8 November 1974, Volume 186, Number 4163

Landscape Development, ForeFires, and Wilderness Managemei

Fire may provide the long-term stability neecto preserve certain conifer forest ecosyster

H. E. Wright,

The major components of the naturallandscape are landforms and vegeta-tion. Their status at any moment is aresult of a complex of interacting fac-tors operating over various lengths oftime. These dynamic systems attainquasi-equilibria lasting tens or hundredsof years but show sequential develop-ments over longer times. Thus, theform of a hillslope represents a short-term equilibrium among processes ofweathering, erosion, and vegetationalstabilization under conditions of crustaland climatic stability, but in the longterm the hillslope is reduced until theexternal controlling conditions change.Similarly, vegetation may progress in ashort time through successional stagesto a climax, but in the long term ex-ternal factors may control the sta-bility.

Recognition of dynamic landscapedevelopment stems from interactionsbetween geology and biology. Lyell's(1) synthesis of paleontology andgeology in 1830 helped stimulateDarwin (2) to develop the theory oforganic evolution in 1859. Evolutionarytrends were then sought in other sci-ences. Davis (3) introduced the erosioncycle in 1889, in which youth is fol-lowed by maturity and finally old age,which persists until the cycle is re-initiated by crustal uplift. In plantecology, Cowles (4) proposed thatvegetation stages parallel those of theerosion cycle. Clements (5) likened

8 NOVEMBER 1974

vegetation developmentof an individual.

These theories of lanement have been severelycriticisms are based larltative analyses of the rrat least the short-termmorphic and ecologic pever, the schemes ofDavis, Cowles, and Cllengths of time that z

evaluate. For all evoluticextrapolation from mcpast rates must be basecof past conditions derivetative paleontology, strgeochronology.The genetic mechanis

is now largely understgenetic change and specnisms such as the fruitdetermined, but directsuch rates to other anijustified. Particle movemand in streams has betensively, so that ratesbadlands and of sedimestreams are known, burates to solve problemrlandscape developmentShort-term early plantnewly exposed landscapdata on nutrient suppl)bilization, are documentational development ondecades and centuries ipredict.

SCIE:NCE

Landscape processes can be studiedin a wide variety of situations. Evenhighly disturbed landscapes can pro-vide some insights. Often experimentalplots can be designed in which con-

St trolling factors can be manipulated.However, studies of long-term land-

nt scape development can be carried outonly on natural landscapes, where thecontrolling factors have been operating

led for hundreds or thousands of years.Large-scale and long-term ecosystemprocesses can best be studied in largevirgin forests.

Jr. Preservation of some natural areasmay require only that exploitation becurtailed. Elsewhere, certain naturalprocesses that have been eliminatedmust be restored. Specifically, in cer-

to the life cycle tain conifer forest areas, including mostof the national parks and wilderness

dscape develop- areas of the western and northern( criticized. The United States and Canada, forest firesgely on quanti- have been effectively excluded for atiechanisms and least 50 years. Yet recent researchE rates of geo- shows that the maintenance of natural)rocesses. How- conditions in these forests depends onLyell, Darwin, frequently occurring fire. Continued[ements involve fire suppression will upset the eco-are difficult to system balance and lead to unpredict-)nary processes, able changes.)dern rates to In this article I evaluate and recon-1 on knowledge cile certain concepts of landscape,d from quanti- developmenit by clarifying the time^atigraphy, and scales for which the various concepts

are applicable. Particular attention is;m of evolution devoted to problems of vegetationaltood. Rates of succession and climax in the cutoverc:iation in orga- forest of the Appalachian Mountainsfly have been and the fire-dependent virgin forest ofapplication of the Boundary Waters Canoe Areasimals is hardly (BWCA) in Minnesota. Inasmuch asient on hillsides many areas of virgin forest, includingen studied in- the BWCA, are part of the nationalof erosion in wilderness system, management must

:nt transport in understand the effects of disrupting theit use of such natural processes that have brought thes of long-term forest into being. For the BWCA ais hazardous. new management plan has been adopted

successions on (6) that not only ignores crucial eco-es, along with logical principles but also contravenesZ and soil sta- the spirit of the Wilderness Act (7, 8).kted, but vege-ver subsequentis not easy to

The author is Regents' Professor of Geology.Ecology, and Botany and Director of the Limno-logical Research Center, University of Minnesota,Minneapolis 55455.

