contributedcomparative information. A mass, with steep slopes and narrowflats. The slopes bibliographyof otherpapers dealingwith the Smokies formmost of the area of the mountains;it has been is given by Mason & Avery (1931). Taxonomicref- estimatedthat less than 10% of the surface has less erences for the area are Small (1933), Shanks & than 10 degrees of slope (Message from the Presi- Sharp (1947), Gleason (1952), and Fernald (1950). dent 1902)'.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF TIHE GREAT SMOKY MOUNTAINS 3

k~~~

FIG. 1. Mature, forest-covered topography of the Great Smoky Mountains, a view of the southeast slope from Frye Mountain near Bryson City, North Carolina. Reproduced by permission of W. M. Cline Co., Chat- tanooga, Tenn.

Most of the rocks of the mountainsbelong to the The southernmountains have played a great part Ocoee series (Safford1869, Stose & Stose 1944, 1949, in the vegetationalhistory of the East (Braun 1938, King 1949) of complexlyfolded, metamorphic sedi: 1941, 1950). While otherareas have been glaciated, mentaryrocks which are resistant to erosion and submerged,and exposed to great climatic change fairly uniformin their reactionto it. Deposited in (Fernald 1931), the Southern Appalachians have Cambrian or late pre-Cambriantime (Keith 1902, offereda sanctuaryfor many species of plants and Stose & Stose 1949,King 1949), they were firstfolded animals. The Blue Ridge System has been con- into mountains in the Appalachian Revolution of tinuouslyoccupied by plants and animals for perhaps the late Paleozoic. The mountains were raised 200 million years (Cain et al. 1937). During the furtherin the Cretaceous and have since that time early Tertiarythe higherelevations of the Smokies been throughthree cycles of erosion,the Schooley, and Blue Ridge probably supported temperatefor- Harrisburg,and present(Wright 1931). While some ests ancestral to those now in the area while sub- higherridges of the Smokies may remain from the tropical floras prevailed at sea level (Cain 1943). Schooley cycle of the earlier Tertiary (Willis 1889, It is in the Blue Ridge Province and other areas King & Stupka 1950), it is probablethat throughout where the Schooley peneplane was never perfected this the area was one of hills or low mountains(Fen- that most typical mixed mesophyticforests, most neman 1938, Wright 1942, Braun 1950). After a closelyrelated to the Arctotertiaryforests, have sur- second elevation,probably in the Miocene,the moun- vived (Braun 1950:505). During the climaticchanges tains persistedthrough the shorterHarrisburg cycle, to whicheastern vegetation was subjected,the topog- the peneplaneof whichmay be suggestedby some of raphy of the SouthernAppalachians offeredvaried the lower ridges (King & Stupka 1950). The moun- conditionsof moistureand elevationin whichspecies tains were again raised at the end of the Tertiary. of diverse climatic adaptations might survive while During the Pleistocenethe Smokies were well south sometimesdestroyed elsewhere. of the ice sheetsand possessedno glaciers; but there It mightbe expectedthat age and maturityof the are indicationsthat climaticcooling produceda tim- mountainswould be reflectedin maturityof their berlineon the highersummits (King & Stupka 1950), plant cover, as well as in antiquityof some of the displacing forestvegetation toward lower elevations. flora. Primarysuccession is nearlycompleted in the It is believed,from distributional evidence discussed Smokies; it is perhaps in progresson a few peaks later (Part III) that high-elevationforests were dis- and ridges, but almost all the vegetationis either placed upward 1000 to 1300 ft above presentlevels topographicclimax or secondary. One of the major duringthe warmdry period followingglaciation. forest trees, the chestnut (CoGttanea dentata), wav

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions Ecological Monographs 4 R. HI. WHITTAKER Vol. 26, No. 1 infectedby the blight in the period from 1925 to United States outside the NorthwestCoast. Series 1930; and mosttrees were killed by the early 1940's. of summerthunderstorms advancing along the ridge Death of the chestnutsaltered many forest stands by and banks of cloud in the canyonsare familiarsights openingcanopies and permittingincreased reproduc- in the mountains. The Smokies share withsurround- tion of otherspecies. Areas of second-growthdue to ing areas the distinctiveSouthern Appalachian pre- human disturbanceare extensive. Many valleys and cipitationregime, with two maxima,one in winteror lower slopes were cleared or cut in the course of early spring and the other in late summer (Henry settlement,and undisturbedvalley stands at low ele- 1902, Ward 1925, Kendrew 1937). Monthlyprecipi- vationsare rare. Some of the higherslopes have been tation data for low and high elevations have been logged. Fires, set by man and by lightning,have previously published (Whittaker 1952:2, Shanks swept many of the drier slopes at all elevations. A 1954). To the humidclimate may be related also the large area of mountainvegetation which is virginor haze which,most of the time,lies over the mountain but little disturbed,however, is protected by the landscape and which gave the Great Smoky Moun- Great Smoky MountainsNational Park. tains theirname. The complexityof the climax forestsis a further METHODS expressionof the the physiographyand historyof Tn Part I data on the vegetationof the Great the mountains.Because of age and diversityof habi- Smoky Mountains will be discussed in relation to a tats, the florais rich and contains high proportion threegroups of questions: (1) How are species popu- of endemic species for an eastern flora. Approxi- lations distributedin relationto environmentalgradi- 130 mately1300 species of floweringplants, including ents and one another? (2) What trends in com- species of native trees, have been found in the munity composition and structure along environ- Smokies (King & Stupka 1950); and Cain et al. mental gradients may be observed'? (3) How are (1937) estimatedthat 3.1% of the flora is endemic community-unitsto be understoodin terms of the to the Unaka Range, an additional to 8.5% the relationsof stands and of species populationsto one Southern Appalachian Mountains. The variety of another'? The third of these, and the objective of habitats, ranging from most mesic valleys to dry studyingthe nature of community-units,required the slopes and ridges in moistureconditions and from use of means sampling unprejudicedby assump- to in ,of austral subalpine temperature,causes the ap- tions about community-units.The two methodsof pearance of many vegetationtypes in differentsites. sampling used were field transects, with samples Since,however, the rocksof the mountainsthemselves taken at fixed intervalsalong environmentalgradi- are relatively homogeneous (predominantlyquart- ents, and site-samples taken at random, so far as zites and slates), the vegetationis not furthercom- possible, throughthe whole vegetationpattern. plicated by great differencesin soil parent-material. Only in some of the northwesterncoves, where FIELD TRANSECTS AND TREE CLASSES fenstersor windowsthrough the overthrustrocks ex- Almostany valley of the Smokiescontains a series pose limestones(Fenneman 1938), do calciphilecom- of vegetationtypes fromthe stream-sideto the open munitiesoccur. Geologically and biologicallythese slope or ridge above. The basic sequence of types coves are inliers from the Great Valley; and they below 4500 ft (1370 m) includes: (1) cove forest are consideredoutside the scope of this study. Aside (mixed mesophyticassociation) in most mesic sites fromthese coves, the soils of the Smokies are acid, of valley bottomsand lower slopes, (2) oak-chestnut with acidities generally increasing toward higher foreston less mesic sites, (3) oak-chestnutheath in elevationsand more exposed sites (Cain 1931). still drier or more xeric sites, and (4) pine forests From the high relief of the Smokies both marked and-pine heaths on the driestsouth- and southwest- variation in local climatesand a humid general cli- facing slopes. At elevationsabove 2500 ft (760 m) mate result. Mean temperaturesdecrease with ele- stands dominated by hemlock (Tsuga canadensis) vation at an average rate of 2.230 F per 1000 ft, so usually appear between the deciduous cove forests that high elevationsaverage 10-15? cooler than the and the oak-chestnutforests, and below 2500 ft oak- base of the mountainsduring the growing season hickoryforests often occur betweencove forestand (Shanks 1954). Sea-levelclimates most nearly equiva- oak-chestnutforest. Analysis of this sequence of lent to those of high elevationsin the Smokies may plant populations and vegetation types along the occur about 1000 mi. to the northeast,in northeastern "moisture gradient" was a major problem of the Maine or adjacent New Brunswick (Shanks 1954). study. The field transects were used as a firststep By the Thornthwaite(1931, 1941, 1948) climatic in this analysis,on whichthe moredetailed composite classifications,low elevationsof the range are humid transects describedbelow are dependent. mesotheymal,and high elevations perhumidmicro- In fieldtransects, vegetation samples were taken at thermal;intermediate elevations are differentlyclas- uniformintervals of distance or elevation from the sifiedby the two systems(Shanks 1954). valley bottomto the ridge above it. At each sampling Annual precipitationvalues range from 50 to 60 point the site was describedby topographicposition, in. (127-152 cm) in the lower valleys (Gatlinburg, elevation,direction of exposure,and degree of slope. Elkmont, and Maryville records) to more than 80 Trees were tallied, with diametersat breast height in. (203 cm) at high elevations,the highest in the from 1 iir. up recordedfor each species; about 100

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 5 stems were the usual sample for each of 7 to 10 maximain the second or thirdstations. At the other stations. Along with the tree count the undergrowth extremeare species which have their maxima in the was recordedby a coverageestimate for each stratum most xeric site, station VII, and do not extend to and a list of major and minorspecies. An example, sites less xeric than station VI. Between these ex- one of six such transectsmade, will show the method treme groups there are a number of species with and its relationto the compositetransects. their maximumpopulations in stations III to VI. On the Bullhead Trail to Mt. Le Conte, seven These mightbe grouped together;but they may also sample counts were made at intervalsof 25 m from be separated into two groups,one having maxima in the valley bottomto the southwest-facingslope, all stationsIII and IV, extendingon the mesic side to at elevationsof 3100 ft. In the tables an additional station I or II but not beyondV on the xeric side, sample from a deep valley forest was added to the and the other having maxima in stationsV and VI beginningof the series,since the small valley of the and extendingto the xeric extreme,but not beyond transectdid not representthe extremeof mesic con- IV on the mesic side. The four groups are used as ditions. Percentagesof stand for the trees and oc- classes of trees along the moisturegradient. They currencesof shrubsand herbs were arrangedfor the may be characterizedas follows: 8 stationsas in Table 1. In analyzingtransects with 1. Mesics-Species with maxima in or near the relativelysmall samples, simple tallyingof numbers most mesic sites and with limited extent into more of stemsfor each species was preferredto basal-area xeric situations,occurring rarely in the part of the computation. Canopy dominance was determined gradient representedby oak-chestnutheath. These separatelyfrom larger samples. species predominatein the cove forests. Individual species of trees are well scatteredalong 2. Submesics-Species whichhave theirmaxima in the gradient,but certain loose groupingsof species fairly mesic sites, but are- uncommonor absent in may be suggested. Some species have theirmaximum mostmesic sites and do not extendto mostxeric sites. abundance in the deep-valleycove forest,station K These species predominatein oak-hickoryforests at of the transect,or are abundantthere and have their lower elevationsand in red oak-chestnutforests at higher elevations. in TABLE 1. Bullheadfield transect of moisturegradient. 3. Subxerics--Specieswhich have theirmaxima Alongtrail fromCherokee Orchard to Mt. Le Conte,in morexeric sites,but occur in mostxeric sites only as a small dry valley at 3100 ft and out onto adjacent minorspecies and are absent from most mesic sites. southwestslope. Trees by percentagesof stand from oak-chestnutheaths and 1-in.class up. These species predominatein at higherelevations in whiteoak-chestnut forests. Tree species K* I II III IV V VI VII 4. Xerics-Species which have their maxima in most xeric sites and have limited extent into less Tsuga canadensis...... 11 2 4 3 .. ... of Halesia monticola...... 29 24 25 24 1 xeric sites, extendinginto the range of dominance Aesculusoctandra ...... 9 3 9 the previous group and no further. These species Acersaccharum ...... 9 14 33 25 predominatein pine forestsand pine heaths. Tilia heterophylla...... 16 27 13 13 are recognizedfor shrub and Fagus grandifolia...... 10 The same "classes" Betulaallegheniensis ...... 9 herb populations. Lists of species for each are given LiTriodendrontulipifera ...... 1 in the "Summaryof DistributionalGroupings." Magnoltaacuminata...... x gradientis one of great complexity; Ilex opaca...... x The moisture Carya cordiformis...... x 1 2 along the gradientfrom stream-sideto south-facing FTraxinusamericana ...... 1 1 slope and ridge many factors of soil moistureand Cladrastislutea...... x 11 3 1 atmospherichumidity vary, along with exposure to Magnoliaf. asert ...... 1 x .. .. 6 af- Acerrubrum...... 4 x x 21 54 20 x wind and insolation,and factors of temperature Quetcus borealis v. maxima. . 1 1. .. .. fected by insolation and by patterns of air move- Carya glabra ...... 1 2 ment. In relationto the "primary"gradients of en- Ostryavirginiana ...... 1 1 4 a sequence of vegetationtypes Acer pensylvanicum...... 14 1 11 6 vironmentalfactors Betulalenta ...... 1 4 7 8 10 and a catena of soils develop; and the composition Hamamelisvirginiana ...... 3 8 28 and physiognomyof vegetationand properties of ilethiaacuminata...... 9 4 gradients of environ- .... 2 soils form other "secondary" Amelanchierarborea .. . factors Robiniapseudoacacia ...... 1 x 4 2 x mentalfactors affecting plants. The primary Castanea dentata(dead) ..... 3 9 10 2 1 are so modifiedby the presenceof plant communities Quercusprinus.. ... 6 8 38 4 and "secondary" factors are not . 1 that "primary" Oxydendrumarboreum ...... 4 10 5 in their effectson plants. Nyssa sylvatica...... 1 x 16 2 really to be distinguished Sassafrasalindum...... 5 1 At any point along the gradient the plant lives in Quercuscoccinea ...... x 2 relationto an environmentalcomplex of interrelated . Pinus pungens...... 11 12 factors of physical environment,soil, vegetation, Pinus rigida ...... 2 77 and animal communities;along the "moisturegradi- Total stems..... 642 205 205 91 90 52 55 122 ent" factors of each of these change. The gradient is thus a complexof factorgradients, or a gradi- x, presentat less than .5%. of environmentalcomplexes, which, in distinc- *Kalanu Flats, a cove forest6 mi. east of transectarea, elevation 2800 ft. ent

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 6 WHITTAKER Ecological Monographs R. H. Vol. 26, No. 1 tion from a factor gradient, may be termed a com- thicketswere not counted in forestswhere they oc- plex-gradient (Whittaker 1954b). The "elevation curred.The 25,000 stemsrecorded in 300 site-samples gradient" is likewise a complex-gradient, involving were the total sample analyzed for the vegetation many factors of physical environment, soils, and pattern. natural communities other than temperatures and Exacting phytosociologicalanalysis of the under- growing seasons. growthwas not an objective. Informationon shrubs The complex-gradient from valley bottoms to dry is largely limitedto presenceand stratal dominance, slopes will be called the "moisture gradient," but that on herbs to visible presenceat the timeof sam- with no assumption that moisture factors directly pling. Stratal coveragesare estimates,intended only control the distribution of any plant population along to permit comparisonsbetween differentstands in it. Measurements of al factors of environment and the Smokies. At 15 sample stationsfor anotherstudy determination of which may be most significant for (Whittaker1952) locationand coverageof individual populations of differentplant species are far beyond plants were mapped in quadrats lOmisquare. the scope of the present work. For the present study The site-sampleswere manipulatedin several ways. it may sufficethat a complex-gradient exists, in re- By comparingseries of them from northand south lation to which the distributions of plant populations slopes or otherexposures, the alterneeffect on vege- mav be studied. tation could be determined,a methodfound particu- Such study is dependent on the definition of rela.- larly effectiveat high elevationswhere alterne effects tive positions along the gradient. Since these could are more conspicuousin undergrowththan canopy. not be determined, for hundreds of site-samples, Groups of samples from sites with similar moisture by direct environmental measurement, approaches conditionswithin limited ranges of elevations were through the vegetation itself were sought. Along the compiledinto compositestand counts. These counts, gradient the four moisture classes of trees rise and usually for about 1000 stems, compensatedfor the fall in sequence, forming a set of curves flowing small size of the site-samplecounts and were used continuously into one another (Figs. 2, 3, 4). If, as for the characterizationof vegetationtypes (Part is shown by the transects, there is progressive shift in III). Data from the site-sampleswere arranged in proportions of trees of differenttolerances along the mosaic form on a chart with elevation and topo- gradient, then it is not unreasonable to turn from graphicsites as axes, to show distributionsof species this fact to its converse and regard the same propor- and vegetation types in relation to elevation and tions as expressions of position along the gradient. topography(Part III). In the following discussion, stands and sites will be The site-sampleswere, finally,arranged in com- termed mesic, submesic, subxeric, and xeric according posite transectsin terms of elevation,or of topo- to which of the moisture classes predominate in stem graphic site, or of moistureconditions as indicated numbers. by the vegetationitself. For the elevationtransects SITE-SAMPLES AND COMPOSITE TRANSECTS some means of comparingstands of equivalentmois- The main reliance in solving the vegetation pat- ture conditionsat differentelevations was needed. The tern was on the site-samples and their manipulation. site-samples were consequentlyclassified into A site-sample was a vegetation sample from a re- four groups, according to which of the moisture stricted site of uniform physical habitat-the floor of classes of trees was predominantin a given sample. a valley, a single hillside slope of the same direction Within each of the four classes of stands, the site- and inclination, or the crest of a ridge. In order to samples were grouped by 200- and 300-ftintervals. obtain an approximately random coverage of the Four compositetransects were thus arrangedto cover the whole vegetation pattern, samples were taken from whole of the vegetation pattern, showing the the many trails at all elevations in the mountains. change in levels of plant populations fromlow ele- The method was to move along a trail recording a vations to high in each of the four classes of stands sample from each new slope exposure, inside or out and sites recognized. of a valley, of sufficientextent to give a homogene- A more sensitiveindication of relative positions ous sample. The site-samples were in no case se- along the moisturegradient is possible throughthe lected to represent either apparent vegetation types use of weighted averages as indicator values (cf. or the transitions between them. The bulk of the site- Ellenberg1948, 1950, 1952, Curtis & McIntosh 1951, samples were obtained from the mountains surround- Whittaker 1954b). In a given stand the number ing Greenbrier, Sugarland, and Cades Coves in the of stems in each moistureclass is multipliedby a National Park on the Tennessee or northwest side weight (0 for mesics, 1 for submesics,2 for sub- of the range. xerics, 3 for xerics), and the total of weightedstem At each site the same data were recorded as in the numbersis divided by the total number of stems. transect stations. Sample size varied with the num- Within elevation belts (1500-2500, 2500-3500, and ber of trees thought necessary to indicate stand com- 3500-4500 ft) the site-sampleswere arranged in se- position: fifty were sufficient in some stands with quence from most mesic to most xeric by these one or two dominants, but most counts included about weightedaverages, and were then grouped for tabu- 100, while 200 or 300 were tallied for some mixed lation into 12 or 13 steps along the gradient. This types. The dense small stems of Rhododendron methodof -arrangingthe transectsinvolves an evi-

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 7 dentcircularity; distribution of treespecies is studied signedto forma grid coveringthe wholeof the vege- in terms of previously determineddistributional tation patternof the Great Smoky Mountains. The classes of these same tree species. The approach is followingsections will discuss distributionsof plant based, however,on the objective data of the field populations and trends in communitycomposition transects;and the patternsof species distributions shownby these transects.The whole body of tables are essentiallythe same in the field transectsand cannot be publishedhere. Two tables have already other compositetransects. been published (Whittaker 1951), and the Othercomposite transects were made for elevations tables for tree populationsare presentedhere (with above 4500 ft in subalpineor spruce-firforests, and extensionof elevationintervals from 200 to 400 ft in high-elevationdeciduous forests outside the range in tables 5 and 6). The full set of tables for tree of spruceand fir. In thesethe sampleswere grouped populations and undergrowthspecies are available by topographic position rather than by weighted to those desiringthem (see Note on Supplementary averages. The various compositetransects were de- Publication).

TABLE 2. Compositetransect of moisturegradient between2500 ft and 3500 ft,distribution of trees along gradient. Transect along the moisturegradient from mesic valley sites (Sta. 1) to xeric southwestslope sites (Sta. 13), based on 67 site counts including 6122 stems from elevationsbetween 2500 and 3500 ft. All figuresare percentages of total steussin station from1-in, diameter class up.

STATION NUMBER Tree species?. -- 1 2 3 4 5 6 7 8 9 10 11 12 13

Acerspicatul. 4 X . Fraxinusamerseana. . 2 1 2 1 1 . . . . . Tilia heterophylla...... 17 .. 14 16 1 ...... Aesculu octandra...... 7 1 4 2 x . . . . . Fagus grandifolia...... 1 2 1 x 1 . . . . . Acer saccharum. 6 3 14 9 1 x . . .. . Magnolia acuminata...... x . 1 x 1 ...... Ilex opaca....3 x . . . . . Prunus serotina...... 1 x x x . . . . . Tsugacanadensis...... ,. 25 49 6 8 9 .. 1 . . Betulaallegheniensis...... ,. 26 10 13 3 5 .. X . . Liriodendrontulipifera...... 2 2 1 3 10 1 1 x . .. . Halesia monticola...... 5 19 28 17 3 7 1 2 1 . Magnoliafraseri...... 2 5 1 2 .4 4 1 1 2 . Acer pensylvanicum. 1 . 2 7 1 3 1 1 .. .. . Betulalenta. 2 2 5 4 5 `'2 2 1 1 Acerrubrum.. x 1 2 9 27 37 35 24 16 16 7 5 3 Ilex montana .. x 1 x x .. x . . Quercusborealis v. maxima. 1 1 1 2 1 4 2 2 . Cornusfiorida...... ,- -,x 1 3 x 2 . . . Hamamelisvirginiana ...... 1 15 6 17 9 3 .. . Ostrya virginiana.. . . x 1 ...... Caryaglabra.. .. 1 x 1 .. 2 8 1 . Clethraacuminata ...... x 4 2 .. 1 . Araliaspinosa...... x x .. 1 . Carya tomentosa.. .x ...... 3 .. . Pyrularia pubera...... x . . Amelanchierarborea ...... X Castaneadentata (dead*)... . 2 1 1 5 10 10 15 7 5 1 Robiniapseudoacacia ...... x 2 1 2 9 6 1 .. 1 Oxydendrumarboreum ...... x 8 9 11 8 16 9 3 1 Quercusprinus...... x 5 5 13 15 23 18 9 5 2 kSassafrasalbidum ...... X 1 1 2 2 2 x 1 1 Nyssasylvatica...... 1 3 X X 3 2 2 3 3 Quercus velutina...... 1 4 .. x 1 Quercus alba...... x 1 1 x Quercuscoccinea...... 1 3 1 25 15 8 10 Pinus rigida...... 2 2 16 20 15 40~ Pinus pungens...... 1 5 3 38 60 41 Percentsby classes Mesie. .97 96 84 61 37 12 4 3 3 Submesic..3 4 13 38 45 64 57 42 22 16 7 3 Subxeric .. *. 2 1 17 24 38 48 66 40 20 12 6 Xeric...... 1 6 8 44 73 83 91

Trees in stations...... 335 480 592 675 418 516 620 357 300 425 438 412 554 Site-samplesused..5 3 6 6 5 6 5 6 5 4 5 6 5

x, Presentbelow .5%. *Dead chestnuttrees were counted in all stands. Sincethe smaller stems had ceasedto he identifiableas such in 1947, the numberof chestnutsin the tablesis smaller thanthe numberof livingstems would have been (see size distributionsin AppendixC). I

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions R. H. WHITTAKER Ecological Monographs, 8 Vol. 26, NO.1 DISTRIBUTIONS OF SPECIES ALONG THE quence of species populations along the moisture MOISTURE GRADIENT gradient is similar at all elevations below 4500 ft. Distributionsof tree populations along the mois- but differsin detail because of the varied relationsof ture gradientare shown in the three tables for dif- species populationsto elevation. From mesic sites to ferentelevation belts (1500-2500 ft, Whittaker1951, xeric, major tree species have their population table 1; 2500-3500 and 3500-4500 ft, present work, maxima in the sequence: Aesculus octan-dra*, Tilia tables 2 and 3). Almostall species show a rounded heterophylla, Betula allegheniensis Britt., Halesia or bell-shapedcurve of population distributionalong monticola (Rehd.) Sarg., Acer saccharum, Lirioden- the gradient (see Figs. 2, 3, 4). Population curves dron tulipifera, Tsuga canadensis, Quercus borealis v. for differentspecies, including many of those in dif- maxima (Marsh.) Ashe, Carya glabra, Acer rubrum, ferentmoisture classes, overlap broadly. Population Carya tomentosa, Castanea dentata, Quercus prinus, centersfor species and limits of their distributions * Nomenclature follows that of Fernald (1950) except where are well scatteredalong the gradient. The basic se- authorities are given. TABLE, 3. Compositetransect of moisturegradient between 3500 and 4500 ft, distributionof trees along gradient. Transect along the moisturegradient from mesic valley sites (Sta. 1) to xeric southwestslope sites (Sta. 12), based on 46 site counts including 4906 stems from elevations between 3500 ft and 4500 ft. All figuresare percentages of total stems in station from 1-in, diameter class up.

STATION NUMBER T ree species -______- ______1 2 3 4 5 6 7 8 9 10 11 12

Fagus grandifolia. 10 5 1 1 1 .. . . . Ilex opaca.. 1 ...... Picea rubens .. . x x .. . . . Cornusalternifolia..1 1 .. x x .. . . . Aesculusoclandra...8 9 4 2 6 1 . . . Tilia heterophylla. 29 11 9. 1 14 3 ...... Acer spicatum.. 16 11 . 17 1 ...... Acersaccharum.. 17 7 1 1 5 1 . . . Prunusserolina.2 .2. 1 1 x 2 . . . Fraxinusamericana..1 1 .. 1 1 X . . . Betulaalleghe niensis...... 5 17 10 15 4 1 x .. . Magnoliaacuminata....x *.1 . . . Magnoliafraseri...2 Tsuga canadensis..20 22 34 62 18 x x 1 . Halesia monticola..5 8 4 1 9 13 3 1 1 Ilex montana..:1 x 1 1 1 2 .. . Acerpensylvanicum..1 x 1 3 8 3 x 1 . Amelanchieriaev~is.. x x x Quercusborealis.. 1 . .. 2 40 10 4 15 11 2 1 Acerrubrum.. 1 .. 1 6 37 21 13 10 8 1 I Prunus pensylvanica.. 2 .. 1 .. . Betulalenta. . 1 4 4 1 2 2 . Clethraacuminata.. 1 x .. Hamamelisvirginiana...... 2 5 17 7 1 .. 2 Cornusfiorida...... 1 .. x 4 . .. Liriodendrontulipif era ...... 2. . .. 1 .. x Rhododendroncalendulaceum ...... 1 .. 1 4 Caryaglabra.. . . 4 x 2 6 5 Carya tomentosa...... 2 . Carya ovali~s.. . . - x Nyssasylvatica.. 1 . .. . 2 4 1 *2 7 Oxydendrumarboreum...... x 1 . 1 3 8 14 16 1 1 Castaneadentata (dead).. . 2 5 7 9 10 12 1 Sassafrasalbidum.. .. . 1 1 1 1 4 x Quercusalba.. . 2 1 8 24 10 x Robiniapseudoacacia...... 4 5 1 3 8 3 x Quercusprinus.. . . 3 4 15 4 16 11 1 Quercusvelutina...... x x 1 1 . Quercus coccinea.. . .1 .. . .1 Pinus rigida...... 7 1 1 11 46 Pinus pungens...... 1 4 54 49 Percentsby classes Mesic..98 98 95 90 78 22 5 3 1 Submesic..2 2 4 9 19 62 70 44 39 26 12 2 Subxeric.. 1 1 2 16 23 46 58 69 23 2 Xeric.. . . . 1 7 2 5 65 96 Trees in stations...377 597 520 232 449 594 472 266 369 378 297 355 Site-samplesused. 1 7 4 3 4 4 4 4 4 4 3 4

x, Presentbelow .5%.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 9.

80? 60 60 0 b d

40? Cl)

1L20. U- 0. 0~ 40- b w 0)30- L)30-

C,. - W42?0 LJ80-z~40-INT - ~~~M1TR -XE < 60- w 4 0 20. a. 2 3 4 5 6 7 8 9 10 1 121 5 6 7 8 9 10 II 12 13 w OSUEGAIN MESICI _- _ _ _ _MOISTURE GRADIENT XERIC MESIC MOISTURE GRADIENT -- XERIC FIG. 3. Transect of the moisture gradient, 2500-3500 FIG. 2. Transect of the moisture 1500-2500 gradient, ft. Top-curves for tree classes: ft. Top-curves for tree classes: a, mesic; b, submesic; a, mesic; b, submesic; c, subxeric; Middle-curves for tree c, subxeric; d, xeric. Middle-curves for tree species: d, xeric. species: a, Balesia monticola; Acer coc- a, Betula allegheniensis; b, Cornus florida; c, Quercus b, rubrum; c, Quercus prinus; d, Pinus virginiana. Bottom-curves for under- cinea; d, Pinus rigida. Bottom-curves for undergrowth growthcoverages: a, herbs; b, shrubs. coverages: a, herbs; b, shrubs.

Quercus alba, Q. velutina, Q. coccinea, and Pinus species include Dryopteris spinulosa v. intermedia, Athyrium thalic- ,virginiana, P. pungens, and P. rigida. thelypterioides, Caulophyllum troides, Cimicifuga racemosa, Comparable data on population levels are not Eucpatorium rugosum, Laportea canadensis, Impatiens and Aster available for shrubs and herbs, but these appear to pallida, diraricatus. These extend varying distances into sub- be distributed in the same manner as the trees. Spe- mesic forests. Other species which are important in cies populations overlap widely along the gradient, mesic and centers and limits of distribution are scattered sites (Smilacina racemosa, Polygonatum spp., along the whole of the gradient. Desmodium nudiflorum, Polystichum acrostichoides) are major herb species also in submesic sites. The Among the shrubs Rhododendron maximum is the latter have been grouped with those more clearly most important species in mesic sites, but it is a centered in submesic sites (Aureolaria laevigata major species in submesic sites also and occurs in (Raf.) Raf., Prenanthes trifoliolata, Medeola rir- subxeric and some xeric ones. Hydrangea arbo- Dryopteris noveboracensis, Veratrum parvi- rescens is the only other shrub species very widely giniana, fiorum Michx.) into a submesic class. A number of distributed through mesic and submesic forests. these species extend widely into subxeric where Leucothoi editorum occurs only locally in mesic for- sites, they are joined by others (Campanula ests; other mesic shrub species are restricted to low divaricata, Chimaphila maculata) of more limited extent into elevations or high ones. In submesic sites a number submesic sites. Galax aphylla, the most of deciduous species make up the shrub stratum along important subxeric herb species, is with the evergreen ericads Rhododendron maximum widely distributed from submesic sites to most xeric ones. Other herb species are and Kalmia. latifolia. Among the major species of centered in xeric sites; most of these submesic shrubs some (Viburnumc acerifolium, Caly- (Pteridium aquilinum v. canthus fertilis, Pyrularia pubera) extend more latiusculum, Tephrosia virginiana, Bap- tisia tinctoria, Gaultheria procumbens) extend widely into mesic sites, but these and others (Gay- widely into subxeric sites, and some of them lussacia ursina (M. A. Curtis) T. & G., Clethra acu- (Epigaea rep- minata, Rhododendron calendulaceum, Smilax rotundi- ens, Panicum sp., Coreopsis major, Andropogon sco- extend into folia) extend varying distances into subxeric and parius) submesic sites, in part of their elevation least. xeric sites. In subxeric sites and some xeric ones, range, at More complete lists of herb and shrub Kalmia latifolia is the principal shrub species. Ly- species assigned to moisture classes are given in the "Summary of Distributional Groupings." onia ligustrina, Smilax glauca, and the widespread There is no point along the gradient at which species Vaccinium constablaei A. Gray may best be either floristic composition or dominance changes grouped with it in a subxeric class. Several shrub abruptly in any stratum. Rather than this, the species (Vaccinium vacillans, V. hirsutum Buckl., V. rounded and tapered distributions of species popula- stamineum, Gaylussacia baccata, Pieris floribunda, tions, the scattering of their distributional centers Hex montana v. beadlei (Ashe) Fern.) are centered and limits along the gradient, and their broad over- in xeric sites and extend varying distances into sub- lap with one another imply gradual and progressive xeric and submesic ones. change in relative importance of species and in total A number of herb species are centered in mesic floristic composition from one extreme of the gradi- forests and dominate the herb stratum there; major ent to the,other.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 10 R. H. WHITTAKER EcologicalVol. Monographs 26,No. 1 Herb species are less easily classified,but trendsin importancemay be observedamong the more numer- 60 d 14 ous growth-formswhich might be recognized. Ferns 40- with delicate foliage (Dryopterisand Athyrium)are 0~ centered in mesic sites and decline in importance I- 2 throughsubmesic into subxeric ones. A group of herbsof moderatestature with broad, thin leaves and a characteristicspreading or umbrella-shapedgrowth- a. 10- form (Caulophyllum,Cimicifuga, Actaea, Impatiens, Trillium, Laportea, Osmorhiza, Thalictrum,Eupa- torium rugosum, Aster divaricatus) prevail along with ferns in mesic sites and are of decreasingim- 460- portancetoward more xeric ones. Other herb forms 0- -rosette plants (Goodyera pubescens, Veratrum parviflorum, Viola hastata) and I 3 4 5 7 8 9 10 II1 12 those with leaves GRADIENT MESC-- -- MOISTURE ----XRC spaced along the stem (Aureolaria, Solidago, Smila- cina, Uvularia, Melampyrum,Coreopsis)-are more FIG. 4. Transect of the moisture gradient, 3500-4500 importantin submesicand subxericsites; and foliage ft. Top-curves for tree classes; a, mesic; b, submesic; c, subxeric; d, xeric. Note expansion of mesic stands, of herbs in these sites is, on the whole,tougher than compared with Figs. 2 and 3. Middle-curves for tree that of the delicate-leavedmesic herbs. Species of species: a, Tilia heterophylla;b, Halesia monticola (both these groups occur also in xeric sites, but a variety the preceding are bimodal, with populations on each of other herb side of the mode of Tsuga); c, Tsuga canadensis; d, types prevail there: grasses (Andro- Quercus alba; e, Pinus pungens. Bottom-curves for pogon, Panicum), ground heaths (Gaultheria, undergrowthcoverages: a, herbs; b, shrubs. Epigaea), legumes (Baptisia, Tephrosia), a tough- leaved fern (Pteridium), and a club-moss(Lycopo- TRENDS IN RELATION TO THE MOISTURE GRADIENT dium obscurum). Of these the grasses are the major Various trends in community composition and herb growth-formin xeric sites at lower elevations; structure can be followed from one extreme of the and the ground heaths are the major herb growth- moisture gradient to the other. These trends, com- form in suhxericsites and in xeric ones at higher parable to those already studied in foliage insect elevations. communities (Whittaker 1952), are in most cases COVERAGES continuous through whatever community-types or In general,tree coverageand densityof the canopy associations may be recognized. decrease along the moisturegradient from cove for- GROWTH-FORMS ests into pine forests; light penetrationto lower Four growth-formsof trees are recognized in the strata consequently increases along the gradient Smokies (see Part III ): pines, abietines (Tsuga (Whittaker1952). Estimatedtree coveragesincrease, canadensis), oaks, and other deciduous trees. A con- however,from subxeric sites (oak-chestnutheath) tinuous shift in proportions of these appears along into xeric ones; the verylow canopy coveragein oak- the moisture gradient (Whittaker 1953 :49). De- chestnutheath is in part a consequenceof death of ciduous trees other than oaks predominate in mesic the chestnuts. Shrub coverage in general increases sites, oaks in intermediate sites, and pines in xeric along the gradienttoward more xeric sites (Figs. 2, sites. Toward higher elevations a belt in which 3, 4). This trend is modified,however, by the pres- Tsuga canadensis is dominant is interposed between ence of a secondarymaximum of shrub coverage in the first two of these. The pattern of growth-form hemlockforests in mesicsites, and by a finaldecrease of composition, and the predominance of the semi-sclero- shrub coveragein mostxeric sites. Herb coverage in phyllous, deciduous oak grouping in intermediate generaldecreases along the gradientfrom mesic to xeric sites, is the same whether or not Castanea dentata, sites. This trend also is modifiedin two re- Fagus grandifolia, and the ericaceous tree Oxyden- spects-by verylow coveragesin hemlockstands, and drum arboreum are grouped with the oaks. Among by a final increase of herb coverage from subxeric the shrubs a comparable shift in growth-form com- sites into xeric ones. Maximumherb coveragesoccur in position appears, involving deciduous and evergreen, mesic deciduousforests, where moisture conditions ericaceous and non-ericaceous species. Deciduous are most favorable,and in xeric pine forests,where non-ericaceous species predominate in mesic sites, light penetrationto the herb level is greatest. Herb with some exceptions; but deciduous species decline and shrub coverages show a clear inverse relation in importance along the gradient as evergreen eri- withinthe set of transectsfor elevationsbelow 4500 ft in cads increase to become strongly predominant in sub- the Great Smoky Mountains (Figs. 2, 3, 4). xeric sites. Toward the xeric extreme, evergreen DIVERSITIES ericads decline and deciduous ericads (Vaccinioideae Diversity of the tree stratum can best be ap- 'or Vacciniaceae) increase to dominate the shrub proachedthrough the alpha values of Fisher (Fisher stratum in most xeric pine heaths. et al. 1943.; Williams 1947, 1950; Whittaker1952).

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 11

These values provide a measurement of richness in shows a secondarymaximum of both diversityand species which is, within limits, independent of coverage in xeric sites. sample size. In Fig. 5 alpha values for composite SIZES AND NUMBERS OF STEMS OF TREES stand counts are plotted on the vegetation pattern for Stature and stem diameterof canopy trees in gen- the Smokies developed in Part III. At all elevations eral decrease along the moisturegradient. In mesic highest diversity values are in intermediate sites-in sites canopy trees are morethan 100 ft high and 3-4 the cove forest transition below 3000 ft and oak- ft or morein diameter;in xeri sites theyare mostly chestnut forests above 3000 ft. The hemlock stands, 50-75 ft highand 1.0 to 1.5 ft in diameter.The num- which provide exceptions to all the trends discussed, ber of tree stems per unit area in general increases are less diverse than the more and less mesic stands along the gradient (cf. Ilvessalo 1921, Lutz 1932), on each side of them. In general, however, species in inverserelation to tree stature. The cove forests diversity of the tree stratum rises along the gradient have mostlybetween 750 and 1000 stems,per hectare from one minimum in most mesic sites to a maximum from the 1-in. class up (except in stands of higher in submesic sites and declines to a second minimum in elevationswhere there are many small stems of Acer xeric sites. spicatum); the more xeric stands have mostly2000 to 2500 stems per hectare. In part the increase in VEGETATIONOF GREATSMOKY MOUNTAINS stem numberstoward xeric sites reflectsthe smaller PATTERNOF EASTERNFOREST SYSTEM stature and denser growthof canopy trees; but the numeroussmall stems in more xeric sites are pre- 65W- dominantlymade up of small-treespecies. These small-treespecies (Carpinus caroliniana, Magnolia 6000-_ BALD] tripetala, Ostrya virginiana, llex opaca; Cornus [BOREAL FORESTS 2.0 2,0 florida, Betula lenta, Acer rubrum, Hamamelis vir- 550 giniana, Clethra acuminata, Acer pensylvanicum; Robinia pseudoacacia, Oxydendrum arboreum, Sassa- IF ESTS 5000 ~~~~BEECH Quercus marilandica) MESIC TYPE [ETYP fras albidum; are relativelyun- importantin most mesic sites (as low as 1-2% of 4500------2 stems in some cove forests) and most xeric sites

Z Cz~~~~~~~~~~~~~~~~TABL (10-15%). In submesicand subxericstands of lower 0 ~~~~~~~~~~~~/MOUNTN 4000- ~ ~ ~ .K 0~ PINL9 and middleelevations, however, the small-treespecies / ~~~~~~~~~~HEATcomprise around 50% of stemnumbers. Treindsin stand compositionhave been much affectedby death of the chestnuts(Castanea dentata). Aft~~~~~~~/4. 2.0 In many submesic and subxeric stands chestnut I PINE 27 5 SI formed30-60%, of the canopystems, and deathof the chestnutsboth removedmany of the largest stems fromthe stand and permittedheavy reproductionof I \ ~~~~~~IZ other species. Effectsof death of the chestnutsare 2000- 4.7 5,3 ~~~~~~~~~~~~~~~VIRGINIA 5. Patter of1trePIi iesiis(TCHdi PINE most evident in chestnut oak-chestnutforests, in whichmaximum numbers of tree stemsper unit area IG.. now occur, and in which 70% of the sternsin somie I COVES FLATS DRAWS SHELTERED PNE SLOPIESW PEAKS stands are now of the small-treespecies. NE CMVS RAIIN SLOPES SW RI)GES& Trends in tree sizes are illustratedin a familyof FIG.. 5. Pattern of tree species diversities (alpha di- curves (Fig. 6), in which steepnessof slope reflects versityvalues, for all tree stems in the composite stand normalsurvival of small treesinto larger size classes. counts of Appendix C). The more xeric the site, the steeper the curve and the smaller the proportionof growthand survival Alpha values cannot be computed for the under- into larger size classes. The oak-chestnutcurve is growth data available. Analysis of the transects altered by death of the chestnutsand increased re- through average numbers of species listed per sample productionof other species; the dotted curve is an provides a more limited indication of diversity trends interpolationof what might be expected otherwise. in the herb and shrub strata. For shrubs the average The hemlockforests are exceptional,for large sizes numbers of species recorded in mesic, submesic, sub- are even moreheavily represented than in cove hard- xeric, and xeric stands are: 5.2, 7.6, 6.2, 6.6. A sub- woods forests. Fig. 7 indicatesthe effectof the same mesic maximum corresponding to that for trees is gradienton growthand survivalin the populationsof stratum in xeric sites thus suggested; but the shrub red maples (Acer rubrum). may be more diverse than that in subxeric ones, often strongly dominated by Kalmia latifolia. Correspond- SELF-MAINTENANCE OF STANDS ing average numbers of herb species are: 19.1, 10.6, Curves such as those illustratedin Figs. 6 and 7 7.1, 8.6. The herb stratum is thus richest in species, are expressionsof the dynamicsof stands, the man- as well as of highest coverage, in mesic sites and ner in wiich the tree population is maintainingit-

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 12 R. H. WHIITTAKER EcologicalVol. Monographs26, No. 1

900 8, 80 T0 60 60 50 40 40- 30

20 * 2c

10 B 9 6 6

(

07 2A

0,9 0.6 0.7-

F2 a 6. Ste nubrdaetrcreor te sad 0.4 Cr 4m ndfeetsts 026

Stveso 194 ) ee Seesn( 93)hv 0.3

tai0.2 thmevs(oe 95% Witkr15).Cei

1-3' 4-6" 7-9" 10-12' 13-I5S 16-I8S 191-21' 22-24k -' 4-6" 7-9 10-42" 13-15' 19-21" 22-24 DIAMETER 16-18" DIAMETER FIG. 6. Stemnumber-diameter curves f or tree stands FiG. 7. Stem number-diametercurves for Acer ru- at middleelevations. bru~min differentsites. self or failing to do so (Paczoski 1928, Meyer & duction and replacementwhich these curves imply Stevenson 1943). Meyer & Stevenson (1943) have is by no means the only way climax stands can main- indicateda relativelysimple relation of diameterand tain themselves(Jones 1945, Whittaker 1953). Cyclic stem number,which plots as a straightline on a log reproductionseems to occur in the Smokies pine and linear graph like those of Figs. 6 & 7. Such a stands (Part III). Otherconiferous stands are "stag- plot implies that growthrate and survival rate are nant" in the sense that stems are concentratedin largely constantwith age; variation of growthrate largersize-classes, with inadequate numbers of smaller with age introducesinto log-linearplots the curva- stems to replace them if a constant survival ratio ture to be observedin Figs. 6 & 7. A curvefollowing is assumed. Some of the stands more stronglydomi- the stand data more closelyhas been developed. As- nated by Tsuga canadensis are of this form (cf. sumingx = arw to be a fitfor the stem number-age Meyer & Stevenson1943), as are some of those of relation,and w = (y + d)C a reasonableapproxima- the spruce-firforests, especially the high-elevation tion for the age-diameterrelation, then the stem stands of Abies fraseri. It seems likely that repro- number-diameterrelation becomes x = ar(Y + d)-C duction in these stands is periodic,partial or com- In this x is the numberof stemsin a diameterclass, plete destructionof the canopy permittingits re- a is the numberof stemsin the initial class, r is the placement at irregular intervals,rather than con- survival ratio between successive classes, y is the tinuously.If such limitationsare kept in mind,how- middle diameterof the class, and d and c are con- ever, analysis of all-age stand compositionmay con- stants relating diameter to age. (An alternative, tributeto the difficultproblems of climax identifi- purelyempirical form which is less difficultto apply cation. is the series a, ar, ar(r-b), ar(r-b) (r-2b) . . . y The basic similarityof the curves for different in whichb is arbitrarilyintroduced to help the curve parts of the gradient may be observed in Fig. 6. fit.) Curves for individual tree species differwidely in The value of such curves is in the possibilityof slope fromthose for whole stands, but Fig. 7 (and recognizingthe self-maintaining,climax condition the stand data for otherspecies, Appendix C) indi- they describe. Many all-age and probably climax cate the same basic similarity. Apart from certain stands whichhave been analyzed show the basic form distortionsof the curves clearly produced by death illustrated.It is also true that the continuousrepro- of the chestnuts,there is no evidence that any of