487

on

Oct

ober

5, 2

013

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Oct

ober

5, 2

013

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Oct

ober

5, 2

013

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Oct

ober

5, 2

013

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Oct

ober

5, 2

013

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Oct

ober

5, 2

013

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Oct

ober

5, 2

013

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Oct

ober

5, 2

013

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

on

Oct

ober

5, 2

013

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

The Erosion Cycle

In the 1880's Davis (3) developeda grand and integrated theory of land-form evolution: the sequential erosionof newly exposed land surfaces throughstages of youth, maturity, and old age,under conditions of stable crust andstable climate. The end form is a pene-plain, which persists until crustal up-lift initiates a new cycle (9).

Davis's theory of the erosion cyclewas conceived for humid temperateregions, for which it received wide ac-ceptance. He later extended the theoryto arid regions (10), but tropical andpolar landscapes were not easily fittedinto the model.

Penck (11) was the first to offer anintegrated theory of landscape develop-ment based on premises different fromthose of Davis. Davis assumes crustalstability during the erosion cycle, butPenck believes that crustal uplift pro-ceeds more or less apace with erosion,and that landforms reflect the relativerates of the two processes.A cornerstone of Davis's erosion

cycle, the concept of the graded stream,was elaborated and effectively defendedby Mackin (12), but Leopold andMaddock (13) showed that streamsbecome graded very early in their de-velopment, not just at "maturity." Per-haps the most comprehensive criticismof Davis came from Hack (14), whodenied the evidence for long intervalsof crustal stability and thus for thedevelopment of peneplains. Fromstudies in the Appalachian Mountains,Hack concluded that the characteristic"ridge-and-ravine" topography does notresult from uplift of an older peneplainor partial peneplain followed by ma-ture dissection, but rather from a dy-namic equilibrium, in which the deg-radation processes of weathering, soilcreep, and stream erosion are opposedby rock resistance and crustal uplift.He pointed out that differences in theseprocesses or factors can result in slopesof differing steepness and thus in land-scape of differing expression, but heimplied that neither the crust nor sealevel (which serves as the erosion base-level for stream systems) is stable longenough to permit the landscape to belowered to a peneplain.Arguments for erosion cycles versus

noncyclic dynamic equilibrium revolvearound the time factor and the rates ofgeologic processes. The real ages oflandforms are difficult to determine be-cause of lack of radiometrically datable

488

material. For the Appalachian Moun-tains, the chronology of past events andthe rates of geomorphic processes arenot well enough established to permitus to evaluate whether the landscapeever did or could reach the peneplainstage. The enigmatic flat crest of partsof the Blue Ridge, strikingly discordantto rock structure-a type of landformnot specifically discussed by Hack-suggests the end form of a past erosioncycle, but neither it nor the basementsurface beneath the coastal plain sedi-ments is extensive enough or wellenough mapped to be identified withcertainty as an ancient peneplain. TheAppalachian Piedmont is also a dis-sected plateau with discordant struc-ture. Its identification as a peneplain byDavis was defended by Holmes (15).The debate has reached an impasse.