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 13 these undisturbed stands are changing toward other are morexeric; and thesemay be grouped into three types. All, from cove forest to pine forest, have the stages: shelteredsouth slopes and south-facingsides self-maintaining properties of climax stands, so far of gaps, supportingbeech forests; more xeric open as can be determined. Evidence of convergence to- slopes supportingred oak-chestnutforests, and most ward a single climatic climax type is thus lacking. xeric open south- and southwest-facingslopes, sup- red and white oak-chestnutor white oak- HIGH-ELEVATION DECIDUOUS FORESTS porting chestnut. Some most exposed summitsof peaks, fi- In the southwestern Smokies, outside the range of nally, are coveredby grassy balds. Distributionsof spruce-fir forests, deciduous forest types extend to tree species may be observed along this five-stage the highest peaks (around 5500 ft or 1680 in). In transect (Table 4); distributionsof shrub and herb order to study the distributions of plants in these species are not publishedhere (see Note on Supple- deciduous forests above 4500 ft. a transect by site mentaryPublication). was arranged for 37 site-samples. The most mesic Relations of tree species to the moisturegradient sites available at these elevations are gaps and con- are in general the same at high elevationsas at low cave slopes of north and northeast exposure, where ones. Halesia monticola,however, which is a highly forests of Fagus grandifolia mixed with other mesic mesic canopy tree at lower elevationsis a submesic trees occur. The south, southwest, and west exposures small-treespecies at these highest elevations; this species comprisestwo population-typeswith separate TABLE 4. High-elevation deciduous forests, transect centers(see Part II and Appendix A). of exposure gradient by topographic sites for Eastern distributional Forest System types above 4500 ft. Distributions of Viburnumalnifolium, Cornus alternrifolia,arid Hy- trees by percentages of stems in stand. Steps in gradi- drangea arborescensare major shrub species at the ent: 1, beech-mixed forests in sheltered north slopes; mesic extreme,Vaccinium constablaei and Rhodo- 2, gray beech forests in sheltered south slopes; 3, red forests. oak-chestnutforests, open slopes; 4, white oak-chestnut dendron calendulaceumin the oak-chestnut forests, open south slopes; 5, grassy balds on exposed Vacciniutmconstablaei spans the whole of the gradi- peaks. ent fromnorth-slope beech stands to grassy balds, as do the less frequentspecies Ribes rotundifoliiumand STATION Rhododendron catawbiense. Those shrub species Tree species (Kalmia latifolia,Lyonia ligustrina,Gaylussacia bWc- 1 2 3 4 5 cata, Vaccinium vacillans, V. hirsutum.)which are Acerspicatum . 14 1 mostabundant in the forest-heathtypes at lower ele- Aesculusoctandra ...... 11 2 x .. vations are limited to the oak-chestnutforests and Betulaallegheniensis...... 10 3 x .. grassy balds in the transect. Aronia melanocarpa Acerpensylvanicum..... 1 1 x Acersaccharum.. 1 x. . (Michx.) Ell. and Viburnum cassinoides, species Tilia heterophylla.. x. whichoccur in the heath balds, were recordedin the Sorbusamericana.... x transectonly from the grassy balds. Cornusalternifolia. . .x . Fraxinusamericana.x...... At themesic extreme, species of themesic and high- Amelanchierlaevis...... 1 1 x elevation mesic herb groupings dominate the herb Fagus grandifolia...... 50 81 7 .. stratum; some of these species extend into south- Ilex Montana...... 1 4 slands. Prunusserotina ...... x x slope beech stands and red oak-chestnut Halesia monticola...... 2 8 11 3 Carex aestivalis is stronglydominant in south-slope Quercusborealis...... 2 1 33 30 beech stands and extendsinto bothmore mesic north- Tsuga canadensis...... 1 2 x beech stands and less mesic red oak-chestnut Acerrubrum... 2 1 11 4 slope Hamamelisvirginiana. . . 10 forests. Athyriumfilix-femina v. asplenioides is a Betulalenta. ..4: ' ' ' ' 4-.. .. major herb species of thesehigh-elevation forests and Vacciniumconstablaei...... 1 extends along the gradient from north-slopebeech Rhododendroncalendulaceum. . 1 Medeola vir- Magnoliafraseri. . . . x stands to white oak-chestnut,as does Magnolia acuminata...... x giniana. Epigaea repens, Galax aphylla, Pediculari& Oxydendrumarboreum. .x.... x. . .. canadensis,Pteridium aquilinum v. latiusculum.,and Castaneadentata (dead).... 1 11 28 Campanula divaricata, species of more xeric forest Sassafrasalbidum. . . . . x 1 Quercusalba...... 1 27 types at lower elevations,are limitedto the oak-chest- Robiniapseudoacacia. . . 1 3 nut stands in the high-elevationtransect. Ecotypic Nyssa sylvatica. . ... 2 populationsof someforest herb species (Angelica tri- Quercuwsvelutina ...... 1 Pre- Prunus pensylvanica...... quinata, Stellaria pubera.,Rudbeckia laciniata, Pinus pungens... nanthes altissima, Houstonia serpyllifolia,Gentiana Pinus rigida...... decora) occur in the grassy balds with a varietyof Pinus strobus.. other species (see Part III). Carex aestivalis and Liriodendrontulipifera. . . . otherspecies centeredin the south-slopebeech stands Total stems. 823 643 1211 349 and red oak-chestnutforests above 4500 ft have been Site-samplesused ...... 11 8 9 4 6 groupedin a high-elevationsubmesic herb union. Stratal trends are less clear-cutin these forests x, presentbelow 0.5%. -, seedlingsrecorded. than in those of lower elevations. Tree-stratumdi-

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 14 R. H. WHITTAKER Ecological Monographs Vol. 26, No. 1 versity decreases from north-slope beech stands into TABLE 5. Composite elevation transect in mesic sites, south-slope ones, increases from these to a maximum distributionof trees. All figuresare percentages of total stems in station from1-in, diameter class in red oak-chestnut forests, and decreases again into up. white oak-chestnut. Coverage of the shrub stratum ElvtoStation----. 1* 2 3 4 5 6 7 5 9 10 11 decreases from north-slope into south-slope beech Elvtohundred feet.. 16 20 24 28 32 36 40 44 48 52 56 stands, increases through red oak-chestnut to white Tree species oak-chestnut stands and the forest-edge of grassy Fagusgrandsfolia.....6 x x 1 2 6 16 44 71 50 balds, and is low in the grassy balds. Herb coverage Tsugacanadensis.....12 27 14 23 10 5 5 5 1 Halesia 1 12 increases from the north-slope to the monticola.....3 13 18 30 4 1 3 14 south-slope Fraxinusamerican ....x 1 2 2 1 6 1 1 beech stands, is lower in the oak-chestnut forests, and Tilsa heterophyllas.....6 3 3 1.5 15 20 22 4 x is near 100% in the grassy balds. As in forests of Liriodendrontulipifera. 1 8 2 4 1 x x lower elevations, herb and shrub coverages are in- Aesculusoctandra ....4 4 2 5 4 11 14 12 1 2 Betulaallegheniensis .. 1 4 5 16 11 5 8 10 10 2 1 versely related. Deciduous trees other than oaks de- Acersaccharsum.....4 2 4 2 12 5 12 4 1 3 x crease from mesic sites into submesic and subxeric Magnolsafraseri.....2 5 3 5 2 x x x ones, where oaks predominate; evergreen tree species Magnoliatripetala ....1 .1 are almost absent from these Magnoliaacuminata.... .2 1 x 1 1 forests. In the shrub Carpsnuscaroltniana .. 10 1 stratum non-ericaceous, deciduous shrubs prevail in Ilex opaca...... 1 X x 3 mesic sites; but deciduous ericads (Rhododendron Caryacordifosmis. ... 2 x 1 X x calendulaceum and Vaccinium constablaei) prevail Cladrastislutea...... 1 4 Acerspicatum .x 1 4 17 2 in ...... 30 7 13 the oak-chestnut forests. Prunus setsotsna...... x 1 1 1 Amelanehserlaevis ...... 1 3 2 3 DISTRIBUTIONS OF SPECIES IN RELATION TO ELEVATION Cornusalternifolia...... 1 X MESIC SITES Coynus flarida ...... 14 6 13 2 x Progressive change in composition of cove forests Quercusborealis & v. is indicated in the maxima...... 3 2 3 2 1 1 2 5 4 2 elevation transect of mesic sites Amelanchserarborea. . x x 1 X 1 (Table 5, Fig. 8). Species distributions show a Betulatents.5...... 10 2 3 5 2 x 2 1 rounded or bell-shaped form in most cases, overlap Aces penoylvanicum....1 x 6 2 6 2 1 2 2 2 x broadly, and have their centers and limits scattered Acerrubrum ...... 12 15 10 3 2 1 2 1 4 x Ilex montana...... X 1 x X 2 x 6 1 along the gradient. Most major tree species occur Caryaglasbra ...... 2 x 5 1 1 throughout the elevations represented in the transect, Caryatomentosa ...... x x but the sequence of their population centers from Carya osaltso...... x low elevations to high is: QuetusuPrinuss...... 2 1 3 1 1 x Fagus grandifolia ("white" Nyssa sylsatica...... 1 2 x population see Part II), Liriodendron tulipifera, Castaneadentata (dead). 1 x 2 1 3 1 1 Quercusalba ...... 1 2 Oxydendrumarboreum.. 2 4 4 1 x x Pinsusstrobsss...... x 2 x Sassafrasalbidum ....x x Robiniapsesudoacacia. . x 1 X x... x Sorbusassericana ......

Total stems...... 600 841 518 639 793 429 358 468 646 360 406 z4O.. Site-samplesused ....2 5 6 6 5 5 5 6 9 5 2

*Stationsgrouped at 400-ft.intervals (1450-1800 ft., 1850-2200 ft., etc.) I0- x, Presentbelow 0.5%.

.0. Betula allegheniensis', Halesia monticola., Acer sac- I-40 charumY Tilia Aesculus and z .. heterophylla,, octandra,, w Fagits grandifolia ("red" and "gray" populations). The decline toward higher elevations of Tsuga canadensis and Magnolia fraseri does not reflect their true distributions (cf. Appendix A), for toward higher elevations these species are increasingly segregated into hemlock stands which were not included in the transect. The most significant change in composition of stands occurs at 4500 ft; at this elevation 500 2000 2500 5000 350 4000 4500 5000 there is a relatively abrupt shift of dominance from ELEVATION IN FEET other cove-forest species to gray beech (Fagus FIG. 8. Elevation transect in mesic sites, smoothed grandifolia). Some small-tree populations (Acer curves for tree species: a, Liriodendron tulipifera; b, and Amelanchier laevis) are centered near Tsuga canadensis; c. Halesia monticola; d, Tilia hetero- spicatum phylla; e, Alcer sacoharum; f, Alcer spicatum; g, Car- 4500 ft along with Aesculuis octandra and one popu- pinus carolPniana;a h, Betula allegheniensis;-i, Alesculus lation of yellow birches (Betula allegheniensis or B. octandra; j, Gmericana; k, white, Fraxinuts 1, red, and Trees and shrubs centered in m, gray populations of Fagus grandifolia (based on iutea). the transition data for 200-foot intervals). from cove forests to gray beech and spruce-firforests

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 115 form the "ecotonal-mesic union" listed in the Sum- b mary of Distributional Groupings. 40- 30 = Among the mesic shrubs two species (Euonymus americanus and Lindera benzoin) are restricted to 20a lowest elevations; certain others (Cornus alternifolia 0o- and Viburnum alnifolium) are centered around 4500 z40- ft and form part of the "ecotonal-mesic" grouping. ~-30- Rhododendron maximum and LeucothoW editorum Y"20 at all elevations up to about 4500 ft; Hy- occur LL'10 e drangea arborescens occurs from some of the lowest 0 - elevations to the highest recorded in mesic deciduous forests (5500 ft). No relatively abrupt change in z 30- the shrub stratum at 4500 ft is indicated in the 20 G~~ transect. Most of the mesic herb species occur over a wide 20 range of elevations. Many of these species are centered in the more extensive mesic stands of higher 30 elevations; and these species are of varied extents toward lower elevations. Among the major species some (Dryopteris spinulosa v. intermedia, Trillium 1500 2000 250'0 3000 3500 4000 4500 5600 erectum v. albiflorum,Aster divaricatus) occur at the ELEVATION IN FEET lowest elevations sampled (1500 ft). Caulophyllum FIG. 9. Elevation transects in submesic and subxeric thalictroides, Cimicifuga racemosa, and Laportea sites, smoothedcurves for tree species. Above-submesic c and c', canadensis extend downward to elevations around sites: a, Cornus florida; b, Acer rubrum; Quercus borealis and var. maxima; d, Carya tomentosa; 2000 ft, Oxalis montana and Impatiens pallida to e, Carya glabra; f, Hamamelis virginiana. Below- about 2500 ft, Aster acuminatus and Monarda subxeric sites: a, Quercus prinus; b, Sassafras albidum; didyma to about 3000 ft, Streptopus roseuts and c, Castanea dentata; d, Quercus alba; e, Oxydendrum arboreum; f, Robinia pseudoacacia. Senecio rugelia. A. Gray to about 4000 ft, Clintonia borealis to about 4500 ft. (A species may occur lo- Pedicularis canadensis) cally outside the normal range of the population racemosa, Medeola virginiana, shown in the transects.) The last five species and extend to high elevations. others largely restricted to higher elevations form SUBXERIC SITES in the Sum- the "high-elevation mesic" union listed One population of Qwercus alba is largely re- Groupings. In some sites a mary of Distributional stricted to elevations below 2500 ft, the other to composition of the herb relatively abrupt change in elevations above 3500 ft (Table 6, Fig. 9). Quercus aestivalis becomes stratum occurs at 4500 ft; Carex prinus is centered at lower elevations and has its dominant in the gray beech stands of south-facing normal upper limit below 4500 ft. Castanea dentata and other species concave slopes at this elevation, extends throughout the elevation range of the tran- and (7arex (Angelica triquinata, Prenanthes altissima. sect. At elevations below 3500-4000 ft Quercus pri- ft. sp.) occur with it which are absent below 4500 nus and Castanea dentata are dominant subxeric SUBMESIC SITES trees; above these elevations Castanea dentata shares the populations of Distributions of some submesic tree species (Whit- dominance with high-elevation alba and Q. borealis. taker 1951; table 2) are indicated in Fig. 9. Carya Quercus Kalmia latifolia is the dominant shrub in the oak- tomentosa is largely restricted to elevations below chestnut heaths from lowest elevations to about 3500 2500 ft, Cornus florida and the low-elevation popu- ft; Rhododendron maximum and Gaylussacia ursinta lation of Carya glabra (see Appendix A) to eleva- also occur. At elevations above 2500 ft Rhododen- significant change in tions below 3000 ft. The most dron calendulaceum and Vaccinium constablaei are between the tree stratum occurs in the elevations important shrubs; at elevations above 3500 ft Kalmia as Quercus borealis Michx. f. in- 3500 and 4000 ft, latifolia shares dominance with these until Kalmia at higher creases to become the major submesic tree becomes a minor species in forests above 4000 ft. of submesic elevations. Some of the major shrubs Galax aphylla is the most important subxeric herb at acuminata, sites (Rhododendron maximum, Clethra all elevations. Chimaphila maculata, Campanula Gaylussacia ursina, Pyrularia pubera, Viburnum divaricata, the submesic Aureolaria laevigata, and upper acerifolium, Calycanthus fertilis) have their xeric Epigaea repens are major herb species in sub- ft. limits of distribution at elevations of 3500-4500 xeric sites; and all extend through most of the ele- Rhododendron calendulaceum and Vaccinium con- vations represented in the transect (1500-5000 ft). stablaei extend to elevations above 5000 ft. Poly- stichum acro~tichoides, Aureolaria laevigata, and XERIC SITES other major submesic herb species are confined to Pinius virginiana is centered at low elevations and lower and middle elevations, hut some (Smilacina is scarce above 2500 ft; Pinus pungens is centered at

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 16 R.R. H. II. WHITTAKERWHITTAKER Ecological~~~~~~~~~Vol. Monographs26, No. high elevations and is scarce below 2500 ft (Table 7, of the Southern Appalachians, and heath strata occur Fig. 10). Pinus rigida extends from lowest eleva- in most vegetation types of open south-facing slopes tions to about 4500 ft, but is centered between 2500 and ridges. Viewing the ericaceous shrub complex and 3000 ft. Quercus coccinea is centered around 2500 as a whole; these shrubs are of increasing impor- ft and is important in middle- and lower-elevation tance toward xeric sites and toward higher elevations, xeric and subxeric sites; Quercus marilandica occurs and have their maximum coverage and richness in only below 2500 ft. Some principal shrub species of species in the heath balds of steep, exposed slopes. xeric sites (Vaccinium vacillans, Vaccinium sta- and ridges of high elevations in the area of spruce- mineum, Kalmia latifolia) extend throughout the ele- fir forests. The stands with high development of a. vation range of the transect (1500-4800 ft). Gaylus- heath stratum have acid soils (Cain 193-1), both in sacia baccata and Vaccinium hirsutum are largely the more xeric sites and in the more mesic ones with limited to elevations above 2000 ft, Lyonia ligustrina Tsuga-Rhododendron stands. The ground heaths or to elevations above 2500 ft, and Pieris floribunda to evergreen ericaceous herbs (Epigaea, Gaultheria, and elevations above 4000 ft. Chimaphila, with which the diapensiaceous Galax Some major herb species of xeric sites (Pteridium may conveniently be grouped) in general parallel the aquilinum v. latiusculum, Epigaea repens, Gaultheria ericaceous shrubs in their greater importance toward procumbens, and Galax aphylla) occur at all eleva- more xeric sites and higher elevations. The ground tions; but of these only Pteridium is important at heaths characteristically form the majority of herh low elevations. Sericocarpus asteroides is largely re- coverage in the sparse herb strata under dense eni- stricted to elevations below 2500 ft, Baptisiac tinc- caceous shrub layers (except in the Tsuga-Rhodo- toria, Tephrosia virginiana, and Panicum sp. to those dendron stands). below 3000-3500 ft, and Andropogon scoparius and Coreopsis major to those below 4000 ft. Shifts in TABLE, 6. Composite elevation transect in subxerie sites, distributionof trees. All figures are dominance are gradual percentages and changes in composition of total stems in station from 1-in, diameter class up. continuous from low elevations to high; and no well- defined sequence corresponding to that for xeric trees Station..... 1* 2 3 4 5 6 7 8 9 10 (P. virginiana and Q. marilandicac,P. rigida and Q. Elevation,hundred feet--- 16 20 24 28 32 36 40 44 48 52' coccinea, P. putngens) was recognized in the lower Tree species strata. Tsuga canadenss...... 1 1 1 X.. . 1 Liriodendrontulipifeia...... - 1 . 2 1 x x x Halesia monticola...... x 2 x 2 x 2 1 10~ Magnoliafra~eri...... x 3 1 2 x Acersaccharum...... x 0I- Magnoliaacuminata ...... x x Ilex opaca ...... x Betula 1F40-: allegheniensis...... x (0 Amelanchierlaevis ...... x x...... 01 Acerrubrum ...... 20 25 31 26 27 13 13 10 7 3 0I- Acerpensylvanicum...... x 1 1 Cornusfliorida...... 5 9 2 2 C) 1_ Amelanhier arborea...... x ...... x Betulalenta...... x 2 1 1 Caryaglabra...... 4 2 2 5 4 6 I 00 25-00 3000 3500 400O0 4500 Caryatomentosa ...... 3 1 x 1 ELEVATION IN FEET Carya ovalts...... 1 Quercusborealis &v. maxima. 1 3 3 3 2 13 15 23 20 31 FIG. 10. Elevation transect in xeric sites, smoothed Hamamelisvirginiana.....3 3 8 2 5 curves for tree species; a, Pinus virginiana; b, Pinus Pyrulariapubera ...... x x x rigida; c, Pinus pungens; d, Quercus marilandica; e, Clethraacuminata...... 1 1 Quercus coccinea. )lex montana...... x Aralia spinosa...... x x TRENDS IN RELATION TO ELEVATION Nyssasylvatica...... 1 2 2 3 3 2 10 Oxydendrumarboreum.....15 10 18 9 14 16 15 3 GROWTH-FORMS Quercusprinu ...... 24 18 14 18 17 15 19 . .. No trends in growth-form composition of com- Quercusvelutina ...... 4 1 X 2 2 1 1 . 4 Quercusalba ...... 5 4 x 9 14 30 munities as striking as those 42 along the moisture gradi- Quercusstellata ...... - x...... ent appear along the elevation gradient. Viewing Sassafrasalbidum...... 1 x 2 1 4 4 1 1 the vegetation pattern as a whole, three growth- Robiniapseudoaeacia .....1 x 3 3 5 7 6 .4 7 forms-abietine trees, ericaceous shrubs, and ground Castaneadentata (dead)....5 6 10 12 10 12 0 23 12 47 Pin=s strobus...... 10 heaths-are of increasing importance toward higher Quercuscoccinea ...... 4 3 1 2 1 1 elevations. Among the abietine trees, Tsuga cana- Pinus virginiana...... 2 3 densis is of increasing importance from low eleva- Pinus rigida...... -x 1 2 x 1 1 Pinus pungens...... -.---1 .X 4 . .. tions to about 4500 ft; and Picea rubens and Abies fraseri dominate most forest stands above that ele- Total stems...... 290 620 690 689 559 606 139 346 84 89 vation in the northeast half of the range. The great Site-samplesused...... 5 5 5 7 8 8 2 3 2 1 development of ericaceous shrub communities is one *St~ationsgrouped at 400-ft.intervals (1450-1800 ft., 1850-2200 ft., etc.) of the most characteristic features of the vegetation x, Presentbelow 0.5%.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 17

TABLE 7. Compositeelevation transect in xeric sites, distribution of trees. All figures are percentages of total stems in station from 1-in, diameter class up. Station.. .. 1* 2 3 4 5 6 7 8 9 10 11 12 Elevation,hundred feet. 14 17 20 23 26 29 32 35 38 41 44 47 Tree species Tsuga canadensis.. .1 ...... Liriodendron tulipifera...... 2 ...... Liquidambar styraciflua.... x...... Amelanchier laevis.. . . . 7. . . . Cornus florida..x x . . . . . Quercus borealis...... 3 1 C7aryaglabra..x ...... Carya tomentosa.1 x x ...... Carya ovalis.. .x ...... Hamamelisvirginiana...... 1 2 .. 3 Acerrubrum... 7 6 5 14 7 10 2 9 3 14 1. Quercus falcata..1 ...... Stewartia ovata.....x ...... Pinus strobus.. . j 2 . . Robinia pseudoacacia...... xxx .. .. 1 2 .. 1 Sassafrasalbidum....1 ... . 1 2 x Nyssasylvatica. 4 2. 4 .. 3 3 2 3 4 3 j Oxydendrumarboreum...... 8 5 9 6 .. 2 4 1 2 2 2 (Castaneadentata (dead) .5 1 .. x 1 1 3 2 1 1 Quercus velutina..2 x 2 . . . . . Quercusaiba. 1 3 .. . Quercusprinus..x x 2 2 14 3 7 6 11 7 11 Quercusmarilandica.7 78 .. 2 Quercuscocainea10 12 2 8 26 14 9 2 1 2 Pinus virginiana....40 56 48 38 1 1 Pinus rigida. 13 12 10 30 34 28 36 16 39 16 10 Pinus pungens... . . 10 41 28 65 27 62 55 87 Total stems...301 217 212 366 88 572 378 410 246 362 122 90 Site-samplesused...2 4 2 5 2 5 3 5 3 3 2 1

*Stationsgrouped at 300 ft intervals(1250-1500 ft ,1550-1800 ft.,etc.) x, Presentbelow 0.5%,.

TREE STATURES AND STRATAL COVERAGES The transect data do not show the relatively slight decrease in tree statures and coverages toward higher elevations which may occur below 4500 ft. Tree 402 sizes in stands above this elevation are distinctly smaller than in the deciduous and hemlock forests -0. below it. Importance of the small-tree species decreases toward higher elevations, except in mesic stands. In submesic and subxeric stands below 3500 W4- - z6t ft the small-tree species make up 45-70% of stem numbers, between 3500 and 4500 ft 30-50%, and above 4500 ft in deciduous stands 10-45%. Shrub coverages (Fig. 11) in .mesic deciduous forests are lower above 3000 feet than below, except as the ecotonal-mesic shrub grouping forms a secondary 60 maximum at elevations around 4500 ft. Maximum zt 6L0_E S _ shrub coverages in submesic and subxeric stands oc- Ir2150 0020 003 40 050 cur at middle elevations (around 3000 ft); but in xeric stands the heath stratum increases in coverage from low elevations to high. In all four transects herb coverages increase toward higher elevations. ELEVATION IN FEET Although the inverse relation of shrub and herb cov- FIG. 11. Elevation tranlsects,curves for undergrowth erage is suggested in three of the transects (Fig. 11), coverages: a, herbs; b, shrubs. From top to bottom,inl the parallel trends of the two strata in xeric sites mesic, submesic, subxeric, and xeric sites. provide an exception to this relation. Community structure changes profoundly along nized here (mesic deciduous forests, abietine forests, the moisture gradient through three or four physiog- oak forests, and pine heaths and forests) extend from nomic types; but major physiognomic types recog- the lowsestelevations to) 4500 ft and above. Floristic

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 18 WHITTAKER Ecological Monographs R. H. Vo.26,No. 1 change also is more conspicuous along the moisture the possible exception of the herb stratum, maximum gradient than the elevation gradient. No species diversities occur not in the "most favorable" condi- spans completely the moisture gradient, but many tions of the cove forests but in the intermediate ones extend from lowest elevations to 4500 ft and above. of cove forest transition or oak-chestnut forest. Within the limits of the transects so far described and Within a given stratum or community-fraction di- excluding the spruce fir forests, community compo- versity and productivity (or coverage) may not run sition and structure change more radically along the parallel (Whittaker 1952), although they often ap- moisture gradient within a single elevation belt than pear partially correlated. with a change of elevation from 1500 to 4500 or It was suggested in the previous study (Whittaker 5500 ft. 1952) that interpretations based on favorableness of environment may have little meaning unless the DIVERSITY AND ENVIRON MENTAL FAVORABLENESS community property or group of organisms for which en- Species diversity of the tree stratum decreases vironments are to be favorable is specified. Data of from low elevations to high (Fig. 5); the trend is the present study may further indicate that "favor- most apparent within a single moisture-class of ableness" of environment is relative to the group of stands. Tree-species diversity is maximal in the cove organisms under consideration, and that for a given forest transition of low elevations and mesic-sub- group diversity and productivity may define differ- mesic sites, and decreases with any departure from eat centers of maximum "favorableness." these stands toward more or less mesic conditions or higher elevations. It is in the cove forest transition SPRUCE-FIR FORESTS that species of the cove-forest or mesic grouping In studying subalpine forests of the Smokies, site- and the oak-forest or intermediate grouping are most samples from three elevation belts (spruce forests highly mixed; in the interpretation of Braun (1950) from 4500 to 5400 ft, spruce-fir forests from 5500 the cove forest transition forms part of the Mixed to 6100 ft, and fir forests from 6200 ft up) were Mesophytic Association, from which some other, less arranged in topographic transects similar to the diverse stand types may be said to "segregate." The one for high-elevation deciduous forests. Sites were more limited data on undergrowth diversities are grouped along the moisture gradient in five stages- based on averaging the numbers of species recorded moist valley sites, north slopes and flats, intermediate per sample in transects of different elevation belts. east and west slopes, xeric south and southwest Herb strata show some evidence of increasing diver- slopes, and exposed ridges. Table 8 is so arranged sity toward higher elevations; average numbers of as to form a grid of transects covering the spruce-fir species for the elevation belts 1500-2500, 2500-3500, forest pattern; distributional relations to the topo- 3500-4500, 4500-5500 ft (the last for deciduous stands graphic gradient within an elevation belt and to only) are 12.8, 15.5, 16.8, and 18.8. Corresponding elevation within a given type of site may be observed values for the shrub strata (6.1, 7.9, 7.7, 7.0) suggest in the table. Data on species distributions have been only that shrub strata may be less rich in species, on compressed for presentation into stratal unions as the average, at lower elevations than at middle and follows: higher ones. 1. Mesic herb union (Solidago glomerata Michx., Two properties of natural communities, diversity Senecio rugelia, Chelone lyoni Pursh, Diphylleia cy- and productivity, may be thought in general cor- mosa, Aster acuminatus, Trillium erectum, Strepto- related with "favorableness" of environment. Thus pus roseus, Rudbeckia laciniata, etc.). Some species Finnish authors have found parallel trends of de- of the low-elevation mesic herb union (Eupatorium creasing productivity and decreasing numbers of rugosum, Impatiens pallida, Cimicifuga racemosa, plant species along ecological series of site-types Aster divaricatus) are of wide extent in subalpine (Cajander 1909; Ilvessalo 1921, 1922; Linkola 1924). forests and have been grouped here with the high- Data frolu the Smokies indicate how much more elevation union. complex than this the relations of these community 2. Deciduous trees other than Sorbus americana properties to one another, and presumptive favor- Marsh. (Betula lutea, Amelanchier laevis,l Acer spi- ableness of environment, may be. catum and A. pensylvanicum, and trees of the cove Major trends in the Smokies are consistent with forest grouping). the general tendency toward increasing stature and 3. Ground moss unions, described in more detail differentiation into strata of communities with in- by Cain & Sharp (1938). creasingly favorable conditions of moisture, tempera- 4. Low-herb layer-Oxalis montana with Clintonia ture, and nutrients (Dansereau 1951; Whittaker borealis. 1953, 1954a). Stature and coverage of canopy trees 5. Mesic fern union, with Dryopteris spinulosa in general decrease, while the development of under- v. americana most important. growth strata increases, toward most xeric sites and 6. Low shrub union-Vaccinium erythrocarputm highest elevations. Relative development of herb, with Menziesia pilosa. shrub, and moss strata are complexly related to one 7. Deciduous high shrubs (ecotonal-mesic union) another as well as to the tree stratum and environ- -Viburnum alnifolium, with Cornus alternifolia and mental "favorableness," however. Diversity trends in Sambucus pubens. differentstrata do not parallel one another; and, with 8. High heath union-Rhododendron catawbiense

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 19

TABLE8. Transect data for subalpine forests. Percentages of coverage or stems for stratal groupingsin different sites along the moisturegradient in five steps: 1, most mesic valley and sheltered-slopesites; 2, open north-facing slopes; 3, east and west slopes; 4, south-facingslopes; and 5, ridges. Figures are based on averages fromseveral site- samples of stem numbersand estimated coverages.

RED SPRUCE FORESTS SPRUCE-FIR FORESTS FRASER FIR FORESTS Elevationbelt.... 4500-5500ft. 5600-6200ft. above 6200 ft. Site. 1 2 3 4 5 1 2 3 4 5 1* 2 3 4 5 Plant speciesor union-Mosses (Hylocomium splendensetc.), coverage% ...... 5 24 17 8 0 30 55 48 13 2 .. 90 85 50 5 Low herb (Oxalis montana)'.... 2 55 29 23 0 25 54 25 5 0 .. 47 8 3 0 Fern (Dryopterisspinulosa v. americana).. . 30 45 22 8 0 20 39 12 5 0 .. 37 1 1 0 Low shrub(Vaccinium erythrocarpum).. 2 8 5 4 0 5 14 4 5 0 .. 5 2 2 0 High shrub(Viburnum alnifolium) ...... 20 18 7 4 0 20 13 9 1 0 .. 2 1 0 0 Total, "five-storyundergrowth". 59 150 80 47 0 100 175 98 29 2 .. 181 97 56 5 Mesic herbgrouping. 50 5 1 1 0 40 3 1 1 0 .. 2 1 1 0 High heath (Rhododendroncatawbiense and R. carolinianum)... 0 0 5 40 84 0 0 10 30 70.. 0 0 37 88 Picea rubens,% of conifers.... 40 33 48 79 100 10 12 7 40 55 .. 1 2 4 12 Deciduous trees(except Sorbus americana); %of stems..... 30 8 13 7 14 15 13 11 4 5.. 2 1 0 2 Sorbusamericana, % of stems. 0 0 0 0 0 1 1 3 0 0 . 8 8 3 2

*Not observed.

and R. carolinianum Rehd., with Kalmia latifolia, ward higher elevations and less mesic sites. The cove- Vaccinium constablaei, and other species of the heath forest tree species which occur in subalpine forests balds. (Aesculus octandra, Tilia heterophylla, Halesia mon- 9. Mountain ash (Sorbus americana), a mesic small ticola, Acer saccharumr,Fraxinus americana, and the tree of highest elevations. subinesic Acer pensylvanicum) are similarly related to the gradients; their numbers decrease toward STRATAL DISTRIBUTIONS higher elevations and more xeric sites. In the tree stratum composition of subalpine for- Three shrub unions with differentdistributional re- ests changes more conspicuously with elevation than lations have been distinguished in the spruce-fir for- along the moisture gradient. At 5000 ft on south-fac- ests. The distributional relation of the deciduous ing slopes 100% of the conifers may be spruce (Picea high shrubs of the ecotonal-mesic union is the same rubens); above 6300 ft 100% may be fir (Abies as that of Betula lutea and Acer spicatum; their im- fraseri), while intermediate sites have mixed stands. portance declines with departure from mesic stands Spruce and fir cross in numbers of stems at around near 4500 ft. The shrub species of the group differ 5600 ft if exposure differences are averaged out. in extent and importance in the spruce-firforest pat- The change in canopy dominance occurs higher, near tern; Viburnuumalnifolum is the most wide-spread 6000 because of the greater size of spruce. Ele- ft, and much the most important of them. The low-shrub vation alone, however, does not determine propor- union dominated by Vaccinium erythrocarpum is tions in the mixture of these two dominants; at any centered in middle elevations of the spruce-firforests, elevation fir increases relative to spruce toward more rather than lower ones, and in mesic sites (north- mesic sites (Table 8). facing slopes and flats) other than the valley sites Sorbus americana is confined to high elevations where the ecotonal-mesic species are most important. and is centered in the more mesic forest stands above The high heath union is most important on ridges, 6200 ft. Fagus grandifola ("gray" population) but the heath species also occur with lower coverage dominates stands in the concave slopes of "gaps" in in south-slope stands and on some intermediate the area of spruce-fir forests, and it occurs also to a slopes. Of these shrub species Vaccinium constablaei limited extent in stands dominated by spruce and fir. occurs more widely in spruce-fir forests than the All other deciduous tree species which occur in the heath dominants Rhododendron catawbiense and R. spruce-fir forests are at least equally important in carolinianum. deciduous stands below 4500 ft. The high-elevation In valley sites the mesic herbs strongly dominate population of yellow birch (Betula lutea) is centered the herb stratum. Most of the herb species listed in the transition from cove forests to spruce forests are of wide extent in spruce-fir forests (other than at 4500 ft, but is of wide extent at higher elevations coverage is low as the major deciduous tree species of spruce-fir those of most xeric sites), but their forests. Yellow birch and other "ecotonal-mesic" outside valleys. The moss stratum, the low-herb species (Acer spicatum, Amelanchier laevis, Cornus stratum, and the fern stratum are, in contrast, cen- alternifolia) decrease in importance with departure tered in stands of north slopes and flats. These three from their centers in mesic stands near 4500 ft to- unions also are of wide extent through the spruce-

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 2020 R. H. WHITTAKERWHITTAKER Ecological~~~~~~~~~~~~~~~~~~~~~~~Vol. Monographs26, No. 1 fir pattern and decrease in importance along the in contrastwith the patternof forestsat lower ele- moisturegradient from north slopes and flatsto south vations. Deciduous trees decreasein importancefrom slopes and ridges. valleys toward xeric sites; ericaceousshrubs increase Layered structureseems especiallyevident in some in importance from intermediatesites to ridges. vegetationof high elevationsand the far north; the Herb coverage and species diversitydecrease along same strata are to be distinguishedelsewhere but the moisturegradient. There is thus a trend of de- are more obvious and give a greater impressionof creasingimportance of non-evergreenelements in all distinctnessin such vegetationas the spruce-firpat- three strata toward more xeric sites. Valley spruce- ternin the Great SmokyMountains. Analysis shows fir forests resemble high-elevationcove forests in that the various stratal unions,and the species which most respects,even thoughthe canopy is dominated make up a given union, are differentlydistributed. by conifers. At the xeric extreme,however, ever- Among the undergrowthunions recognized,however, green elementsare stronglydominant both in canopy several are of wide extentin the spruce-firpattern and in the dense ericaceousshrub stratum, and ground and are of decreasingimportance from more mesic heathspredominate in the sparse herb stratum. sites toward more xeric ones. Five unions dominate Along the elevationgradient within the spruce-fir the undergrowthof north-,east-, and west-facing forests,moss coverage increasesstrongly from lower slopes and flats,each of the unions being strongly elevations toward higherones (Table 8). Coverage dominatedby a single species. Togetherthey form of shrubsother than the heath is in general lower at a five-storyundergrowth complex (moss, Hyloco- highestelevations, but undergrowthcoverage trends mium splendens; low herb, Oxalis montana; fern, with elevationare not otherwiseconspicuous. Repre- Dryopteris spinulosa v. americana; low shrub, Vac- sentationof deciduoustrees otherthan Sorbus amer- cinium erythrocarpum; high shrub, Viburnum alni- icana decreases toward higherelevations, and (since folium) which is most characteristicof the spruce- Sorbus is a small-treespecies) the fractionof decidu- fir forests of the Smokies, although other under- ous trees in the canopy decreases toward higherele- growth unions predominatein most mesic and in vations. Species diversityof the tree stratum de- more xeric sites. creases toward higher elevations,and the same is probablytrue of other strata of vascular plants. TRENDS The most conspicuous change in canopy of sub- Major structuralfeatures of the spruce-firpattern alpine forestsis the increase of fir and decrease of along the moisturegradient are shown in Fig. 12, spruce toward higher elevations,but the most con-

VEGETATION PATTERNS IN THE GREAT SMOKY MOUNTAINS Change of vegetation along the moisture gradientat lower and higher elevations BOREALFOREST SYSTEM - 5500 FEET SPRUCE - FIR FORESTS HEATH MESIC 5- STORY SPARSE BALD UNDERGROWTH * UNDERGROWTH UNDERGROWTh

EASTERN FOREST SYSTEM- 3000 FEET E FOREST HEMLOCK OAK-CHESTNUT OCHESTNUT PINE HEATH { 4 ~~~~~~~~~~FOREST ~ ~ H1EATH

MIDDLE VALLEY LOWER N.E., S L 0 P E 3 S.W.,UPPER RIDGE

KEY: SUMESIC TREES e EVERCREEN HEATH vGRASS 4 RED SPRUCE 4 CHESTNUT, DEAD VPVACCINIUM RACKEN RASER nR OAKS 9 MESIC HERB GROUND 4EATHS 4 EASTERN HEMLOCK t YELLOW PO" MUSC FERNS OXAES

? MESIC ROADLEA TREES Ay ESIC DECIDUOUS SHRUSS SJMSIC HERBS _ ... FIG. 12. Vegetation patterns along the moisture gradient.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 21 spicuous change in the undergrowth is that along the fraseri and Sorbus americana) show varying degrees moisture gradient. The canopy and understrata thus of distributional similarity. It was not found possible seem at cross purposes, as it were, in the patterning to define vegetation types through groups of tree of these forests. Extensive patterning, or change in species having closely similar distributions in the community composition and structure in relation to transects and nomograms. All species groupings used environmental gradients, appears in these subalpine in the present work are considered to be primarily forests, as in the deciduous, pine, and hemlock for- arbitrary groupings of species with different distri- ests of lower elevations. Differences in site along the butions. moisture gradient at low elevations are reflected as The moisture classes are species groupings which much in species and grpwth-formcomposition of the correspond in part to traditional associations in the canopy as of the undergrowth. In the spruce-fir American sense-mixed mesophytic, oak-hickory, oak- forests, however, as in other subalpine and boreal chestnut, and pine-oak (cf. Braun 1950). Species of forests, canopy trees are broadly tolerant of differ- the oak-hickory and oak-chestnut associations do not ences in sites; and site differences are reflected pri- fall naturally into submesic and subxeric classes, how- marily in undergrowth. ever. Quercus borealis v. maxima and three hickory RELATIONS OF SPECIES TO UNIONS AND ASSOCIATIONS species may be regarded as "submesic"; but other A final group of questions to be considered relate major oak species of the oak-hickory forests (Q. alba to the nature of community-units. The transects offer and Q. velutina) are slightly more xeric than Cas- evidence bearing on three aspects of the community tanea dentata and Quercus prinus and must be classed problem: continuity and discontinuity of stands and with these as "subxeric." Although Castanea dentata community-types,naturalness or arbitrariness of spe- and Quercus prinus are subxeric, the third dominant cies groups, and relation of dominant species to of oak-chestnut forests (Quercus borealis) is dis- minor species and undergrowth. tinctly less xeric and must be classed as "submesic." One of the major species of the pine forests (Quercus CONTINUITYOF VEGETATIONTYPES coccinea) is intermediate in its relation to the xeric Three relative discontinuities affect the vegetation pine and the subxeric oak-chestnut groupings. Spe- pattern described-those between cove forests and cies of the Mixed Mesophytic Association of Braun beech-gap stands of south-facing slopes, between (1950) are widely scattered along the moisture gradi- grassy balds and the forests surrounding them, and ent and have been classed with the mesic, submesic, between heath balds and spruce-fir forests. Apart and subxeric groups. There is thus no clear and from these, community-types are continuous with simple relation between the moisture classes (and one another. The transect data thus suggest that other groupings attempted by the author) and the some vegetation types are relatively discontinuous regional vegetation types or "associations" in the with one another along continuous gradients of "pri- sense of American authors. mary" environment, but that many vegetation types The essential problem may be whether the group- exist within a continuum of populations (cf. Whit- ings of species used by ecologists are "arbitrary" or taker 1951, 1952; Curtis & McIntosh 1951; Brown "'natural,'" and some definitionof these expressions is & Curtis 1952). Problems of continuity and discon- necessary. Any classification of stands or of species tinuity will be further discussed in terms of "grada- is a product of interaction (or of "transaction" in tion" and "zonation." the sense of Dewey & Bentley 1949) between vege- NATURE OF SPECIES GROUPINGS tation and ecologist through observations and sample An alternative approach to the recognition of vege- data. A classification is, like human knowledge in tation types is through groups of species which are general (Carmichael 1930: 182-200, 223), a joint similarly distributed and have their boundaries to- product of the observer and the observed. Distri- gether, presumably at the boundaries of vegetation butional relations of species in the field set limits on types. Inspection of the transect data reveals no the groupings which an ecologist will attempt; some such groupings of species with closely similar dis- combinations of species which rarely occur together tributions, with the possible, partial exception of the are obviously "unnatural." On the other hand, a discontinuous types already mentioned. In general, combination of species which do occur together, al- no two species have quite the same distribution in a though "natural" in the sense that it does not violate set of transects for elevation, or for the moisture obvious distributional relations, may be "arbitrary" gradient. Dissimilarity of distributions is even more in the sense that this combination is only one which evident when species are followed through both sets the ecologist has chosen to recognize from the many of transects or compared in the population charts of which are equally justified by distributional relations Appendix A. Even such familiar pairs of "asso- of species. ciates" as Tilia heterophylla and Aesculus octandra, If groups of species which have closely similar dis- Quercus prinus and Castanea dentata, Quercus alba tributions occur, clearly separated from other such and Carya glabra are distributedquite differently,groups of species, then the ecologist's choice of dis- especially in relation to the elevation gradient. Other tributional groupings is narrowly restricted by dis- pairs of species (Tsuga canadensis and Magnolia tributional relations of species in his data. The dis- fraseri, Pinus rigida and Quercus coccinea, Abies tributional groupings which he recognizes should be

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions R. H. WHITTAKER Ecological Monographs 22 R. H. WHITTAKER ~~~~~~~~~~Vol.26, No. 1 "natural" in the sense that they are clearly deter- sion given by the transect data is one of approximate mined by the data in his possession. If species are randomness, when the distributions of many species quite variously distributed, forming no well-defined are being compared (see Whittaker 1951, Fig. 1). groups distinctly separate from other groups, the The moisture classes of trees are consequently re- groupings recognized are free or "arbitrary." They garded as arbitrary groupings, justified by their use- are not clearly determined by the data, but repre- fulness rather than by any correspondence to distinct sent one among a series of groupings which might clusters of species existing in the field. The locations be recognized without violating the more evident dis- of population modes along the moisture gradient pro- tributional relations of species in the data. Relative vide a basic sequence of species which should not be arbitrariness and naturalness of species groupings violated in forming the groupings; but within these may thus depend on the degree to which the ecolo- limits various ways of grouping species seem equally gist is clearly guided and limited in his choice of reasonable. When major species only are considered, groupings by actual distributional relations of spe- these sometimes appear to have a. low degree of clus- *ies in the field and in adequately representative tering in the moisture-gradienttransects. Thus in the sample data. transect for elevations between 2500 and 3500 ft The basis of groupings in the present study is the (Table 2), six major species (Acer saccharum, Tilia location of centers of maximum population level for heterophylla, Aesculus octandra, Betula allegheni- species along gradients, and the problem of natural- ensis, Halesia monticola, and Tsuga canadensis) have ness vs. arbitrariness of groupings may be further their ecentersin steps 1 to 3 of the transect. Four defined on this basis. Various manners in which spe- major species (Quercus borealis, Acer rubrum, Cas- cies population centers might be related to one an- tanea dentata, and Quercus prinus) are centered in other may be conceived in terms of degrees of cluster- steps 6 to 9, while the two pines (P. rigida and P. ing along a hypothetical gradient from 0 to 10: (1) pungens) are centered in step 12. It is thus possible Species populations are absolutely clustered, so that to regard these as three "natural" groups along the species of one group are centered at 2.0 and those of gradient if the intermediate positions of other, minor another at 8.0, and none are centered between these species and one major one (Quercus coccinea) are points. (2) Species populations are relatively clus- not considered. Such groupings of dominants are, tered along the gradient, so that in the sequence of at best, weak or of low degree among the trees of the their centers along the gradient (1.3, 1.6, 2.0, 2.3, 3.5, Smokies. It may also be observed that the more at- 6.0, 7.2, 7.5, 8.1, 9.0) groupings may be recognized, tention is limited to a few major species, the more even though the centers are somewhat scattered and the irregularity inherent in approximate randomness sonmespecies are intermediate. (3) Species are distrib- may give an appearance of clustering which is not uted irregularly along the gradient (0.9, 1.4, 2.1, 3.5, supported by the distributional relations of other 4.2, 5.5, 6.0, 7.3, 8.1, 9.0), and the spacing of their species. centers approaches randomness. (4) Distribution of Distributions of some species affected by vegeta- centers along the gradient is less irregular than ran- tional discontinuity suggest a degree of natural clus- domness would require, and the centers are approxi- tering. The ecotonal-mesic group of shrubs and trees inately equidistant from one another (1.0, 2.0, 3.0, are centered in mesic sites near 4500 ft. Although no 4.0... two of the species are otherwise alike in distribution The problem of relative "naturalness" of species (see Table 5 and Fig. 13), the group may form a groupings may thus be approached through degrees "natural" cluster. Other transitional groupings will the degrees of "conta- of clustering suggestive of be discussed below in relation to ecotones (Part II). gion" or clustering of individuals in a stand. Con- Some species of other discontinuous types, notably dition (1) would imply that associations of species the grassy balds, may form natural clusters in rela- are "natural" groupings (though community-types tion to the rest of the vegetation pattern of the might still be continuous with one another). Condi- The data for the Smokies thus suggest: tion (2) would define varying degrees of relative Smokies. is absent "naturalness" vs. relative "arbitrariness" of group- (1) Natural clustering of species probably ings which would not necessarily represent either or of low degree in most of the vegetation pattern. obligate associations of species or discrete com- (2) Recognized types and associations are for the munity-types. Groupings of species distributed ae- most part definable by single dominant species or by cording to conditions (3) and (4) would be primarily distributional overlaps of two or more dominant spe- arbitrary. The transect data and nomograms (Ap- cies, but not by natural clusters. (3) A tendency to- pendix A) apparently exclude conditions (1) and ward natural clustering may exist among some of the (4) for the distributions of species in the Smokies. species of some communities, at least if the distribu- The data show no evidence of high degree of cluster- tions of species concerned are considered only within ing according to condition (2). No sensitive statisti- the limits of the Great Smoky Mountains. Such cal tests are available at present to distinguish con- relative clustering as may occur in these cases in- '.ition (3) from low degrees of clustering in condi- volves the relatively discontinuous communities of tion (2) and slight departure from randomness to- more "extreme" environments and the transitions be- ward the regularity of condition (4). The impres- tween these and other communities.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 23

DOMINANCE IN RELATION TO COMMUNITY COMPOSITION growth species occur in several types of stands, The tradition of regarding dominants as the keys dominated by different canopy species. One excep- to natural communities is well established in Ameri- tion to the apparent independence of strata is in the can ecology. The American association was defined close correspondence of the range of gray beech by Nichols (1923) as essentially constant in floristic dominance in the canopy with that of Carex aesti- composition, at least with regard to dominant species; valis in the herb stratum in south-slope beech gaps. and associations of American authors are in general In other cases in which there is difference in domi- defined by dominant species or genera. It was con- nant growth-form between stands, this difference is sidered by Clements that dominant species deter- reflected in undergrowth composition. Stands strongly mined the conditions of life for other species, char- dominated by hemlock differ in undergrowth compo- acterized the community-type,and indicated the habi- sition from those dominated by mesic deciduous trees, tat of the community (Clements 1928 :236, 253; and stands of drier sites dominated by spruce and fir Weaver & Clements 1938:91, 478; Clements & Shel- differ in undergrowth from stands of comparable ford 1939 :238-239); dominants were the master-keys elevations and sites dominated by deciduous trees to the understanding of communities. If such is the outside the range of spruce and fir. In other cases, case, the effectsof dominant species on conditions of however, quite similar undergrowths appear under life for other species should be clearly shown in the canopies dominated by different growth-forms. In distributions of undergrowth and minor canopy spe- the elevations around 4500 ft, similar shrub and cies. Boundaries of other species might be expected herb strata occur in valley spruce forests, deciduous in general to coincide with the distributional limits cove forests, and north-slope beech forests. Similar of dominants and the boundaries of the community- shrub and herb strata occur in oak-chestnut heaths types they define. The transect data provide means and pine heaths of the same elevation, when the of evaluating the hypotheses that (1) Dominant shrub layers of the latter are not dominated by Vac- species through effectson conditions of life determine cinioideae. the distributions of other species, and (2) dominant These evidences of partial independence of strata species characterize community-typeswhich are rela- may give some support to the synusial approach of tively homogeneous in floristic composition. Gains (1918, 1927) and others (Bolleter 1921; Lipp- One of the significant findings of the transect maa 1933, 1935, 1939; Du Rietz 1930, 1932, 1936). analysis is the manner in which floristic composition Distributions of canopy and undergrowth species of communities changes continuously through and cannot be unrelated when both occur along the same between many of the vegetation types defined by complex-gradients, and distributions of undergrowth dominance. In the elevation transects of submesic species can scarcely be unaffected by factors of light and subxeric sites, the dominant species of oaks and soil conditions influenced by growth-forms of change at around 3700 ft, but the limits of minor dominants. The approach to communities through canopy species and undergrowth species are scattered stratal associations or unions has been criticized through the whole of the transect rather than con- (Beger 1922-3, Cain 1936, Gleason 1936), espe- centrated around 3700 ft. In the elevation transect cially on the basis that canopy and undergrowth spe- of xeric sites, dominant pine species change at ele- cies are parts of each others' environments, at least vations around 2300 and 3300 ft; but limits of other during some stages of life-histories. On the other species are again scattered along the gradient rather hand, there is no close and exact correspondence of than concentrated at these elevations. In the de- undergrowth composition and community dominance. ciduous cove forests, various combinations of domi- Probably no single, simple key to composition of the nant species occur which are not, so far as could be whole community exists; certainly community domi- determined, reflected in corresponding combinations nance often fails to provide such a key. Both "inter- of major undergrowth species or *in total under- dependence" and "independence" of strata are rela- growth composition. In the gray beech forests and tive; distributionally they are expressed in partial hemlock forests, two quite different,but continuously correlations of varying degrees in differentsituations, intergrading types of undergrowths occur under rather than in full correlation or the absence of canopies dominated by the same species. In spruce- correlation. Because these correlations are only par- fir forests, rates of change of canopy and under- tial, the author has chosen to follow the synusial growth along gradients, and directions of most con- tradition in seeking distributional groupings within spicuous change are different; and a variety of strata as part of the present study. undergrowth subtypes or site-types occur under can- The transect data suggest these conclusions on the opies dominated by the same species. Quite different theory of community-units: (1) Many vegetation undergrowths can occur under similar canopies and types are fully continuous with one another, though closely similar undergrowths under quite different others are relatively discontinuous. (2) Species popu- cinioideae. lations are distributed variously or "individualisti- Distributions of undergrowth species do not ap- cally" (Gleason 1926), no two alike. (3) "Natural" pear closely related to those of dominant species. clustering of species is probably of limited and spe- Association of dominant species with other green cial, rather than general, significance. (4) Dominant plant species is not obligate, and almost all under- species dQ not, in many cases, define clearly delim-

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions Ecological Monographs 24 R. H. WHITTAKER Vo.26,No. 1 ited and homogeneous community-units. These state- 7. Picea rubens, dominatingsubalpine forests be- ments may imply (5) that, in general, community- tween 4500 and 6000 ft. units are more "arbitrary" products of classification 8. Abies fraseri and Sorbus americana, centeredin than "natural" units clearly defined in the field. subalpine forestsof highestelevations. This possibility must be further considered in rela- II. Shrub strata tion to other problems of ecological theory and logic 1. Low-elevation Euonymus in following sections (Part II). mesic shrub union: americanus and Lindera benzoin. SUMMARY OF DISTRIBUTIONAL GROUPINGS 2. Mesic heath union: Rhododendron maximumn Distributional groupings of plant species will be and Leucothoe editorum. listed to summarize some results of the gradient anal- 3. Hydrangea arborescens, widespread in mesic ysis and reduce repetition of species lists in type de- and submesicsites. seriptions (Part III). The conception on which the 4. Submesic shrub union, identical with submesic groupings are based is that of the "commodium" (see tree union listed above but also including shrubs: Part II). Species are grouped together whose modes Pyrularia pubera, Rhododendron calendulaceum, Gay- or distributional centers are relatively close to-ether lussacia ursina, Viburnum acerifolium, Calycanthus in relation to environmental gradients. Species thus fertilis and f. nanus, Ceanothus americanus, Aralia grouped tend to occur together in many stands; but spinosa, and Smilax rotundifolia. they will also occur separately in other stands, since 5. Subxeric heath union: Kalmia latifolia with no two have the same distributions. For reasons al- l-yonia ligustrina, Vaccinium constablaei, and Smilax ready indicated, the groupings are considered pri- glauca. marily arbitrary in the sense that the number recog- 6. Xeric shrub union, with low Vaccinioideae nized, their limits and relations to the gradients, are dominant: Vaccinium vacillans (including V. palli- determined by subjective choice and convenience. The ditm), V. hirsututm, V. stamineum (including V. groupings will be termed unions and will in most candicans (C. Mohr) Sleumer and V. melanocar- cases be named for environmental relations rather pum), Gaylussacia baccata, Pieris floribunda, Ilex than major species. Herb-layer groupings may be montana v. beadlei. considered also groups of unions, since species are 7. Ecotonal-mesicshrub union, identical with eco- not separated according to life-fornis. tonal-mesicunion listed above but also includinb I. Tree strata shrubs: Viburnum alnifolium, Sambucus pubens, Ribes cynosbati. 1. Mesic tree class or union, centered in cove for- 8. Low-shrub union of subalpine forests: Vac- ests below 4500 ft: Tilia heterophylla (including var. cinium erythrocarpum, Menziesia pilosa, Diervilla michauxii and T. truncata Spach.), Aesculus oc- sessilifolia. tandra, Halesia monticola, Betula allegheniensis, 9. Lower-elevationheath bald shrub union: Kal- Tsuga canadensis, Acer saccharum, Liriodendron mia latifolia, Viburnum cassinoides, Aronia melano- Fagus grandifolia ("white" and "red" tulipifera, carpa. populations), Fraxinus americana and v. biltmore- 10. High-elevationheath bald shrub union: Rho- ana, Magnolia acuminata, M. fraseri, M. tripetala, dodendron catawbiense, R. carolinianum, Leiophyl- Cladrastis Prunus Carya cordiformis, lutea, serotina, litm lyoni Sweet. Hex opaca, Carpinus caroliniana, Ostrya virginiana. 2. Submesic tree class or union, centered in oak- III. Herb strata hickory and oak-chestnut forests: Quercus borealis 1. Mesic herb union: Dryopteris spinulosa v. in- and v. maxima, Carya glabra, C. tomentosa, C. ovalis, termedia, Athyrium thelypterioides, Eupatoriutm Acer rubrum, Betula lenta, Acer pensylvanicum, rugosum, Cimicifuga racemosa, Impatiens pallida, Cornus florida, Hamamelis virginiana, Amelanchier Caulophyllum thalictroides, Laportea canadensis, Tril- arboreal, Hex montana, Clethra acuminata. lium erectum f. albiflorum, Aster divaricatus, Adian- 3. Subxeric tree class or union, centered in oak- tum pedatum, Actaea. pachypoda, Thalictrum dioicum, chestnut heath and drier oak-chestnut forests: Cas- Hepatica acutiloba, Viola rotundifolia and spp., tanea dentata, Quercus prinus (Q. montana Willd.), Stellaria pubera, Tiarella cordifolia, Mitella diphylla, Q. alba, Q. velutina, Nyssa sylvatica, Oxydendrum Euonymus obovatus, Galium triflorum,Mitchella re- arboreum, Pinus strobus, Robinia pseudoacacia, Sas- pens, Osmorhiza. longistylis. safras albidum. 2. Mesic high-elevationherb union: Rudbeckia 4. Xeric tree class or union, centered in pine forests laciniata, Circaea alpina, Dryopteris spinulosa v. and pine heaths: Pinus virginiana, P. rigida, P. americana, Allium tricoccum, Solidago glomerata, pungens, Quercus coccinea, Q. marilandica. Chelone lyoni, Clintonia borealis, Diphylleia. cymosa, 5. Fagus grandifolia ("gray" population) in mesic Lilium superbum, Oxalis montana, Senecio rugelia, deciduous stands above 4500 ft. Monarda didyma, Aster acuminatus, Streptopus ro- 6. Ecotonal-mesic union, centered in mesic sites at seus, Trillium erectum. elevations around 4500 ft: Betula lutea, Acer spi- 3. Submesic herb union: Polystichum acrosti- catum, Amelanchier laevis, Cornms alternifolia. choides, Smilacina racemosa, Viola hastata, Aureo-

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 25 laria laevigata, Solidago sp., Prenanthes trifoliolata, vival against factors of environment which prevent Uvularia pudica and U. sessilifolia, Pedicularis cana- germination or kill increasing proportions of the densis, Desmodium nudiflorum, Medeola. virgini- population with increasing departure from the op- ana, Veratrum parviflorum, Goodyera pubescens, timum. It seems likely that the curve of population Galium latifolium, Dryopteris noveboracensis, Thas- level along the gradient would be of the classical pium trifoliatum, Parthenocissus quinquefolia, Poly- bell-shaped binomial, or Gaussian, form. It has been gonatum spp. suggested by Gause (1930, 1931) and others (Whit- 4. Submesic high-elevation herb union: Carex aesti- taker 1951, 1952; Brown & Curtis 1952) that the dis- valis, Carex sp., Prenanthes altissima, Solidago sp., tributions of species approximate this form. The Arisaema quinatum (Nutt.) Schott., Houstonia. ser- majority of the curves indicated by transect data pyllifolia, Smilax herbacea, Athyrium filix-femina in the present work are of this form, if one allows v. asplenioides, Angelica triquinata, Thalictrum poly- for occurrence of very wide dispersions and some- gamum. times of more than one peak or mode. 5. Subxeric herb union: Galax aphylla, Chima- It is suggested that curves approaching the bi- phila maculata, Campanula divaricata. nomial form are the most general pattern in the dis- 6. Xeric herb union: Coreopsis major, Andro- tribution of species populations along environmental pogon scoparius, A. virginicus, Tephrosia virginiana, gradients. Various departures from the ideal form Baptisia tinctoria, Pteridium aquilinum v. latinscu- may be expected as conditions depart in various ways lum, Sericocarpus asteroides, S. linifolius, Aster sp., from the ideal situation considered above. In the Antennaria sp., Panicum sp., Gaultheria procumbens, Smokies data four conditions (which are interre- Epigaea repens. lated and may differonly in degree) affecting popu- 7. Heath bald herb union: Trillium undiulatum, lation curves may be recognized: genetic diversity Melampyrum lineare, Lyco podium obscurum. of a relatively homogeneous -species population, ge- 8. Grassy bald herb union: Danthonia compressa, netic complexity in a relatively heterogeneous spe- Potentilla canadensis v. caroliniana Poir., Stachys cies population, effects of interspecific competition, clingmanii, and introduced species (Agrostis alba, and direct environmental limitation of the population. Phleum pratense, Hieracium scabrum, Rumex aceto- Seldom does a population in the field consist of a sella, Prunella vulgaris). single biotype (i. e. of individuals which are geneti- cally identical), though such populations can be pro- II. DISCUSSION: AN INTERPRETATION OF duced through apomixis. Usually the population con- VEGETATION PATTERNING tains some range of variation, and genotypically different individuals encounter different environments DISTRIBUTIONS OF SPECIES AND THE OF STUDY in the range of conditions occupied by the species. GENECOLOGY Probability of survival of differentgenotypes differs In Part I a number of curves for distributions of at differentpoints along a gradient, and progressive plant populations along environmental gradients were change in genotypic composition of the population illustrated. Such curves are doubtless of fundamental along a gradient is consequently to be expected. Since significance in ecology, clarifying as they do the re- genotypic differentiationwithin the population per- lations of species populations to one another, to habi- mits different individuals to occur in different en- tats, and to community-types. Real understanding of vironmental conditions, the range of environments phenomena of population dynamics which underlie occupied by the species is in part a reflection of its these curves must await future research. Present r'ange of genotype diversity. In relation to disper- knowledge of the genetics of species populations sions or amplitudes of species distributions along may, however, permit at least partial interpretation gradients we may distinguish: a) the range of tol- of some distributions observed in the Great Smoky erance and corresponding environmental dispersion Mountains. For much of the genetic information used of a single biotype of the species, and b) the addi- in this discussion the author is indebted to W. H. tional range of tolerance and dispersion resulting Camp, who has given the author the benefit of his from diversity of biotypes. experience with plant populations in the Smokies Among the distributions of species along the ele- through personal communication as well as published vation gradient (Figs. 8, 9, 10; tables 5, 6, 7; Whit- work (Camp 1945, 1951; Camp & Gilly 1943). taker 1951, table 2), three intergrading types of An ideal situation may first be imagined: an un- curves may be observed: a) sharply peaked binomial broken complex-gradient of environment along which curves, e.g. Acer spicatum, Quercus coccinea, b) a population consisting of a single biotype is distrib- widely dispersed binomial curves, occupying much uted and is maintaining itself. A maximum popula- of the gradient, e.g. Acer saccharum, Acer rubrum, tion level should occur at that point of the gradient Pinus rigida, and c) curves extending along the which represents optimum conditions for the biotype, whole of the gradient studied, with population levels and the number of individuals should decrease with changing gradually from one extreme to the other, increasing departure from the optimum in either di- e.g. Tsuga canadensis, Castanea dentata. The spe- rection. The curve of distribution for this population cies of the firstgroup appear in the field to be rela- would be essentially a curve of probability of sur- tively homogeneous, so far as may be judged from

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 26 R. H. WHITTAKER Ecological Monographs their morphologies.In most species of the second one or more adaptive peaks which a given species and third groups there is some morphologicalevi- population has succeeded in occupying within the dence of geneticheterogeneity. limitsof the Great Smoky Mountains. As the great A genetictrend in a populationalong an environ- diversityof possible gene combinationsis reduced mental gradient may be termed a dine (Huxley to a representationon two dimensionsby Wright, 1938, 1940, 1943; Gregor 1939, 1946). Since its in- so the whole range of environmentalgradients en- troductionby Huxley, the term dine has been used counteredby a species is reduced in the nomograms in two senses which are related but should be dis- to two complex-gradientswithin the limits of one tinguished.Probably the majorityof authors using mountainrange. In the nomograms"adaptive peaks" it have applied the termto continuouschange in an are represented by maximum population levels individual characterin populations observed along reached by the biotypes occurringin an environ- an environmentalgradient. Some authors have re- mentaloptimum, and these peaks are surroundedby jected the dine concept as an approach to ecotypic slopes and valleys of lower populationsand absence. variation because of the irregularor zig-zag varia- Many species with particularlywide distributions tion in some morphologicalcharacters (e.g. Clausen in the Smokies are complexes of population-types, 1951: 26-28). Irregularityin some morphological races, or ecotypes. It may be thoughta general prin- traits should not obscure the trendsin more funda- ciple that species occupying a wide range of en- mentalphysiological characters of populationsalong vironmentalconditions are able to do so because gradients,such as are illustratedin the studies of they have evolved series of physiologicallydistinct Clausen,Keck, and Hiesey (1940,1948) and Clausen races (Clausen 1951:29). The red maples (Acer ru- (1951). Because genes are interrelated,it is prob- brum), whichhave probablya wider ecologicalrange ably seldom the case that a single characterof a in the Smokies than any other "species," form an plant population changes quite independentlyof array of high-polyploidraces. Geneticdifferentiation othersin relationto environment.Changes in many in the sugar maples (Acer sa.ccharum), whichoccupy genes, affectingin differentbut related ways the a wide range of elevationin the Smokies,has been physiologicalpattern of the whole plant, may be described by Desmarais (1952). Quercus falcata, expected along environmentalgradients. It is the which extends widely along the moisture gradient trendsin genotypiccomposition of populationsalong fromcove foresttransition to pine forestas a scarce environmentalgradients which are of real ecological species of lowest elevationsin the Smokies,is well significance,and in the presentwork the term dine known as a memberof a complex. The white bass- will be applied in the second sense, to these trends. woods, here designated Tilia heterophylla, form a Even withoutexperiments, the morphologiesand complexin whichT. heterophylla v. michauxii and T. distributionsof tree species in the Great Smoky truncata have also been named. The squaw-huckle- Mountainsprovide limitedevidence that wide range berries are likewise a complex in which Vacciniumn of environmentaltolerance of a continuousspecies candicans and V. melanocarpum have been named, population may be based on a wide range of clinal as well as V. stamineum, the name here applied to variation in that population. Some indications of the whole complex. Two species which are familiar this effectare provided by the nomogramsof Ap- as part of the forestsof the northeasternstates appendix A, as well as by the transects. The nomo- pear in the Smokies with contrastingdistributions. grams outlinewith population contourlines the dis- Pinus strobus is confinedto low elevations,but Tsuga tributionsof species in relation to the two major canadensis extends from lowest elevations to high complex-gradientsstudied in the Great SmokyMoun- ones. Presumablythe high-elevationpopulations of tains, elevation and moisture-balance.In each spe- Tsuga canadensis are ecotypicvariants related to the cies one (at least) environmentaloptimum for the northernones, and equivalent ecotypic variants are population may be recognized in the peak of the lackingin the populationof P. strobus in theSmokies. population figure; and as one departs from this It is of interestto observethat Tsuga canadensis is optimumin any directionthe population levels de- confinedto low elevationsin the nearby Cumberland cline with the tapering slope of a binomial curve. Mountains (Braun 1950), as Pinus strobus is in The area of the diagramcovered by the whole popu- the Smokies. lation representsits range of tolerancesof combina- In otherspecie>. partial or completedivision of the tionsof elevationand moisture-balance,and henceits populations into two or more population peaks ap- full ecological amplitudesin relation to these com- pears. The two birches of the Smokies are both plex-gradients. high-polyploidcomplexes showing evidence of differ- In an essay on evolutionWright (1932) has de- entiationalong the elevation gradient. The yellow scribeda symboliclandscape of adaptation,in which birchesare designatedBetula allegheniensis, and one a given favorable combinationof genes forms an mode or point of highestpopulation densityin rela- "adaptive peak" for the species, surroundedby "val- tion to elevationis around 3,000 ft. At higher ele- leys" of less favorable genetic combinations. The vations the yellowbirches extend into spruce-firfor- diagrams of Appendix A may be compared with ests from a second mode near 4,500 ft, and these Wright's landscape of adaptation, with the restric- high-elevationyellow birches are probably to be tion that the popluation figuresrepresent only the comparedwith the northernB. lutea. The sweet or

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 27 black birches (B. lenta) have a wide range through forestsand cove foresttransition the "white" beech, community-types,with a principal population center of primarily southern distribution,may be recog- at lowest elevationsand a secondarymode at high nized. About 1000 ft above the upper limit of the elevations. The birches of the yellow complex,and whitebeech, other, "red" beechesappear in the upper probablythose of the sweetor black,are facultatively cove forests (3500-4500 ft). The third beech type, apornietic. Establishmentof apomictic races with the "gray," appears in the beech gap forestsof still characterssegregated from the polyploid complexes higherelevations, above 4500 ft. The red and gray is a probable means of their occupation of wide beechesin the northeasthalf of the Smokies are dis- ranges of environmentsin the Smokies. Apornixis tinctivepopulation-types, each with its own habitat does not createnew genotypes,but can fix and main- and population mode; but they are only partially tain favorable ones already achieved which permit discontinuous.Where the two types come in contact the occupation of an environment,especially one they are connectedby dlines leading throughinter- becomingnewly available. Both the birch complexes mediate types, and genes have apparently crossed are importanttrees of firesuccessions, as well as of betweenthem extensively.To a more limitedextent climax forests,in the Smokies. genes may also have crossed in both directionsthe In AppendixA thebimodal populations of Quercus gap separating the white from the red and gray borealis, Carya glabra, Halesia monticola, and Hama- beeches. At high elevationsoutside the range of the melis virginiana may be observed. In these four spruce-firforests, in the southwesternpart of the cases the populations are continuousalong the ele- Great Smoky Mountains,the habitat distinctionbe- vation gradientand appear to gradate into one an- tween "gaps" (cols of the high ridges and the steep other morphologically(morphological differentiation concave slopes extending down from them) and in C. glabra and H. virginiana was not observed). "coves" (bottomsof V-shaped valleys with flowing In the firstthree of these,at least, the high-elevation streams) breaks down. In this area, slopes of the ecotype appears relativelyless mesic than the one mountainsabove 4500 ft occupied by beeches corre- of low elevations (so far as can be inferredfrom spond to the habitat of neitherred nor gray beech; theirrelations to the whole vegetationpattern). The and correspondinglythe partial discontinuitybe- two modesin each case representdifferent, favorable tweenthem appears to break down. The beech popu- genetic combinationsand adaptive peaks within a lations of the southwesternSmokies appear to be species poplation,which also occupies the valley be- formed by thoroughmixture of characters of the tweenthem at lower population levels. gray and red, variously segregatinginto different The whiteoaks (Quercus alba) in the Great Smoky individualsand stands. Mountainsconsist of two fullydiscontinuous popula- Several patterns of response have been observed tion-types,separated by an elevationgap of 1000 ft among species of wide extent along the elevation or more. In the Cumberlandsthe high-elevationwhite gradient in the Great Smoky Mountains. Some ex- oaks are lacking,and the upper altitudinallimit for tend from low elevationsto high with a continuous whiteoak is about 2000 ft (Braun 1950). Low- and population embodyingclinal variation. Others show high-elevationecotypes may be recognized also in partial discontinuity,with the appearance of two or the red oaks, althoughthese are not separated by a moremodes along the gradientwithin the population. comparablehiatus. The low-elevationpopulation of In two species population-typeswhich are fully dis- northernred oaks may be identifiedas Quercus bore- continuousmay be recognized;but one of these,the alis v. maxima; in the high-elevationpopulation beeches,forms a complex combinationof continuity charactersof Quercus borealis v. borealis predomi- and discontinuityinvolving three or fourmajor popu- nate, though somewhat mixed with those of var. lation-types. It may thus be apparent that there maxima. From lowestelevations to highestthe char- are all degreesof continuityand discontinuityin the acteristicsof populations change fromthose of var. relations of natural populations to environmental maxima toward those of var. borealis, although a gradients. There are all degreesof segmentationof a range of variationin these charactersoccurs in each clinal population into partially or wholly discon- population observed. An extended dline thus continuous population-typesor ecotypes. nects Quercus borealis and var. maxima along the ele- The term ecotype has been variously interpreted vation gradient (and the moisture gradient, var. since its introductionby Turesson (1922a, 1922b) maxima being relativelymore and var. borealis less for the genotypicresponse of the plant species to mesic). It is believed that this dline is one formed habitat. If the ecotypicvariation of a species is in throughintrogressive hybridization (Anderson 1949, the formof a fullycontinuous dline, then any ecotypes 1953) between two ancestral oak population-types, whichmay be recognizedare arbitrarilydistinguished thoughit is not assumed that the red oaks of lowest parts of the dline (Gregor 1944). The termecotype and highestelevations in the Smokiesare representa- may also be restrictedto thosepopulation-types which tive of these ancestraltypes. are whollyor partially discontinuousand are differ- A gap of about 1000 ft also exists in the distribu- entiatedin relationto habitat,within a species popu- tion of beeches (Fagus grandifolia), among which lation. In the terminologyused here, cbne is a gen- three population-typesand modes are to be distin- eral term for genetic trends through populations, guished (Camp 1951). At lowest elevationsin cove trends which may extend as "intergroup lines"

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions Ecological Monographs 2828 R. H. WHITTAKERWHITTAKER ~~~~~~~~~~~~~~~~~~~~~~~~~~Vol.26,-No. 1 (Huxley 1938, 1940, 1943) throughseries of par- for the low-bush blueberrieswhich occur in pine tially or wholly discontinuouspopulations. Cline forestsat all elevationsfrom 1500 to 4500 ft forma is customarilyapplied to the populations through dine gradatingupward from populations approach- which a genetictrend exists as well as to the trend ing Vaccinium vacillans to those approaching V. itself. Along a segmenteddline ecotypes may be pallidum (Camp 1945). The population of black recognized as relatively discontinuouspopulation- oaks (Quercus velutina) has a ridge extendingup- types,marked by a distinctpopulation mode, at least. ward in subxeric sites from the main population The termecotypic population is convenientlyapplied center toward the high-elevationpopulation of Q. to a population-typewhich is recognizedas distinc- borealis (Appendix A). It is believedthat this ridge tive in its adaptationto habitat,when it is not known resultsfrom the introgressionof genes fromQ. bore- whetherthis is a clinal' population continuouswith alis, permittingan extensionof the Q. velutina popu- otherpopulations along a dline,or an ecotype rela- lation toward higher elevations. Introgressionmay tivelydiscontinuous with other ecotypes. thus permitextension of range of a population,as in Because of the separationof populationsinto local the more conspicuousexample of Helianthus bolan- habitatsnot arrangedalong gradients,and because of deri, a serpentine endemic turned weedy (Heiser the possibilityof genetic drift withinlocal popula- 1949; Anderson 1949 :76). By hybridizationalso, tions,these will not always formdlines. In some spe- essentiallynew population types may be produced cies theremay be muchecotypic variation which can- from the genetic materials of old. Thus the high- not be reducedto dlinesand relatedto environmental elevation beeches of the southwesternSmokies may gradients. It may furtherbe observedthat ecotypes be interpretedas a new ecotype produced from the occur withinecotypes (e.g., edaphic within climatic geneticmaterials of the otherbasic beech types. The ecotypes,Zherebina 1931, Kruckeberg 1951, 1954), importanceof introgressivehybridization has been that ecotypes occur along elines and dlines within emphasizedespecially in relationto the new habitats ecotypes,that elinesmay crossone anotherin various produced by man (Anderson 1948, 1949). The few ways, and that minor elines may occur within seg- examples given from the Smokies may suggest the mentsof wajor ones. Ecotypes and dlines,as defined importanceof hybridizationsamong woody plants in here, occur in all possible combinations.Clines and this patternof climax vegetation(cf. Muller 1952). ecotypes are thus alternative and complementary Certain distributionalpatterns are affected by means of abstractingfrom the ecotypicvariation of polyploidy. Among the maples the diploid mountain species, emphasizingrelative continuityand relative maple ( Acer spicatum), with a restricteddistribu- discontinuity,respectively, applicable according to tion and sharplypeaked binomialcurve, may be com- the particularcombination of continuityand discon- pared (see Appendix A) with the widely ranging tinuityexisting in a given species population (cf. polyploid red maples (Acer rubrum). Among the Stebbins1950, Whittaker1954b). magnoliasthe relatively sharply peaked and restricted Both dlines and ecotypes may be thoughtfunda- diploids Magnolia fraseri and M. tripetala may be mentallyphysiological (in a broad sense) in meaning compared with the tetraploid M. acuminata. The though sometimesrecognizable also in morphology. high-bushblueberries of the Smokiesare a hexaploid The ecotypemay be thoughtto representa relatively complex (Vaccinium consta-blaei) derived by allo- favorablegenetic combination within the species, an polyploidyfrom V. altomontanum and V. simulatum adaptive peak in relation to environmentwhich has (Camp 1945). V. constablaei occupies a particularly actually been attained. It seems likely that such wide range of habitats,ranging from lower elevations favorablegenetic combinations are usually expressed to highones, from oak-chestnut forests to pine heaths in local maximaof population density,in population along the moisturegradient, and into types as physi- peaks such as are illustratedin Appendix A. The ognomicallydifferent as deciduous forest and pine occurrenceof dlinesand the ridge-likeshape of some heath,subalpine forest,grassy bald, and heath bald. of the population figuresin Appendix A (granting The geneticcharacters derived from its parentssegre- that the relativelengths of the ordinateand abscissa gate into individualsresembling the parent species, are whollyarbitrary) suggestthat adaptive "peaks" as well as in othercombinations. V. constablaei has may take the form of "ridges" in relation to com- been regarded by Camp (1945) as a typical allo- binationsof gradients. The genetic pattern and the ploidion (Camp & Gilly 1943) because of this utili- more abstractadaptive landscape of a complex spe- zation in differenthabitats of genetic materials de- cies, or of a genus or comparium,may be visualized rived fromtwo ancestralspecies. as a complextopography of hills, peaks, and ridges There is no reason why a diploid species with ex- of differentheights and extents,variously related to tensive ecotypic variation should not have a wider one anotherand variouslyseparated by cols, ravines, range than a polyploidderived from it. The Fagaceae and level valley bottoms of differentdepths and of the Great Smoky Mountains form an instructive widths. series on the diploid level: from Quercus prinus, Q. The role of introgressivehybridization in modify- coccinea, Q. marilandica, and Q. stellata, which are ing the geneticpattern of species has been referred relativelyhomogeneous and of restrictedecological to. An introgressivedline suggestiveof that in the amplitude (withinthis studyarea), throughQ. velu- northernred oaks appears among the blueberries, tina with a limitedextension of its range by intro-

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 29 gressionand Castanea dentata withan extendeddine quite differentedaphic conditions(see Part III). A of unknownderivation, Q. alba and Q. borealis with series of deciduoustree species show evidencein the two distinct ecotypes, discontinuousor connected nomograms(see also Fig. 4) of partial division of by introgression,to the complex pattern of Fagus their populations by hemlock. The partially sepa- grandifolia. Allopolyploidy,however, combines the rated populations may well be ecotypesor incipient genetic stock of two species from which characters ecotypes; some genetic differentiationmay be ex- may be segregatedin relationto environmentand is pected to accompany partial separation of popula- a meansby which,in some cases, species of wide eco- tions into differenthabitats. Because of effectsof logical amplitudesmay be produced. "competition"on populations,the physiologicalop- Genetic knowledge of the plants of the Great timumfor a species or ecotype,as it may be deter- Smoky Mountainsis very limited. It may neverthe- minedin the laboratoryor experimentalgarden, and less be evidentfrom the precedinghow closelyinter- the population mode in relationto environmentsoc- relatedare the geneticpattern and distributionalpat- cupied by other plants, should be clearly distin- tern of a species. Althoughthese develop together guished. The necessity of the distinctionmay be in relationto environmentalpossibilities, the distribu- illustrated with special effectivenessby serpentine tional patternmay to some extentbe regardedas an plants which are largely confinedto serpentinesoils expression of the genetic pattern. In terms of by intoleranceof competitionon othersoils, on which Wright'stopographic analogy, the population center theymay show bettergrowth in the absence of other or centersoccupied by the species are expressionsof plants than on serpentine(Kruckeberg 1954). the one or more favorableadaptive combinationsin Most of the mature forests of the Smokies are relation to environmentpossible for that species mixedstands of severalor manyspecies, whose popu- whichhave actuallybeen occupied. Selection,restrict- lations are in competitionfor resourcesof environ- ing the diversityof biotypes,may tend to confineor ment. Since the space, water, and nutrientsof a reduce the area of these population peaks; but habitat are limited,and since several species grow differentialselection in differentenvironments may togetherthere in competitivebalance, none of these tend also to produce lines and to cause the differ- species can reach the abundance whichits reproduc- entiation of ecotypes. Mutation, cross-fertilization,tion, growth,and survival might permit if it were and introgressionmay tend to maintain or increase alone in the site. When several species of rather the area of population peaks by maintainingor in- broad amplitudes are distributedtogether along a creasing the diversityof biotypes. Other reproduc- gradientas the subxerictrees are (Fig. 9), they are tive patternswhich limit the recombinationof char- competingwith one anotherat or near theiroptima; acters (self-fertilization,apomixis, structural hy- and the peaks of their distributionsmust conse- bridity) may increase the range of environments quentlybe mutuallydepressed. Such depressed dis- effectivelyoccupied by quite differentmeans, facili- tributionsmay be contrastedwith those of the pines tatingthe maintenanceof favorablegenetic combina- illustratedin Fig. 10, in whichthe modes of a series tions whichhave already been achieved. Hybridiza- of dominantspecies are spaced along the gradientat tion, involving either introgressionor polyploidy, differentelevations. Two effectsmay thus be distin- may permitthe formationof new favorable genetic guishedin producingdistributional curves which ap- combinationsand hence the occupation of different pear "depressed" in contrastwith the ideal, sharply habitats and the developmentof new population peaked binomial-biotype diversitywhich increases peaks. dispersion,and competitivelimitation of the height Effectsof environmentallimitation and the pres- of the curve. ence of otherplants on these distributionalpatterns When environmentalresources more directlylimit have not yet been discussed. For those effectsof a single species, an even more distinctlyflattened plants whichreduce growthof individualplants and curve may result. If a species has such competitive the populations of species, the convenientlyvague advantage that it approaches single-speciesdomi- term "competition" exists. The forests of the nance along part of a gradient,height of the curve Smokies are (excludingdisturbed stands and allow- may be limitedby the maximumdensity of that spe- death of the chestnuts)mature and ing for effectsof cies possible in the environment.The distribution self-maintainingforest communities,fully occupied may then be verticallytruncated, flattened under a by plants. The population patterns described are ceilingin the formof a plateau, withfairly constant populationpatterns maintained against a background of the gradient. A distribution of relativelyrigorous competition. abundancealong part from cove One of the tree species in the Smokies may be of this type is shown for a fieldtransect thoughta particularlyeffective "competitor." East- forestinto gray beech forest (Fig. 13). Implications ernhemlock (Tsuga canadensis) formsstands strongly of such "plateau" distributionswill be discussed dominatedby itselfin sites whichare relativelymesic, furtherbelow. "Competition"may thus affectdis- thoughless so than those of cove hardwoodsforests. tributionalpatterns in various ways: formationof a The hemlockthus dividesthe deciduousforests along plateau withcompetition occurring within the species, the moisturegradient by a belt of coniferousforests depressionof population levels throughcompetition which approach single-speciesdominance and offer with other species, competitivesplitting of the dis-

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 3030 R. H.H. WHITTAKERWHITTAKER Ecological~~~~~~~~~~~~~~~~~~~~~~~~Vol. Monographs26,No. 1 tribution,and displacement of the environmental THE ASSOCIATION-UNIT THEORY AND INDIVIDUALISTIC mode from the physiologicaloptimum. HYPOTHESIS It is sometimessaid that distributionsof species One of the most nearlyuniversal ideas among dif- are ultimatelyto be understoodthrough their physi- ferentschools of ecologyis that of vegetationunits, ologies, and that autecologybecomes in the end ap- whichhave most commonlybeen termedassociations. plied physiology. It should be clear that there is Some representativestatements on associationsfrom far more to the distributionsof species than is to be differentschools are the following: Flahault & learned from physiologicalexperiments with small Schr6ter (1910 :152, Pavillard 1935), "An associa- numbersof individuals. The ecological amplitudeof tion is a plant communityof definitefloristic compo- a populationis in large part also the geneticampli- sition,presenting a uniformphysiognomy, and grow- tude of that population,and not only the physiologi- ing in uniformhabitat conditions. The association cal toleranceof a few individuals. The population is the fundamental unit of synecology." Braun- mode is the environmentaloptimum for the species, Blanquet (1921), "The association is a plant com- but this is an optimumin relation to competition, munity,characterized by certain floristicand socio- etc.,in naturalcommunities and not simplya physiological marks,which reveals a certain independence logical optimum.Species populationsin the Smokies throughthe presence of character-species."Nichols did not in general have sharp boundariesat particu- (1923), "Viewed in the abstract,an associationmay lar physiologicallimits, but tapered gradually to be describedas a vegetation-unitcharacterized by its scarcityand absence with decreasingprobability of essentially constant physiognomy and ecological survivalof individuals. Outlinesof distributionswere structureand by its essentiallyconstant floristic com- rounded,as the population levels were affectedby position,at least with regard to dominantspecies." differentcombinations of interrelatedmoisture and Conard (1939), "The association of individualsand temperaturefactors, rather than by particularlimits species is much more than a chance meeting. It is for one factor or another. part of the order of nature. . . . Associations of Distributionsof species are to be understoodnot plants on land are definableentities, susceptible of fromlimits of toleranceof experimentalindividuals naming and classifying." Du Rietz (1929), ". . . the alone, but fromvarious types of information,which association is a concretepart of actual vegetation, may include history,physiology, ecological life his- thoughmostly split up into a large numberof iso- tory,population dynamics,genetics, etc. To be ade- lated parts. I am quite unable to regard each of quate for distributionalinterpretation of complex these isolated parts as a fundamentalunit of vege- species, knowledgeof physiologyand life histories tation analogous to the individuals,the real funda- should include differencesin these among ecotypic mental unit being the sum of all the parts showing Tansley (1920), populations. Population dynamics of plants is a essentiallythe same composition." "But if we admit, as everyonewho has worked at major area of neglectin ecology,and could be given the subject does admit,that vegetationforms natural little discussionhere except in terms of "competi- units which have an individualityof their own, and genetics tion." Ecological genetics and population that these units owe theirexistence to the interaction have not been broughtextensively into autecological of individual plants of differentspecies with their interpretation,but theirimportance should be clear. environment,then it becomesclear that a mere study The implicationsof genetic understandingin syne- of the distributionof species as species cannot form cologymay be less direct,but scarcelyless significant, the basis of the science of vegetation. We have in- than those in autecology. The vegetationpattern of stead to focus our attentionon the vegetationalunits the Smokies would, for example, be scarcely in- themselves." telligibleif population complexitiesof the red and The centralidea of theseconceptions of vegetation whiteoaks and beeches were not recognized. may be termed the association-unit theory, the belief A particular type of genetic knowledge,relating that vegetation consists of distinct,natural units the geneticsof populations to habitats and distribu- called associations. With the association,so far as tional patterns,has been emphasizedhere, an area of basic philosophyis concerned,may be grouped such formationor study termed genecology by Turesson (1923), its other units as the biome, sociation, etc. We may pioneer. In spite of Turesson'sname, genecological or union and socion, site-type,life-zone, regard the association-unittheory as a form of a ecogeneticknowledge has developedprimarily in the broader community-unittheory, which assumes that the of evolution,and fieldsof systematicsand study natural communitiesare made up of units and that not in ecology. Genecologymay appear as a scientific these units are themselvesthe proper objects of crossroadsrelating the four major fieldsof genetics, study. Like the cell theoryand the moleculartheory, ecology,systematics, and the study of evolution. As the community-unittheory is a conception of the an essential part of understandingboth in ecology structureand organizationof the phenomenawith and in the systematicson whichthe ecologistis de- which the scientist deals. Whatever the present pendent,genecology may well be a major concern status of the community-unittheory, for its role in of ecologists(cf. Constance1953, Whittaker1954b) the developmentof ecology and for its influence