There is as much (or as little) histori-cal evidence of Appalachian crustalstability with occasional uplift as thereis of uplift that nearly keeps pace witherosion. Shifts in baselevel resultingfrom Pleistocene sea level fluctuationsdid not affect streams far enough in-land to control erosional processes inthe Appalachian Mountains. The in-effectiveness of Pleistocene baselevelcontrols indicates that Appalachianpeneplain formation was much olderthan Pleistocene, perhaps as old asCretaceous, 60 to 100 million yearsago. However, the stratigraphic recordis inadequate to solve the problem.Schumm and Lichty (16) attempted

to evaluate the time factor, and con-cluded that the controversy resultsfrom analyzing geomorphic processeson different time scales. The processgeomorphologist may measure the fac-tors in a dynamic equilibrium, butextrapolation of an erosion rate todetermine the age of a landform maynot be justified because modern ratesmay be unnatural, owing to human dis-turbance of the system. The historicalgeomorphologist may be able to evalu-ate such extrapolation over a shortterm if a datable sedimentary record ofthe erosional process is available. Butlong-term evaluations are difficult be-cause such geomorphic factors asclimate, uplift, and bedrock resistancechange independently.One can conclude with Carson and

Kirby (17) that evidence for long-termcrustal stability in certain regions issufficiently strong to support the con-cept of the erosion cycle and the de-velopment of peneplains, althoughmechanisms of slope retreat and de-

velopment of stream profiles may notbe as simple as Davis envisioned. Muchcare is necessary in historical analysis,not only of the structural and hydro-logic controls on the development ofterraces and other landforms, but alsoof the time scale and the rates ofgeomorphic processes.

Vegetational Development

While Davis and his followers usedthe erosion cycle to interpret landformsworldwide, Cowles (4) and Clements(5) developed the theory that all vege-tation progresses through successionalstages to a stable climax, which persistsuntil some interruption occurs. Cowles(4, 18) insisted on a precise resem-blance between the climaxes of theerosional and vegetational cycles.

Clements, however, deemphasized therelation to landform evolution, arguingthat vegetational succession is limitedby climate. He acknowledged that localbare areas produced by vigorous ero-sion are more common and providemore opportunities for early stages ofplant succession in youthful than inmature landscapes. But he noted thatsuccessional stages proceed to a climaxmuch more rapidly than Davisian ero-sion cycles reach old age. Also, a singleerosion area might transect variousclimatic belts with different climaxvegetation. Clements used the Darwin-ian model to postulate that the climaxformation actually is an organism,which is generated, grows, matures,reproduces, and dies.

Gleason (19) opposed climax theoryby introducing the notion that a plantassociation represents merely an acci-dental juxtaposition of individualsrather than a- naturally interrelated as-semblage of species. This controversycontinues today. Whittaker (20), inreviewing the entire climax conceptliterature, rejected two points: the con-cept of a vegetation assemblage as anorganism, and the rigid connection be-tween climax and climate. The notionof a polyclimax (21) or climax patternis still supported. Local ecologic condi-tions such as high soil moisture or dis-turbance permit the development of aquasi-stable plant association that re-produces itself short of true climax. InClements' view, the climax representsintermediate edaphic conditions towardwhich all landforms and soils eventuallytend.

Virtually all evaluations of the climax

SCIENCE, VOL. 186

concept are based on studies of mod-ern vegetation. Despite the use of newtechniques of measurement and statisti-cal analysis, results are inconclusivebecause the historical record has notbeen considered in the proper timeframework, and the vegetational rela-tions may have been significantly af-fected by timber cutting, agriculture,fire suppression, wildlife extermination,introduction of exotic insects, and otherunnatural factors. These elements aretaken into consideration in the follow-ing examination of the vegetation andhistory of Appalachian forests and fire-dependent Great Lakes forests. Espe-cially considered are the natural forcesin long-term stability, and problems inmanagement of virgin forests.

Forests of the Appalachian Mountains

The complexity of Appalachian for-est types has long been a challenge asan historical problem. Braun (22)found a solution by applying the climaxconcept to the erosion cycle. The mixedmesophytic forest type (beech, tulip,basswood, sugar maple, sweet buck-eye, chestnut, red oak, white oak, andhemlock) thrives on diverse topog-raphy, so it is said to have expanded

after the uplift and dissection of amid-Tertiary peneplain, perhaps 30million years ago. This and other climaxassociations are thus thought to havepersisted for millions of years. Theirdistribution and extent at various timeswere controlled by Appalachian erosioncycles. And, in Braun's view, they wereessentially unaffected by Pleistoceneclimatic changes, except for the south-ward infiltration of northern foresttypes along coastal plains freshly ex-posed by sea level depression, or alongyouthful glacial outwash in river valleyslike the Mississippi.