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 31 within and beyond the field,it may be recognized stand of one association meets a stand of another as one of the major theoriesof science. association; it may further imply that many, if not An alternativeconception, the "individualistichy- all, species populations will have their limits in the pothesis,"has been held by some authors. The prin- boundaries of their associations. Any interpretation eiple of species individualitywas advanced in Russia of associations may poorly represent the views of by Ramensky (1924) and withina few years inde- some authors, but a formulation of the association- pendently recognized in this country by Gleason unit theory based on distributional relations of spe- (1926) and in France by Lenoble (1926, 1928) and cies may be attempted (cf. Du Rietz et al. 1920, Fournier (1927). In the Americanliterature it has Du Rietz 1921). been supportedby Gleason (1926, 1929, 1939), Cain A few essential ideas are implicit or explicit in the (1947), Mason (1947), 'Egler (1947), Whittaker statements quoted and in most ecological thinking on (1951, 1952), Curtis & McIntosh (1951), and Brown associations. The first is that vegetation is made up & Curtis (1952), and has been criticizedby Clements of units, the associations, which are "fundamental" et al. (1929:315), Nichols (1929), and Cain (1934). and not subjective in nature. Second is the belief By this conceptionthe species of plants are regarded that the unity of associations is intrinsic and dy- as "individualistic,"each responding to environ- namic, that species are organized into associations mental factors according to its own characteristics, by common response to environment and by neces- and hence no two alike; and stands are regardedas sary relations among them. Third is the conception essentiallyindividual phenomena,varying combina- that species are distributed in terms of these units. tions of species populations which may only arbi- If the units are fundamental and the organization of trarily and artificiallybe grouped into associations. species into units real, the reality of the units should The significanceof bioticfactors may be granted,but appear in the distributions of species. Interdepend- species populations respond individualisticallyto ent species cannot very well be independently dis- biotic, as well as physical,factors; and interaction tributed. The third proposition is a consequence of among species is not, in general,of the obligatechar- the second and an explanation of the first; and we acter necessary to accomplish the organization of may formulate the association-unit theory as a logi- complex communitiesinto distinctunits. The essen- cal triangle of three interrelated propositions: (1) tial propositionsare two: (1) Species are distributed Vegetation is made up of units, the associations, individualistically,each accordingto its own physi- which are real because (2) their unity is intrinsic ology and biotic relations; and (2) communities, and dynamic, based upon necessary relations among formedby the overlap of species distributions,are species, with the result that (3) species are distrib- not organizedinto discreteunits or associations. uted in terms of the units, so that similarity of dis- The controversyhas centeredon the question of tributions of associated species and correspondence whether or not associations are "real," "natural," of their distributional limits with the boundaries of >"valid,"or "fundamental"units. The association is associations may justify statement (1). an abstract type or class of communities;and the In all three points such a formulation is in con- statementthat such an abstractclass is, or is not, a trast with the individualistic hypothesis. The point real, natural, valid, and fundamentalunit conveys which is most directly subject to controlled research littlethat is not subject to varied interpretation.The is the third. Choice between the two conceptions discussionof such a problemis likelyto be a semantic may rest primarily on the decision as to whether or skirmishunless the issue is so definedthat research not analysis of natural communities reveals them can be done and evidence broughtto bear upon it. to be units of species having closely similar distri- For effectivediscussion it seems necessaryto formu- butions. The individualistic hypothesis may be relate the questionin termsof propertiesof stands in jected if organization of groups of species into relationto one anotherand distributionsof species bounded units appears generally in their distribu- in relationto one anotherwhich may satisfyone hy- tions. Evidence may also be considered for the first pothesisor the other. point: whether or not vegetation samples taken by Relations between stands which would make the unprejudiced means fall into clearly defined units association an ideal, "natural" unit are simply de- which are homogeneous within themselves. For the fined: the association should be homogeneousinter- second point, it may be unnecessary to argue whether nally, its stands being very much like one another, or not the web-of-life should imply dynamic organi- and discontinuouslyexternally, its stands being quite zation of species into units when research can reveal differentfrom those of other associations. Transi- whether or not it actually does so. tions betweenassociations should be, if not absent, It has not proved possible to establish the existence of very limited extent compared with that of the of associations as natural units by statistical means homogeneousassociations themselves. Homogeneity (Clapham 1936, Ashby 1948, Curtis & McIntosh of an association,in the sense of close similarityof 1950, Goodall 1952, 1954a). The present research its stands, implies that its species are similarlydis- was undertaken with the assumption that existence of tributed,having similarpopulation levels in the dif- "real" or "natural" vegetation units could be estab- ferentstands. Discontinuityof associations implies lished in two ways: First, by gradient analysis which abrupt change in species populations wherever a revealed the natural distributional groups of species,

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 3232 R. H.H. WHITTAKERWHITTAKER Ecological~~~~~~~~~~~~~~~~~~~~~~~Vol. Monographs26, No. 1 second, by study of random samples which should a complex,gradating pattern of populations,within fall into relatively homogeneous and discontinuous which a numberof vegetationtypes may be recog- groups. Results are recounted in Part I: Species are nized. Accordingto the conceptiondeveloped here, not organized into distributional units, and stands and the types recognized are "arbitrary" rather than vegetation types are for the most part continuous "real" or "fundamental"units. Nevertheless,a major with one another. Although limited clustering of problem in the interpretationof community-types species and limited vegetational discontinuity are remains: How are these types to be understoodin suggested, the degree of approach toward these con- termsof the distributionsof species populations? ditions of natural units is slight. The Smokies Certain vegetationtypes (see Part III) are de- sample is a limited one; but stand tables compiled finedby a single dominantspecies-Tsuga canaden- by Braun (1950: 53, 62, and 64; 83, 95, 97, 139, 167, sis, Fagus grandifolia, Pinus virginiana, P. rigida, 191, 219, 221, 311) suggest that the phenomena of P. pungens, Abies fraseri, Picea rubens. These are species individuality and stand continuity may be easily recognizedand definedas types, and they ap- generally valid in the eastern forests. Especially the pear essentiallyconstant in floristiccomposition with concurrent, independent studies of Curtis and his regard to dominantspecies accordingto the associa- associates in the widely differentconditions of Wis- tion definitionof Nichols (1923). They may conse- consin forests (Curtis & McIntosh 1951, Brown & quently appear at firstglance to be truly distinct Curtis 1952) indicate that these phenomena are not and homogeneouscommunity-types (consociations in peculiar to the Great Smoky Mountains. Americanusage). It has been shown,however, that A conception of vegetation quite different from thesetypes are heterogeneousin undergrowthcompo- that of a mosaic of natural units may be suggested. sition and intergradecontinuously with other types. Vegetation forms a complex population pattern in Othervegetation types are definedby a pair of domi- relation to environment. In this pattern species (of nant species. Thus Castanea dentata shares domi- green plants, at least) distribute themselves "indi- nance with Quercus prinus in some low-elevation vidualistically"; they occur together wherever their stands,but with Q. borealis or Q. alba in other,high- distributions overlap, and associate and dissociate elevation stands. Castanea and the three oaks are freely in space and in evolutionary time (Mason quite differentlydistributed in the vegetationpat- 1947). Neither in time nor in space does a species tern; and they definevegetation types not by a high occur in terms of associations, but where it may- degreeof "association,"but simplyby distributional where at a given time some combination of genetic overlap in relationto environmentalgradients. and environmental possibility for that species is Withingradating vegetation the area of dominance realized in one of its populations. In relation to the of a single species, or the area of overlap of two population pattern of vegetation, outlines of species dominantspecies, may be used to characterizepart distributions are most varied and centers are well of the vegetationpattern and definea community- scattered. Most population distributions take the type. Gradation within the types thus definedand form of tapered curves, and the pattern is largely between them and other types implies that these continuous. Because of environmental interruptions community-typesare neithertruly homogeneousnor and some relative discontinuities inherent in vege- discrete. It should be understoodthat definitionof tation itself, the pattern may also be considered a community-typesby one or two dominant species complex mixture of continuity and relative discon- representsan ecologist'schoice of a single criterion tinuitv. In this it is important to recognize that of vegetationby which stands otherwiserelatively continuity which does exist, rather than overlooking heterogeneouscan be grouped. it in an emphasis of "natural" units. It is not in- A more general problem in the interpretationof frequently the case that approaches based on the community-typesis the groupingof several or many study of continuity and relation are more productive species whichare to be considered"associated." The of understanding than those which treat an object wide ranges and geneticcomplexities of some species, of study in terms of units, assuming discontinuity discussedin the precedingsection, render such group- or self-sufficiencyof these. ing difficult.Widely distributedspecies normally consistof diverseecotypic populations, and oftenof THE DISTRIBUTIONAL BASIS OF COMMUNITY-TYPES some numbers of ecotypes partially discontinuous GRADATION AND THE GROUPING OF SPECIES with one another. Such species, when they occur in When population curves of species are drawn to- differenthabitats with different sets of associates,may gether along a gradient (Whittaker 1951, Fig. 1; presumablybe grouped with each of their sets of 1954b, Fig. 3), the resulting picture is not one of a associates (e. g. Castanea dentata with the threeoak series of associations marked off from one another, species). A species may in this sense "belong" to two but of a pattern of curves flowing into one another, or more associations or other community-typesin a continuum of populations. This vegetational char- whichits ecotypicpopulations occur. acteristic of continuous change in proportions of The traditionalmeans of grouping species is by populations along an environmental gradient may be the limits of their distributions(Transeau 1905, thought a fundamental and general condition. and Hulten 1937, Bocher 1938, Meusel 1939, Raup 1947). termed gradation. The forests of the Smokies form Useful as limits of distributionmay be for some

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 33 kinds of ecological analysis, they cannot easily be (e. g. Acer rubrum) to group these with species of determined for binomial curves which taper continu- narrower amplitudes, and (3) grouping of bimodal ously from scarcity to absence. While arbitrary populations (e. g. Halesia monticola) by their major limits can be defined in the tails of the curves, these population centers only. As distribution groupings limits may still not effectivelyindicate the relations arbitrarily defined in relation to a single gradient, of species to one another. Populations with similar these classes correspond to the ecological groups of distributional outlines may be important in different Ellenberg (1950, 1952, cf. the biological plant types parts of their common area; along the gradients their of Iversen 1936). distributions may be differentlyskewed. Plant associations, when defined by several species, If maximum populations of two species appear in are best conceived as commodia of character-species, the same habitats, however, these species appear or of dominants. The association in the sense of the associated even if their distributional limits differ. school of Zirich-Montpellier (Braun-Blanquet 1921, Since these species usually occur together when they 1951) is normally defined by a group of character- are sufficientlyabundant to characterize stands, their species, which have narrow ecological amplitudes and "association" (in this sense of distributional rela- are largely restricted to, or at least centered in, a tion) is the basis of our subjective recognition of an given association. Since the species of high Treue- "association" (an abstract stand-type). Association grad, or degree of fidelity,to the association are those of population-types (species, ecotypes, etc.) may be having their modes within the association, the char- defined for some purposes by coincidence of popula- acter-species form a comninodalgrouping. The "char- tion centers in relation to environment. Such asso- acteristic species composition" of an association in ciation is, clearly, relative. There are all degrees of the sense of Braun-Blanquet (1951 :108) comprises association and non-association, from species with these character-species, plus other species of wider quite similar distributions (Tsuga canadensis and amplitudes (companions or Begleiter) which have Magnolia fraseri), through those with distributions ecotypic populations reaching constancy levels of 60% related along one gradient but different along an- or more in the stands chosen to represent the associa- other gradient (Castanea dentata and Quiercus tion. It is probably not necessary for success of the prinus), and those with distributions overlapping method that the character-species groupings represent only toward their limits, to those which do not occur natural clusters. The distributional individuality of together. species and the scattering of their modes should iiii- The appropriate measure for locating distribu- ply considerable freedom in the recognition of char- tional centers of populations is the mode. The mean acter-species groupings and in the corresponding nay also be used to express the average environ- definition and delimitation of associations. mental adaptation of a relatively homogeneous popu- The associations of American authors (Clemnents lation. But the mean mnaybe difficultor impossible 1928, Weaver & Clemients1929, Oosting 1948, Braun to determine for skewed or bimnodaldistributions and 1950) are in general based on very broad groupings those which are cut short by absence of more extreme of dominant species. In somnecases (Quercus-Carya, habitats into which they might extend in an area. Qnercus-Pinus Associations) the species grouped to- If the nmodeis the datum to be used, associated popu- gether are so widely differentin ecological relations lations are those with their modes near the same loca- and geographic distributions that somie dominants of tion along the gradients of environment. Species the association may not occur together in the s une (ecotypes, etc.) having their modes close together stands. It is in general impossible to define the may be described as commodal, and a group of spe- Nnierican associations quantitatively through their cies (etc.) related by the approximation of their species composition (Braun 1950:11). They may be niodes in relation to environment niay be designated interpreted, however, as very loose comnmodalgroup- a commodium.* ings of dominant species, at least some of which are The conimodium is proposed as a conception of sufficientlyrelated distributionally to be important distributional association and as a term which may in the same stands. be applied to any grouping of plants or animals in The distributional groupings used in the present natural communities by the location of their popula- paper are considered primarily "arbitrary"; it is be- tion centers. The groupings of species in the present lieved that they are not, for the miiostpart, based on work may illustrate the concept. The "classes" of natural clusters of species which are clearly sepa- trees are groupings in relation to the moisture gradi- rated from one another in the field. The same may be ent of species with quite different distributions in said for the grouping of character-species or domi- relation to the elevation gradient. The grouping was nants into associations in general; such grouping is accomplished by (1) grouping together species of possible without any assumption of discontinuity be- narrower amplitudes whose population centers oc- tween species-groups or vegetation types. It has also eurred in the same span of the moisture gradient, been suggested that a degree of natural clustering of exist. (2) use of the location along the moisture gradient species in natural communities may sometimes In relation to the theory of comninunity-typesit of maximum population levels of widespread species should be observed that such relative clustering is in * Latin con, together, plus modus, in the sense in which it has entered statistics. no way inconsistent with either the individualistic hy-

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions Ecological Monographs 34 R.. H . WHITTAKER Vol. 26, No. 1 pothesis or the frequent continuity of types with one another. Whether some associations already recognized are supported by natural clustering would seem 60b0 a matter for research, rather than assumption. For 1-40- the present, the "comminodal"conception provides a 0~ means of grouping species in vegetation analysis and the definition of community-types without assuming natural clustering which cannot now be demonstrated. U 30- W'r20- ZON ATION

Most of the vegetation pattern of the Great Smoky 2500_ 36000 3500 4000 4500 500 55.00 6000 Mountains is one of continuous gradation; but eer- ELEVATION IN FEET tain relative discontinuities in the pattern are also to be recognized. Some of these, as the edges of burns 801 z60- and old fields, are products of disturbance. Others ~-40- appear as interruptions or teleseopings of the gradation where there are environmental discontinuities, 20 as along sharp ridges. A few vegetation types of the F- Jto!z 40- Sniokies, however, are separated from the rest of the INFE ELVTO pattern by more abrupt transitions not produced by 030- W20 g environmental discontinuity. Some characteristics of these relatively discon- 2500 36000 '3500 4000 4500 5000 5500 tinuous types may suggest their meaning. It may ELEVATION IN FEET first be observed that they are types strongly dominated by one (or two) species. The beech gap forests, FIG. 13. Distributions of plant populations in relation to "'zones.'" Above the transition from cove for- which are relatively discontinuous with the cove forests below 4500 ft to mesic spruce-firforests above, ests below theiii, are dominated by Fagus grandi- based on composite transect data. Zoned distributionof folia. The grassy balds, separated from surrounding evergreen and deciduous growth-formsrelative to one b, all ever- forests by a are dominated another: a, all deciduous,broad-leaved trees; shrubby forest-edge, by green, needle-leaved trees (Picea rubens, Abies fraseri, mountain oat grass (Danthonia compressa). The and, below 4700 feet, Tsuga canadensis); c, Picea ru- subalpine forests, which may be interpreted as a bens; d, Abies fraseri. Distribution of other species in zone above the deciduous forests, are dominated by relation to the transitionand spruce-firforests: e and e', Betidla allegheniensis and B. lutea; f, Acer spicatum; Picea rubens and, at higher elevations, Abies fraseri. g, Cormas alternifolia; h, Amelanchier laevis; i, Vi- In the heath balds, which may be abruptly separated bitrnumtalnifolium (estimated coverage percents) ; j, froim the subalpine forests, one or two of several Sorbus americana. Below the transition from cove 4500 ft to gray beech forests above, based major shrub species dominate a given stand. forests below on a field transect in Trillium Gap. Zoned distribution Transect data are available for the distributions of of gray beech relative to other tree species: a, major species in two of these relatively discontinuous types. mesic tree species (Halesia monticola, Betula allegheni- Tsuga canadensis, Acer sac- The distribution of gray beech (Fagus grandifolia) ensis, Aesculus octandra, charum, Tilia heterophylla,Fraxinus americana, Lirio- in relation to elevation (Fig. 13) shows a "plateau" dendron tulipifera); b, red, and c, gray populations of distribution, in which the species strongly dominates beech (Fagus grandifolia). Distribution of other spe- some stands above 4600 ft, but is rather sharply cies in relation to the transition: d, Betula allegheniensis; e, Cornus alternifolia; f, Aesculus octandra; g, limited between 4600 and 4400 ft. In the coniferous Acer spicatum; h, Prunus serolina. zone of high elevations, Picea rubens and Abies fraseri appear to have ordinary, tapered curves in re- ties (heath balds), and coniferous forests (spruce-fir lation to elevation; no tendency toward plateau dis- zone). The gray beech is a deciduous tree, but differs ti'ibution has been established for these two species. from many of the cove forest trees in its small stature The relative discontinuity is in this case between and manner of reproduction. At a contact between growth-formiis;grouping all deciduous and all conifer- differentdominant growth-forms,such as that of the ous species together, the subalpine forests form a grassy balds and forests, environmental conditions with coniferous zone which is relatively discontinuous related to dominance of one growth-formmay largely not with the the deciduous cove forests (but hemlock prevent individuals of the other from establishing char- forests) below them. These "zones" are thus themselves. The two dominant growth-forms may of a acterized by plateau distributions species popula- thus, in some cases, replace one another abruptly tion or a growth-formalong one gradient which has rather than forming mixed stands along a wide span been studied. of the gradient. Because stands of mixed dominance The discontinuous types of the Smiiokiesare mostly are limited, the two growth-formsappear, in human characterized by dominant growth-forms differing interpretation, relatively "incompatible." from the predominantly deciduous forests. The rela- The relation of the relatively discontinuous types tivelv (liscontinuous types include grassland coirn- to the gradients is of interest. All are types of high itiunities (grassy balds), evergreen shrub communi- elevations; and the two which are most clearlv dis-

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 35 continuous, the heath bald and grassy bald, are types as is illustrated by the continuity of the xeric, pine both of high elevations and of most exposed situa- forests of the Smokies with one another and with tions. To these observations may be added geo- deciduous forests. graphic ones. In eastern North America, well-defined Since zones are relatively discontinuous, they con- zones seem more characteristic of the Far North form better to traditional ideas of distinct com- than of more temperate climates; and in the West, munity-units than do the types recognized in con- zones are more apparent in some more arid regions tinuously gradating vegetation. Evidence on the na- than in the mesophytic coastal forests. Zones, we ture of zones as vegetation units may be considered may think, are more characteristic of those condi- in terms of four types of questions: (1) How sig- tions interpreted by ecologists as "extreme." A con- nificant is the boundary as a population disconti- verse interpretation must be rejected; it is by no nuity, what part of the species populations of the means generally true that community-types of more zone are limited by it? (2) To what extent are spe- extreme environments are discontinuous with one cies of a zone restricted to that zone, not occurring another. The data of Molholm Hansen (Hansen elsewhere with differentassociates? (3) To what ex- 1930) for the vegetation of Iceland offer some of tent is the zone homogeneous, or to what extent does the most impressive evidence of vegetational con- gradation occur within it? (4) To what extent does tinuity available (see also Ostenfeld 1908, Cooper the zone intergrade with other communities along 1942). gradients other than the one on which discontinuity The term zone has been variously applied to vege- has been observed-is the zone bounded in all direc- tation, but often with the implication that relatively tions, or only in one observed direction? discontinuous belts of vegetation could be recognized When we examine the zones in the Smokies in along a gradient. For present purposes a zone will terms of (1), we find the ericaceous dominants of the be defined as a vegetation type dominated by a spe- heath balds ranging extensively into forest types, cies (or growth-form) possessing a plateau distribu- and most forest species ranging through the bound- tion, and consequently relatively discontinuous in at ary between cove forests and other types below, and least one observed contact with another vegetation- beech and spruce-fir forests above, 4500 ft. The for- type along a presumably continuous gradient of "pri- est-edge of the grassy balds seems more effectiveas a mary " environmental factors. The condition of discontinuity and limits some of the species, but by zonation, or partial discontinuity of types, may thus no means all of them. Very few species are restricted be distinguished from the continuous gradation al- to a zone as asked in (2). Gray beech and spruce ready discussed. Evidence discussed above suggests and fir are (since the zones are defined by their dis- these interpretations of the condition of zonation: tributions);, and some undergrowth species are re- 1. Under some conditions plant communities are stricted to the high elevations of these forests. No "simpler" or of limited species-diversity; this is true species is restricted to the heath balds (Cain 1930b), especially of some successional communities and those and apparently all species of the grassy balds except of some environments interpreted by ecologists as the introduced weeds occur also in forests. Zones are "extreme" or "unfavorable." In contrast to the mixed not homogeneous, but show the gradation of (3) dominance of many communities in "favorable" en- within themselves. A marked elevation gradation ap- vironments, these communities are often dominated pears in the heath balds (Cain 1930b and below); strongly by one or two species; and the dominants of the grassy balds are not homogeneous and appear to adjacent commiunity-typesare more often of different intergrade with sedge balds (Wells 1937). Grada- growth-forms. tion within the gray beech and spruce-fir forests is 2. Populations of dominants of these communities described elsewhere (Parts I & III). may have "plateau" distributions in relation to a Finally, according to (4), it may be observed that given environmental gradient, and the communities zones defined by their boundaries in one direction may consequently appear relatively discontinuous gradate into other vegetation in other directions. with other communities. Continuity of the gray beech forests along the miois- 3. Contrast of dominant growth-formsbetween com- ture gradient with high-elevation oak-chestnut forests, munities may steepen and render more obvious the which in turn are continuous with the rest of the relative discontinuity between them. "Zones" which pattern of deciduous and pine forests, has been indi- are not definable by discontinuity of species popula- cated. The gray beech zone is discontinuous with tions may sometimes be defined by relatively abrupt other deciduous forests in one contact onlv. The change in dominant growth-forms. heath balds appear to gradate into oak-chestnut 4. Zonation is not a general characteristic of vege- heath at lower elevations, although no transect data tation; it is a special and local condition which some- are available on this transition. The spruce-fir for- times results from the manner in which a species is ests of the Appalachians gradate northward into the distributed, or from a contrast in dominant growth- different spruce-fir forests of Canada (Oosting & forms, or both. Neither single-species dominance, nor Billings 1951). contrast in dominant growth-forms, nor occurrence "Zones" should not be taken for granted as ideal in "extreme" environments, nor any combination of vegetation units, nor should their existence be inter- these necessarily implies vegetational discontinuitv- preted as evidence that vegetation in general consists

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 36 R.R. H. II. WHITTAKER WHITTAKER ~~~~~~~IVol. Ecological Monographs26, No. of well-defined units. As the plateau distributions objective means of defining ecotones. As a supple- by which zones are here defined may be regarded as mientto the demonstration of a plateau distribution modifications of the more general binomial form, so of gray beech, however, the percentage differences zonation may be regarded as a modification of the may support the interpretation that there is a rela- general condition of vegetational gradation. Grada- tively steeper rate of change in those populations to tion occurs within the zones but is steepened between which the measurement has been applied in the ele- them; and apparent zones are to varying degrees de- vation interval on each side of 4500 ft. Considered finable by plateau distributions-some not at all, together with the fact that most populations occurring while others have poorly defined or well-defined pla- on either side of the ecotone extend through it, the teau distributions. Zonation may be interpreted, not percentage differences may further emphasize that as a state apart from gradation, but as a tendency, this is a relative discontinuity only. more or less developed under some circumstances, Competition has often been assigned the major toward partial or apparent segmentation of the role in producing discontinuities between natural fundamental vegetational continuum. communities (Cajander 1909, 1925; Elton 1947; Allee et al. 1949: 476-8), and these discontinuities ECOTONES have been regarded as "tension zones" between com- Transitions between vegetation types have usually peting communities. A few cases of mutual exclu- been called "ecotones" by English-speaking ecologists. sion have been observed amiong animal species, when Like the term "zone," "ecotone" has been widely and pairs of species which are ecological equivalents come loosely used, with little attention to what distribu- in contact (Beauchamp & Ullyott 1932, Hairston tional phenomena were involved. In gradating vege- 1951). It seems clear from the transect data of the tation, an ecotone is any part of the pattern which an present study that the change from one vegetation ecolGgist chooses to regard as transitional between type to another is not in general a matter of mutual other parts. For the purposes of the present paper, exclusion of species. Only in the case of the grassy transitions between types recognized in gradating balds is there evidence that one dominant growth- vegetation will be termed simply "transitions"; and form occupies the environment to the near exclusion the term "ecotone" will be applied only to the rela- of the other. Competition among plants occurs as tive discontinuities bounding zones. much in the stands on each side of the ecotone as in If the boundaries of zones are relative discontinui- the ecotone itself, and it seems difficultto conceive of ties, they should be demonstrable as such by suitable a fundamental differencein the character of competi- quantitative methods. The data for the gray beech tions among plants in stands which are interpreted as zone offer a test of this possibility. Percentage dif- ecotonal or transitional and in those which are not. ference (Odum 1950, Whittaker 1952) may be used Neither should the ecotone be, interpreted as a as an approach to rate of change of plant popula- boundary between two communities and a line or tions between stands sampled at fixed intervals along tension zone along which the two are in competition the elevation gradient. Comparing successive stand as communities. Natural communities do not form counts taken at 200-ft intervals in a field transect units of many species bound together by obligate from cove forest up into a beech gap (Whittaker relations, integrated wholes which can meet and 1948, Appendix B, table 6) yields percentage differ- compete as wholes. Examination of species distribu- ences for 2600 and 2800 ft, 2800 and 3000 ft, etc., tions in transects is revealing in this respect. Species of 34, 19, 35, 33, 6, 41, 17, 29, 33, 77, 7, 2, 6. The populations show in relation to the ecotone between difference of 77% is for the interval 4400-4600 ft cove forest and gray beech forest the same individu- and expresses the change of dominance between gray ality of distribution which is demonstrated through- beech and the cove forest species in these elevations. out the vegetation pattern. Species also respond va- The apparent discontinuity in this transect may be riously to the forest-edge of the grassy balds. In this sharpened, however, by an environmental change case the abrupt change of physiognomy implies a from V-shaped ravine to concave "gap" in the same strong telescoping of gradients of light, evaporative elevations. Applying the measurement to the com- conditions, etc. Plant populations which extend into posite transect (Table 5) and grouping samples for or through the ecotone respond individualistically to 400-ft intervals gives values for 1800 and 2200, 2200 the very steep gradient of environmiients-within-com- and 2600 ft, etc., of 40, 42, 46, 28, 40, 50, 62, 30, 23. munities in the forest-edge. Some extend different The differenceof 62% is for the interval 4200-4600 distances into or through the ecotone; others have ft and may be contrasted with the less rapid change population modes within the ecotone. of populations indicated by the values for higher Phytosociologists recognizing the extent of gradual elevations, within the gray beech zone. transition between community-typeshave tended also These percentage differences are strongly influ- to regard the transitions as community mixtures. enced by irregularity in the distributions of popu- There seems to be no sense in which the ecotones are lations (ef. Whittaker 1952). Because of the wide mixtures in which the communities on each side of fluctuation of these values in other transects for them are not also mixtures. Ecotones are not simply which they have been computed, it -is believed that intermediate between the stands on each side of them such measurements do not vet provide a reliable and (cf. Odum 1953 :207 on "edge effect"). A union or

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 37 commodium of species centered in or near the eco- and of a distinct biotic comimunity with its own tone is illustrated for the transition from cove forest physiognomy and characteristic bird species. Ken- to mesic spruce forests (Fig. 13) ; and a related com- deigh (1948, 1954) has shown that a characteristic modium appears between cove forest and gray beech assemblage of bird and mammal species occurs in forest. Several woody species appear in the forest- the forest-edge, and considered this as in part a edge between deciduous forests and grassy balds. No stabilized biotic community-type (biociation). Many such grouping appears between subalpine forest and ecotones may be regarded as community-types in heath bald, although Sorbus americana may have an their own rights by any of the traditional criteria- ecotonal mode there. These ecotones are characterized character-species, dominant species, total species com- by smaller woody species; and another grouping of position, physiognomy, and stratal structure and shrubs and small trees appears in the Smokies, the stratal dominance. submesic commodium between cove forests and oak- In the occurrence of gradation within them and, chestnut forests. Species of the three transitional so far as is known, in the manner in which "com- commodia which are best developed are: petition" operates within them, communities desig- Ecotonal-mesicunion Grassy bald edge Submesicunion Betuia lutea Betula lenta Betula lenta Acer spicatum Acer pensylvanicum Acer rubrum Acer rubrum Amelanchier laevis Amelanchier laevis Amelanchier arborea Viburnum alnifolium Viburnum acerifolium Cornus alternifolia Cornus florida Rhododendron calen- Rhododendron calendu- dulaceum laceum Vaccinium constablaei Vaccinium constablaei Lyonia ligustrina Crataegus macrosperma v. roanensis Rubus canadensis Salix spp. Pyrularia pubera Hex montana Calycanthus fertilis Ceanothus americanus

Some of the species of each group are centered in nated as ecotonal do not differ fundamentally from the transition, so far as occurrence within the communities 'not designated as ecotonal. "Transi- Smokies is concerned; the transitions thus have their tion," "boundary," -"tension zone," and "mixture" own character-species. Other species appear to link represent human attitudes or interpretations imposed these transitional groupings, in the sense that they on vegetation. It is suggested that ecotones and have populations centered in more than one of them. transitions are not phenomena distinct and different Certain genera of shrubs and smaller trees are espe- from communities, and that no "ecotonal principle" cially characteristic of the ecotones in the Smokies, of the general occurrence of distinct boundaries be- and apparently elsewhere also. The transitional tween communities (Allee et al. 1949 :476) is actu- groupings also share species with seral communities; ally known to ecologists. In the sense of the present Betula lenta and B. lutea, Acer rubrum, Amelanchier paper, "ecotones" are communities and parts of the laevis, and Rubus canadensis are major successional whole vegetation pattern in which the gradation oc- species in the Smokies, and some others listed occur curring throughout the pattern is relatively steeper in seral stands. All the species listed, except some of at the limits of a few of the recognized community- those of the grassy-bald edge, occur also as under- types, those interpreted as "zones." the influence growth species in climax forests beyond CLIMAX PATTERNS AND THEIR COMPARISON of the ecotones. The ecotones are thus populated in It has been indicated that undisturbed vegetation part by "versatile" species which have ecotypic popu- Smokies is in a self-maintaining or climax lations (or apomictic races) in a wide variety of in the the moisture gradient habitats. condition at all points along has unbal- It was observed by Gleason (1926) that some (Part I). Recent death of the chestnuts in their species tend to colonize transition zones more freely anced many stands and set in motion changes vegetation than either contiguous association so that, measured composition, but there is no evidence that climatic by component species, such a transition zone is almost of these mountains is converging toward a the climax an independent association. In studying the decidu- climax (Whittaker 1951). According to Phil- ous chaparral of the Wasatch Mountains, Hayward theory as stated by Clements (1916, 1928, 1936; forests (1948) observed that this has properties both of an lips 1934-5, Cain 1939, Braun 1950)) the cove in the ecotone between desert scrub and montane forest. and pine forests of most mesic and xeric sites

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 3838 R. H.II. WHITTAKERWHITTAKER Ecological~~~~~~~~~~~~~~~~~~~~~~~Vol. Monographs26, No. 1 Snmokiesmight be regarded as postelimax and pre- range, however, a pattern in which deciduous broad- climax, respectively. Oak-chestnut forests, occupying leaved trees form 100% of canopy stems occurs at many of the lower slopes of the mountains, might elevations between 4500 and 5500 ft. then be interpreted as climatic climax. Alternatively, 3. Major climax tree species at low elevations are, by the polyclimax theory (Nichols 1917, 1923; Tans- in order of over-all importance with percentages of ley 1935), the five types distinguished along the canopy stems for the whole transect (Whittaker 1951, moisture gradient at low elevations are so many poly- table 3): Quercus prinus 14, Tsuga canadensis 14, climaxes. Of these oak-chestnut forest would pre- Castanea dentata 13, Quercus borealis v. maxima 6, sumably be designated climatic climax and the re- Pinus virginiana 6, Acer saccharum 4, A. rubrum 4, mainder physiographic or topographic climaxes. Tilia heterophylla 4, Pinus rigida 4, Liriodendron In the present study these five types are regarded tulipifera 4, Pinus strobus 3, Aesculus octandra 3, riot as discrete associations but as parts of a single Quercus alba 3, Carya glabra 3, Halesia monticola vegetation continuum. If stands at all points along 2, Betula allegheniensis 2, Quercus coccinea 2, Carya the gradient are accepted as self-maintaining, and tomentosa 2. No other species is represented by more therefore climax in this sense, the continuum itself than 1 per cent of canopy stems. should be regarded as the climax. Rather than a 4. In contrast to these diverse stands of low ele- single climatic climax or set of polyclimaxes, climax vations, composition of canopy trees at highest ele- vegetation in the Smokies comprises a whole vege- vations, above 6100 ft, is simple: Abies fraseri 90, tation pattern ranging from cove forest to pine forest Picea rubens 9, and Betula lutea 1% of canopy stems. along the moisture gradient and from these low-ele- Undergrowths of these stands are varied, but in the vation types to spruce and fir forests in one part canopy there is profound simplification in composi- of the range and to other deciduous types in the tion from lowest elevations to highest. other part of the range at high elevations. 5. If a criterion is chosen for the oak-forest group- In such conditions as those of the Smokies, the ing (all stands in which oaks plus other submesic pattern of vegetation along the moisture gradient and subxeric species make up at least 50% of stems), provides a reasonable description of climax vegeta- then 57% of the climax stands sampled at low ele- tion in a given elevation belt. Through the moisture- vations are oak forests, as defined. Among the same gradient patterns (and the transect data representing samples only 28% are oak-chestnut forests, if these them), composition of vegetation at different eleva- are defined as all stands in which Castanea dentata tions may be compared. Population levels of species and Quercus prinus form at least 50% of canopy can be compared from one elevation to another; thus stems. the changing role and importance of hemlock at 6. The relative importance of oak stands in the differentelevations may be revealed (Fig. 15). Within climax pattern decreases somewhat with elevation. the climax pattern for a given elevation, relative im- For the elevation belts 1500-2500, 2500-3500, 3500- portance and relative position along the moisture 4500, and 4500-5500 ft outside the range of spruce, gradient of species may be compared. When differ- percentages of stands classified as oak forests were ent climax patterns are compared, statements may -57, 46, 44, and 41. The relative importance of be made about the relative importance of a given mesophytic stands correspondingly increased-11, 28, species population or communityfraction in different 41, 59-while that of pine stands decreased-32, 25, patterns, relative importance of growth-forms or 15, 1. life-forms in different patterns, relative variety of 7. Stand composition at the mid point of the imois- species or of dominants in differentpatterns, relative ture gradient may be determined-by the middle sta- extent of change in composition from most mesic to tion of the transect, by a "median" stand so chosen most xeric sites in differentpatterns, etc. Some of that half the samples have more highly mesic and these possibilities may be illustrated in the following half more highly xeric moisture-class formulas, or by statements on climax composition in the Great Smoky percentage-similarity comparison with stands repre- Mountains: sentative of the extremes of the gradient. More com- 1. The climate of low elevations (1500-2500 ft) is plicated means of finding an "elective mean" in a a deciduous forest climate, as expressed in composi- multi-dimensional pattern are described by Ramensky tion of climax stands. Deciduous trees form 73%, (1930). The three techniques mentioned give equiva- hemlock 14%, and pine species 13% of total canopy lent results in the Smokies. Using the first as -the stems in the low-elevation moisture-gradient transect simplest, intermediate stands at different elevations (Whittaker 1951, table 3). are: stands dominated by Quercus prinus at 1500- 2. The climate of high elevations (4500-5500 ft) 2500 feet, stands in which Q. prinus and Castanea in the northeastern half of the range is a coniferous dentata share dominance at 2500-3500 ft, and stands (taiga) climate. Two abietine dominants, Picea ru- in which C. dentata and Quercus borealis share domi- bens and Abies fraseri, make up 87% of canopy nance at 3500-4500 ft. More limited data for high stems in the sample, deciduous trees only 13%. To- elevations give stands in which C. dentata and Q. ward still higher elevations the deciduous trees de- borealis share dominance (45 and 40% of canopy cline to 7 per cent (5600-6100 ft) and 1% (6200-6600 stems), with Acer rubrum, Halesia monticola, and ft) of canopy stems. In the southwestern half of the Fagus grandifolia as major subordinate species.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 39

Moisture-class formulas (Mesics: Submesics: Sub- for the elevation belts 1500-2500, 2500-3500, and xerics: Xerics) change in a manner which indicates 3500-4500 ft, and 4.6 for deciduous forests above increasingly inesic conditions for these intermediate 4500 ft. The tree stratum of the spruce-fir forests is stands toward higher elevations-2 :50 :46 :2, 4:57: much less rich in species, and the diversity values 38:1, 5:70:23:1, 15:73:12:0. are 2.2, 1.3, and 0.7 for 4500-5500, 5600-6100, and Three quantitative approaches to the "climatic cli- 6200-6600 ft. max" may be recognized in the above statements. One approach to measurement of type diversity of Nuitibers 1 and 2 define the over-all character of cli- a vegetation pattern is through percentage similarity. max vegetation in terms of dominant growth-forms. These values give some indication, necessarily subject Numibers 3 and 4 may also express the over-all, cli- to sampling error, of "ecological distance" (Whit- matic character of the vegetation in terms of "aver- taker 1952), the extent to which community popula- age climax composition" for different elevations. tions differ in response to separation along environ- Number 5 determines the stand type to which the mental gradients. When cove forests and pine forests largest proportion of climax stands at a given ele- as extremes of the moisture gradient are compared, vation belong; number 6 describes the differingrela- the percentage similarity is zero. An alternative tive importance of types at differentelevations. It is means of comparing "ecological distances" along suggested that the type to which the majority of gradients is the determination of "half-changes." If climax stands in an area belong be termed the "pre- percentage similarities are measured between an ex- vailing climax type" as an approximation to the treme and successive points along a gradient, a point climatic climax, but with no implication that types is reached at which the value is 50%. If this point other than the prevailing are not also climax. The is then used as a standard and further points along fact that recognition of a prevailing climax type is the gradient compared with it, a second point of contingent on the subjective delimitation of types 50% change may be determined. The number of should be evident. At high elevations in the north- these "half-changes" occurring along a gradient may eastern Smokies a prevailing climax type is easily thus indicate the extent of change in species popula- recognized in the spruce-fir forests; at low elevations, tions along the gradient, for the particular stratum it seems of much more limited meaning to describe or group of organisms to which the measurement has the oak-chestnut forest as a prevailing type. Finally, been applied. Half-change, like half-life, may serve number 7 defines an "intermediate climax stand" to as a useful measurement along a curve without a which other stands are presumably more or less post- sharp lower limit. From the nature of the data, the climax or preclimax in Clementsian nomenclature. values obtained here must be very approximate. In- Complications in the definition of average climax dicated numbers of half-changes along the moisture composition, prevailing climax type, and intermediate gradient are-7, 6.5, and 9 for 1500-2500, 2500-3500, climax stand in other areas may be evident. Their and 3500-4500 ft. Computed in terms of moisture- recognition in the Smokies was much simplified by balance classes rather than species composition, the the mature and stream-eroded topography, absence of values become 4, 4, and 5. hydric sites and near-absence of primary succession, Published data of Braun (1950) permit more lack of contrast in parent materials, and existence of limited application of these approaches to climax extensive undisturbed stands. Where it is possible to characteristics. Extremes of the pattern in the Cum.- use them, they offer quantitative, definable abstrac- berlands are taken to be the sugar maple-basswood- tions from the climax pattern in place of the more buckeye forest (Braun 1950, table 2), and the oak- subjective "climatic climax." They permit quantita- chestnut subelimax (Braun 1950, table 7) occurring tive comparisons of climax patterns whieh are scarcely on some of the driest slopes and ridges. Pine stands possible on the basis of the climatic climax. As com- in the Cumberlands (Braun 1950, table 8) occur only parison between patterns of differentelevations was in special situations of rocky summits and sandstone possible in the Sinokies, so comparisons between dif- outcrops. The following comparisons are possible: ferent soil parent-materials, different climates, etc. 1. Average climax composition in the Cumberlands should be possible elsewhere. Their use will be is approximated by the tabulation for Mixed Meso- further illustrated below in drawing comparisons be- phytic on all sites (Braun 1940a:237), although the tween the Smokies and the related vegetation pattern sample is biased for this particular purpose by the of the Cumberlands. exclusion of hemlock segregates. Average climax A further measurable property of climax patterns composition given above for low elevations in the is that of diversity, in two senses-diversity or rich- Smokies was a melange of species, some of which do ness in species of stands and the pattern as a whole not occur together. That for the Cumberlands is, in (see also Part I), and diversity or variety of stand contrast, recognizable as a type, the Mixed Meso- types occurring along the moisture gradient in the phytic forest, though more highly mixed than individ- pattern. Measurement of species-diversity of pat- ual stands. terns is possible through application of the alpha in- 2. Mixed mesophytic may also be regarded as the dex of Fisher et al. (1943) to the whole sample of prevailing climax type in the Cumberlands, in con- tree steins included in the composite transect for a trast to the prevalence of oak forests of various types given elevation. Values derived are 7.4, 5.6, and 6.1 at low elevations in the Smokies.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions Ecological Monographs 4040 R. II.H. WHITTAKERWHITTAKER ~~~~~~~~~~~~~~~~~~~~~~~Vol.26, No. 1 3. Percentage similarity comparisons with the ex- ent. Oak-hickory stands are of limited extent in the tremes suggest a stand of the chestnut-sugar maple- Southern Appalachians, where mixed mesophytic pre- tuliptree segregate (Braun 1950, table 5) or corre- vails; mixed mesophytic stands are very limited in sponding drier mixed mesophytic (Braun 1950, table the Ozarks, where oak-hickory prevails; between the i B) as intermediate climax stand in the Cumber- Southern Appalachians and the Ozarks there is con- lands, in contrast to the oak-chestnut intermediate tinuous shift of emphasis between the populations of climax of the Smokies. these groupings. It may have little meaning to say 4. In composition and diversity the all-deciduous that one association is the climax here and another mixed mesophytic forest (Braun 1940a :237) is closely association the climax there, with an ecotone between similar to the most highly diverse stands in the them. Fundamentally, the difference in climaxes is Sniokies, those described as cove forest transition. one of relative importance of populations at different Diversity conditions of stands in the two ranges ap- points along gradients. Along the moisture gradient pear to be similar, although the samples are not at a given elevation in the Smokies, the stand popu- strictly comparable. The sum of a set of tables which lations shift progressively as species and groups of approximate a moisture-gradient pattern (Braun species decrease and drop out while others enter 1950, tables 2, 1A, 5, 7, 8) gives a diversity value and increase. Along the elevation gradient the whole (5.5) scarcely different from the 5.8 obtained for moisture-gradient pattern changes in character as canopy stems at low elevations in the Smokies (Whit- species populations and vegetation types change in taker 1951, table 3). relative importance from lower elevations to higher. 5. Indicated half-changes along the moisture gradi- Along the climatic gradient west from the Cumber- ent in this same set of tables are slightly more than lands species populations, vegetation types, and 2 in terms of species and approximately 2 in terms whole groups of species, or associations, drop out and of moisture classes, if the pine stands of special sites enter, decrease and increase, or replace one another are excluded. in relative importance. The gradient west from the Comparing the vegetation pattern of the Cumber- Cumberlands provides an especially favorable op- lands with that of the Smokies, it may be said that portunity to observe how the climax pattern of one the former is more clearly mesophytic in character area may be gradually transformed along a cli- and of comparable species diversity, but less diverse matic gradient to that of another area. in terms of stand types along the moisture gradient. CONSIDERATIONS OF LOGIC AND METHOD The Cumberlands pattern largely corresponds to the mesic half of the Smokies pattern (cove forest, cove In the phenomena with which the student of vege- three levels may be recog- forest transition, and oak-chestnut forest), while the tation deals, at least major for- more xeric types of the Smokies (oak-chestnut heath nized. The first of these is that of environment, gradients. A second is and pine forest) are reduced in the Cumberlands to mulated here as a pattern of distrib- stands of special edaphic situations. The two climax that of populations of individual organisms, these A third is that of patterns are thus related through a shift of relative uted in relation to gradients. composi- emphasis of the mesophytic grouping and the oak- community phenomena as such-community metabolic and energetic forest grouping. tion, diversity, productivity, individual organism Such shift in relative importance of parts of cli- function, etc. TWhelevel of the consideration be- max patterns is further illustrated in Braun's treat- itself is hardly to be omitted from but the present study ment of vegetation of the Southern Appalachians, the tween the first and second; distributions of populations to environmental Ozarks, and the area between. Three regions are relates recognized from the Cumberlands west: a region of gradients. thought that interpretation in science mixed mesophytic climax ii and around the Cumber- It is generally from lower to higher levels, as from lands, a region of oak-hickory climax in the Ozarks should proceed the individual to the and west and north from them, and between these the chemical to the biological, mutual interrelations of the Western Mesophytic Region, regarded as eco- social. Because of the levels, must sometimes in tonal, and forming "a mosaic of like and unlike cli- study and interpretation proceed of the maxes" (Braun 1947). Of this last region Braun the opposite direction; thus the individuality be understood in relation (1950:123) writes, "In its eastern part, mixed meso- person is scarcely to except the of distribu- phytic forests are of frequent occurrence; westward to a social milieu, and individuality is to be understood in they become more and more limited in extent and tion of a species population of communities. Some of the more closely dependent on very favorable habitat relation to a pattern well be considered, how- conditions. ... Because of the many mixed forest com- methods of synecology may to the ideal of from lower munities, and of the gradual change from east to ever, in relation proceeding west in extent of mixed mesophytic communities levels to higher. com- and in composition of forests, and of the increasing When a vegetation study proceeds from frequency of communities in which oaks are domi- munity-types to a consideration of species distribu- nant, this is considered a transition region." tions, it is proceeding from the third to the second The description seems most expressive of grada- level. In relation to species distributions, community- tion on a grand scale, along a major climatic gradi- types are secondary or derivative phenomena; for it

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 41 is the distributional relations of species which cause communities are interpreted. In sampling, an ecolo- ecologists to recognize community-types. The ap- gist must usually choose certain stands as "typical" proach to species distributions through community- or "representative" while other stands are passed by types thus involves a methodological inversion, pro- as atypical, transitional or mixed, or disturbed. When ceeding from a higher to a lower level and from sec- a number of stand samples are presented in a com- ondary phenomena to the distributional relations piled table for a community-type,the table and the which underlie them. If, on the other hand, prob- properties of the community-type indicated by it lems of the validity and meaning of associations are represent not merely "the community," but an ecolo- *approached by way of the associations themselves, gist's conception of the community as this influences other difficulties of method appear. Samples are his choice of samples. A constancy value of 100% -taken from the vegetation, and it is difficultto take for species A may sometimes mean little more than such samples without preference for those which con- that an ecologist regards as typical only stands con- form to subjective ideas of associations and some ex- taining species A. Consistent differenceand apparent clusion of those which do not. If, now, the samples discontinuity between compiled tables for two com- are used to study validity of associations, a clear cir- munity-typesmay imply only that intermediate stands cularity is introduced into the procedure; associations were not sampled. Statement that a community-type are being studied in terms of samples taken in terms is homogeneous or heterogeneous may reflect the nar- .of associations. The student is thus caught in a circle rowness or broadness of the ecologist's conception, as of studying secondary phenomena and his own ab- this directs his sampling. The author would empha- stractions from them in terms of themselves, in place size both the value of such compiled tables and the of the more reasonable procedure from lower levels fact that sample choice, and therefore statements to higher. The contrary viewpoints of Du Rietz about the community-typeand constancy and fidelity (1921 :215) and Tansley (1920,) may be noted. of species, can seldom be wholly independent of a It has been indicated that the approach to species given ecologist's conception of the community-type. distributions along the moisture gradient (but not In practice compiled tables are often presented and along the elevation gradient) used in this paper also inferences drawn from them as if the table repre- involves a circularity. It is further clear that the sented "the community" in nature, wholly independ- approach to the moisture gradient proceeds from the ently of the ecologist's interpretation of it. higher level of the community gradient to the lower Two very general difficultiesappear here, in the one of the assumed gradient of physical environment failure to allow for the ecologist's interpretation of -the approach is inferential. The point is not vegetation as guiding choice of data on which further that such procedures are excluded from ecological interpretation is based, and the confusion of levels methods; it is that they should be understood, allowed involved in the word "community." At least three for, and made explicit. The author has sought to levels of phenomena may be recognized among the state such limitations in his own methods and feels applications of this term (cf. Dice 1952 :426): (1) that certain limitations of other methods which may communities or stands which are essentially homo- be advanced as alternatives should also be observed. geneous over some area of the earth's surface and The relation of abstract types to the stands from comprise all organisms above and below the soil sur- which they are abstracted involves further difficulties. face in this area, (2) community-fractionsand micro- In preliminary field work a type or association is communities, comprising some organisms of the com- recognized, and a number of samples conforming to munity or stand which are for some reason grouped it are taken. The samples are relatively uniform and together, as in a stratal community of plants or as a group distinct from those taken to represent animals, a community of animals of a particular other associations; the further sampling thus rein- taxonomic category, or the community of a special forces the conception of the association. This, now, micro-environment, and (3) community-complexes, is likely to be seen as a "good," "real," or "valid" unit, comprising a number of distinguishable communities verified by the sampling. The abstract conception is or stands arranged in a mosaic or pattern in space, likely to be projected back into nature and regarded such as landscapes, biotic provinces, the complexes as a real unit of structure of that which has been of bogs and frost-determinedlandforms, topographic sampled; the unit in the ecologist's mind is reified patterns, etc. On each level abstract types of com- or hypostatized (cf. Tansley 1929). The initial choice munities may be recognized, such as: (1) formations, of a "climatic climax" miay similarly be reinforced associations, sociations, dominance-types, site-types, by observations of successions and used as a basis for (2) unions, socions, micro-associations, (3) landscape- further interpretation, until the "climatic climax" types, bog types, climax-complexes, the generalized seems an established and necessary fact. It is all too topographic patterns of the present work. The am- .easy to move from the view that such abstractions biguity of the term community results not from the have real usefulness to the view that they are part of, accepted convention that it may apply on any level of ,or must correspond to, reality. community-size from the inhabitants of a termite's The reification of the community-typeis often ap- gut to those of an ocean basin, but from its applica- -parent in the manner in which quantitative data on tion on at least two levels of abstraction: the con-

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions Ecological Monographs 42 R. H. WHITTAKERWHITTAKER Vol. 26, No. 1 ception of a particular "concrete" stand, and that of product of an ecologist's study of natural communi- an abstract type derived from several more or less ties and formulation of a class-concept. similar stands. In a complex pattern such as that formed by na- Other fields are not similarly affectedby use of the tural communities, the creation of class-concepts is sanmeterm for such differentconcepts as, say, organ- in general "free" or "arbitrary" (see also Part I). ism and taxon, individual and social class, personality No properties of vegetation force an ecologist to and character-type. An association is not a com- choose a single manner of defining community-types. munity; it is a type of community. Various authors Emphasis of different properties of vegetation will have interpreted the association as a "concrete" unit lead to differentclass-concepts of vegetation types; (Du Rietz 1921:15, 1928, 1929, Alechin 1925, Clem- these class-concepts in extension will have differing ents 1928 :128); many others have recognized asso- memberships, variously related to one another. The ciations as abstract units (Nordhagen 1922, 1928, differentdefinitions of community-types in different Wangerin 1925, Kylin 1926, Liidi 1928, Wendel- schools, the nature of the data presented in this study, berger 1951, 1952, Dice 1952 :425). Abstract char- and the recent research of Goodall (1953) may show acter of the association has in general been empha- how variously stands may be grouped and class-con- sized in the school of Ziirich-Montpellier, and the cepts derived from them. The concept of the "com- stand has often been termed the association-indi- plex continuumr"may be modified in two respects- vidual (Pavillard 1912, 1935, Braun-Blanquet & granting that a degree of discontinuity exists in vege- Pavillard 1922, Braun-Blanquet 1921, 1951). Never- tation and that a degree of species clustering may theless, the confusion of levels in the term community also occur-without altering this understanding that exists as an encouragement to ambiguity and the un- ecologists enjoy considerable freedom in the creation critical transfer of information and inferences about of their class-concepts for associations and other communities in one sense to communities in the other units. sense. In the author's opinion, this usage is a major A further means of clarifying relations of com- factor of confusion, misunderstanding, and ineffec- munities and community-types is through the em- tiveness in discussions of the nature of communities phasis of non-identification in general semantics and types of comnuunities. It is suggested that clarity (Korzybski 1933, Johnson 1946, Vogt 1948, Rapo- of discussion is greatly enhanced by a simple and port 1952). A manner of thinking based upon classi- fundamental distinction of terms: Community should fication and limited by that classification, tending to be applied only to the concrete community or stand, identify classes with the phenomena from which they and any abstract type or class of communities may are abstractions, to accept the classes as inherent in be termed a community-type. these phenomena and to approach the phenomena The general conception most appropriate to the only through the classes, may be termed "Aristote- theory of community-typesis that of class (Langer lian" in the sense of the general semanticists. Funda- 1953). A stand is an individual member of that class mental to the classificatory approach and to ways of which is an association or other community-type; in thinking related to it are three "Aristotelian laws." this relation it need be assumed neither that the 1. The law of identity: A is A, beech-maple is stand is discontinuous with other stands nor that the beech-maple, beech-maple stands Al, A2, A3, . . . An class is discontinuous with other classes. The mem- belong to or are Beech-Maple Association A. bership of the class forms its extensional or denota- 2. The law 8f the excluded middle: Anything either tive definition; those criteria by which it may be de- is or is not A; each of a set of stands Al, A2, A3, cided whether a given stand is, or is not, a member . . . An either is or is not a stand of Beech-Maple of the class constitute its intensional or connotative Association A. definition. (E. g. Members of association A shall 3. The law of non-contradiction: Something can- contain at least 4 of the following character-species not be both A and not-A; stands Al, A2, A3, * * . An . . . or, Members of association B shall have canopies cannot both belong and not belong to Association A. containing at least 50% Fagus grandifolia and Acer The limitations of themelaws may be observed in saccharum). relation to the diversity and intergradation of stand Intensional and extensional definitions are coupled types. A is not A; beech-maple stand Al is not and normally develop in relation to one another in beech-maple stand A2; the Beech-Maple Association the interaction or transaction between ecologists and is not beech-maple stands Al, A2, A3, . . . An; the vegetation. The relation of members to the class is Beech-Maple Association is not beech-maple vegeta- not, however, one of parts to the whole; for the com- tion; the beech-maple concept of one ecologist is not inunity-class is not a unified, functional whole of the beech-maple concept of another ecologist; beech- which stands are the interacting or interrelated parts. maple, in general, is not beech-maple. Something The class, in extension, is the collection of stands or can just as well be both A and not-A; a stand can stand-samples to which a class-concept (the inten- be both the Beech-Maple Association of one ecolo- sional definition) applies. The class is thus a con- gist and not the Beech-Maple Association of another ceptual entity; and membership in the class is the re- ecologist, to whom it may be the Maple-Basswood lation of falling under, or corresponding to, a con- Association. If the class-concept is so stated as to cept. An association or other community-typeis the provide a 'sharp limit (50% Fagus plus Acer), then

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 43 a stand either is, or is not, a member; but the limit CONCLUSION is arbitrary. If (as is more generally the case) the The thesis of this discussion, developed through the class-concept is that of an ideal stand approached by five preceding sections, is that (1) species popula- actual stands (full dominance of Fagus plus Acer, tions are variously distributed, each according to its or, possession of all or most of the following charac- own physiology and genetic pattern, so that (2) ter-species . . . ), then stands are not simply beech- community-types are not organized units of many maple or not-beech-maple; they may be 90, 65, 40, 15, species, but (3) can be interpreted and defined in 5% beech-maple by the chosen criterion of beech- terms of population distributions along gradients and mapleness. (4) understood as part of the complex population The justification of the classificatory approach is patterns of climax vegetation, to which (5) the not that classes are inherent in, or necessarily appro- method of gradient analysis is appropriate. In simile, priate to, the phenomena being studied, but that they vegetation may be compared with a pattern of col- serve the needs of investigation and scientific com- ored lights cast upon a screen, a pattern formed by munication. In the latter of this work the part au- innumerable overlapping individual beams of light, thor will venture to classify stands of the Great some bright, some faint, some broad, some thin, some Smoky Mountains. Up to the present, however, a of simple and some of complex form, but most having different and non-classificatory conception has been a brighter center and dimming gradually away from developed. This interpretation which is, like the this, all somewhat mobile, changing position and association-unit theory, a conceptual construction per- intensity through time, moving somewhat independ- mitting the relation to one another of data on en- ently and thus changing their relations to one an- vironments, organisms, and communities, may be other, composing all together a blended, intricate, brieflv formulated. Physical environment forms a and fluid pattern. Parts of the pattern may be classi- complex pattern of gradients. At each point in this fied; but an essential complement to the classificatory pattern, communities develop through succession; approach is one treating vegetation patterns as and as community and soil develop and physical en- wholes, regarding vegetation units not as objects in vironment is modified, a successional gradient of themselves but as fragments abstracted from the de- community composition corresponds to a successional sign within and in relation to which they exist and complex-gradient of environments. At the culmina- have their meaning. tion of successions, a complex climax vegetation pattern corresponds to the complex pattern of environ- III. VEGETATION TYPES AND THEIR mental gradients. In this pattern species populations DISTRIBUTIONAL RELATIONS distribute themselves individualistically, each maintaining itself where it can in relation to gradients BASES OF RECOGNIZING AND DESCRIBING TYPES which are, now, complex-gradients of environments Vegetation of the Great Smoky Mountains has been in communities. Species distributions mostly take the interpreted (Parts I and II) as a complex popula- form of tapered curves; and the vegetation itself tion pattern, but it is difficultto describe as such. For forms a complex continuum of populations, in which the descriptions to follow, the author has resorted to local steepenings and interruptions of gradation a system of vegetation types. The differentschools occur. of ecology offer a variety of vegetation units from One approach to this pattern is through classifica- which to choose according to the purposes of a par- tion of its parts. An effectivealternative is through ticular study. The most widely used system, that of study of gradients of species populations and of com- Braun-Blanquet (1921, 1932, 1951), is not well suited munity characteristics in relation to gradients of en- to the present study because data on herbs, bryo- vironment (Whittaker 1951, 1952, 1954a, 1954b). phytes, and lichens are incomplete. Rather than fol- These two approaches, emphasizing relative discon- low this or any other established system, the author tinuity vs. relative continuity, may be regarded as has sought to let vegetational conditions of the mutually complementary and as serving different Smokies and the kind of data available determine the needs, rather than antagonistic. It is the latter ap- choice of units. In the recognition of types physiog- proach, of relating gradients to gradients, which nomy and dominance are emphasized, following the has been explored in the preceding parts of this traditions of English-language ecologists. paper. With varied details of technique the basic Two vegetation patterns or complexes are first to method may be recognized also in work of Ramensky be distinguished. At elevations over 4500 ft (1370 (1924, 1930) and other Russian authors studying m) in the northeastern half of the Smokies, forests ecological series, in Finnish work with ecological se- of spruce and fir prevail. These may be regarded ries of site-types, and in studies by M0lholm Hansen as the subalpine forests of this range, and as the (Hansen 1930, 1932) ,Tuomikoski (1942), Ellenberg local representation of the taiga or boreal forests of (1948, 1,950, 1-952), Curtis & McIntosh (1951), high latitudes and altitudes. At elevations below Brown & Curtis (1952), and Hale (1955), Major 4500 ft (and above 4500 ft outside the range of (1951), Goodall (1954a, 1954b), and Beard (1955). spruce), forests are predominantly of deciduous trees, It is suggested that this approach of relating gradi- as are the forests of the eastern United States south ents to gradients be termed gradient analysis. of the spruce-firbelt. Each of these patterns includes

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions Monographs 44 R. R.H. H.WHITTAKEREcological WHITTAKER ~~~~~~~~~~Vol.26,No. I more than one physiognomic type; they are conse- treue for vegetation types, unless these are defined quently not formations. The word "system" has been more broadly than in the present work. In certain used for them here, with no intention of establishing cases species have been designated character-species this as a new vegetation unit. These two major parts for two vegetation types in which their varieties or of the vegetation pattern are named, for their geo- distinguishable population-types are centered. No graphic relations, the "Eastern Forest System" and efforthas been made to distinguish local and regional the "Boreal Forest System." character species, and character-species indicated can- Physiognomic types within each system are a logi- not be assumed to have more than local usefulness. cal basis of further division. Vegetation of the The units used differ also from the very broad Smokies includes types dominated by four major dominance-types, characterizing extensive regions as growth-forms-needle-leaved evergreen or coniferous prevailing climaxes, which are termed associations by trees, deciduous broad-leaved trees, evergreen-sclero- many American authors (Clements 1928, Weaver & phyllous shrubs, and grasses. It is useful for the Clements 1929, Oosting 1948, Braun 1950). What- present study to divide the tree growth-forms ever their value for geographic treatment of vegeta- further. The pines of the Smokies, forming open, tion, the American associations are considered too xeric, lower-elevation stands are conveniently distin- broad and ill-defined to be suitable as units in the guished as a growth-form from the abietine trees Great Smoky Mountains. In dealing with the vegeta- (Picea, Abies, Tsuga) forming denser and less xeric tion of this range it is surely a question not of "oak- stands, predominantly of higher elevations. Oaks hickory," "oak-chestnut," and "oak-pine," but of affect lower strata through late leafing and type of particular types dominated by certain species from leaf-litter differentlyfrom many other deciduous trees these genera. (Braun 1935b, Kucera 1952) and may be distin- The fifteentypes listed below are by no means the guished as a growth-form. Along the moisture gradi- only ones which might be recognized. In a vegetation ent in the "eastern" forest pattern, non-quercine trees pattern as complex as that of the Smokies, a large predominate in mesic sites, oaks in intermediate ones, number of dominance-types may be distinguished in and pines in xeric ones (see Part I). Ericaceous variations in stand composition. The author has shrubs are a major growth-form in the Smokies, subjectively chosen those types which seem most sig- dominating heath balds and sharing dominance with nificant in the vegetation pattern. The types recog- an open tree stratum in forest-heaths. The heath nized are those which occur most frequently in the strata are predominantly evergreen but contain, and pattern and occupy largest areas of the, mountain are in some types dominated by, deciduous Vac- surface. They are those types which seem to be cinioideae. Physiognomic types recognized in the combinations of dominants usually appearing in a Smokies are thus: abietine forest, non-quercine de- particular range of environmental conditions, rather ciduous forest, oak forest, pine forest, forest-heaths than occasional variations in stand composition. Only (pine heaths and oak-chestnut heath), heath bald, those types considered "climax" in the sense of ap- and grassy bald. parent stability in their present sites will be de- Vegetation types are further distinguished accord- scribed. Major successional types which may be men- ing to their dominant species. All but two of the tioned include Prunus pensylvanica, Betula lutea, and vegetation types described below are defined by domi- Amelanchier laevis forest types at high elevations, nance of one species or a pair of species. The two Betula lenta and Acer rubrum forests in intermediate exceptions (cove forest and heath bald) are domi- sites of most elevations, stands of Pinus spp. in drier nated by several important species, the proportions sites of middle elevations and many sites of lower of which vary from one stand to another. These types ones, and Liriodendron tulipifera in more mesic sites are consequently defined by distributional groupings of lower elevations. A type list for the Smokies, in- or commodia (Part II) of species. cluding some successional types, is given by Cain Since the vegetation units are not defined primarily et al. (1937). by character-species, according to the system of The use of vegetation units is in contrast to the Braun-Blanquet (1951), they are not necessarily as- approach developed in parts I and II. The funda- sociations in the sense of that term which most nearly mentally differentapproach through units has been has international acceptance. They consequently will chosen for Part III strictly as a matter of conveni- be designated only as "types" and will not be given ence in describing parts of the vegetation pattern and formal, latinized names. In deference to phytosocio- relating the pattern to topography. It should be logical practice, however, character-species have been emphasized that the classification developed is only designated for these types as far as possible. To in- one of many, equally valid, ways of classifying vege- dicate these without repetition, names of character- tation of the Smokies. It should not be supposed that species are marked with asterisks (*) in the descrip- the units to be described are "real" units, already tion of the type to which they apply. In view of the marked out in the vegetation, which the author has distributional data already discussed, it need hardly recognized by studying it. Classifications of vege- be emphasized that these character-species are of low tation are human creations, resulting from the study degree (mostly preferential or holde). Very few spe- of vegetation for certain purposes, emphasizing cer- cies of the Smokies can be considered exclusive or tain properties of vegetation which become criteria

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 45 of classiflcation.Continuity of mostvegetation types has been previouslydiscussed and may well be kept in mind; but, for brevity'ssake, most transitionsbe- tweentypes will not be described. The descriptionsare based on data compiledin the composite stand tables (Appendix C). For these, the 300 site-sampleswere -classified into types accordingto dominantspecies. Within each vegetation type the site-sampleswere furtherdivided by 1000-ft elevationbelts and, in some cases, by relativeimpor- tance of moistureclasses of trees. By this classification,the site-sampleswere grouped into 27 composite tables for vegetationtypes and their subdivisions, each table includingabout 1000 tree stems and 100 canopy stems. The tables cannot be published here, but are otherwise made available (see Note on Supplementary Publication). Numbers of tables in Appendix C on whichdescriptions are based are indicatedafter the names of types. The classificationof climax vegetationtypes in the Great Smoky Mountainsis as follows: I. Eastern Forest System A. Mesic, non-quercineforests 1. Cove hardwoodsforest (1-7) 2. Eastern hemlockforest (8, 9) 3. Gray beechforest (10, 11) Flats. B. Intermediate,oak forests FIa. 14. Cove hardwoodsforest, Porter Creek Reproducedby permissionof Thompsons,Inc., Knox- 4. Red oak-pignuthickory forest (12) *ville,Tenn. 5. Chestnutoak-chestnut forest (13-15) 6. Chestnutoak-chestnut heath (16, 17) and the small stemsof the reproducingtrees are not 7. Red oak-chestnutforest (18, 19) dense. The foresthas an open and spacious appear- 8. White oak-chestnutforest (20, 21) ance, with a high deciduous canopy borne above a. C. Xeric, pine forests rich herbaceouscarpet and little interveningwoody 9. Virginia pine forest (22) undergrowth. 10. Pitch pine heath (23) 9ix tree species share dominancein most of the 11. Table mountainpine heath (24) cove forests-Tsuga canadensis, Halesia monticola', D. Extremeexposure type Aescutlus octandral, Tilia heterophylla', Acer sac- 12. Grassy bald charum', and Betula alleghenieneis>'-, while Lirio- and Fagus grandifoliaare impor- IL. Boreal Forest System dendrontulipifera tant in some stands. Each of theseforms 10 or 20% A. Subalpine forests of the canopy in some stands and is locally more 13. Red spruce forest(25, 26) Togetherthese 8 species make up 80 or - important. 14. Fraser firforest (27) 90% of the canopy of cove forests,dominating stands B. Extremeexposure type in differentcombinations. With them occur a num- 15. Heath bald ber of otherspecies. Fraxinus americanaand v. biltmoreana,F. pennsylvanicav. subintegerrima,Quer- VEGETATION TYPES Of THE GREAT SMOKY MOUNTAINS cu48borealis v. maxima,Magnolia acuminata,Carya 1. COVE HARDWOODSFORST cordiformisa,and the large mesic populationof Acer The cove forestsof the SouthernAppalachians are rubrumare widely distributedcanopy species which doubtlessamong the most beautifuldeciduous forests are seldom numerous. Prunus serotina is locally in the world. Trunks of the canopy trees, 3 and 4 importantbetween 3500 and 5500 ft. Species of ft or more in diameter,rise to bear crowns 75 and small and mediumstature include Magnolia tripetala- 100 ft fromthe forestfloor. Tops of the trees reach and Carpinu8 caroliniana at low elevations,Mag- 100 to 150 ft withoccasional great tulip trees (Lirio- nolia fraseri, Cladrastis Wtea, Hex opaca, and deudrotstulipifera), 6 and 7 ft in diameter,ap- Ostryavirginiana at mostelevations, and Acer spica- proaching200 ft. Below the canopy smaller trees tumand Amelanchierlaevis at high elevations. Ca8- forma broad, but not dense, foliage layer. In sum- tanea dentata and other subxericspecies are absent mer a verdantgrowth of mesophyticferns and other fromthe most mesic stands of deep valley flats (see herbs nearlycovers the forestfloor. Small tree spe- tables 2 & 3; Appendix C, tables 3 & 4; Whittaker cies and shrubsare almostabsent frommany stands, 1951, tables 1 & 3).

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 46 R.R. H. H. WHITTAKER WHITTAKER Ecological~~~~~~~~~~Vol. Monographs26, No. The shrub stratumis poorly developed or absent gests, a naturalistwandering, unknown to him, into in standsof valley flatsat middleelevations. Euony- some Tertiaryforests might see no very conspicuous mus americanus* and Lindera benzoin* occur in low- differenceexcept the presence of Ginkgo and Se- elevationstands, Cornus alternifolia, Viburnum alni- quoia (or Metasequoia). folium, and Ribes cynosbati in high-elevationones. Various combinationsof dominantsappear locally Hydrangea arborescens occurs locally at all eleva- in the cove forests: Acer saccharum-Halesia monti- tions; Rhododendron maximum occurs along streams cola, A. saccharum-Tilia heterophylla, Tsuga cana- and (often with Leucothoe editorum) in some other densis-T. heterophylla, Aesculus octandra-T. hetero- sites,particularly under hemlocks. There is, however, phylla, A. octandra-Betula allegheniensis, T. cana- no shrub groupingcharacteristic of the cove forests densis-Liriodendron tulipifera, T. canadensis-Fagus as a whole. grandifolia, T. canadensis-B. allegheniensis, A. sac- The herb stratumis the richestin the mountains. charum-F. grandifolia. None of these (cf. Cain 1943) Cain (1943) lists over a hundred species for the representsa well-definedand extensive type. The spring and summeraspects, and Braun gives com- most significantchange in the cove forests,other parable lists in her papers (1935b, 1940a, 1942). than the transitionto oak types, is the gradual Summerherb coverage is as high as 80 percent in rise of Aesculus octandra, Tilia heterophylla, and some sites, with a luxuriantgrowth of mesic ferns Betula allegheniensis toward their differentmodes and herbs of spreading or umbrella-shapedfoliage at higher elevations,where they form the so-called form. Most species of the mesic herb union are cen- northern hardwoods dominated by these three tered in the cove forests and may be regarded as southern tree species (cf. Braun 1950 :207). The character-speciesfor them. The most abundant spe- upper cove forestsmay be regardedas a sub-typeof cies of the union in summerare Eupatorium rugo- the cove forestswith the same dominantsin different sum*, Cimicifuga racemosa*, Caulophyllum thalic- proportions,but with Liriodendronand Tsuga often troides*, Impatiens pallida*, Laportea canadensis*, absent,and with Acer spicatum, Amelanchier laevis, Trillium erectum v. albiflorum*, Aster divaricatus*, Viburnum alnifolium, and Cornus alternifolia domi- and the ferns Dryopteris spinulosa v. intermedia*, nating the stratumof shrubs and low trees. Mo- and Athyrium thelypterioides*. Beneath them grow narda didyma, Rudbeckia laciniata, Oxalis montana, smaller species-Viola spp., Stellaria pubera, Ti- and other species of the high-elevationmesic union arella cordifolia', Galium triflorum,and Euonymus appear in the herb stratum. obovatus. The beautifullydeveloped herb community An additional subtype may be recognizedin the is stratifiedas the forest is, with a canopy of tall cove forest transitionbetween the mesic forest and spreading herbs and ferns and an undergrowthof oak forest groupings. As less mesic sites are ap- lower and prostateplants. Some species of the sub- proached from the cove forests,increasing percent- mesicherb union occur,and towardhigher elevations ages of submesicspecies occur in the stands. Sub- species of the mesichigh-elevation union forma con- xeric species (Castanea dentata, Oxydendrum ar- spicuous part of the stratum. boreum, Quercus alba and Q. prinus, Nyssa sylvatica) The relationof cove forestand mixed mesophytic also appear in significantnumbers. In the cove forest to the Arctotertiaryforests has been indicated by transition25 or more tree species, representingall Braun (1935a, 1947, 1950). Cain (1943) studiedthe classes but the xeric,may be encounteredamong 200 Tertiarycharacter of the cove forestsin the Smokies, or 300 stems. In the shrub stratumRhododendron using fossil recordsof genera and distributions,par- maximum is mostimportant, and many of the shrub ticularly the frequent eastern Asia-eastern North species of oak-chestnutforests appear. With their America disjuncts. Cain consideredthat all genera heaviergrowth of small trees and greatershrub cov- of ferns and shrubs and most genera of trees and erage (20-50%), thesestands are less open and spa- herbswere of Tertiaryhistory. His figuresfor trees cious in appearance than the cove foreststhemselves. and floweringherbs were 86 and 75% of genera with Herb coveragesin cave foresttransition vary widely known or indicated Tertiary history,including the (5-50%), the extremesof this range corresponding minorspecies in his lists. On the basis of his con- to the low coveragesin oak-chestnutforests and the stancv-classV or the most abundant species of the high ones in cove forests. low-elevationmesic herb union listed here, all prin- Mixed Mesophytic in the Smokies and Cumberlands cipal herb species are of Tertiary The union history. The cove forestsof the Smokiesare closelyrelated is related, and linked by its predominantgrowth- to the mixed mesophyticforests of the Cumberland form, to Lippmaa's (1939) Galeobdolon-Asperula- Mountains (Braun 1935b, 1940a, 1942, 1950), but Asarum Union of herbs surviving from Eurasian the mesophyticcommunities of the two ranges differ Arctotertiaryforests. Of the trees listed by Cain, in-some respects (see also Part II). With the greater if Tulipastrum is grouped with Magnolia, and area occupied by mixed mesophyticin the Cumber- if Robinia pseudoacacia and Oxydendrum arboreum, lands, thereis a greater differentiationof the asso- whichare extraneousto the cove forests,are set aside, ciation into segregates; some of the combinations all are of Tertiaryhistory. The Tertiaryhistory of mentioned by Braun were not observed in the the genera certainlydoes not imply that extensive Smokies. Notable among these are the beech (Fagus changes have not occurred. However, as Cain sug- grandifolia) ravine communities,Quercus alba-F.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 47 grandifolia, F. grandifolia-Castanea dentata, Tsuga these mesophyticforests in the Smokies have dis- canadensis-F. grandifolia, and T. canadensis-Q. alba couragedthe authorfrom following this course: forests. In the Smokies the low-elevation(white) 1. Tilia heterophylla and Aesculus octandra are indi- beech is a minortree, and the low-elevationpopula- cated by Braun (1950:43) to be characteristicspecies tion of Quercus alba is importantonly in oak-hickory of Mixed Mesophytic.In the Smokies,at least, other stands which are limitedin area. In the somewhat character-speciesmay be added to these,notably Ha- lower elevationsof the Cumberlandsthese two low- lesia monticola, Betula allegheniensis, Cladrastis lu- elevation populations are major tree species and tea, and the local populationof Acer saccharum. The dominantsof types. populations of these character-speciesare centered In general compositionmixed mesophyticof the not in the cove foresttransition but in cove forests Cumberlandsmay best be comparedwith cove forest proper. transitionas recognizedin the Smokies. Thus the 2. The optimumdevelopment of mesophyticforests, average compositionof mixed mesophyticon Black mesophyticherb stratumas well as Mountain,based on 2,000 stems (Braun 1940a: 237) with the richest shows a striking correspondencewith cove forest strongestdominance by fiveof the character-species transitionbelow 2500 ft (Appendix C, table 6) in above, is in the cove forestsof deep valleys, not in the Smokies. Principal differencesare in the occur- the cove foresttransition. rence of Tsuga canadensis as a major species in the 3. Castanea dentata and other tree species charac- Smokies, and the appearance of Halesia monticola teristicof the oak forestsare almostabsent fromthe and Pinus strobus in the Smokies stands. The for- cove forestsproper and extendinto cove foresttran- mulas for compositionin terms of moisture-classes sition only with the tails of theirdistributions. It is (Part I) permitfurther comparison of forestsin the difficultto interpretthese species as belongingto the two areas. These formulas(Mesics: Submesics: Sub- cove forestsand segregatinginto oak forests. Rather xerics: Xerics) are, for Braun's table of mixedmeso- than this their distributionsappear simply to over- phytic,67:12:20 :0, and for the cove forest transi- lap in part withthose of cove forestspecies. tion in the Smokies,65:14:21:0. Table 1 of Braun 4. Applied to the Smokies, the Mixed Mesophytic (1950:53) indicatesthe differencesin compositionof in Braun's sense seems less a definablevegetation all-deciduousmixed mesophyticforests of the Cum- type than a range of stand conditionsfrom cove for- berlandMountains according to slope. The stands of ests to more mesophytic oak forests. Mixed Meso- mesic slopes (81 :9:10:0 to 70:12:18 :0) correspond phyticseems too broad and heterogeneousa group- to more mesic cove forest transitionstands in the ing, bringing togetherinto one association species Smokies,while in stands of southslopes in the Cum- whose relations to the moisture gradient are too berlandsthe mesics are slightlyin the minority(49- widelydifferent. 15:35:0, 42:20:39:0). In the Cumberlands,stands of mesic species mixed Cove foreststhemselves in the Smokies are more with oak-forestspecies form the prevailing climax distinctlymesophytic in composition.For threecom- type. It is consequentlyreasonable to emphasize posite stand counts from differentelevations (Ap- these highly mixed stands and to regard the mesic pendix C, tables 1, 2, 5) canopy compositionwas- and xeric extremesas secondaryto them. The diverse 88:10:2:0, 89 :8:3 :0, and 94:6:0:0. The cove forest stand types of the Cumberlandsmay then be given stands of Porter Creek and Kalanu Prong gave unityin termsof the monoclimaxtheory, very broad formulas of 98:2:0:0 and 96:4:0:0, respectively. definitionof associations,and the conceptionof asso- These cove forestsare most nearly comparablewith ciation-segregates.In the Smokies it seems prefer- one of the segregatesof the mixed mesophytic,the able to identifythe mesophyticassociation in ques- sugar maple-basswood-buckeyeforest (Acer saccha- tion withthe cove forestsand to regardstands highly rum-Tilia heterophylla-Aesculus octandra), except mixedwith oaks and Castanea dentata as transitional. for the absence of Halesia monticola fromthe latter Applied to the Cumberlands,the author's interpre- (see Braun 1950:56). In this segregatemesics may tationwould suggestthat the Mixed Mesophyticcom- formmore than 90 percentof the stand; composition prises: (1) trulymesophytic forests (sugar maple- in examples given (Braun 1950 :58) ranges from basswood-buckeye),(2) a wide range of highlymixed 93:6:1:0 to 78:15:7:0. In some examples of beech transitional stands between these and oak forest ravine communitiesalso the mesics comprise more types, and (3) various segregate types dominated by than 90 percentof the stand (Braun 1950:60). For species which are representedalso in stands of the comparisonbetween the two ranges, the cove forest transitionis best identifiedwith the all-deciduous preceding. in natureto tell us whether mixed mesophytic(and equivalentswith some Tsuga There are no signposts canadensis), the cove forests themselveswith the we should see mixtureoccurring in one directionor sugar maple-basswood-buckeyemesic segregate. segregationin the other,where in a vegetationalcon- In Braun's schemeof interpretation,the cove for- tinuumwe should locate our associationsand where est transitionshould perhaps be given primaryem- our transitions. The interpretationsof Braun and phasis and the cove forestsand oak forestsadjacent the authorare different,possible patternsof abstrac- to it regarded as segregates. Certain features of tion from the vegetation of the Southern Appa-

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 4848 R. H.H. WHITTAKERWHITTAKER VEcological~~~~~~~~~~~~~~~~~~~~~~~Vol.Monograph.1 26, No. 1 lachians. It is not asserted that the interpretation Magnolia fraseri*, and sometimesLiriodendron tu- of Braun (1950 :195-215) is in any way erroneous lipifera or Fagus grandifolia. A low-treelayer of when applied to the Smokies. For the present study, Ilex opaca*, Betula lenta, Acer pensylvanicum, and however, "cove forest," and "cove forest transition" most other small-treespecies of the submesic and have been preferred to the broader conception of ecotonal-mesicunions, is well developed in some "Mixed Mesophytic" for locally recognized, quan- stands but relativelyunimportant in othersin which titatively distinguished types. Rhododendrondominates the undergrowth. On steep slopes the shrub laver may be strongly 2. EASTERN HEMLOCK FOREST developedas a highheath. At higherelevations heath Toward higher elevations hemlock (Tsuga cana- coverage may be 60 to 80% with Rhododendron densis*) has an increasingly high population peak maximum or R. catawbiense dominant. The Tsuga- in mesic sites (Fig. 15). The heavily hemlock- Rhododendron combinationresembles the type de- dominated stands of upper slopes and ridges have scribed by Oosting & Billings (1939), but the com- been called the hemlock ridge forest and suspected bination with Vaccinium (Polycodium) was not en- of being segregated by thin soils (Cain et al. 1937). countered in the Smokies. Leucothoe editorum is frequent in these forests, Hydrangea arborescens, Kalmia latifolia, and others are occasional. Some stands on lower flats,more mixed with hardwoods, have little shrub undergrowth. The herb stratum may be nonexistentin dense stands with heavy heath undergrowth.The effectof hemlockneedles and the superficialroot systemis to produce a surface soil much more acid than that of hardwoodsstands and to reducewater penetrationto lowerlevels, drying the soil (Daubenmire1930, 1931). Althoughhemlock stands occur in mesic sites, they are relativelyxeric and acid for undergrowthplants. FIG. 15. Rise of Tsuga canadensis toward higher ele-. A dense Rhododendronheath may develop under vations, distribution in moisture-gradienttransects: a, these conditions,and the forestfloor may be a sterile 1500-2500 ft; b, 2500-3500 ft; c, 3500-4500 ft. surface of hemlock and Rhododendronleaves. In stands with less heath an herb stratumof low cover, Some broadleaf cove forests occur on equally steep age may appear. Dryopteris spinulosa v. intermedia, slopes, however, and some hemlock stands occur on Mitchella repens*, withGoodyera repens v. ophioides* valley flats. Hemlock stands appear in sites which often growingamong its stems,and Tiarella cordi- are somewhat less mesic than those of the cove for- folia are the most successfulherbs. Polypodium vir- ests themselves, whether on open valley flats at ginianum is frequent on exposed rocks and roots. middle elevations, or slopes above the valleys at Herb coverage varies from zero up, related to the high elevations. proportionof hardwoodsand development-of heath. Hemlock stands are so different from deciduous The hemlockforest type is most distinctive,with cove forests as to appear almost unrelated. The heaviest hemlock dominance,densest heath growth, stands* have the dense, dark appearance of some and most impoverishedherb stratumon steep slopes abietine forests; the herb layer may be almost absent at higherelevations. At lower elevationsstands are and the shrub layer a tangle of Rhododendron. The more mixed with hardwoods,with less heath and forests have a somber aspect, unrelieved by the ver- more herbs. Below 2500 ft the hemlockforest gradu- dant green of deciduous forests, with no colors but ally mergeswith the cove forests; and hemlockbe- the brown floor and tree trunks and dark evergreen comes only one of the. dominants,though the most tree and shrub foliage, except for flower clusters of importantone, of lower-elevationcove forests. Rhododendron shining white and purple where they may be caught by sunlight down the slopes. The 3. GRAY BEECH FOREST trees are as large as those of broadleaf cove forests; Above 4500 ft the high-elevationor "gray" popu- stems 3 and 4 ft in diameter bear the canopy crowns, lation of beech (Fagus grandifolia*) is the most im- reaching heights above 100 ft. The large stems are portant tree of mesic deciduous forests (see also more clos~ely spaced than in the cove forests, and Russell 1953). Gray beech formssmall-tree forests. small tree stems are fewer. The beech itself seldom exceeds 15 in. diameter; Hemlock may form more than 70 or 80% of the and the bulk of canopy stems may be in the 8- to canopy at higher elevations, but stands may include 12-in.classes, though cove-forest hardwoods occurring most other mesics and some other tree species. Be- with it may reach fair size. In two quadrat studies cause of the large size of hemlock, stands in which it canopy stems (8-10 inches) were only 25 to 40 comprises only 20 or 30%O of the stems may be domi- ft high, with crowns12 to 20 ft from the ground. nated by it. The most important associates of hem- Small stems may be numerous,but frequentlythe lock are ilalesia m~onticola, B etula allegheniensis, undergrowthis slight,with few seedlingsand few or

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 49 no shrubs above a floor of sedges, ferns, and other tered plants of most of the herb species of north herbs. The small, low trees with rounded crowns slopes and upper cove forests except Cimicifuga. and the grass-like floor of sedges, open and free of With the mesics occur a few submesics, notably woody undergrowth, led the natives to call these Medeola virginiana, and members of the submesic woods "beech orchards," a fair description of them. high-elevation union, including Prenanthes altissima', Trunks of the trees are mottled gray from the lichens Arisaema quinatum', Angelica triquinata*, and Soli- which cover them and are so lichen-coated that barks dago sp. The abundant fern,as on northslopes, is as distinctive as those of beech and buckeye cannot be Athyrium filix-femina v. asplenioidesK. The south- distinguished. Compared with the superb cove for- slope beech stands are in many respectsintermediate ests, the trees have a stunted appearance; and other to north-slopebeech-mixed stands and high-elevation species share with beech the "orchard" growth-form. red oak-chestnutforests. In form, flora, and single-species dominance the beech 4. RED OAK-PIGNUT HICKORY FOREST woods suggest a vegetation of extreme conditions, a deciduous type in a limiting and stunting subalpine At lower elevations,usually below 2500-3000 ft, environment. oak-hickoryforests occupy the next position toward xeric sites beyond cove forest transition. In many The gray beech forests of north and south slopes places theyare poorlydeveloped or absent as a type, are so differentthey may best be described separately. theirplace being taken by forestsof mixed oaks and Stands of north-facing slopes, occurring on north and chestnutwith some hickories. In their best develop- northeast sides of gaps within the range of spruce mentoak-hickory forests are dominatedby combina- and on many open slopes of the same orientations in tions of northern red oak (Quercus borealis v. maxi- the southwest half of the mountains, are a continua- ma*), pignut hickory (Carya glabra*), white oak tion and modification of the upper cove forests. Al- (Q. alba*), and mockernuthickory (C. tomentosa*). though beech is most numerous in stems, two trees of Stands usually include Q. velutinaand C. ovalis*, Q. the upper cove forests, Betula allegheniensis (or B. prinus, Castanea dentata, Nyssa sylvatica, and Lirio- lutea) and Aesculus octandra are equally important dendron tulipifera. Acer rubrum, Cornus florida, in the canopy and dominate some stands. The low- and Oxydendrumarboreum are importantsmall trees. tree stratum usually includes numerous small stems The lower strata of oak-hickoryforests in the of Acer spicatum, A. pensylvanicum, and Ame- Smokies may be much like those of oak-chestnut lanchier laevis. Viburnum alnifolium, Cornus alterni- forests. Where oak-hickorystands occur on lime- folia, Hydrangea arborescens and, infrequently,Ribes stone soils on the northwestedge of the range, they rotundifolium occur in the shrub stratum. Shrub cov- are quite unlike those of the mountainsthemselves. erage is quite low, usually 2 to 10%. The herb Juglans nigra, Carya ovata, and Cercis canadensis stratum is essentially like that of upper cove forests, occur, and on limestone outcrops Juniperus vir- but with sedges nearly always present in small cov- giniana, Rhamnus caroliniana, and Viburnum rufidu- erage and with greater representation of herbs of lum, with conspicuouslydifferent lower strata of a the high-elevation mesic union-Senecio Rud- rugelia, middle-westernaspect. On acid soils of the moun- beckia laciniata, Streptopus roseus, and others. Herb tains, however,shrubs of the oak-chestnutforest ap- coverage is 40 to 60%, of the same order as that in pear, with variable coverage (5 to 50%). The eri- cove forests. The north-slope beech forests are in all cads Gaylussacia ursina and Kalmia latifolia are strata a distinctly mesic type, closely related to usually most important,with Rhododendronmaxi- upper cove forests and to be distinguished as a type mum, R. calendulaceum, Pyrularia pubera, and most only by larger numbers of beech stems, limitation other shrubs of the submesicunion sometimespres- of tree size, and more subalpine cast of the flora. ent. Most herb species are shared with oak-chestnut When one crosses a gap from the north to the forests: Polystichum acrostichoides, Solidago sp., south side, one enters quite different beech forests, Aureolaria laevigata, and Galax aphylla are most transitions to which occur in the intermediate south- frequent; Viola hastata, Chimaphila maculata, Des- east and west exposures. The south-facing, sedge- modium nudiflorum, Goodyera pubescens, Prenanthes floored beech woods are among the most limited forest trifoliolata, Smilacina racemosa, and Uvularia pudica types in the Smokies, largely restricted to gaps and all occur,along with a few mesics (Cimicifugarace- concave slopes within the range of spruce. These mosa, Aster divaricatus, Actaea pachypoda). Herb stands differ significantly in canopy composition coverageis generallylow, in the range of 1 to 10%. from north-slope ones-in smaller number of species, greater preponderance of beech, and occurrence of 5. CHESTNUT OAK-CHESTNUT FOREST Halesia monticola as second most important species. The most extensiveforest-s of middle and lower The ecotonal small trees are absent or unimportant; elevations in the Smokies are those dominated by the shrub stratum is insignificantor absent, with only chestnut oak (Quercus prinus) and chestnut (Cas- Hydrangea arborescens appearing occasionally. It is tanea dentata). Oak-chestnutheaths and forestsoc- in the herb stratum that the south-slope beech forests cur on almost all slopes except those facing south are most distinctive, however. Sedges (Carex aesti- rnd southwest,and are markedunmistakably now by ralis* and sp.) dominate the herb stratum with cov- the gray striae of dead chestnuttrunks and branches. erage up to 80 and 90%. With the sedges occur scat- From accounts of old-residentsand from trends in