Deevey (23) discussed some unten-able aspects of this picture of Appa-lachian vegetation history. Braun (24)countered vigorously, but much of herstory collapses in the face of extensivepaleobotanical studies of late Pleisto-cene lake sediments, which indicatethat current forest types are only a fewthousand years old. During the heightof the last glacial period, 22,000 to13,000 years ago, not only the moun-tains but also the piedmont and thecoastal plain were characterized by anorthern conifer forest (25, 27). Sprucewas a dominant -tree as far south asVirginia and North Carolina, and jackpine down to northern Georgia (Fig.1). Alpine vegetation occurred on some

mountain crests (28). Components ofthe mixed mesophytic forest may havebeen present in small amounts-pollenanalysis is not a sufficiently sensitivetechnique to establish their absence.They probably survived in favorablelocalities in this area of diversifiedtopography and geology, especially inthe more southerly parts. They wereapparently not pushed south en masseto Florida, which featured an openvegetation with oak and herbs andxerophytic shrubs at this time (25). Inany case, northern forest probably pre-dominated in the Appalachians, withenclaves of southern types, rather thanthe reverse.For Holocene time, significant vege-

tational changes are also documented.From Virginia south, oak gave way topine about 6000 years ago (Fig. 2),when modern mesic forests becameestablished (29, 30). Some of thesechanges may be related to climaticshifts.

In the central and northern Appa-lachians, the postglacial pollen sequenceshows slight changes in the proportionsof hemlock, beech, hickory, and chest-nut, with oak dominating throughout.This was originally interpreted as repre-senting gradual changes of climax for-est types caused by a relatively warm,

QUICKSAND PONDNorthern Georg ia

l"T"rm

LAKE LOUISESouthern Georgia

cA\)

I . I

VI LL[. [ RLb)-8540BPr"--I rmrnr r r r r r-''r, r-r- r mr r O0 2 4 60

r r20 0 20 40 60 02 024 2

0 20406080 0 204060 020 Percent of total pol lenPercent of pol len sum

Fig. 1 (left). Selected pollen curves for Quicksand Pond, Bartow County, northern Georgia (26). The pine pollen type before13,000 years before present (B.P.) is determined from pollen size and needle morphology to be jack pine (Pinus banksiana),which now does not grow south of New England or the Great Lakes area. Fig. 2 (right). Selected pollen curves forHolocene sediments of Lake Louise, near Valdosta, southern Georgia (30). Before about 6000 years ago, dry forests of oak(Quercus) and hickory (Carva) dominated, with openings marked by prairie plants, including ragweed (Ambrosia), sage(Artemisia), and other composites and chenopods. The pine after 6000 years ago is probably longleaf pine (Pinus palustris).Various mesic deciduous trees, such as sweet gum (Liquidambar) and iron wood (Ostrya type), expanded about the same time,and swamp trees such as bay (Gordonia) and cypress (Taxodium) spread over the lake margin soon after. Beech (Fagus) appar-ently did not reach the area in quantity until later.