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions Ecological 5050 R. H.H. WHITTAKERWHITTAKER ~~~~~~~~~~~~~~~~~~~~~~~~Vol.Monographs26, No. 1 secondary forests, it seems that forests of great less of their structure. Not all that is dominated by chestnut trees once covered most of the lower coves oak and chestnut is forest, however; and no one who and broad open valleys away from stream-banks. The has mapped quadrats or traveled cross-country slope oak-chestnut stands which have survived settle- through oak-chestnut heath and forest would wish ment and the blight are not impressive forests. Only to overlook the distinction. The two types are dis- a few of the surviving trees, mostly chestnut oak. tinguished by Conard (1935) in his study on Long exceed 2 ft in diameter, although chestnuts with Island as the Quercetum kalmietosum and Quer- diameters of 4-5 ft and more are still standing in cetumrprini. Submesic stands in the Smokies are places (1947). The death of the chestnuts, by open- unquestionably forests, with closed canopies; but in ing the stands, permitted heavy reproduction of other the oak-chestnut heath total tree coverage, including trees formerly limited by competition and shading, small trees, may fall below 40 or 50%. Large chest- and has to date strongly favored small-tree species nut oaks and chestnuts are widely scattered, and (cf. Keever 1953). The forests now have dense under the open tree stratum the shrub layer is closed; stands of small stems of Acer rubrum, Oxydendrum the mountain slopes are covered by a continuous ever- arboreum, Cornus florida, and others. One who green heath of Kalmia and other species. Travel crosses a flat or slope in the oak-chestnut forest finds through the oak-chestnut heath is difficult;effort and his vision limited and progress retarded by the heavy the force of either exertion or gravity are necessary undergrowth of small trees and shrubs and the spines to push one's way through the heath, and only the of the catbriers (Smilax rotundifolia), particularly pressure of the shrubs supports one and prevents abundant here. frequent tripping among the dense stems. Chestnut was formerlydominant in the forests, with Chestnut and chestnut oak make up 80% or more probably 30 to 70% of canopy stems. Chestnut oak of the "canopy" of tall trees. The chestnuts are was the second dominant, and with chestnut dead is now dead; and the coverage of large chestnut oaks, now the first. There is no other very important can- while difficultto measure, is probably no more than opy species; these two made up 75%

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 51 heaths Gaultheria procumbens and Epigaea repens, becomes strongly dominant on soiie exposed south- Panicum sp., Campanula divaricata&', and Aureolaria west ridges, with a rather open growth of small trees laevigata are fairly frequent. Herb coverage is quite and a fairly rich herb layer with grasses often con- low, 1 to 5%, with herbs other than Galax and the spicuous. In structure and appearance these forests ground heaths well below 1%. suggest the white oak woods of hills far west of the Smokies, though very differentin their floristics. In 7. RED OAK-CHESTNUT FOREST some stands chestnut (Castanea dentata) is dominant The red oak-chestnut combination is submesic in (cf. Cain 1931). Qtercits borealis is almost always composition of both tree canopy and lower strata, present; and Carya glabra and Quercus prinus oc- and is the high-altitude equivalent of chestnut oak- cur in the canopy at lower elevations. Of the small chestnut forest. Northern red oak (Quercus bore- trees Acer rubrum is most important; Oxydendrurni alis*) and chestnut (Castanea dentata) form 70 or arboreulm and Robinia pseutdoacacia are fairly nu- 80% of the canopy, with the former usually more im- merous toward lower elevations. portant. Although most of the chestnuts are dead, The shrub layer is predominantly ericaceous, with many large chestnut trees in these and other high- Rhododendron calendulaceum and Vaccinium con- elevation forests had at least a few branches living in stablaei most important. Kalmia latifolia is frequent 1947. No canopy associate reaches any substantial at lower elevations and is often the most important percentage in most stands; of other species Acer ru- shrub species there. Shrub coverage is fairly high- brum is most important. The lower-elevation stands generally 30 to 60% below and 20 to 50% above may have a fair percentage of mesics-Tilia hetero- -1500 ft. The herb stratum is a mixture of ferns phylla, Prunus serotina, Betula allegheniensis, Aes- (Dryopteris noleboracensis, Athyrium fl'x-fernina v. culus octandra, Fraxinus americana-, but few mesic (asplenioides) with other herbs of the submesic. sub- trees except Fagus grandifolia enter the canopy of xeric, and high-elevation submesic unions. Grasses, stands above 4500 ft. The submesic small trees, with Pteridium aquilinum v. latiusculum, and other xeric Halesia monticola joining them here, form 30 to 50% species also are prominent. Herb coverage is 10 to of the sterns,having taken advantage of the death of 50% below and 20 to 50% above 4500 ft. the chestnuts along with northern red oak. The chestnuts were less numerous in submesic stands of higher Pine Stands and Their Maintenance elevations, however, and reproduction and appearance- Most steep, open south- and southwest-facing have not been altered as much by their death as in slopes on the Tennessee side of the mountains are the chestnut oak-chestnut forests. occupied by pine forest and pine heath. Usually the Shrub coverage above 4500 ft is low, with Rhodo- population of a single pine stand shows a bimodal dendron calendulaceum and Vaccinium constablaei size distribution; often other modes are represented the principal species. Vaccinium pallidum and V. by a few old trees of earlier generations (Fig. 16). hirsutum occur in some stands, especially on ridges, In these dry sites occasional severe fires destroy the and Vaccinium erythrocarpum and Viburnum alni- forests, and following the fires dense stands of voung folium are occasional. In forests below 4500 ft pines develop. As these trees mature, age, and die, Rhododendron calendulaceum is most important; but the canopy is opened; and a new generation of pines most other submesic shrubs appear in some stands, establish themselves. When these in turn mature, a and Hydrangea arborescens is occasional. Kalmia third generation of seedlings appears, while some latifolia is almost absent from submesic stands above scattered stems of the first are still present. The 3800 ft. Shrub coverage varies from 20 to 50% below effects of fires in permitting the establishment of 4500 ft to only 0 to 20% above. initial single-age stands thus introduces into the The herb layer above 4500 ft often resembles closely maintenance of the pine stands a rhythm of genera- that of south-slope beech stands, but with lower tions which may persist for some time. The conmpos- counts from mature stands smooth out the coverage. Sedges (Carex spp.) and ferns (mostly ite stand composition of individual stands, and the Athyrium filix-feminav. asplenioides) are abundant. double-age form A few mesic species, notably Eupatorim rugosum, composite curves then take on the characteristic The dense Laportea canadensis, and Aster divaricatus, appear of self-maintaining forests (Fig. 6). along with the submesic and high-elevation submesic small-tree stands on disturbed and recently burned unions and Potentilla canadensis v. caroliniana. Be- sites were not included in these counts; but repro- to the stands in- low 4500 ft the herb flora is much the same as in duction appears adequate replace chestnut oak-chestnut forests, with onlv the subalpine definitely. chance may the of pine species Trillium undulatum, Solidago glomerata, and Since determine species a a fire, chance, Diphylleia cymosa distinguishing the list from that of seeds available to restock stand after as Cain et al. (1937) noted, may determine which of the lower type. Herb coverage is 10 to 40% in the elevation first doini- lower stands, 20 to 60% above 4500 ft. the pines available at a given nates the stand. Once dominant, the species tends to 8. WHITE OAK-CHESTNUT FOREST maintain its dominance through several generations Below 4500 ft white-oak chestnut forests are sel- while other pines and oaks may gradually increase to dom clearly distinguishable from red oak-chestnut. approach a balanced stand for that elevation and Above 4500 ft, however, white oak (Quercus alba") site. Most immature pine stands in the Smokies are

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 5252 R. H.H. WHITTAKERWHITTAKER Ecological~~~~~~~~~~~~~~~~~~~~~~~Vol. Monographs26, No. 1 in climax stands of south-facing slopes and in the many seral stands of old fields on other sites. Pinus rigida ordinarily occurs in the mature climax stands. Quercus coccinea is most important, next to the pines. Quercus prinus and Castanea dentata are usually present; and Pinus strobus, Nyssa sylvatica, Quercus velutina, and Q. alba occur and may enter the canopy. Oxydendrum arboreum, Acer rubrum, and Quercus marilandica* are the only important small-tree species. The trees of these forests are not large; stems over 18 in. diameter are rare, and the dominants in most stands are only 12 to 15 in. Canopy heights in a sample were 50 to 75 ft. The forest is open, with tree coverage about 70% and much light admitted through the open canopy to lower layers. With their southwest exposures, open structure, and partly bare soil, these forests have in FIG. 16. Stand chart for Pinus plungens in a pine heath at 4150 ft on GreenbrierPinnacle, showing peri- the afternoon a bright, hot, and arid climate which odic reproductionof stand. is in palpable contrast with that of forests in nearby valleys. single-species stands; most mature ones are mixed. Shrub coverage is variable with 10 to 40% the In the mature stands the oaks, also, are reproduc- usual range, but pine-heath stands with shrub cov- ing well; but it is by no means certain that the oaks erage over 50% occur even at low elevations. The andi their associates are capable of taking dominance shrub stratum is strongly dominated by ericads. from the pines on these sites. It seems likely that the Kalmia latifolia is the usual shrub dominant, but nature pine stands are in balance, the pines, with Vaccinium vacillans is almost always present and in a small population of oaks, being all that the dry places is dominant. Vaccinium stamineum is fre- sites will support. The near-absence of oak-chestnut quent, and Gaylussacia baccata is occasional above forests from open south and southwest slopes below 2000 ft. The only non-ericads are Ilex montana y. 4500 ft on the Tennessee side of the Smokies is one beadlei and the catbriers, Smilax glauca and S. ro- indication that such may be the case. Only on shel- tundifolia. tered slopes and those of small inclination are oaks The herb layer is low in coverage, with 2 to 10% and chestnuts dominant, or apparently replacing the the usual range. Herb species include Andropogon pines as dorninants. Even in many of the' stands in scoparius', A. virginicus, Panicum sp., Pteridium which oaks share dominance with pines, the latter aquilinum v. latiusculum, Tephrosia virginiana*, Bap- are reproducing well. But whether or not the oaks tisia tinctoria*, Coreopsis major, Sericocarpus aste- might ultimately succeed pines in most sites, the time roides* and S. Aster Solidago sp., necessary to effectthe change exceeds the normal ex- linifolius*, sp., Antennaria sp., Epigaea repens, and Galax aphylla. pectation of fire on dry mountain slopes in an area Of these Andropogon scoparius leads in coverage. of summer thunderstorms. The fires which have swept the drier slopes of the 10. PITCH PINE HEATH mountains have certainly favored the pines in some Between elevations of 2200 and 3200 ft pitch pine sites where pines form stands seral to oak-chestnut. (Pinus rigida*) most frequentlydominates pine stands. The same fires that may favor pines in most xeric Quercus coccinea* sometimes shares dominance with sites, however, favor the shrubs of oak-chestnut heath the pine and is almost always present along with in other sites; fire alone is not an explanation of the Quercus prinus, Castanea dentata, and one or both pine stands. Fires are part of the environmental of the other pine Nyssa sylvatica occurs in complexes in which the pine stands and oak-chestnut siecies. small numbers, and Acer rubrum and Oxydendrum heaths exist, and the stands are scarcely the same as arboreum are important small trees. In tree coverage those hypothetical stands which might exist in their and stature the pitch pine stands differ little from sites in the absence of fire. Both types are, how- the Virginia pine forests; but shrub coverage is ever, self-maintaining in their xeric and subxeric higher at middle elevations. Shrub coverage is 40 to sites under present conditions; they are not seral. 70% and is in most cases over 50%; the type is con- Pine stands have often been relegated to subclimax sequently termed a pine heath. Kalmia latifolia and status in the eastern forests, but there are good rea- Vaccinium vacillans or V. hirsutum* share domi- sons for accepting the Appalachian pine stands as nance in different proportions in different stands; topographic or edaphic climaxes of their dry and ex- and V. stamineum, Gaylussacia baccata, and Lyonia posed sites (cf. Braun 1935b, Cain et al. 1937). ligustrina are also important shrubs. Species of the 9. VIRGINIA PINE FOREST xeric union (Andropogon scoparius, Pteridium aquili- In low-elevation pine stands Virginia or scrub num v. latisculum, Epigaea repens, Gaultheria pro- pine (Pinug lirginiana*) is usually dominant, both cumbens) dominate the herb stratum; but legumes

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 53 and composites,except Coreopsis major, are less rious herbsare scatteredthrough the grass,Potentilla commonthan in Virginia pine forests. Herb cover- canadensis v. caroliniana* being muchthe most wide- age is usually 5 to 20%. spread and abundant. Stachys clingmanii*, Solidago sp., Prenanthes altissima, Smilax herbacea, and 11. TABLE MOUNTAIN PINE HEATH Houstonia serpyllifolia, which occur in the high-ele- Toward higher elevations table mountain pine vation deciduous forests, are frequent; and some (Pinus pungens*) is moreand morefrequently domi- mesic herb species (Rudbeckia laciniata, Gentiana the pine heaths it is nant; at the upper limit of decora, and Stellaria pubera) occur here. With these stronglydominant, and other trees occur only in are many introducedweeds-the grasses mentioned, small percentages. The stands are small, low, and Hieracium scabrum, Rumex acetosella, Prunella vul- in open. Canopy stemsare usually only 10 to 15 in. garis, etc. in was only diameter,canopy tree height a sample Althoughthe balds are primarilyherb-layer com- is usually 70 or 80%. 40 to 50 ft, and tree coverage munities,shrubs occur at low coverage over most of Pinus rigida is the second most importantspecies, their surfaces. Rhododendron calendulaceum, Vac- often absent at higher elevations. Quercus but is cinium constablaei, and Rubus canadensis are the coccinea, Q. prinus, Castanea dentata, and Nyssa most importantspecies. Tree seedlings are almost in numbers,with Acer sylvatica usually occur small always present (see Part I, table 4). The scattered Robinia pseudo- rubrum, Oxydendrum arboreum, seedlingsdo not necessarilyimply that the balds are Sassafras albidum as small trees. acacia, and becomingforest; theirsuccessional status will be dis- is dominatedby Kalmia latifolia The shrubstratum cussed below. or by Vaccinioideae-Vaccinium pallidum, V. hirsu- tumr,or Gaylussacia baccata. Vaccinium stamineum The Southern Appalachian Subalpine Forest and Lyonia ligustrina are usually present,and Vac- Center cinium constablaei and Rhododendron catawbiense In the northeasternhalf of the Great Smoky occur in some stands. Pieris floribunda* occurs to- Mountains,forests of spruce and fircover nearly all ward higherelevations and is occasionallydominant. slopes and valleysdown to 4500 ft. In structureand Shrub coverageis high,60 to 90%; and most stands compositionthese forestsresemble those of Canada; have an open pine canopy above a dense,low heath. but they are floristicallyquite different.Of the Herb coverage is usually 5 to 20%. Andropogon dominants,Abies fraseri is a southernendemic, while scoparius is important;but Galax aphylla, Epigaea Picea rubens, thoughit ranges far to the north.is repens* and Gaultheria procumbens* lead in cover- primarily Appalachian rather than Canadian and age, and species of the heath bald herb union occur. occurs mainly in the mountains above 1000 ft 12. GRASSY BALD throughoutits range (Harlow & Harrar 1941). The dis- The mostlimited vegetation type of the Smokiesis southernpopulation of Picea rubens has been A one of the mostdistinctive and celebrated.In an area tinguishedas P. australis Small. large proportion where forests continuouslyclothe the valleys and of the plants are southern species or distinctive ridges and most of the peaks, a few of the latter southernpopulations; in addition to Abies fraseri ery- bear patchesof grasslandon theirsummits. In reach- these include the importantshrubs Vaccinium ing the balds one may climb3000 or 4000 ft through throcarpum and Menziesia pilosa and the herbs So- forests,to emergeat last in an open meadow with lidago glomerata, Senecio rugelia, Geum radiatum full view of the mountainsin all directions. The Michx. Parnassia asarifolia, and Chelone lyoni. are change from forest to bald is abrupt; except that Others (Viburnum alnifolium, Clintonia borealis) tree stature may decrease up the slopes and along species of the easternUnited States and Canada, while the ridges toward the balds. The trees do not thin a few (Oxalis montana, Sambucus pubens, Sorbus out gradually into the bald, however,but end in a americana) are widespreadboreal species whose South- been distin- forest-edgewhere ericads (Vaccinium constablaei, ern Appalachian populations have not Rhododendron calendulaceum, Lyonia ligustrina, guished. Cain et al. (1937) consideredthat 25% of extendinto Vaccinium pallidum and V. hirsutum) are mixedwith the woodyspecies are endemic,while 60% other shrubs and small trees (Betula lenta, Ame- the Northeast,although many of these are primarily lanchier laevis, Crataegus macrosperma v. roanensis, Appalachian. Salix sp., Rubus canadensis, and dwarfed trees of It consequentlyseems inappropriate to refer to the surroundingbeech and oak-chestnutforests). these forestsas eitherboreal or Canadian. The term Beyond the forest-edgethe balds are meadows boreal is used in this paper, but only in the loose wheremountain oat grass (Danthonia compressa*) is sense in whichit may be applied to coniferousforests strongly dominant. Grasses introduced when the or taiga of both high latitudes and high altitudes. balds were used for pasture (Agrostis alba and Although they gradate into the Canadian forests Phleum pratense) may also occur. Sedges have been through the Northeast,the Southern Appalachian reportedas dominantin balds in otherareas (Wells spruce-firforests seem distinct. Far from being 1937) in contactwith spruce-firforests, and are im- merelya southernfragment derived from the Ca- portant in Andrews Bald, in contact with both de- nadian forests,they may well be a centerfrom which ciduous and spruce-firforests, in the Smokies. Va- somespecies have spread into the latter. Theyshould,

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 54 R. H. WHITTAKER Ecological Monographs in any case, be regarded as an additional cordilleran high herb and high shrub strata are of moderate member of the boreal and subalpine forests of North coverage (around 30 and 20%), and various species America, with status equivalent to those of western of the mesicand high-elevationmesic unions occur in ranges (Whittaker 1948, Oosting & Billings 1951). the former. The subtype of north slopes and flats The Southern Appalachian subalpine forests are of has high coveragesof moss and Oxalis (25-55%) and small extent-originally perhaps 1,000,000 acres, now moderate ones (5-20%) for low and high shrubs. much reduced (Korstian 1937)-but admirable vir- The high herb stratumis stronglydominated by a gin stands are preserved in the Smokies. fern (Dryopteris spinulosa v. americana*); other herbs are Clintonia borealis*, Aster acuminatics, Se- 13. RED SPRUCE FOREST necio rugelia*, Trillium erectum*, and Streptopus In their lower elevations the subalpine forests of roseus*. These stands show a characteristiclavering the Smokies are dominated by red spruce (Picea (see Part I), with a five-storystructure of ground rubens*). Fraser fir occurs in most stands except moss (Hylocomium splendens), low herb laver (Ox- those of dry lower slopes, but is a much smaller spe- alis montana), fern (Dryopteris spinulosa v. ameri- The cies than spruce and scarcely enters the canopy. cana), low shrub (Vaccinium erythrocarpum') and only important canopy associate of spruce is high- high shrub (Viburnum alnifolium). Acer spicathm elevation yellow birch (Betula lutea), with 10 to formsa low tree stratum. birch the small 15% of canopy stems. Along with Coverages for all these strata are less in the in- Ame- trees Acer spicaturm, A. pensylvanicum, and termediatesubtype of east- and west-facingslopes; lanchier laeris are important, increasingly so toward moss, Oxalis, and fern coveragesin this are usually sites. Aesculits oc- lower elevations and more mesic between15 and 30%, low and high shrub coverages monticola, and t~indra, Fagus grandifolia, Halesia between 5 and 10%. In the south-slopesubtype, occur in some spruce other cove forest species also coveragesfor all five strata are usually below 10% of lower elevations and more mesic sites. forests (Oxalis may be higher). Species occurringon south 18 30 in. Canopy trees are of only medium size, to slopes, which are rare in more mesic subtypes,in- diameter for the most part, and of medium height, clude Medeola virginiana, Lycopodium lucidulum, 50 to 80 ft in two samples. and Monotropa uniflora, Vaccinium constablaei, Ilex Spruce and fir forests of the Smokies have been montana, and Rhododendron catawbiense. In the Cain (1935, Cain et al. 1937) and described also by fifthsubtype of ridges and steep upper south slopes, con- Oosting & Billings (1951), and these studies the Rhododendronheath approaches full coverage. tain useful data for stands representing only part of Under the heath, coverages for species of the five- the spruce-fir pattern. The more extensive coverage storyundergrowth approach zero, and thereare few of the present work (51 samples) is intended to pro- herbs (Trillium undulatum, Gaultheria procumbens, vide limited information on the striking variations Epigaea repens). Floristicallythe undergrowthsof in with site in these forests. Trends forest composi- such stands are related to the heath balds. On some de- tion along environmental gradients have been ridgesthe canopy of spruceis open, forminga phvsi- I. is relatively scri'bed in Part Canopy composition ognomicallydistinctive spruce-heath transitional to of lower eleva- uniform, compared with the forests the heath balds. In most cases, however,the bound- in structure and tions. but undergrowths vary greatly ary betweenspruce forest and heathbald is relatively floristic composition. Five "subtypes" will be dis- abrupt. tinguished in the undergrowth as it changes continuously along the moisture gradient. U~ndergrowth 14. FRASER FIR FOREST variations are comparable to those which occur in Most subalpineforests of the Smokiesare mixtures other boreal and subalpine forests and have been of red spruce with southernbalsam or Fraser fir recognized in the site-types of Finnish and other (Abies fraseri*); but above 6200 ft (1890 m), on forest ecologists (Cajander 1909, 1926, 1949; Linkola the upper slopes of the highest peaks, forests are 1924; Ilvessalo 1929; Heimburger 1934; Arnborg dominatedby fir alone. Tree compositionof these 1940, 1953), sociations of Scandinavian phytosociolo- forestsof extremeelevations is the simplestin the gists (Du Rietz 1932), and associations of Russian Smokies; thereis one dominant,the fir,and one de- phytocenologists (Katz 1930, 1933; Sukatschew 1928, ciduous tree, mountain ash (Sorbus americana*). 1932; Sokolowa 1935, 1936). The "subtypes" of the These are the only tree species at the extremede- author differ in manner of definition from these, velopmentof the type; but many stands have red however, and do not include all the sociations which spruce, and some have yellow birch and other de- might be recognized (see also Cain 1935). ciduous species in small numbers. Although undergrowth is variable in any topo- Undergrowthvaries as strikinglyas in the spruce graphically defined type of site, average under- forests. Valley sites were not observedat these ele- growth coverage and composition are correlated with vations. The subtypeof north-facingslopes has very site in the series of topographic subtypes: valleys, high coveragesof moss and Oxalis (around 90 and north slopes, intermediate slopes, south slopes, and 50%). The Dryopterisstratum is of high coverage ridges. In the valley subtype, coverages of moss, (40%) ; and other herbs include Aster acuminatus, Oxalis, and low shrubs are generally below 5%. The Clintonia borealis, Solidago glomerata, Senecio ru-

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 55 gelia, Chelone lyoni,and Geum radiatum. In the in- 15. HEATH BALT-D termediatesubtype moss coverage is high (80-90%); The heath balds are one of the most attractive but coveragesof the otherstrata are low, usually less vegetationfeatures of the Smokies, becomingwhen than 10%. Undergrowthof mesticand intermediate Rhododendronsand Kalmia are in bloom the cele- forestsof highestelevations consists largely of the bratedfloral epaulets on the shouldersof thesemoun- lowest strata of the flve-storyundergrowth-moss, tains. In the northeasternpart of the range, in the Oxalis, and fern. Shrubs may be almost absent from area of subalpine forests,many of the ridges and these stands, although Viburntimalnifolium, Sam- steep faces bear theseshrub communities, which were bucus ptibens,Vaccinitim erythrocarpum, Menziesia studied by Cain (1930b). They are dominatedby pilosa, and Diervilla sessilifoliaoccur. evergreenericaceous shrubs which approach not only In the subtypeof south facing slopes, moss cover- full coverage but complete impenetrabilityin their age is 60% or less, and heath coverage is usually dense thicketsof tough stems. The shrub canopy substantial(40-60%). In the ridgesubtype the, heath may be 10 ft high, but on ridges and summitsis (Rhododendroncatawbiense and R. carolinianum) more frequently3 or 4 ft. On sharp crestsof ridges approaches full coverage; and other undergrowthis or peaks the heathmay be open, withscattered cush- limitedto moss coverageof a few percentand scat- ions of Leiophyllumlyoni. teredherbs of the heathbald herb union. The balds occur throughoutthe elevationsof the In appearance the fir forestsare like no others; subalpine forestsand down to 4000 ft and show con- they are even unlike the spruce forests. The trees siderable differencein composition with altitude. are quite small-the bulk of canopy stemnsmay be Kalmia latifolia is most importantat lower eleva- 7, 8 and 9 in. in diameter with correspondingly Rhododendronmaximum low height(30 to 40 ft). Stemsas large as 12 in. are tions, with and R. catawv- rare. Many of the forestsare stagnantpole stands, biense, Aronia melanocarpa', Clethra acuminata, withseedlings and smallerstems numerous, but mostly Vacciniumconstablaei, Hex montana,Viburnum cas- dead. Undergrowthmay be sparse; and the forests, sinoides', and Smilax rotundifolia,and with Gay- withtheir small and ratheropen canopy,have an im- lus.saciabaccata in more open parts. Rhododendron poverished and stunted appearance. Mosses and maximumis absent above the lower balds, Kalmia lichens,however, flourish. They may carpet the floor; and mostof the otherspecies listed are absent above and trunksand branchesof trees and fallen poles 6000 ft. The high-elevationbalds are somewhatlower are coated,crusted, and tuftedwith them, giving the in canopy heightand moreopen. Rhododendronca- forestsa gray-green,hoary aspect. From this, and tawbienese and R. carolinianum'are dominant,with from the small trees and dead seedlings,the forest Vaccinium constablaei. The low shrubs Vaccinium appears appropriate to the subalpine extremityin erythrocarpum,Menziesia pilosa, and Leiophyllum the Appalachians, a dwarf coniferforest which on lyoni' occur in the more open parts; and Sorbus a fair day has a charmof its own but at othertimes, americana appears along the edges. The balds may with the summitsinvolved in cloud and a raw wind well be dividedinto two intergradingtypes, the lower whippingfog throughthe firs,assumes the bleakness dominated by Kalmia latifolia and Rhododendron of the climatein whichit lives. maximumand related to the oak-chestnutheath, the

FiGt. 17. Fraser firrforest, near summnitof Mit. Le~onte. Reproducedby permissionof Thompsons,Inc., Knoxville, TennI.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions EcologicalMonographs 5656 R. H.H. WHITTAKERWHITTAKER ~~~~~~~~~~~~~~~~~~~~~~~Vol.26, No.1I upper dominatedby heaths of the subalpine forests and relatedto them. The herb stratumof the balds is limited,with coverage below 5% and in denser stands approaching zero. Gaultheriaprocumbens is the leading species, with Galax aphylla, Melampyrumlinearel, Medeola virginiana,and Trillium undulatum*. A few other species occur along trails. Tree seedlings are com- monly present-Abies fraseri, Picea rubens, Sorbus americana,and Prunus pensylvanicaat higherelevations, Betula lenta, Acer rubrum,Castanea dentata, Oxydendrumarboreum, Sassafras albidum, Pinus pungens,Robinia pseudoacacia, and Hamamelis virginiana at lower. As Cain indicates,there is no reason to think that the deciduous trees could take dominancefrom the heath,though the conifersmay 41 be successfullyinvading some balds. The Balk as Topographic Climaxes? Much discussionhas centeredon the various bald types of the SouthernAppalachians and the reasonrs for theirexistence. In relationto the climaticclimax theoryit has been temptingto apply logic to them: Forests are climax in the region; the balds are not forests; thereforethey are not climaxes. Some evi- denee may be consideredon whetherthey are necessarily seral types,or whetherthe balds may cap the climaxforest patterns of the Appalachians withnon- forestclimax types. The problemof the heathbalds in the Smokies has been consideredby Cain (1930b). In general interpretation Cain consideredthem secondary replace- mentsof forests,in whichthe treeswere destroyedby fire,windfall, or landslide, permittingthe heath already beneath them in these sites to close in and largelyexclude tree reproductionby densityof cover and effectson soil. Since theyare fairlystable shrub communities,the balds were interpretedas, in this sense, postclimax. Dead spruces and firscan be seen Fia. 18. Heath bald and succession,on a spurof Mt. in some balds, suggestingin some cases retrogression Le Conte. Reproducedby permissionof Thompsons,Inc., fromforest to heath. Knoxville,Tenn. It seems likely that some of the balds represent primarysuccession. The high balds of Mt. Le Conte invading the edges of some of the balds. In most and its spurs have bare rocks along the ridges,with of the balds, however,such invasion is proceeding lichens, mosses, and tufts of grass invading them. very slowlyat best; and in many of the balds there Scatteredcushions of Leiophyliumlyoni surround the is no convincingevidence that the conifersare, or bare area. Farther fromthe crest of the ridge these are capable of, occupyingthe site. have closed in; still fartherthey are over-toppedby The balds are thus probablysuccessional, in part. the higher heath species, which form a continuous Some representprimary successions and others re- shrubcanopy and excludeLeiophyllum from the dense sult fromdestruction of the forestcanopy; some of heath of the slopes. The soil grades from absent these will probablydevelop, or will return,to forest. or a thin crust on the ridges to a thick peat below. Many of the balds, however,are relatively stable It seems unlikelythat a soil whichonce supporteda communities;they seem clearlyto be self-maintaining spruce-heathwas secondarily lost from under the stands under present conditions,and thereforecli- shrubs. The soil and vegetation structureof the max in this sense. In relationto Clementsiantheory balds seem better explained by assuming that pri- they may be consideredlong-arrested seral types- marysuccession is still in progresson some of these i.e. serclimaxes. The author would preferto regard high ridges. Once the heath is established,the highly themas topographicclimaxes in accordancewith the acid, podsolizedsoil describedby Cain and the dense polyclimaxtheory or, in the pattern interpretation shrubcanopy should retardforest invasion and tend of this paper, simply as parts of a complex climax to maintainthe heath. Spruce and firappear to be pattern.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 57 The heath balds are characteristicallycommunities tion to adjacent forestssuggest natural causes. They of exposed summitsand ridges,especially ridges with are suggestiveof the balds and false timberlineson steeperslopes and thosewith axes sloping towardthe isolated ranges of the West (Rydberg1913, Dauben- southor west. Althoughheath balds occur on slopes mire& Slipp 1943, Merkle1951), wherethe tree line of all orientationsbelow peaks and ridges, their is depressedbelow its normal elevationfor the lati- greatestextent is on slopes of south and west expo- tude. Grassy mountainsummits above a pattern of sure. This patternof occurrencesuggests that effects deciduous (beech) and pine forestsare reportedalso of exposureare critical. Precipitationis high at the fromthe Crimea (Poplawskaja 1933). elevations of the balds; but on these summitsand All grassy balds of the Smokiesare at fairlyhigh ridges much of the water from summer thunder- elevations,on peaks or points,in contactwith high- showersis lost by surface run-offand much of the elevationdeciduous forests. Usually the balds cap the rest by sub-surfacedrainage down the slopes. The peaks; but, in six of themwhich were paced offand balds are fullyexposed to dryingeffects of winds. It mapped, the longer axis or greaterextent down the seems likely that these edaphic and climatic conse- slope was always toward the south, southwest,or quencesof theirexposed positionsare responsiblefor west. This orientation,together with the fact that the the existenceof the heath balds. most stunted and open oak-chestnutforests occur A number of naturalists-Harshberger (1903), on southwestridges below some of themand the fact Davis (1930), Camp (1931, 1936), Fink (1931), thatin the clearedfields tree invasion is moreevident Cain (1931), Wells (1936a, 1936b, 1937, 1946), and along northand east edges and in draws than along Brown (1941, 1953)-have reported on the grassy south and west edges, suggest that climaticfactors balds and speculated on their history. Clements connectedwith south and west exposure are critical. (1936) held the unshared opinion that they were Even more than the heath balds these communities producedand maintainedby fire; Gates (1941) sug- of most exposed summitsare affectedby water loss gestedkilling of oaks by gall-fliesat high elevations throughrun-off and exposure to the full force of as their cause; Wells considered them man-made, winds,with no obstructionto the west. It seemslikely cleared as camp-sitesand lookoutsby the Cherokee that resulting conditions of unfavorable moisture- Indians. Otherauthors have consideredthem natural; balance have excluded trees from the grassy balds. Camp suggested that exposure to drying westerly To this may be added the fact that the deciduous winds during the late-summerdry season was criti- forestsinvolved are high-elevationmembers of their cal. Clearing by Indians has been doubtedby Fink group, of reduced stature,and perhaps well beyond because of the labor involvedfor stone-agemen and favorable conditions of temperatureand growing by Brown (1941) because of the lowland habits of season. the Indians. Cain observed that the grassland soil Relations of the grassy balds to deciduous forest is rich and deep, too deep for the balds to have been in the southwesthalf of the range and of the heath formed in recent history,and that the eccentric balds to spruce-firforest in the northeasthalf of the growthrings of trees on theirmargins indicate that range are similar. The heath balds, however,are thesetrees have been in theirpresent positions in re- much more extensiveand contact deciduous forest lation to the forest-edgefor nearlya century.Brown types as well as spruce-firforests. Withoutdrawing (1953) foundconifers capable of growingin a grassy too close a parallel betweenthe two bald types,both bald on Roan Mountain,and suspected that failure may be thoughtclimax vegetationtypes of the two of seeds to germinateor of seedlings to survive in "systems" developing in related conditions of ex- the balds may inhibittheir invasion by forest. tremeexposure. has that the grassy balds were Camp suggested DISTRIBUTIONAL RELATIONS originallymeadows with ericaceousislands and over- growth,the ericads and grasses sharing dominance. THE MOSAIC CHART A numberof introducedEurasian plants appear in As a final step in the analysis, dispositionof the thebalds now; and, whetheror not theyare "climax," types describedupon the topographyof the Smokies the balds are not the same natural communitiesthey may be studied. The pattern,with the complexityof were beforethey were grazed. Close similarityof the sites and varietyof types,cannot be a simpleone but vegetationof the balds with the cleared fieldsof the may be based upon the two major gradientsof ele- Smokies (Spence and Russell Fields) mightsuggest vation and site. Reduction of environmentto two that the grassy balds also were cleared by man at complex-gradientsrequires various assumptionsand sometime in the past. This same similaritymay only simplifications;but, for the special conditionswithin imply that similar grassland communitiesare able the Smokies,these seem withinreason. to exist at high elevationsin these mountainsin the The method of developing the pattern may be absence of trees,whether the trees are naturallyex- describedas the mosaic chart. The two major gradi- eluded or artificiallyremoved. Tree invasion is evi- ents were used as axes, with elevationfrom 1500 to dent in the cleared fields,but is proceedingslowly, if 6600 ft (460-2010 m) on the ordinate,moisture-bal- at all, in the grassy balds and probablyin response ance as indicated by site on the abscissa. The sites to recent clinmaticamelioration (Brown 1941). The in the Smokieswere arranged along the gradientin positions of the balds, their forest-edges,and rela- fivesteps, by topographicand climaticconsiderations:

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 5858 R. H. WHITTAKERWHITTAKER Ecological~~~~~~~~~~~~~~~~~~~~~~~Vol. Monographs.26, No. 1 (1) deep coves and valley flats, deep canyons; (2) of the complexvegetation; the final charts are pre- open shallow valleys and flats, smaller ravines and sentedas Figs. 19 and 20. About 40 of the 300 site- draws; (3) lower sheltered slopes of valleys; (4) samples were outsidethe lines drawn for theirtypes. open slopes; and (5) ridges, summits, and peaks. A few limitationsof the pattern may be noted. Open slopes were further subdivided by orientation Gentle slopes often support vegetationa step less from the mesic northeast to xeric southwest. The xeric than indicatedon the pattern. Althoughsharp overlap of the site groups, due to complexity of to- peaks are morexeric than slopes below them,rounded pography, differentsizes and forms of valleys, differ- or level summitsmay support more mesic vegetation ent degrees of exposure of slopes and ridges, etc., than their south slopes. The beech gaps in certain is evident. But on the whole, these steps form a concave slopes in the subalpine forestsare not indi- series with roughly graded conditions of exposure cated in the boreal forest pattern. The two bald and inoisture-balance, a topographic gradient along types appear in similar topographic positions for which distribution of vegetation types can be studied. the two systems,but are shownin contactwith only With the framework of the chart set up in terms one other type in each case. The contact indicated of the major gradients, the 300 site-samnpleswere is the critical one of the most closely related type, plotted by topographic position and elevation. The normally adjoining the bald on the south slope. samples had been classified into the 15 types and Other types meet the balds on other slopes through transitions between them. Subtypes distinguished by abrupt transitions,and the grassy balds thus con- lower strata were plotted in the high-elevation for- tact red oak-chestnutand beech forests. The heath ests; and degree of slope and percentages of key tree balds may comein contactwith oak-chestnut and pine Species and groups were entered for each sample. heaths and occasionallywith oak-chestnutand hem- Isorithins for each of the moisture-balance classes of lock forests. Wherever the environmentalgradient the eastern forests were drawn, and lines for certain is telescoped, as along a sharp ridge, two types important species were drawn in-hemlock and beech, whetheradjacent ones or not may meet with a tele- spruce and fir. The percentage values were well scoped transitionbetween them. scattered; but intermediate positions for the iso- One noteworthyfeature of the patternis the ex- rithms, reflecting average conditions of many sites, pansion of mesic forests toward higher elevations. were sought. The two systems at high elevations were Vertical lines on the chart extend through topo- plotted separately because of their overlap along the graphicallysimilar sites at differentelevations, but gradients. not throughsites of similarmoisture-balance. Change The result was a gratifying reduction to pattern of moisture conditions in topographicallysimilar

VEGETATIONOF GREATSMOKY MOUNTAINS VEGETATIONOFGREAT SMOKY MOUNTAINS PArTERNOF EASTERNFOREST SYSTEM PMTERNOF BOREALFOREST SYSTEM

6500

FHEATH 6000 E~~~~~~~~~~~~~~BALD1 [BOREAL FORESTS]

5500X) MESICBEECHI FORESTS / MESICTYPE SEDGETYPE / 650(0RED SRUCE FOREST

*34 ~ ~~~~~~/ BZ / I \N [EASTERNFORESTS] C I RAVINES MUWNTAIN ,-'~~~~,'/ ~~~~~I PINE -004- ~~~~I / ~~~~,// HEATH

MESC----RE SPRUCE------I

I I I ~~~~~~~~~~~~PINE

2000- - I 2 OES FAT DRM SETRD OPNSOEI NE E W S RIDGESNE of EasternIForestCANYONS RAVINES SLOPES SWI 15002000 __SysHEATH ViRSNII~~~~~ MEIC ------FO REST COVES IFLATS DRAWSI SHELTERED I~~OPNSLOPSES RIDGES K PEAK~S FORVgeainofGetEll MutispT- IN NWtSE SWl FIG.~~~~~~~~~~UFIG.19. (Vegetat~~~~~~~~io ofGetSoyMutinpt CANOwI RAVJINES I LPS

FIG. 19. FIG. 20. (Vegetation of Great Smoky Mountains, pat- (VsegnFoestainoGrseatSok.ontis,)t ternlof Boreal Forest System.)

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 59 sites toward higher elevations is indicated by the Throughout its range, as the cove forests expand to- curving boundary of the cove forest grouping. The ward higher elevations, hemlock moves ahead into boundary between spruce and fir dominance is sites which are one step less mesic. oblique, and the lower boundary of the subalpine 3. Gray beech forest. The south-slope subtype with forests is slightlyoblique. Since valleys may be some- sedge floor occupies gaps, draws, and concave shel- what cooler than open south-facing slopes at a given tered slopes facing southeast through south to west elevation, occurrence of spruce 2-300 ft lower in val- above 4500 ft. The north-slope subtype with beech leys would be expected. If the lower limit of spruce or beech-mixed canopy and herbaceous floor occupies were largely dependent on moisture conditions, the comparable sites of northerly orientation. Outside boundary should be much more steeply oblique than the range of spruce forests the north-slope subtype is the case. Factors of temperature or growing sea- occupies open northerly slopes. son closely correlated with elevation appear to be of 4. Red oak-pignut hickory forest. Oak-hickory primary importance in determining the lower limits forests in the Smokies are restricted to the lower of spruce, rather than moisture conditions as sug- slopes of the range, where they occur in draws and gested for subalpine forests in the Rockies (Dauben- ravines, on sheltered northerly slopes, and on lower inire 1943). sheltered southeast and west slopes. 5. Chestnut oak-chestnut forest. Sheltered south- DISTRIBUTIONS OF TYPES erly and open northerly slopes are occupied by chest- the north- Topographic distributions of types on nut oak-chestnut forests at lower elevations, but to- west slope of the Smokies may be summarized: ward middle elevations the oak-chestnut heath ex- forest. At low elevations cove 1. Cove hardwoods pands to replace them on most open slopes and some slopes, forests are restricted to valleys and lower sheltered southerly ones. The oak-chestnut forests with cove forest transition occupying smaller ra- expand at the same time to take the place of oak- vines and flats of shallow valleys. Toward higher hickory on sheltered slopes and in draws, and to ad- cove forests spread outward on sheltered elevations join cove forest transition. At 3700 ft chestnut oak- ex- and northerly slopes, occupying all valley sites chestnut forests occupy a wide range of sites from The cept dry, open draws on south slopes. upper draws through sheltered slopes to open northerly limit is near 4500 ft, where either spruce or beech slopes; above that elevation they are replaced by red usually becomes dominant. The cove forest transition oak-chestnut forests. extends from its lower sites upward on the xeric side 6. Chestnut oak-chestnut heath. At lower eleva- of hemlock, to become above 2500 ft a transition be- tions the type occupies intermediate slopes. Toward tween hemlock and oak-chestnut forests. middle elevations, above 2000 ft. it expands to occupy most open slopes except south and southwest ones. Toward higher elevations chestnut oak-chestnut forests occupy sonmeof the open slopes, and above 3700 ft the oak-chestnut heaths drop out with expansion of oak-chestnut forest and replacement of chestnut oak by red and white oaks. 7. Red oak-chestnut forest. Red oak-chestnut forests occupy sheltered southerly slopes, more xeric than those occupied by hemlock, and open northerly slopes above 3700 ft. Toward higher elevations they FIG. 21. Topographic disposition of vegetation types. occur on open slopes while hemlock occupies the View of idealized mountain and valley, looking east, sheltered and above 4500 ft red oak-chestnut with 6500-ft peak bearing subalpine forest on left, ones; lower 5500-ft peak covered up to summitbald with de- occupies open intermediate and southerly slopes as ciduous forest on right. Vegetation types: beech forests occupy open northerly slopes. Outside BG-Beech Gap OH-Oak-Hickory Forest the range of spruce, high cols and ridges, except the CF-Cove Forest P-Pine Forest and Pine most exposed ones, bear red oak-chestnut. F-Fraser Fir Forest Heath GB-Grassy Bald ROC-Red Oak-Chestnut 8. White oak-chestnut forest. Slopes more xeric H-Hemlock Forest Forest than those covered by red oak-chestnut are occupied HB-Heath Bald S-Spruce Forest by red and white- or white oak-chestnut forests. Be- OCF-Chestnut Oak- SF-Spruce-Fir Forest Chestnut Forest WOC-White Oak-Chestnut low 4500 feet white oak-chestnut occurs on some open OCH-Chestnut Oak- Forest southeast and west slopes. Above 4500 ft open south Chestnut Heath and southwest slopes and some exposed ridges and summits support white oak-chestnut. 2. Eastern hemlock forest. At elevations below 3500 9. Virginia pine forest. South- and southwest-fa6- ft hemlock stands occupy valley flats which are ele- ing slopes and summits of low hills bear Virginia vated above or remote from the stream. Toward pine forests, and successional stands are frequent higher elevations, as cove forest occupies these sites, on other sites. Virginia pine extends above 2300 ft hemlock moves up on sheltered northerly slopes and, but seldom dominates stands above that elevation. still higher,onto sheltered southerly slopes and ridges. 10. Pitch pine heath. Open south and southwest

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions Ecological Monographs 60 R. H. WHITTAKER Vol. 26, No. 1 slopes and some ridges support pitch pine stands Of the historical interpretationswhich might be between2200 and 3200 ft. Individual stands domi- offered,one seems most adequate and is based on to- nated by pitch pine may be found at any elevation, pography of the mountains. Observationof a map but are uncommonat higherand lower elevations. of the Smokies reveals a suggestive correlation: 11. Table mountainpine heath. Pine heathsabove Clingmans Dome, southwestof which the last red 3200 ft occupy open south and southwestslopes spruce grow,is the high point of the range and the and some ridges,and are usually dominatedby table firstof a series of peaks extendingabove 6000 ft mountainpine. Few pine stands occur above 4500 along the ridge to near the northeastend of the ft, but some appear on exposed slopes up to 4700- range. The spruce-firforests grow on these high 4800 ft. peaks and on the adjacent ridges,slopes, and valley 12. Grassybald. The grassybalds are restrictedto sites down to 4500 ft. The highestpoints south of the highestelevations and most exposed points out- Clingmans Dome are Thunderhead Mountain and side the range of spruce,where they appear on sum- Silers Bald, 5530 and 5620 ft,which support decidu- mits and extenda shortdistance down the south and ous forestsonly and, on their peaks, grassy balds. west slopes of peaks. The spruce-firforests are limitedto that part of the 13. Red spruce forest. The spruce forestsare re- range wherepeaks above 5700 ft occur (Fig. 22). strictedto the northeasthalf of the range and occur thereabove 4500 ft, occupyingall sites except those of beech gaps and heath balds. Red spruce is dominant in subalpine foreststands at lower elevationsbelow 5500 to 6000 ft, depending on site. Under- growthsubtypes of red spruce and Fraser firforests have been relatedto topographyin the precedingde- in I scriptions. 14. Fraser fir forest. Above about 6000 ft fir is dominant;and all available sites except the extreme exposures of heath balds are occupied by it, with subtypessimilar to thoseof red spruce forests. 15. Heath balds. Heath balds occur on sharp exposed ridges, peaks, and points above 4500 ft and locally down to 4000 ft. Ridges pointingsouth and 0 10 20 k04 SOUTHWEST DSTANC IN MILES NOMTEAST west are especiallylikely to bear them,and they extend fartherdown south and west slopes on other FIG. 22. Profile of Great Smoky Mountains to illustrate hypothesis on southern limit of Appalachian ridges. They may occur, however,wherever slope is spruce-firforests. Distribution of spruce-firforests on steep and exposure great. higher peaks in northeasthalf of the range is indicated by diagonal shading. Grassy balds are indicated by DISTRIBUTION OF SUBALPINE FORESTS black caps on lower peaks of the southwesthalf of the One aspect, a mostinteresting one, of the distribu- range, otherwise covered by deciduous forest. Climatic the warming during the xerothermicperiod sufficientto dis- tion of types remains. In the northeasthalf of place the lower limits of spruce-firforests upward from range sites above 4500 ft, with some exceptions,are 4500 ft to approximately 5700 ft would account for occupied by forestsof spruce and fir. Southwestof present distributionof these forests. the middleof the range, not far beyond Clingnians Dome, spruce-firforests are absent fronmthe high The survival of spruce-firforests in the area of mountainseven thoughconsiderable areas above 4500 higherpeaks may be thoughtto have been dependent ft exist there. In the southwesternSmokies the place on these peaks, where the forests could find sanc- of the subalpine forests is taken by high-elevation tuaryduring the xerothermicperiod. If, in the warm deciduous forests-beech-mixedand beech in mesic period following the last glaciation, temperatures sites,red and whiteoak-chestnut in more xeric ones. rose sufficientlyto displace the spruce-firforests up- The change betweenthe two halves of the range is ward by 1000 ft or a little more implyinga mean not one of gradual taperingaway and dilutionof the temperaturerise of 20 or 3? F (Shanks 1954)-their spruce-firforests toward the south but of theirpres- absencesouth of ClingmansDome wouldbe accounted ence, with full dominance,in one area and complete for. A somewhatgreater warming, with a displace- absence in the other. Climatesin the two halves of mentupward of 1300 ft, would still permitspruce to the range cannot differgreatly, for vegetationpat- surviveon summitsnear 6000 ft and firand moun- ternsbelow 4500 ft are the same. Precipitationabove tain ash to exist with it on northeast slopes of the 4500 ft may be less in the southwesternhalf of the peaks only a few hundredfeet higher. range,but thereis no reason to doubt that the high Of the history of high-elevation vegetation in the slopes of these mountainscould support spruce-fir Smokies, it may be suggested that the spruce-fir forests. The fact that theydo not do so is a striking forests extended farther south during glaciation than example of failure of physiognomicconvergence of at present-how much farther can scarcely be vegetationin similarclimates (cf. Beadle 1951). guessed. During the last xerothermic period they

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 61 were pushed upward to 5600-5800 ft elevation and importance of species, groups of species, types, and were pushed offthe tops of the lower peaks south of parts of vegetation patterns (see also Part II). Se- ClingmansDome. As the climate cooled again, the mantic limitations of such interpretation may be evi- forests advanced down the slopes from the higher dent. Expansion in one direction is, with equal northeasternpeaks where they had had found sanc- justice, contraction in the other. The types and parts tuaryand reoccupiedthe land above 4500 ft. South- of vegetation patterns which expand and contract west of ClingmansDome absence of spruce forests are subjectively chosen fractions of the whole pat- would leave mountainsurfaces above 4500 ft avail- terns. A type or part of a pattern which is treated able for those species of the eastern forests best as the same in differentmountain ranges is not really adapted to high-elevationconditions-gray beech, the same; and the further apart the two ranges are, northernred and white 6aks, and chestnutparticu- the less meaningful it is to treat it as the same. Such larly-while other deciduous species were displaced interpretation offers an alternative to that based on downwardby the coolingclimate. The spruce forests climax regions and ecotones, however, and may do should have been movingsouthwest along the ridge more justice to the complexities and relativities of from ClingmansDome in the 4000 years since the vegetation patterning (cf. the mosaic interpretation peak of the xerothermicperiod (Flint 1947), but are of Krause 1952). perhaps retarded or halted by the extensivebeech Five major forest groupings of the eastern United forestsof Double Spring Gap. States nmeetin the Smokies and contribute to the variety of their vegetation pattern. Vegetations of REIATION OF THE VEGETATION PATTERN TO THOSE OF other ranges north and south along the Appalachians OTHER MOUNTAIN RANGES are suggestive in part of reduced versions of the Most of the site-sampleswere for the northwestor pattern of the Smokies (or of the Cumberlands or Tennesseeside of the Smokies,and the pattern de- Blue Ridge). Thus Raup (1938) describes a pattern rived applies to that side. The vegetationpattern of in the Black Rock Forest of New York which may be the southeastside is closely similar; but mesic for- compared with that of the Smokies: mixed hard- ests are morenarrowly restricted to valleys,and pine woods and hemlock hardwoods in mesic sites (swamp stands are much more limited. The apparent incon- types also occur), red oak and chestnut oak forests sistency,with both mesic and xeric types reduced, on slopes, white oak-pignut hickory on northerly should be due to effectsof the mountainsthemselves faces of hill tops, and pitch pine-scrub oak (Q. ilici- on climate. Because of the rain-shadoweffect, there folia) on southerly faces. Southern exposures of is less precipitationto the southeastto supportinesic summits have small areas of natural meadow inter- valley types; but the open slopes and ridges are spersed with scrub oak, suggestive of grassy balds. somewhatprotected from the westerlywinds which Because in the Smokies an Appalachian vegetation contributeto the drynessof pine sites on the north- pattern of maximunmdiversity appears, intermediate west side. The resultis an expansionof intermediate, to the mixed mesophytic and oak-chestnut prevailing oak-chestnuttypes at the expense of the extremeson climaxes of the Cumberlands and Blue Ridge, the the southeast,as compared with the northwestside author has regarded the vegetation of this range as of the range. Farther east of the Smokies in the "central" to that of other Appalachian Mountains Blue Ridge, the oak-chestnutgrouping is further not too distant. expanded as a prevailing climax, and nmesophytic The Southern Appalachians have long been recog- communitiesare very restricted. nized (Adams 1902; Harshberger 1903, 1911:198, West fromthe Smokies in the CumberlandMoun- 209-211) as a geographic center for the forests of the tains, the mesophyticportion of the patternexpands eastern United States. In the interpretation of Braun to becomeprevailing climax, while oak-chestnutand the Appalachian center has been given special promi- pine types are much restricted(see also Part II). nence, and the Mixed Mesophytic Region in and Seen fromthe Smokiesthe vegetationof the Cumber- around the Cumberland Mountains is regarded as the lands is a somewhatreduced, more highly mesophytic center. It is Braun's thesis (Braun 1950 :39) that versionof the Smokies pattern,to which are added "The Mixed Mesophytic association, which charac- low-elevationtypes not recognizedin the Smokies. terizes this region, is the most complex and the oldest Seen from the Smokies the vegetationof the Blue association of the Deciduous Forest Formation. It Ridge is a comparabletransformation by expansion occupies a central position in the deciduous forest as of the oak-chestnuttypes and contractionof the a whole, and fronmit or its ancestral progenitor, the mesophyticcommunities. Seen from the Cumber- mixed Tertiary forest, all other climaxes of the de- lands the pattern of the Smokies representsa re- ciduous forest have arisen." The choice of which of duction of the mesophyticpart of the pattern and two ranges of the Southern Appalachians is to be expansion of oak-chestnut,and a transitiontoward designated as central is not a matter of great mo- the oak-chestnutprevailing climax of the Blue Ridge ment. With due respect for the case for the Cumber- and beyond. lands developed by Braun, certain other considera- Changes in vegetation patterns along climatic tions related to the Smokies and the Unaka Moun- gradientsand from one mountainrange to another tains of which they are part, may be mentioned. maybe interpretedon the basis of changesin relative 1. "Centers" are products of human interpretation

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 62 R. H. WHITTAKER Ecological Monographs VnV 2&; Nn~1 and, like such other concepts as "association" and fixed intervals along environmental gradients, and "climatic climax," should not be taken for granted. site-samples taken widely through the vegetation Designation of a center may be thought meaningful pattern as an approach toward randomization, were only as the criterion or criteria (Adams 1902, Cain used as means of vegetation sampling unprejudiced 1944) one chooses to define a center are stated. by assumptions about associations. 2. There is no reason, a priori, for locating a center 2. Arbitrary classes or ecological groups of tree in an area possessing a well-defined prevailing or species and weighted averages based on these classes climatic climax. If, on the other hand, diversity is were used for indication of relative positions of regarded as a primary criterion, then location of stands along the moisture gradient. The site-samples centers in areas like the Smokies with vegetation too were arranged into composite transects, showing the varied to possess a well-defined prevailing climax is relation of plant populations to environmental gradi- perhaps to be expected. ents. Major environmental gradients, those of eleva- 3. Designation of one area, the Smokies, as transition and moisture conditions of site, are conceived tional between other areas which are non-transitional as complex-gradients of many interrelated factors seems of little intrinsic meaning. Areas of great affecting plants. vegetational diversity are most conveniently regarded 3. The sequence of vegetation types from mesic as transitional in a framework of thought which in- sites to xeric at low elevations is: cove forest (mixed cludes climatic climaxes and climax regions as major nmesophytie),oak-hickory forest, oak-chestnut forest, concepts. These concepts should not necessarily be oak-chestnut heath, and pine forest. Deciduous trees applied to the biogeographic problems of "centers." other than oaks predominate in mesic sites, oaks in 4. A traditional primary criterion of a center, intermediate sites, and pines in xeric sites; but rela- though not the only possible one, is biotic diversity- tive proportions of these growth-forms change con- richness in species in general, or in a particular tinuously along the moisture gradient through the group being studied. Some secondary criteria are various types. Along the same gradient, stature of features usually associated with areas of maximum trees and canopy coverage in general decrease; but biotic diversity-age of the area, wide range of habi- the number of tree stems is greater in xeric sites than tats, and occurrence of narrow endeniism among its in mesic. At all points along the gradient, undis- species. turbed forest stands are self-maintaining; and a 5. The greater ranges of elevation and moisture curve relating stem numbers and diameters in these is or conditions of the Smokies imply greater floristic di- all-age, climax stands suggested. Diversity, in of the tree stratum is maximal in versity, as well as greater variety of vegetation types. richness species, low-elevation stands transitional between cove for- This fact, as well as the secondary criteria mentioned, may point to the Smokies rather than the Cumber- ests and oak forests; and diversity values decrease lands as "central" for the Southern Appalachians. with increasing departure from these stands toward more mesic or more xeric conditions and toward Both ranges are doubtless to be regarded as parts higher elevations. of the broader Southern Appalachian center. Within the this area the differentcriteria of Braun and the au- 4. Coverage of shrub stratum in general increases along the moisture gradient toward xeric thor lead rather naturally to their differentemphasis sites. Deciduous, non-ericaceous shrubs predominate of the Cumberlands and the Smokies. A major con- in mesic sites, evergreen ericaceous shrubs in interme- cern of the present paper is with vegetation patterns diate sites, and deciduous ericads (Vaccinioideae) in in relation to environmental gradients, and it is xeric sites. With increase of shrub coverages along thought that in the Snmokiesthe eastern forests show the gradient, herb coverages decrease; coverages for their greatest diversity and differentiationinto types the two strata are inversely related. Ferns and deli- within a limited area. The pattern is doubtless one of cate, umbrella-shaped herbs prevail in mesic forests the most complex on the continent. Also in the Siski- but decrease in coverage along the moisture gradient you Mountains and Klamath Region, somewhat com- as other herb growth-forms (rosette plants, grasses, parable forest center of the West, an observer is at ground heaths, etc.) increase. Herb coverages in gen- firstoverwhelmed by variety of vegetation types; but eral increase toward higher elevations, but shrub this variety is expressive of climatic gradients and coverages are more complexly related to elevation. variety of soil materials as well as of elevation and 5. In subalpine forests Picea rubens is dominant local moisture conditions. The author knows no pat- from 4500 ft up through the lower part of the for- tern of climax vegetation, occurring within a limited ests; Abies fraseri is dominant above 6000 ft. Sub- area on relatively uniform soil materials, more varied alpine forests show limited differencein canopy com- than that of the Great Smoky Mountains. position along the moisture gradient but striking dif- SUMMARY ference in undergrowths. The same herbs and shrubs as those in mesic deciduous forests appear in valleys; I. Gradient Analysis a comnplexundergrowth of. five strata (Hylocomium, 1. Vegetation of a range of the Southern Appa- Oxalis, Dryopteris, Vaceinium, and Viburnum) oc- lachians, the Great Smoky Mountains of Tennessee, curs on north and east slopes and decreases in cover- was analyzed. Field transects, with samples taken at age around to west and south slopes; and a dense

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 63

Rhododendron heath occurs in spruce-fir stands on any very real sense boundaries between communities; ridges. they are communities themselves and may often be 6. Species populations have for the most part characterized by commodal groupings of smaller rounded and tapered curves of population distribu- woody species. whole tion along environmental gradients; these curves ap- 11. Climax vegetation is interpreted as a a single pear to be of binomial form. Both centers and limits continuum or climax pattern, rather than as moisture- of population curves for different species are well climatic climax or set of polyclimaxes. The climax char- scattered along the gradients. Natural clustering of gradient pattern permits comparison of in terms populations into groups of associated species is ab- acteristics of different elevations or areas rela- sent or weakly developed. Vegetation types are, with of dominant growth-forms, species diversities, etc. some exceptions, fully continuous with one another. tive importance of species and species groups, approaches to the "climatic cli- Distributions of undergrowth species do not corre- Three quantitative average climax composition, spond to those of dominant species, and change in max" are suggested: type, and intermediate climax floristic composition of communities is continuous prevailing climax through vegetation types defined by dominant species. stand. 12. Associations are not communities; they are of Vegetation II. Discussion: An Interpretation community-typesor classes of communities,and hence Patterning conceptual units which should not be thought in- 7. The distributional pattern of a species to some herent in vegetation itself. A necessary complement extent expresses the genetic pattern of that species. to the classificatory approach is one considering Certain major species in the Great Smoky Mountains vegetation patterns as wholes. The approach of re- consist of two or more ecotypes; and distributions of lating gradients of plant populations and community species are significantly affected also by occurrence characteristics to gradients of environment is termed of lines, introgressive hybridization, polyploidy, and gradient analysis. apomixis. "Competition" may affect distributional III. Vegetation Types and Their Distributional patterns in various ways: (1) formation of a plateau Relations distribution with competition occurring within the types of the Great Smoky Moun- species, (2) depression of population levels through 13. Vegetation of subjectivity competition with other species, (3) competitive split- tains are defined as Xunits recognized and dominance. Charac- ting of a species distribution, and (4) displacement by criteria of physiognomy the so far as of the environmental mode from the physiological ter-species are designated for types, vegetation types are de- optimum. possible. Fifteen climax The relation of cove forests to mixed meso- 8. Species are distributed not in terms of associa- scribed. forest in the Cumberland Mountains is con- tions, but each according to its own physiology and phytic of driest and most exposed sites in genetic pattern, and hence no two alike. The as- sidered; and types (pine stands, heath balds, and grassy sumption that species are organized into associations the Smokies are interpreted as topographic climax. as real units, designated the association-unit theory, balds) were plotted on a mosaic chart is rejected in favor of the individualistic hypothesis 14. The site-sainples relation of vegetation types to elevation of Ramensky and Gleason. Vegetation niay be inter- to show the The southwestern limit of the Ap- preted as a complex and largely continuous popula- and topography. forests is reached in the Smokies, tion pattern. palachian spruce-fir in the northernhalf of the range where peaks of 6000 9. The condition of continuous population change ft and above occur. It is believed that the spruce- along continuous environmental gradients in natural fir forests found sanctuary on higher peaks during communities is termed gradation. In gradating vege- the xerothermic period, when a displacement up- tation, some community-types are conveniently de- ward of 1000-1300 ft destroyed these forests in lower fined by one dominant species or the distributional parts of the range and farther south. overlap of two dominant species. Associations de- 15. The vegetation of the Smokies is interpreted as fined by several species may be conceived as group- an Appalachian pattern of maximum diversity and ings of character-species (or of dominant species) differentiation of types, intermediate in some re- with their population modes close together in relation spects to the patterns of the Cumberlands and Blue to environmental gradients. Species thus related may Ridge. Vegetation patterns of different mountain be termed commodal, and a group of such species ranges are often related by the relative expansion may be termed a commodium. and contraction of different,recognized parts of the 10. Some vegetation types of more extreme en- patterns. Both the Smokies and Cumberlands are vironments are separated from the rest of the vegeta- parts of the Southern Appalachian center of the by relative discontinuities. These dis- tion pattern eastern forests. continuous types or zones may be characterized by flattened or "plateau" distributions of dominant spe- LITERATURE CITED cies or growth-forms. Zonation is distinguished from Adams, C. C. 1902. Southeastern United States as a gradation as a tendency toward partial segmentation center of geographic distributionof flora and fauna. of the vegetation continuum. "Ecotones" are not in Biol. Bull. 3: 115-131.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 6464 R. H. WHITTAKERWHITTAKER Ecological~~~~~~~~~~~~~~~~~~~~~~~Vol. Monographs26, No. 1 Alechin, W. W. 1925. Ist die Pflanzenassoziation eine . 1932. Plant sociology; the study of plant com- Abstraktion oder eine Realitait Bot. Jahrb. 60: munities. Translation of "Pflanzensoziologie" by G. (Beibl. 135): 17-25. D. Fuller & H. S. Conard. New York: McGraw-Hill. Allee, W. C., A. E. Emerson, 0. Park, T. Park & K. P. 439 pp. Schmidt. 1949. Principles of animal ecology. Phila- 1951. Pflanzensoziologie. GrundzUgeder Vege- delphia: Saunders. 837 pp. tationskunde. 2nd edn. Wien: Springer. 631 pp. Anderson,E. 1948. Hybridization of the habitat. Evo- Braun-Blanquet, J. & J. Pavillard. 1922. Vocabulaire lution 2: 1-9. de sociologie v6getale. Montpellier: Roumegous et . 1949. Introgressive hybridization. New York: Dehan. 16 pp. Wiley. 109 pp. Brown, D. M. 1941. Vegetation of Roan Mountain: A * 1953. Introgressive hybridization. Biol. Rev. phytosociologicaland successional study. Ecol. Monog. 28: 280-307. 11: 61-97. Arnborg,T. 1940. Der Vallsj6-Wald, ein nordschwedis- . 1953. Conifer transplants to a grassy bald on cher Urwald. Acta Phytogeog. Suecica. 13: 128-154. Roan Mountain. Ecology 34: 614-617. - . 1953. Det nordsvenskaskogstypssehemat. Stock- Brown, R. T. & J. T. Curtis. 1952. The upland conifer- holm: Svenska Skogsv.-f6ren.F6rlag. 20 pp. hardwood forests of northernWisconsin. Ecol. Monog. Ashby, E. 1948. Statistical ecology. II.-A reassess- 22: 217-234. ment. Bot. Rev. 14: 222-234. Cain, S. A. 1930a. Certain floristic affinitiesof the Ashe, W. W. 1922. Forest types of the Appalachians trees and shrubs of the Great Smoky Mountains and and White Mountains. Elisha Mitchell Sci. Soc. Jour. vicinity. Butler Univ. Bot. Studies 1: 129-150. 37: 183-198. -. 1930b. An ecological study of the heath balds Ayres, H. B. & W. W. Ashe. 1902. Forests and forest of the Great Smoky Mountains. Butler Univ. Bot. conditions in the Southern Appalachians. Appendix Studies 1: 177-208. A, Message from the President. Pp. 45-109. -. 1931. Ecological studies of the vegetation of Beadle, N. C. W. 1951. The misuse of climate as an in- the Great Smoky Mountains of North Carolina and dicator of vegetation and soils. Ecology 32: 343-345. Tennessee. I. Soil reaction and plant distribution. Beard, J. S. 1955. The classificationof tropical Ameri- Bot. Gaz. 91: 22-41. can vegetation-types.Ecology 36: 89-100. -. 1934. Studies on virgin hardwood forest: II, Beauchamp, R. S. A. & P. Ullyott. 1932. Competitive A comparisonof quadrat sizes in a quantitative phyto- relationships between certain species of fresh-water sociological study of Nash's Woods, Posey County, triclads. Jour. Ecol. 20: 200-208. Indiana. Amer. Midland Nat. 15: 529-566. Beger, H. K. E. 1922-3. Assoziationsstudien in der -. 1935. Ecological studies of the vegetation of Waldstufe des Schanfiggs. Jahresber. Naturf. Gesell. the Great Smoky Mountains. II. The quadrat method Graubjindens,N. F. 61 (Beilage): 1-96; 62 (Beilage): applied to sampling spruce and firforest types. Amer. 97-147. Midland Nat. 16: 566-584. Bicher, T. W. 1938. Biological distributional types in . 1936. Synusiae as a basis for plant sociological the flora of Greenland. A study on the flora and field work. Amer. Midland Nat. 17: 665-672. plant-geographyof South Greenland and East Green- -. 1939. The climax and its complexities. Amer. land between Cape Farewell and Scoresby Sound. Midland Nat. 21: 146-181. (Danish summary.) Medd. -om Gronland 106(2): 1- -. 1943. The Tertiary character of the cove hard- 339. wood forests of the Great Smoky Mountains National Bolleter, R. 1921. Vegetationsstudienaus dem Weisstan- Park. Torrey Bot. Club Bull. 70: 213-235. - nental. St. Gall. Naturw. Gesell. Jahrb. 57(2): 1-140. -. 1944. Foundations of plant geography. New Braun, E. Lucy. 1935a. The undifferentiateddeciduous York: Harper. 556 pp. A the forest climax and the association-segregate. Ecology -. 1945. biological spectrumof the flora of 16: 514-519. Great Smoky Mountains National Park. Butler Univ. Bot. Studies 7: 11-24. . 1935b. The vegetation of Pine Mountain, Ken- -. 1947. Characteristics of natural areas and tucky: An analysis of the influenceof soils and slope factors in their Ecol. 17: 185- exposure as determinedby geological structure. Amer. development. Monog. 200. Midland Nat. 16: 517-565. Cain, S. A. & A. J. Sharp. 1938. Bryophyticunions of -. 1938. Deciduous forest climaxes. Ecology 19: certain forest types of the Great Smoky Mountains. 515-522. Amer. Midland Nat. 20: 249-301. . 1940a. An ecological transect of Black Moun- Cain, S. A. et al. 1937. A preliminaryguide to the tain, Kentucky. Ecol. Monog. 10: 193-241. Greenbrier-BrushyMountain nature trail, the Great -- -. 1940b. Mixed deciduous forests of the Appa- Smoky Mountains National Park. Knoxville: Univ. lachians. Virginia Jour. Sci. 1: 1-4. Tenn., Mimeographed. 29 pp. . 1941. The differentiationof the deciduous for- Cajander, A. K. 1909. Ueber Waldtypen. Acta Forest. est of the eastern United States. Ohio Jour. Sei. 41: Fenn. 1(1): 1-175. 235-241. -. 1925. Der gegenseitige Kampf in der Pflanzen- --. 1942. Forests of the Cumberland Mountains. welt. In ";Festschr. Carl Schr6ter," Ver6ff. Geobot. Ecol. Monog. 12: 413-447. Inst. Rfibel Zurich 3: 665-684. 1947. Development of the deciduous forests of -. 1926. The theory of forest types. Acta Forest. eastern North America. Ecol. Monog. 17: 211-219. Fenn. 29(3): 1-108. -. 1950. Deciduous forestsof eastern North Amer- -. 1949. Forest types and their significance. Acta ica. Philadelphia: Blakiston. 596 pp. Forest. Fenn. 56(4): 1-71. Braun-Blanquet, J. 1921. Prinzipien einer Systematik Camp, W. H. 1931. The grass balds of the Great Smoky der Pflanzengesellsehaftenauf floristischerGrundlage. Mountains of Tennessee and North Carolina. Ohio St. Gall. Naturw. Gesell. Jahrb. 57(2): 305-351. Jour. Sei. 31: 157-164.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 65

. 1936. On Appalachian trails. N. Y. Bot. Gard. Daubenmire, R. & A. W. Slipp. 1943. Plant succession Jour. 37: 249-265. on talus slopes in northern Idaho as influenced by . 1945. The North American blueberries with slope exposure. Torrey Bot. Club Bull. 70: 473-480. notes on other groups of Vaceiniaceae. Brittonia 5: Davis, J. H., Jr. 1930. Vegetation of the Black Moun- 203-275. tains of North Carolina: an ecological study. Elisha -. 1951. A biogeographic and paragenetic analysis Mitchell Sci. Soc. Jour. 45: 291-318. the of the American beech (Fagus). Amer. Phil. Soc. Desmarais, Y. 1952. Dynamics of leaf variation in Yrbk. 1950: 166-169. sugar maples. Brittonia 7: 347-387. the Camp, W. H. & C. L. Gilly. 1943. The structure and Dewey, J. & A. F. Bentley. 1949. Knowing and origin of species. Brittonia 4: 323-385. known. Boston: Beacon. 334 pp. R. 1952. Natural communities. Ann Arbor: Carmichael, R. D. 1930. The logic of discovery. Chi- Dice, L. 547 pp. cago: Open Court. 280' pp. Univ. Mich. Press. Du Rietz, G. E. 1921. Zur methodologisehenGrundlage Clausen, J. 1951. Stages in the evolution of plant spe- der modernen Pflanzensoziologie. Wien: Holzhausen. cies. Ithaca: Cornell Univ. Press. 206 pp. 267 pp. Clausen, J., D. D. Keck & W. M. Hiesey. 1940. Experi- . 1928. Kritik an pflanzensoziologischenKriti- mental studies on the nature of species. I. Effect of kern. Bot. Notiser 1928: 1-30. varied environments on western North American -. 1929. The fundamental units of vegetation. plants. Carnegie Inst. Wash. Publ. 520: 1-452. Internatl. Cong. Plant Sci., Ithaca, Proc. 1926, 1: . 1948. Experimental studies on the nature of 623-627. responses of climatic species. III. Environmental 1930. Classificationand nomenclatureof vege- races of Achillea. Carnegie Inst. Wash. Publ. 581: Svensk Bot. Tidskr. 24: 489-503. 1-129. tation. -. 1932. Vegetationsforschungauf soziationsanaly- Clapham, A. R. 1936. Over-dispersionin grassland com- tischer Grundlage. Handb. Biol. Arbeitsmeth.ed. E. munities and the use of statistical methods in plant Abderhalden 11, 5: 293-480. ecology. Jour. Ecol. 24: 232-251. -. 1936. Classificationand nomenclatureof vege- Plant succession: An analysis of Clements,F. E. 1916. tation units 1930-1935. Svensk Bot. Tidskr. 30: 580- Carnegie Inst. Wash. the developmentof vegetation. 589. Publ. 242: 1-512. Du Rietz, G. E., T. C. E. Fries, H. Osvald & T. A. Teng- --. 1928. Plant succession and indicators. A de- wall. 1920. Gesetze der Konstitution natiurlicher indi- finitive edition of plant succession and plant Pflanzengesellsehaften.Vetensk. och prakt. Undexs. i cators. New York: Wilson. 453 pp. Lappland, anordn. av. Luossavaara-Kiirunavaara . 1936. Nature and structureof the climax. Jour. Aktiebolag. Flora och Fauna 7: 1-47. Uppsala & Ecol. 24: 252-284. Stockholm. Clements F. E. & V. E. Shelford. 1939. Bio-ecology. Egler, F. E. 1947. Arid southeast Oahu vegetation, New York: Wiley. 425 pp. Hawaii. Ecol. Monog. 17: 383-435. Clements, F. E., J. E. Weaver & H. C. Hanson. 1929. Ellenberg, H. 1948. Unkrautgesellsehaftenals Mass Plant competition: An analysis of communityfunc- fur den Sauregrad, die Verdichtung and andere tions. Carnegie Inst. Wash. Publ. 398: 1-340. Eigenschaften des Ackerbodens. Ber. uiberLandtech- Conard, H. S. 1935. The plant associations of central nik 4: 130-146. Long Island: A study in descriptive plant sociology. 1950. Landwirtschaftliche Pflanzensoziologie. Amer. Midland Nat. 16: 433-516. I. Unkrautgemeinschaftenals Zeiger fur Klima und -. 1939. Plant associations on land. Amer. Mid- Boden. Stuttgart/z.Z. Ludwigsburg: Eugen Ulmer. land Nat. 21: 1-27. 141 pp. Constance, L. 1953. The role of plant ecology in bio- -. 1952. Landwirtschaftliche Pflanzensoziologie. systematics. Ecology 34: 642-649. II. Wiesen und Weiden und ihre stand6rtlicheBewer- Cooper, W. S. 1942. Vegetation of the Prince William tung. Stuttgart/z.Z. Ludwigsburg: Eugen Ulmer. 143 Sound region, Alaska; with a brief excursion into PP. post-Pleistocene climatic history. Ecol. Monog. 12: Elton, C. 1947. Animal ecology. 3rd impression. Lon- 1-22. don: Sidgwick & Jackson. 209 pp. Core, E. L. 1929. Plant ecology of Spruce Mountain, Fenneman, N. M. 1938. Physiographyof easternUnited West Virginia. Ecology 10: 1-13. States. New York: McGraw-Hill. 714 pp. identi- Curtis,J. T. & R. P. McIntosh. 1950. The interrelations Fernald, M. L. 1931. Specific segregations and the of certain analytic and synthetic phytosociological ties in some floras of eastern North America and characters. Ecology 31: 434-455. Old World. Rhodora 33: 25-63. botany; a handbook of . 1951. An upland forest continuumin the prai- -. 1950. Gray's manual of and ferns of the central and rie-forest border region of Wisconsin. Ecology 32: the flowering plants northeasternUnited States and adjacent Canada. 8th 476-496. (centennial) edn. New York: American Book. 1632 Dansereau, P. 1951. Description and recordingof vege- basis. Ecology 32: 172-229. PP. tation upon a structural Fink, P. M. 1931. A forest enigma. Amer. Forests 37: of certain eco- Daubenmire, R. F. 1930. The relation 538, 556. logical factors to the inhibition of forest floor herbs Fisher, R. A., A. S. Corbet & C. B. Williams. 1943. The under hemlock. Butler Univ. Bot. Studies 1: 61-76. relation between the number of species and the num- -. 1931. Factors favoring the persistence of a ber of individuals in a random sample of an animal relic association of eastern hemlock in Indiana. But- population. Jour. Anim. Ecol. 12: 42-58. ler Univ. Bot. Studies 2: 29-32. Flahault, C. & C. Schroter. 1910. Rapport sur la no- . 1943. Vegetational zonation in the RockyMoun- menclature phytogeographique. Actes du 3me Colngr. tains. Bot. Rev. 9: 325-393. Internat. Bot., Bruxelles, 1: 131-164.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions WHITTAKER Ecological Monographs 66 R. H. Vol. 26, No. 1 Flint, R. F. 1947. Glacial geology and the Pleistocene Harlow, W. M. & E. S. Harrar. 1941. Textbook of den- epoch. New York: Wiley. 589 pp. drology. New York: McGraw-Hill. 2nd edn. 542 pp. Fournier, P. 1927. Qu'est-ce que l'association du hetre? Harshberger,J. W. 1903. An ecologic study of the flora Soc. Bot. de France Bull. 74: 416-429. of mountainousNorth Carolina. Bot. Gaz. 36: 241-258, Frothingham,E. H. et al. 1926. A forest type classifi- 368-383. cation for the Southern Appalachian Mountains and -. 1905. The plant formationsof the Adirondack the adjacent plateau and Coastal Plain regions. Jour. Mountains. Torreya 5: 187-194. Forestry 24: 673-684. - -. 1911. Phytogeographicsurvey of North Amer- Gams, H. 1918. Prinzipienfragen der Vegetationsfors- ica. "Die Vegetation der Erde," ed. Engler & Drude chung. Ein Beitrag zur Begriffsklirungund Methodik 13: 1-790. New York: Stechert. der Biocoenologie. Naturf. Gesell. Zurich Vrtljschr. Hayward, C. L. 1948. Biotic communitiesof the Wa- 63: 293-493. sateh Chaparral, Utah. Ecol. Monog. 18: 473-506. -. 1927. Von den Follateres zur Dent de Morcles. Heimburger, C. C. 1934. Forest-type studies in the Vegetationsmonographieaus dem Wallis. Beitr. Geo- Adirondack region. N. Y. (Cornell) Agr. Expt. Sta. bot. Landesaufn. 15: 1-760. Mem. 165: 1-122. Gause, G. F. 1930. Studies on the ecology of the Or- Heiser, C. B., Jr. 1949. Study in the evolution of the sun- thoptera. Ecology 11: 307-325. flower species Helianthus annuus and H. bolanderi. . 1931. The influenceof ecological factors on the Univ. California Publ. Bot. 23: 157-208. size of population. Amer. Nat. 65: 70-76. Henry, A. J. 1902. The climate of the Southern Ap- Gates, W. H. 1941. Observations on the possible origin palachians. Appendix D, Message from the Presi- of the balds of the Southern Appalachians. Baton dent. Pp. 143-153. Rouge: Louisiana State Univ. Press. 16 pp. (Biol. Holmes, J. S. 1911. Forest conditionsin western North Abstr. 16: 6295, 1942). Carolina. N. C. Geol. & Econ. Surv. Bull. 23: 1-111. Gleason, H. A. 1926. The individualistic concept of the Hulten, E. 1937. Outline of the history of arctic and plant association. Torrey Bot. Club Bull. 53: 7-26. boreal biota during the Quaternary period. Stock- -. 1929. Plant associations and their classifica- holm:Bokf6rlags Aktiebolaget Thule. 168 pp. tion: A reply to Dr. Nichols. Internatl. Cong. Plant Huxley, J. S. 1938. Clines: An auxiliary taxoiiomic Sci., Ithaca, Proc. 1926, 1: 643-646. principle. Nature [London] 142: 219-220. 1936. Is the synusia an association? Ecology -. 1940. Towards the new systematics. In "The 17: 444-451. new systematics" ed. J. S. Huxley, pp. 1-46. Oxford: -. 1939. The individualistic concept of the plant Clarendon. association. Amer. Midland Nat. 21: 92-110. -. 1943. Evolution: The modern synthesis. New . 1952. The new Britton and Brown illustrated York: Harper. 645 pp. flora of the northeasternUnited States and adjacent Ilvessalo, Y. 1921. Die Waldtypen als Grundlage der Canada. New York: N. Y. Bot. Garden. 3 vols., 482, neuen Ertragstafeln Finnlands. Part II of "Ueber 655, & 589 pp. Waldtypen II" by A. K. Cajander & Y. Ilvessalo. Goodall, D. W. 1952. Quantitative aspects of plant dis- Acta Forest. Fenn. 20(1): 42-63. tribution. Biol. Rev. 27: 194-245. . 1922. Vegetationsstatistische Untersuchungen fiberdie Waldtypen. Acta Forest. Fenn. 20(3): 1-73. -. 1953. Objective methods for the classification -. 1929. Notes on some forest (site) types in of vegetation. I. The use of positive interspecificcor- North America. Acta Forest. Fenn. 34(39): 1-111. relation. Austral. Jour. Bot. 1: 39-63. Iversen, J. 1936. Biologische Pflanzentypen als Hilf- -. 1954a. Vegetational classification and vegeta- smittel in der Vegetationsforschung.-Ein Beitrag tional continua. (German summary.) Angewandte zur 6kologischen Charakterisierung und Anordnung Pflanzensoz. [Wien], Festschr. Aichinger 1: 168-182. der Pflanzengesellsehaften. Medd. Skalling-Labora- -. 1954b. Objective methods for the classification toriet (Kopenhagen) 4: 1-224. of vegetation. III. An essay in the use of factor Johnson,W. 1946. People in quandaries: The semantics 2: analysis. Austral. Jour. Bot. 304-324. of personal adjustment. New York: Harper. 532 pp. Gregor,J. W. 1939. Experimentaltaxonomy. IV. Popu- Jones, E. W. 1945. The structureand reproductionof lation differentiationin North American and Euro- the virgin forest of the north temperate zone. New pean sea plantains allied to Plantago maritima L. Phytol. 44: 130-148. New Phytol. 38: 293-322. Katz, N. J. 1930. Die grundlegendenGesetzmaissigkeiten 1944. The ecotype. Biol. Rev. 19: 20-30. der Vegetation und der Begriff der Assoziation. 1946. Ecotypic differentiation. New Phytol. Beitr. z. Biol. der Pflanz. 18: 305-333. 45: 254-270. . 1933. Die Grundproblemeund die neue Rich- Hairston, N. G. 1951. Interspecies competitionand its tung der Phytosoziologie. Beitr. z. Biol. der Pflanz. probable influence upon the vertical distribution of 21: 133-166. Appalachian salamanders of the genus Plethodon. Keever, Catherine. 1953. Present composition of some Ecology 32: 266-274. stands of the formeroak-chestnut forest in the south- Hale, M. E., Jr. 1955. Phytosociology of corticolous ern Blue Ridge Mountains. Ecology 34: 44-54. cryptogams in the upland forests of southern Wis- Keith, A. 1902. Topography and geology of the South- consin. Ecology 36: 45-63. ern Appalachians. Appendix B, Message from the Hansen, H. M. 1930. Studies on the vegetation of Ice- President. Pp. 111-122. land. In "The botany of Iceland," ed. Rosenvinge & Kendeigh, S. C. 1948. Bird populations and biotic com- Warming 3 (10): 1-186. Copenhagen: Frimodt. munitiesin northernlower Michigan. Ecology 29: 101- . 1932. N0rholm Hede, en formationsstatistisk 114. Vegetationsmonografi.(English summary.) K. Danske - . 1954. History and evaluation of various con- Vidensk. Selsk. Skr., Nat. og Math., Ser. 9, 3 (3): cepts of plant and animal communities in North 96-196. America. Ecology 35: 152-171.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 67 January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS Gips- Kendrew, W. G. 1937. The climates of the continents. Meusel, H. 1939. Die Vegetationsverhiiltnisseder Harzvorland. 3rd edn. Oxford: Clarendon,473 pp. berge im Kyffhduserund im siidlicheni 2: 1- King, P. B. 1949. The base of the Cambrian in the Ein Beitrag zur Steppenheidefrage. Hercynia SouthernAppalachians. Amer. Jour. Sci. 247: 513-530, 372. structure 622-645. Meyer, H. A. & D. D. Stevenson. 1943. The for- King, P. B. & A. Stupka. 1950. The Great Smoky and growth of virgin beech-birch-maple-hemlock Res. 67: Mountains-their geology and natural history. Sci. ests in northern Pennsylvania. Jour. Agr. Monthly 71: 31-43. 465-484. Korstian, C. F. 1937. Perpetuation of spruce on cut- Muller, C. H. 1952. Ecological control of hybridization of evolution. over and burned lands in the higher Southern Ap- in Quercus: a factor in the mechanism Mountains. Ecol. Monog. 7: 125-167. Evolution 6: 147-161. palachian and application Korzybski, A. 1933. Science and sanity: An introduc- Nichols, G. E. 1917. The interpretation the classi- tion to non-Aristoteliansystems and general semantics. of certain terms and concepts in ecological Plant World 20: 305- Lancaster, Pa. & New York: Science Press. 798 pp. fication of plant communities. Krause, W. 1952. Das Mosaik der Pflanzengesellschaften 319, 341-353. for the ecological classi- und seine Bedeutung fur die Vegetationskunde. -. 1923. A workingbasis Ecology 4: 11-23, 154- Planta 41: 240-289. fication of plant communities. Kruckeberg, A. R. 1951. Intraspecificvariability in the 179. and their classifiea- response of certain native plant species to serpentine 1929. Plant *associations Plant Sci., Ithaca, Proc. 1926, soil. Amer. Jour. Bot. 38: 408-419. tion. Internatl. Cong. The ecology of serpentine soils. III. 1: 629-641. . 1954. auf der Insel Plant species in relation to serpentine soils. Ecology Nordhagen, R. 1922. Vegetationsstudien Bergens Mus. Aarbok, 35: 267-274. Utsire im westlichenNorwegen. 1-149. Kucera, C. L. 1952. An ecological study of a hardwood Naturv. rekke 1920/21(1): Vegetation und Flora des Sylenege- forest area in central Iowa. Ecol. Monog. 22: 283-299. -. 1928. Die Norske Vidensk.-Akad. i H. 1926. tber Begriffsbildungund Statistik in bietes. I. Die Vegetation. Kylin, 1-612. Bot. Notiser 1926: 81-180. Oslo, I. Mat.-Nat. Kl., Skr. 1927 (1): der Pflanzensoziologie. of the Highlands Langer, Susanne K. 1953. An introductionto symbolic Odum, E. P. 1950. Bird populations in relation to plant succes- logic. 2nd edn. New York: Dover. 367 pp. (North Carolina) Plateau Ecology 31: 587-605. lenoble, F. 1926. A. propos des associations vegetales. sion and avian invasion. of ecology. Philadelphia: Soe. Bot. de France Bull. 73: 873-893. - . 1953. Fundamentals . 1928. Associations vegetales et especes. Arch. Saunders. 384 pp. of the plant Bot. [Caen] 2 (Bull. Mens. 1): 1-14. Oosting, H. J. 1942.' An ecological analysis Carolina. Amer. Mid- Linkola, K. 1924. Waldtypenstudienin den Schweizer communitiesof Piedmont, North Alpen. Verbif. Geobot. Inst. Rilbel Zurich 1: 139-224. land Nat. 28: 1-126. An in- T. 1933. Grundziige der pflanzensoziologis- -. 1948. The study of plant communities: Lippmaa, Freeman. chen Methodik nebst einer Klassifikation der Pflan- troductionto plant ecology. San Francisco: zenassoziationen Estlands. (Estonian with German 389 pp. summary.) Acta Inst. et Horti Bot. Univ. Tartu. Oosting, H. J. & W. D. Billings. 1939. Edapho-vegeta- hemlock 3(4): 1-169. tional relations in Ravenel's Woods, a virgin Amer. Mid- . 1935. Une analyse des forktsde l'ile estonienne forest near Highlands, North Carolina. d'Abruka (Abro) sur la base des associations uni- land Nat. 22: 333-350. forest strates. Acta Inst. et Horti Bot. Univ. Tartu. 4(1/2): -. 1951. A comparison of virgin spruce-fir system. 1-97. in the northern and southern Appalachian -. 1939. The unistratal concept of plant com- Ecology 32: 84-103. the munities (the unions). Amer. Midland Nat. 21: 111- Ostenfeld, C. H. 1908. The land-vegetation of plants. 145. Faer6es, with special reference to the higher Danish investiga- ILidi, W. 1928. Der Assoziationsbegriffin der Pflanzen- Botany of the Faerbes based upon Bog- soziologie, erldutert am Beispiel der Pflanzengesells- tions. Copenhagen & Christiania: Gyldendalske 867-1026. chaften des Tanzbodengebietes im Lauterbrunnental. handel & Nordisk Forlag. Part III: Biblioth. Bot. 96: 1-93. Paczoski, J. 1928. La vegetation de la Forkt de Ser. E Lutz, H. J. 1932. Relation of forest site quality to Bialowieza. Edition du Minist. de l'Agrie. numberof plant individuals per unit area. Jour. For- (Warsaw) (V. I. P. E. 1928). 87 pp. estry 30: 34-38. Pavillard, J. 1912. Essai sur la nomenclaturephytogeo- 35: 165-176. Major, J. 1951. A functional, factorial approach to graphique. Bull. Soc. Languedoc. Geog. associa- plant ecology. Ecology 32: 392-412. . 1935. The present status of the plant Mason, H. L. 1947. Evolution of certain floristicasso- tion. Bot. Rev. 1: 210-232. the cli- ciations in western North America. Ecol. Monog. 17: Phillips, J. 1934-5. Succession, development, of con- 201-210. max, and the complex organism: An analysis 22: 23: 210- Mason, R. L. & M. H. Avery. 1931. A bibliography for cepts. Parts I-III. Jour. Ecol. 554-571; the Great Smokies. Appalachia 18: 271-277. 246, 488-508. Einige Ergebnisse der Merkle, J. 1951. An analysis of the plant communities Poplawskaja, G. I. 1933 (1934). in Hin- of Mary's Peak, westernOregon. Ecology 32: 618-640. Erforschung von Bergwiesen geobotanischer (Russian with German Message from the President of the United States. 1902. sicht und auf ihren Futterwert. Inst. Bot. Acad. Scient. A report of the Secretary of Agriculture in relation summary.) Geobotanica, Acta 195-249. to the forests,rivers, and mountains of the Southern U.R.S.S., Ser. 3, 1: Grundgesetzmaissigkeitenl Appalachian region. Washington: Gov. Printing Of- Ramensky, L. G. 1924. Die (Russian.) Wjestii. fice. 210 pp. im Aufbau der Vegetationsdecke.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions WHITTAKER Ecological Monographs 68 R. H. Vol. 26, No. 1 opytn. djela Woronesch. 37 s. (Bot. Centbl., N. F. Transeau, E. N. 1905. Forest centers of eastern Amer- 7: 453-455, 1926.) ica. Amer. Nat. 39: 875-889. . 1930. Zur Methodik der vergleichendenBear- Tuomikoski,R. 1942. Untersuchungenfiber die Vegeta- beitung und Ordnung von Pflanzenlistenund anderen tion der Bruchmoorein Ostfinnland. I. Zur Methodik Objekten, die durch mehrere, verschiedenartig wir- der pflanzensoziologischenSystematik. (Finnish sum- kende Faktoren bestimmtwerden. Beitr. z. Biol. der mary.) Ann. Bot., Soc. Zool.-Bot. Fenn. Vanamo Pflanz. 18: 269-304. 17(1): 1-203. Rapoport, A. 1952. What is semantics? Amer. Scient. Turesson, G. 1922a. The species and the variety as eco- 40: 123-135. logical units. Hereditas 3: 100-113. Raup, H. M. 1938. Botanical studies in the Black Rock - . 1922b. The genotypical response of the plant Forest. Part I. Black Rock Forest Bull. 7: 1-107. species to the habitat. Hereditas 3: 211-350. . 1947. Some natural floristic areas in boreal . 1923. The scope and import of genecology. America. Ecol. Monog. 17: 221-234. Hereditas 4: 171-176. Reed, F. W. 1905. Report on an examination of a forest Vogt, W. 1948. Road to survival. New York: Sloan. tract in western North Carolina. U. S. Dept. Agr. 335 pp. Bur. Forestry Bull. 60: 1-32. Wangerin, W. 1925. Beitrige zur pflanzensoziologischen Russell, N. H. 1953. The beech gaps of the Great-Smoky Begriffsbildungund Terminologie. I. Die Assoziation. Mountains. Ecology 34: 366-374. Repert. Spec. Novarum Regni. Veg., Beih. 36: 3-59. Rydberg, P. A. 1913. Phytogeographical notes on the Ward, R. D. 1925. The climates of the United States. Rocky Mountain region-I. Alpine region. TorreyBot. Boston: Ginn. 518 pp. Club Bull. 40: 677-686. Weaver, J. E. & F. E. Clements. 1929. Plant ecology. Safford, J. M. 1869. Geology of Tennessee. Nashville: New York: MeG~raw-Hill.520 pp. 2nd edn. 1938. 601 Mercer. 550 pp. PP. Shanks, R. E. 1954. Climates of the Great Smoky Moun- Wells, B. W. 1924. Major plant communitiesof North tains. Ecology 35: 354-361. Carolina. N. C. Agr. Expt. Sta. Tech. Bull. 25: 1-20. Shanks, R. E. & A. J. Sharp. 1947. Summer key to the . 1936a. Origin of the Southern Appalachian trees of eastern Tennessee. Tenn. Acad. Sei. Jour. grass balds. Science, N. S. 83: 283. 22: 114-133. . 1936b. Andrews Bald: The problem of its Small, J. K. 1933. Manual of the southeastern flora. origin. So. Appalachian Bot. Club. Jour. (Castanea) New York: Small. 1554 pp. 1: 59-62. Sokolowa, L. A. 1935 (1937). Die geobotanischen Be- -. 1937. SouthernAppalachian grass balds. Elisha zirke der Onega-Dwina Wasserscheide und der Onega Mitchell Sci. Soe. Jour. 53: 1-26. Halbinsel. (Russian with German summary.) Geo- . 1946. Archeological disclimaxes. Elisha Mitch- botanica, Acta Inst. Bot. Acad. Scient. URSS, Ser ell Sei. Soe. Jour. 62: 51-53. 3, 2: 9-80. Wendelberger,G. 1951. Das vegetationskundlicheSys- . 1936. Die Vegetation des Rayons der Louchi- tem Erwin Aichingers und seine Stellung im pflan- Kestenga Chaussee (Karelien). (Russian with Ger- zensoziologischenLehrgebiude Braun-Blanquets. Ange- man summary.) Geobotanica, Acta Inst. Bot. Acad. wandte Pflanzensoz. [Wien] 1: 69-92. Scient. URSS, Ser. 3, 3: 241-306. . 1952. Vom Wesen und Inhalt der modernen Stebbins, G. L., Jr. 1950. Variation and evolution in Pflanzensoziologie. Schr. Ver. z. Verbreitung naturw. plants. New York: Columbia Univ. Press. 643 pp. Kenntn. i. Wien, Ber. 92: 46-70. Stose, G. W. & Anna J. Stose. 1944. The Chilhowee WhittakeT, R. H. 1948. A vegetation analysis of the group and Ocoee series of the Southern Appalachians. Great Smoky Mountains. Ph.D. Thesis, Univ. of Ill., Amer. Jour. Sci. 242: 367-390, 401-416. Urbana. 478 pp. --. 1949. Ocoee series of the Southern Appalach- . 1951. A criticismof the plant association and ians. Geol. Soe. Amer. Bull. 60: 267-320. climatic climax concepts. NorthwestSci. 25: 17-31. Sukatschew, W. N. (Sukachev) 1928. Principles of . 1952. A study of summer foliage insect com- classificationof the spruce communitiesof European munitiesin the Great Smoky Mountains. Ecol. Monog. Russia. Jour. Ecol. 16: 1-18. 22: 1-44. . 1932. Die Untersuchung der Waldtypen des . 1953. A consideration of climax theory: The osteuropiiischenFlachlandes. Handb. Biol. Arbeits- climax as a population and pattern. Ecol. Monog. meth. ed. E. Abderhalden 11, 6: 191-250. 23: 41-78. Tansley, A. G. 1920. The classification of vegetation -. 1954a. The ecology of serpentinesoils. IV. The and the concept of development. Jour. Ecol. 8: 118- vegetational response to serpentinesoils. Ecology 35: 149. 275-288. . 1929. Succession: The concept and its values. . 1954b. Plant populations and the basis of plant Internatl. Cong. Plant Sci., Ithaca, Proc. 1926, 1: indication. (German summary.) Angewandte Pflan- 677-686. zensoz. [Wien], Festschr. Aichinger 1: 183-206. . 1935. The use and abuse of vegetational con- Williams, C. B. 1947. The logarithmic series and its cepts and terms. Ecology 16: 284-307. application to biological problems. Jour. Ecol. 34: Thornthwaite, C. W. 1931. The climates of North 253-272. America according to a new classification. Geog. Rev. - - . 1950. The application of the logarithmicseries 21: 633-655. to the frequency of occurrence of plant species in . 1941. Atlas of climatic types in the Un-ited quadrats. Jour. Ecol. 38: 107-138. States 1900-1939. U. S. Dept. Agr. Misc. Publ. 421: Willis, B. 1889. Round about Asheville. Natl. Geog. 1-7. Mag. 1: 291-300. 1948. An approach toward a rational classi- Wright, F. J. 1931. The Older Appalachians of the ficationof climate. Geog. Rev. 38: 55-94. South. Denison Univ. Sci. Labs. Jour. 26: 143-250.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 69

1942. Erosional history of the Southern Ap- Tables 20-22, elevation gradient in subxeric sites, 200- palachians. Jour. Geomorphol.5: 151-161. ft intervals. Wright, S. 1932. The roles of mutation, inbreeding, Tables 23-25, elevation gradient in xeric sites, 300-ft crossbreeding, and selection in evolution. 6th In- intervals. be purchased ternatl. Congr. Genet.,Ithaca, Proc. 1: 356-366. Photostats of these tables (35 pages) may Zherebina, Z. N. 1931. Essay of a botanical-agronomi- from the American Documentation Institute for $5.00, cal study of awnless brome grass (Bromus inermis microfilmsfor $2.25, as Document No. 4722. Leyss.). (Russian with English summary.) Bull. Appendix C comprises the composite stand counts on Appl. Bot., Genet. & Plant-Breed. 25(2): 201-352. which the descriptions of Part III are based. Data on trees include numbersof stems for each species in 3-in. Appendix A. Population Charts for Major Tree Species. diameter classes, percentage of stand from the 1-in. In the following population charts, stand percentages diameter class up, and percentage of canopy stems, of tree species in composite stand counts (Appendix C) based in each case on approximately 1000 stems and have been plotted on the vegetation pattern in relation 100 canopy stems. Presences of shrub and herb species for stand to elevation and site (Fig 19). Isorithms are given. The 27 tables represent 13 climax forest have been drawn in, forming nomograms percentages types recognized in the Great Smoky Mountains; tables of tree population levels in relation to the two major gradients studied. The population figures thus outlined available for the various types are indicated in the may be visualized in three dimensionsas hills with their classification of vegetation types at the beginning of summitsrepresenting environmental optima for species, Part III. Photostats of these tables (46 pages) may be and with slopes of decreasing population density with purchased from the American Documentation Institute increasingdeparture from the optima. Population charts for $6.25, microfilmsfor $2.50, as Document No. 4723. are included for 44 major tree species; charts are avail- The full set of population nomogramsfor tree species (see able also for canopy populations and minor species (Appendix A) may also be obtained. Nomogramsare for Note on SupplementaryPublication). 54 species of trees and arborescentshrubs; of these the Note on Supplementary Publication of Appendixes B 32 marked with an asterisk (*) are represented by and C nomograms both for total stems and for canopy stems only. The charts thus permit study of the vegetation as of publishing the The author regrets the impossibility a population pattern in terms of all tree populations, is based. full body of tabulations on which the study and of canopy stems alone. Photostats of these nomo- are made avail- Tables which cannot be presented here grams (22 pages) may be purchased from the American Docu- able to those desiring them throughthe American Documentation Institute for $3.75, microfilmsfor $2.00, mentation Institute, Auxiliary Publications Project, as Document No. 4721. The set of charts includes: Photoduplication Service, Library of Congress, Wash- Abies fraseri*, Acer pensylvanicum, A. rubrum*, A. ington 25, D. C. Advance payment is required. Make saccharum*, A. spicatum, Aesculus octandra*, Ame- checks or money orders payable to: Chief, Photodupli- lanchier arborea, A. laevis, Aralia spinosa, Betula alle- cation Service, Library of Congress. gheniensis*,B. lenta*, Carpinus caroliniana, Carya cordi- Appendix B comprises the full set of transect tables formis, C. glabra*, C. ovalis*, C. tomentosa*, Ca.stanea which were to accompany Part I. Transect tables show dentata*, Cercis canadensis, Cladrastis lutea*, Clethra population levels (stand percentages) of trees, presence acuminata, Cornus aIlternifolia,C. florida,Fagus grandi- of shrubs, and presence of herbs along environmental folia*, Fraxinus americana*, IHalesia monticola*,Hama- gradients in the following sets of transects: melis virginiana, Ilex Montana, I. opaca, Juglans nigra, Tables 2-4, moisturegradient between 1500 and 2500 ft. Liriodendron tulipifera*,Magnolia acuminata*, M. fra- Tables 5-7, moisturegradient between 2500 and 3500 ft. seri*, M. tripetala, Nyssa sylvatica*, Ostrya virginiana, Tables 8-10, moisturegradient between3500 and 4500 ft. Oxydendrumarboreum, Picea rubens*, Pinus pungens*, Tables 11-13, moisture gradient in deciduous forests P. rigida*, P. strobus*, P. virginiana*, Prunus sero- above 4500 ft. tina*, Quercus alba*, Q. borealis and var. maxima*, Tables 14-16, elevation gradient in mesic sites, 200-ft Q. coccinea*, Q. marilandica, Q. montana* (Q. prinus), intervals. Q. velutina*, Rhododendron calendulaceum, Robinia Tables 17-19, elevation gradient in submesic sites, 200- pseudoacacia*, Sassafras albidum, Sorbus americana, ft intervals. Tilia heterophylla*,Tsuga canadensis*.

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions Monographs1 70 R. R.. H.WHITTAKER WHITTAKER ~~~~~~EcologicalVol. 26, No.

Abies froseri Acer pensylvanicum stems stems F FRASERFIR I FOREST I 90 89 183' I ~~~~~~~~~~~~~~~~~~~~~5O~~~~~~~~~~~HEATH 90 [BOREAL FORESTS]

/ TY P E: SE D S E TY PE 5- - - - 4 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~4

1 I F /\ 8 [E400; WASTERNFORESTS]

H1, 1 ',Si I',,--, "2 19 WA~~~~~~~~~~~~~~~~~~~~~.50 5~~~~~~~~~00 2 .9 .\

RD SPRUCE FORES q 1 ; \ / 0 s | l 35 / 0 4.1 RIc 34 4.6 ,22,1\ t24 IJ-

BOEAL TER LFORE[ORALSOESf |SHELTERED PEN RD3ES PEAR SHLEE |NE OEFEWU5 5 FDE15165 EARS COVES |FLATS ORAWS O S*LOPES5____ &i COES LATS DRA*S| N___ SLPE__ SE_ tNSERECA RW/INES | SU)PES |N 1W ? CANYONS RAVNE " \ - - -'- Tt X Ee___------| A___ ,, < X

PIN I / ~~~~~~~~~~~~~~~~~~~ ,/~~~~~~~~~~~~~~~~~~~ Acer2500- rubrum2500~I -I EBATN j |Acer saccharum

I PINE .7 ".8 ,"'15.9 14.4/ I PIES.I 10.1 " 6 '1 2 / M~~~~~~~~~~~~~~~~ORNESN 1500- 1500- / ' stm i \ i4 stemsX / *4 /iSI | ; 5I\ % 0 i 4 J2 0 TH IMPINE-- XI9

1 HEATH | ; 7 ,/<5.9 14.4S4X~~~~~~~~HEA.S 4X

V F 4J~~~~~FHAHIFEAH

<5- HE~~

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 71

Acer spicatum Aesculus octondra stems stems 650 65 600( ~~~~~~~~~~~~~~~~~~FEBATH1600HEEATH-1 OREAL [BOREAL FORESTS] FORESTS

/ SS(T 13.9L ) / S/ \ 55?+0A 1 // 17 17 /RAM|0}f,#. ? 55TM/ // 31 - Z A #BEECH,7 IFOR / 2 T4E FORS / / 50004X20/ 5000YP TYE ,'SIC / 10.9 T S~~

_ 8 _ - / , _ _4Q-- / - RE X X 0 S , z RNE / TABLE 1 1 / / ! 2 IABLE 1EE / X ' MOUNTAIN F 1 n / n //}F, A ~~~~/ I MOUNTAIN Z - 20 18.5 ~~~~~~~~7,",' 2 ~~~~~~PNE 11.0.3",'.3INE I 1./ E 5 / I HEATH 45000 CANYONSI RV I SOPE N WS SI CANYONSERAVINESASLOPESTNHNW SE - 3500 / / _

PITCH PITCH 3 30001 1.2.2 / ~' .3-2 5.3 sf PINE II \ ~I IN

HEATH I HEATH / ~ ~~~~~~~~~~5~ - 2000 J V~8 ~I ViMIA PINE PINE I FOREST I FOREST

00 EC OIET / ! 250004Betula allegheniensis25M} Betula lenta stems stems37 < AL I~~~~~~~~~~~~44 _7 /AL H~--~ 650 650 I FET1FHEATH]ATH ~~~~~~ FWE ~~~~~~~~~~~~~~~~~~LBALD] FORE~~~~~~~LAD]W0ijOEAJORSS 6 BREAL EBRELORRESJ 1 5 / 1 550 /

12 80 I FORESTS 5 Ilk~ HIFORESTS "' ~~~~~~~~~~~~~~~~~~~~~~~~~00BEECHEDETP 24 MESICT ~~~~SEDGETYPE ' &MESIC TYPESEGI TP

Z 24 TABLE ZTAL 24 I '~~~~~~~~~~~~~~~TA LE II ' to2.,' 4MOU~NTN MOUNTAIN A 12 / ~~~~~~~~~~~~~~ ~1 2 9.4 ~ 'Il7' EAHHEATH I

PITCH / / ~~~~~~~~~~/ I PTC 13 2.7 ~2.4 .8 1 300 3. /9 . 3 1 .2 I, 3000H 3.6 PINE I / ~~~~~~~~~~~= I PINE HETET / / ~~~~~~~I !R ~ HEATH / I

.3 2000- I VIRINIA 2000 4.01 VFONI PINE~~~~~~~~~~

FOREST ORS

______N5LOE1RDES2EK OE SLPS ~~~~~~~~~~~~~~~5 ~ COVES FLATS DRAM SHEL-TERED RIDGES& COVES IFLATS DRAVA SHIf.rERED INE RIDGS& E PEAK~S NES SOPES RAI RIS ? ___ ESIC-CA~tVNS RA~iNES SLOPES?N WRCNYNS

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 72 R. II. WHITTAKER Ecological Monographs R. II. WIIITTAKER ~~~~~~~~~Vol.26, No.I

Carya cordiformis Carya glabra stems stems 6s 6s~ 6oae- [B 1600C - [AS DHEATH] HEATH]

[BOREAL FOREST] LOREAL FORESTLAL

ssa , so 5500550/ ~ ~ * / / ~ ~ I, /10

1000- BEECHI FORESTS BBEECH I FORESTS /,// MESIC TYPE:SEDGETYPE / \ MESIC TYPE:SEDGETYPE / /

ZF~~~ ,TE;/ 1 , | ?i45 /------AE//

'' ?F48 , // ~~~~~~~~ABETABLEw { I M0UPNINTEAN 1 1 ?4XL , W //X-_ X//HEAT/ 1 o T / // / / MOUNTAN MOUNTAIN /4 x / I / PINE 7,..z PINE 1 7tX // I; \ Q 1\ W | HEATH W HEATH

| ? / 3000- 30i 1.5 1 P MfSIC RAVNE/4 | S~)PS N NS I P IN NSI PINE / Iw 350

2000 / .PICH5 200 / 3 .5 /IG AVRII PITC II /*iI I HEATH1.5IN /6I~~~2 ~I.7/ PINE

-- I i - I I 1500 \ PINE ~~~~~~~~~~~~~500COE-LTSFRS ~ 1.0 1.0 11.5 . 1 5 5 2 COVESANOSIFLATS DRAWSISHELTERED O'N ~ RIDGES& PEAKS CVS FAS DRAWSISHEL.TERED JNE NSLOPESj REIGES&PEAKS CAYN RAVINES E I SLOPES NEN AWVES CANYONS RAVINES SLOPES WEEN ASE -- - 6500 _lMESIC- ? ?- -? --SERIC I____2500 _ESIC- ? - ? -?SERI

ssoa 1/I8 ~ ~ OEi ~ ~ HEATHso C9R u ~~~~~~~~ ~ ~ uB~A~ Ofi ~ ~ ~~~~~~~~~~ 2 ~1 // /1 / 25 / Carya ovatis Carya tomentosa xwtstems 6EECH|FOREST3 / f t;~EiEECH|FORFOREST \zt | 52 stems FOREST

FHEATiH1 [ ] LBAI.QJ ~~~~~~~~~~~~~~~[BOREALFORES-jLBLD

/ A/ I / ,/~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~..~~~~~~~~~~~~~~~~~~~~~~//~~~~~~~~r - IFORESTS / BEECHIFORESTS 5000 ~~BEECH ,- ' 00 - / MESIC TYPE SEDGETYPE / ~MEDIC TYPE SEDGETYPE . /

I -,//~~~~~~~~~~TBL I TABLE

?4000 /7' MOU'AI/

- 3500~~~~~~~~~~ 350 - ~ A?/

,',, 1/ MSUNTAJN~~~~~~~~~~~300- , 4, PI/OUTCIN ' A -'' I I ~~~~~~~~~PINIE " . ' I PINE HET HEATH

I PK I~ wow~~

-'tel ~~~~~FRSTI FOREST

1 190,.. 15 COVES FBIOjLATS DRAVAj SH9ELTEREDIOE A CANYONS 15 -5 E FLATS DRAWS SHELTERED OPENI:nL:S PRWES& PEARS CANYON PWANI SLOPES RAVINES SLOPES NEN RWS&PASCORESCAY S INENEWS SW tE~~~~~~iC- ? ? ? ?~~~~~~~~~~~-R-ERICI .IESIC ?? ? ? -XERIC

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 73 Ja-nuary, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS

Castanea dentata,dead Cladrastis lutea stems stems 6500 6500

[HE 0 FHEATHI LBAD 1 [EBOREAL FORESTS [BOREAL FORESTS] 10 2 0 I 10 2 1~~~5 I/

FORESTS BEECHI FORESTS - ~~~~~~~~~~~~~~BEECH". 0 5 5000 MESIC TYPE SEDGETYPE / ,' t 50 MESIC TYPE SEDGETYPE ,

- - -- // .4|500------/ / z4|i / -- / f 1TABLE 1 ?41 / MNT 1 W/ // F '0vG /F 1MOUNTAIN '?(T / I ,1/ 4 , S AI / / I A /.1 1 1/ 3 PINE 4000 .4 I '/ I I HEATH IC J71.33 2.. HEAT, i

F~~~~~~ PITCH

3000 3.3 11.91 11.6 30T E , I 1 3EECHEF9 //t'///B H 1 35t /;,//'<;e~o//t1RTCH1135 / 2 .2 \ 5 F 2 2 tA | HT1 I8 | /// 2S H, .2/ t NEA

.8 6 4 I I. 6\ I R~~~~~~~~~~~~~~~NIA 2000I ~~~~~~~~~~~~~~I RGINIAPINE PINE i~ ~ FORESTT - 150- 35001500 I ~ ~ 3500 FLT DRV PEK OPEN & PEAKS & FLT DRAV.'-'~''E E SL-OPESW I~S RWnB(1 COVES SHETERED RIDGESRO II /A FLATNES Tw4S |NE CANS //.IES SL OPES -| COVES3. , I ~ IAVW~~~~ CBRA FOES / 2. \ /BRA IOET

Ctethraacuminata Cornu's florida stems stems 6500- 650 FHEATH] FHA1 20 EBOREAL ]FLBESDJ .2i- 1lfhi250 [BOREAL FORESTS]LAD8 / T 5 !5/ t I FCET1/0 iV55 7 /z \ 55M X- ~~~~~~~~~~/ ~ ~~

I FORESTS /~ BEECHI BEEH 5M - ~~~~~~FORESTS 5000- CZ WISI TYP'SOE TYE C MESIC TYPE SEDGETYPE

- -_ _ _ L_ . - - -ON - SLPE--- ^E "' ~~~~~~~~~~~N ~5'--NWSE ~ SW| ~ NEI I --SEIC---j -_ __R_ ESI,-! / - TABLE Z K.' ~~~~~~~~~~~/ I Z/ ,, MOUNTAIN ~~~~~~~TABLEUMOUNTAIN PNE// //PINE 4000 4000,/ 4 1 ,Z HEATH '' I HEATH I ~~~~~~~~~I 4? t~~ /// t~~

I I PINE PN HET HEATH

- 4 3 2000 3 1 V~~~~IFIGI~NIA 8.00-/ ~ 73 V~3 ~~~FORESTI ~~~~~~~~~~ FRS PS LOE OMREAST FLT DAA S ERD RE DRV IEOE W SWRELPEFLAS N IonCAYN AKCVSCWONS ME LPE PAMNE SHLTPERE PEN IRMNESSE MSICs-SLOPES N- MN-??E

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions Ecological Monographs 74 R. H. WHITTAKER Vol. 26, No. 1

Fagus grandifolia Fraxinus americana and stems var. biltmoreana stems

6500-

FHEATH FHEATH LBAD BOREAL FOREST REAL FORESTLD 20 50 75 75 50 20 10 5 1

FORESTS / / BEECH FORESTS 5000 ~~~~H 6. SIC TYPE SEDGETYPE / '.

I ; TABLE Z re6 /I TABLE.; ' / o / -40 MOUNTAIN 4000 MO/UNT1HEAINTH 1 ;=4000/-/56XI SI E I i 1 PPTINHEPINE | W .5 /I / 2 PINE ; { t lI Q I HEATH ?CZ # \ WT1/IHEATH t

3 / PITCH 3 . 2 ~PITCH ..~~~~~~~~~~~~~~I~~~~~ R*INES SLOPES N N SE SW CANYONS RAVINES SLOPES PINWSEPswlIcNE HEATH 1Q |7x O / 1G 2000L~ /Q ~ | /F !MGNA1 5 8\ I l\ \ I fRS / w h| t e Z a | \ \ , fOREST \ ; i z~~~~~

VIRGINIA~ ~ ~ ~ ~ 0~. IV~I

T white AL FORESfflFOREST FORA 150( _ I_ _ _ _ _ ' IIS0C - I I \.I PEAKS COVES S FLATS DRAWS SHELTERED OPEN SLOPES RDGESI PEAKS COVES FLATS DRAW SHELTERED OPEN SLOPES ROMES& E S CANYONS RAVINES SLOPES NE W CANYONS RAIVNES SLOPES N NWSE ESIC-K-- -RI(?ESIC------f ?R

- - 2000 [ 1.3 .9 2 [E VANA- Halesia monticola Hamnamelisvirginiana stems stems 650

FHEATH FHEATH - LBALDI [~~~~BOREALFORESTS] E00 BOREAL FORESTS] LBALD]I

1/ 5500 hit 1 f, // /5 I .3 4~~/ / IFORESTS/ / BEECHIFO ST ,'0- BEECH 5000 MEESCITYPEPTYPEEDGE/IC TYPE TEYCTYEPEEDE TYPEYP

4500 4500 ------~~~~~~~~~~~~~~~---/

/ / TABLE TABLE .Z ~~ ~~~IMOUNTAIN o/3 23/MOUNTAJN I HEATH ~ 1HEATH

300 4. i. ,ii 69 22 ~ I PITCH 30 2.6 / 3.4 1. PITCH tR HEATH tR HEATH

5 1 Ii~~~~~~~~~~~~~~~

COVS FATS I DRWS HELERE SLOPES R~GS&PMLPESTPFOREST&PE WS FAS DAS SETRD ~ O EK C ANYONSORAVNESOSLOPE ~SLOES RAVINES CANNON SOPESSNSLOPES W E W SE MESIC- ???? ~-XERIQ I MESIC ? ??? -CR

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 75

Ilex opaca Liriodendrontulipifera stems stems _ 6500 6500_

HEAT ZHEH

B[OREAL FORESTS LBALD [OREALB FORESTS] LBADJ

5500 - ~~~~~~~~~~~~~~0 550/

MSCBEECHFORESTS / k'5000BEECH 5000 IFORESTS ,/ II X // 5000 S MESIC TYPE SEDGETYPE T/ W UMEDICTYPE SEDGETYPE

--- ~450 / ATMli 450 7--- / l , / TABLE // I TABLE z ' , / ' / / MOUNTAIN 7 // ,9 ' 3 / MOUNTAIN 4000 ~ ~~ ~ ~ ~ ~ ~ ~ ~ // PINE / ''PINE / 7 /I <,,HEATH | I HEATH

3500I 3500

. I / I / 21 PC / F/ PNE ] 3 4 .1

2500,In- 7 Is 2500I

.3 2 VIRGINIA VIRG3INIA12\ ~ PNE PINE I FOREST I ,z. ~FOREST IS0 I I1500 6 COVES FLATS DRAWS SHELTERED OPEN SLOPES RIDGES& PEAKS COVES FLATS DRAWS SHELTERED OPEN SLOPES RIDGES PEAKS SLOPES SEGE CANYONS RAVINES SLOPES N GESW M CA RAVINES ----- EN -NKW ? - -SW ??------NW IESIC------? - V-ERI' ____ ESIC- ?--XERid

Magnolia acuminato Magnolia fraseri stems stems

FHEATH rHEATA1

[BOREAL FORESTS] CDj [BOREAL FORES-fl LBALDJ

/ j 5Wssoo Ij // // 1a,, sstj~~~~5w

BEECH I FORESTS EECH I FORET MEICEDICTTYPEE / . SEDGETYPE B

LZ - / ITABLETAE ZRBPE ~~1 ,' /~~~~~MOUNTAIN MOUNTAIN I PIE.N 3. 9 PITNE 34000 .33 ,' C / i I 1l HEATH HEATH

*~~~~~/ I ~~~~~~~~~~~I PITCH 3000 -30 14PITCH1 3 3 I 0 I PINE 00 0 5/ .5. I I PNEPIN 2 .5fES FLATS HRAWSSHELIERED OPENSLOPES RDGES& PEATH /NES FLATSDRAZI SHELTEREDEN SLES RKES&PEATH

8 ~/7 'I|*\ -s 8 mE 1 18 | 21 s

.8 I7 32000 2.1 .9 VISRGINIA 'nNI2 PN PINE N FOREST FOREST

FAS OPEN SLOPE~SRIDGES L PEARS COVES IFLATS DRAMR SHLERED OPEN SLPS RIDGES& PEARS CORES DRAWS SHELTERED E CANYONS RAVINES SLOPES NEE S MEGEORCNNS---NES SLOPES N I( _E ME MES/C- ? ??---GEP-IMEIC? ? RI-

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 76 R. H. WHITTAKER Ecological Monographs

Magnolia tripetola Nyssa sylvatica stems stems 6500 6E0EE

R-HATH1 rH-EATH-1 LB LBALDJ [BOREAL FORESTS] J 60[BOREAL FORESIS

- 5-5( 5500

I / BEECHI FORESTS 65100 1EECHFORESTS 2. 500t-SMESICBTYPE|SEDGE TYPE , MESICTYPE SEDE ETYPE

L._------.'/ " \ \ | 450 _------

X / 7 / ; TABLE /UN // 4 1, TABLE 1 - - K //-' //- MOUNTAIN 0- - --- 4 3. i I PNET3.4 4000 ~ ~ ~ 1 ~ / PINE MOUA. I I HEATH . ///I HEATH

PITCH 3000 3000 .4 NE I I PINE /PI HEATH ~~ I ~~~I~~~~~~~~I I~~t HEATH !R I - 250 .4 /2500

1 2.6 24 1 55 ~~~~771 ~~~~~~~~~~~~~~~~~~~~~~~~~~~ .5 8 .5 V~~~~~~~IRGAINIA 2000I 2. PINE

1500-10 !.5 PNSOE COVES FLATS DRAWS SHELTERED OPEN SLOPS RIIDGEL PEAKS COVES FLATS DRA~ISS SHETERED, EK CANIONS SOE WS W XR EN ??BR MESIC-l--- ?

Oxydendrumnarboreum Picea rubens stems stems

650 FRSRFR1FOETl

FHEATH S(XI S I IJI~ ~~~I I 10/ I D 5HoE IE [BOREAL FORESTS]JLALJ~0 / X NS|R41S/ ,1/1 LOESFNI/SRNSR E IK 5EE SIRE - 550/ / I0

4 BEECH FORESTS /'50021030 5000 ~~~MEDIC,TYPE SEDGE TYPE - / I I / ~~~~~~~~~~~~ 10 ItO~~~~~~~~~~~1

FORIESTS] z N ~ ~~~~~~~~~~~~~~~~U4 [EASTERN 0~~~~~~~~ .1 57 138 157 1 i~~~~~~~TH

.8 /~~.4' 11113 I ~I2.71 6.4

-BER ? -- -

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January,1956 VEGETATION OF THE GREAT SMOKY MOUNTAINS 77

Pinus pungens Pinus rigida stems stems 6500_-50

fHE~~~~~~~~~~~~rH-j 6000- ~~~~~~~~~~FHETH1LBALD] 6000_ bLB&D [BOREAL FORSTS] [BOREAL FOREST]

550(_50 -/ / 5 I {' / // \k I ~~~/ /\ BEECHI FORESTS 1500 MESKCTYPESEDGE TYPE ME/2 BEcTYPE |S SICT SEDGETYP ME't /// / / e ; X q ~ ~~~~ >

4//X1 3500 35004359// HEATH'| MN / 8 I 8 / I 4 | H t/N S o / 8.TH i iz/ w ~~~~~~~~~IF I.9 j' I It' / ! '3 1 [ (;1 4MESI----AS - SLPE N4WS5W--00EI AYN AME LPS IN WS W E

.1 I~~~~~~~~~~~~~~~~3

40ins0tobs0300 < cPINE Pin400r0nan3000

2000 - S AVNE SOPSVIRGWS IANONNRVIES SLPE N S

2 | 2 [% 2 | 2 [E~~~~l 55CX l, / I / \\ 55CS~~~~ //I !

u- FOREST~~~~~~~~~~~~~BRELFOE~

FORESTS LOE , COVSBEECH SORESTS RD OPN IDE PEAK COVES FLTSEECHV IEPE RIDGEPA K DAWI NE I 1 / 'O1 I 8 CAYO S IRATYPES' {// SEDG ------'PESy / 'w I 1 12 MAYN /W/AIES TP LOGS TPES

E W I N W SE/ ~ N - ,; N~ /

z50 - //05

BEECH 5000 - BEECH FORESTS 5000 ORESN M I TYESEG TYP ~ M /I~~~~ /YESDETP ~~~ ,11-*

7' ~~~2.8 ~~~7.l 2.8 412~~9400 .6 98

LO LORAIE LPE i E W S AYN SOE

E35-0-0? XEr ___lEI-?3-5-? -E

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions Ecological Monographs 78 R. H. WHITTAKER V ol. 26, No. 1

Prunus serotina Quercus alba stems stems 6500-_

FHEATH-1

[BOREAL FORESnj [BOREAL FORESTS] 55A 1 5 / I'~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 5-ca - ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ I I550 I

.3/ I I EECH FORESTS / 500 - IBEECHFORESTS 5000 B I MESICTY1E SEDGETYPE ~ EMESIC TYPE SEDGETYPE / '

-- -E-S-I-C--7-/------2 a ABU TABLE /00 "/ MOINTABI MOUNTAIN | 40.0 EI3ET L 0 -20 19S 7 PINE aXT /' R}/7WX/EATH 1i --HT'B 1 ,'- / B HEATH

1 35000- I ,''s/Of,//O\/v A \4 LB I7 ,B | /HAHEr N 350101n t I / ET 1.0 /[ EE/PN PITCH 3000 2ITCH 23 ! 0 2( 0 j HEATH IR HEATH

" 'I '~~~~~~~~~~~ ~~FCSAVIRGINIA /11 I

I PEREST ~~~~~~~~~~~~~~~~~.. ~~~~FOREST I 5~~~~~~~~~~~~~~~~~~~~ PE M OPN RIDGES CI)VES FLT R1IS SH EE RO EK COVES FLATS DRAVA SHELTERED N SLPS &PEAKS CANVINS JNL MHY, SjISL FES E IN S5 CANYONS RAVINES SLOPES

Quercus borealis and Quercus coccinea var.W; maxima,4 stems 0 SS 0 2z = V RNE1 1 X stems# ~~~~~~~~~/ // )

FlHEAO1] 6000- E~~BoREALFORESTS] [BOREAL FODRESTS] LBALDJ

| C YON j AMN | SOPE IN N NWWSE SW|CI INPNKENW I ANYNS J RAYNES I SOPE ? PINE 18 t ~~~~~5 10h/ : % ~ / /

BEECHI FORESTS I ' 5000 ~~~BEECHFFOR SO - MEOC TYPEt YE,. EI TYPEISEDGE TYPE -

- 35M~~~~~~~~~~~~~~~~~~~~~~~~~~.L-- 3500 ' - 3000- i / N-- 3000 // 5~ .8 .3 .0 3.0 11 2.2, PINE ~~~~~~~~~~//P3.& 1 IE? ~~~HEATH //.'~~~~.//

/ PITCHI~PNI /4 ~~~~~~~~~.2 35

25 I -'.1~N-N I VIRGINIA / ' \N35 7.2E -%EI

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions January, 1956 VEGETATION OF THE GREAT SMOKY AJOUNTAINS 79

Quercus marilandica Quercus prinus stems stems

6000- EiwCBOREAL FORESIS LBEALN [BOREALFORESli]BALD

,/ / ssoo ~~~~I ~ ,, I,ss' ,- /

BEECHIFORESTS BEECH FORESTS / , /? , 5000 TYP MSCTYPESEDGE MESIC TYPE SEDGETYPE '9 ~\ L * ?- - / X l 4 .. ~~/ / I / TABLE

;=4000- ~ / I /1 6I IN 40 PNE

' HEATH HEATH X (E { / / 1 1 /\ 3

2.5 .|7 1 2\ l8\ 8l7O .6 X ;wBE''\'\'' ''!QE~1 fLATS SHELERED CPEN s ROGES& PpFS S OPEN COPES ORAWS COVES FLATS DRA HSKLTERED SOE RGS P N N ?S - - C FLATS HE ---- F E LPS SLOPES RR S ORAN N SE SW I y w ~~~~~~~HEATH]

Quercus velutina ||Robinia pseudoacocia 6..~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~41 _stems5 BEECHIFORES ~ ~ ~ ' '~t; stems BEECH FORESTS I /l 65 6500

T We - ITO6BAL6 2000 AL F 6000 [BO R E A L FO R E S TS ] CB ORE OR E S T S ]

CAYO NE CANYON RAN I MESCTYPERSESSEOPE T EW ;; I SEIEOPES IN PSE '

, , e,,, , g ze-----__;t/s -1 N'------/ - ---X' ffiz ~ \ ~ aNER] ( -2Z I 5 I /EI TYPESEG TP N WS C-.S MESI/ T//S EGET3E.NONWS[BSLCPES SE /OPES

I PINE T/. I / ~~HEATH IT /- BEECHIFORESTS ~~~~~~~ BEC FORES T S.0 ~

/iI TABL B @,, \ | X XE I\ X W> 24 1.81I. I .2E " '.I25 P42 2

,.6 2.9 \ 2. '20 1.0

COPES FLATS OPN ROMGEL PEAKS COVES FLATS ORAVA SIELTERED OPEN SLP RIDGESL PEARS DRAWSI SHLTERED %SLO E CANYiONS RPEINES, SLO3PES HE E W CANYONS RAVINES SLOPES I SESI M ? --- ~~~~~~~~~N --SERBMESIC -?-----N E- - --E

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions 80 R. H. WHITTAKER Ecological Monographs

Sassafras albidum Sorbus americana stems -stems

FRASERFIR IFOREST 8 8 311

[BOREALFORESTSJ E

CRBEECHFORESS .

R-EASTRNEFORESTS-

- / 2- _ / 9 4 g? ,/ I HEH "EL| / / >12~~~~~.0 41g-{e r 4

3000- .9 T ~~~~~3.2 3000 MEKTPESDETP ',------/I 5 IL------RI

2000I / T VI 1AIA0

I EST 5 ITYPE SEDGETYPE -XI~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 5OMX 0~Y 0 A X I~~~~~~~~EIISO 1E ISOPEN 'OI MML PEAKS COV/ES FLATS DRAWS SHELTERED CPEN SPES RIDGES& PEAKS COVES TLAS DRA SHEIERED OPEN SL SE SWI CANYONS RAVINES SLOPES N E W S CAN RAVINES SLOPES N W

F1.T ,/ I 9B ,.7/g | .5 4 l l | Tilb30+ 18.t/heterophyllo, / .8 5,?l / w etc. [ 8 RTCH S | 30+Tsuga canadensis stems stems 6 85M

FHEATH FHEATH ww- ~~~~BOREALFORESTS] EBL]600-BOREAL FORESTS] 5BLD

5500~~~~~~~~~~~~~~~~~~~~~~~~~~I' 550-

BEECHI FORESTS / ' TYPE~~~~~SEDGETYPE7 ,/ '~~~~~~~~~~~.'42cSEDE T u AESICXR-MS--lR -- Z 20 'I Z TABLETAL MOUNNTAINZ~~~~~~~~~~~~~ / MorN ;=4OD() /'2 - 0 ~~~~~~~~~~~~~114 ~ 50 I1 HEATH/ HEATH

PIE IN tz I I 2 EATH 2 / I/ 'HAT

8 2500 ~ ~ ~ ORS

SLOPES I N ~~m (.AFORNSTSLORES ------?-KERI ___ ~~~~~~~~~~~~~~~~~~~~~~~ES~~~~-? ~ C

This content downloaded from 198.21.188.189 on Thu, 03 Sep 2015 19:05:27 UTC All use subject to JSTOR Terms and Conditions