I

I

I

I

8 NOVEMBER 1974 489

Fig. 3. Pollen (46)and charcoal (43) LAKE OF 1diagram for Lake Of Northeastern tthe Clouds, north-eastern Minnesota,showing substantial iuniformity of region- aal pollen rain since XI i.,<development of the ' (4) k'Qpine forest 9000 F F H rrtiyears ago. The majorchange was immigra- -tion of Pinus strobus(white pine) about - l6700 years ago. In- 4crease in pollen of liPicea (spruce) and -Cupressaceae (prob- 60 'ably northern white -cedar) and decreaseof Pinus strobus oooabout 3000 years ago - lmay reflect a trendto cooler climate.The charcoal curveshows that fires have hoccurred for at least r

r r mt-r-r9000 years. Time 0 40 0 4cscale derived fromannual laminations.

dry interval 4500 to 1500 years ago,supposedly indicated by an increase ofhickory. However, the climatic re-quirements of the species are not wellenough known to support such conclu-sions (31). The changes might repre-sent delayed migration of particulartree dominants into already establishedforest, a phenomenon that Braun wouldnot accept. Knowledge of hemlock'sresponse to fire and human distur-bances is particularly lacking. Davis(31) commented that many maps ofpresent forests show not actual speciesfrequencies but rather theoretical cli-maxes thought appropriate by someecologists.

The pollen record thus indicates amajor climatically controlled vegetationchange throughout the AppalachianMountains at the end of the Pleisto-cene, about 11,000 years ago, as wellas significant changes in the middle ofthe Holocene. The present forest typesin the Appalachians are closer to 10,-000 years old than 10 million, indicat-ing that they depend on the climate,which has changed in major ways dur-ing the last 15,000 years, rather thanon the erosional cycle, which may nothave progressed appreciably in 10 mil-lion years.

Similar changes undoubtedly oc-curred during earlier Pleistocene inter-glacial episodes. Interglacial records insouthern Georgia and central Floridareveal a cyclic development similar tothe incomplete Holocene cycle (25).

490

THE CLOUDSMVinnesota

-r r rrnnr-rr r- rr r ' 'Percentoges of pollen sum 0 5 40 15

x1o5 p2/cm2/year

Vegetation changes in the Southeastwere fully as pronounced as those muchcloser to the ice front.

Braun's failure to relate vegetationclimax to an Appalachian erosion cycledoes not vitiate the climax concept forthat region, however. Braun simplymisconceived the relative time scales ofgeomorphic and vegetation processes,and she lacked knowledge of the mag-nitudes of Pleistocene and Holoceneclimatic and vegetation changes.To determine the time scale of Ap-

palachian forest development, we mustconsider succession and reproductionmechanisms that perpetuate climaxpatterns. Hack and Goodlett (32)studied an area near the ShenandoahValley of Virginia, where the vegeta-tion types in the ridge-and-ravine to-pography form a distinctive pattern.The pattern is of a definite polyclimax.This interpretation is supported bydetailed geomorphic work, which showsthe sensitive response of certain treespecies to moisture, slope, and soil con-ditions. Ridge crests and noses arecharacterized by pitch pine or table-mountain pine, ravines by yellow birch,basswood, and sugar maple, and inter-mediate slopes by the absence of thesespecies and the presence of certainoaks. Their study shows a dynamicequilibrium between geomorphic pro-cesses and vegetation. The crucial ques-tion now is: How long after the lastmajor disturbance-i 1,000 years ago-did it take the present forest to reach

climax or equilibrium? Hack (14) im-plied that the ridge-and-ravine topogra-phy has persisted virtually since theoriginal Appalachian uplift many mil-lion years ago. Rock weathering anderosion proceed at a finite rate, supply-ing and renewing the soils that helpdetermine forest type. The system isalso always losing a finite amount ofmaterial by solution transport instreams and groundwater. Even thougherosion rates have increased in manyareas because man has disturbed theforest and soil cover, measurements oflong-term erosion rates indicate thatlandscape changes controlled by geo-morphic processes proceed extremelyslowly compared to vegetation changescontrolled by climate. In small solutiondepressions in the Appalachians, forexample, very few meters of sedimenthave been deposited over tens of thou-sands of years, even though each drain-age basin may be several times largerthan its catchment area (26). Theselandforms have been essentially stable,therefore, while forest vegetation wentthrough drastic transformations con-trolled by Quaternary climatic change.The distribution and extent of poly-

climax components are thus controlledby geomorphic processes that providesteady nutrient input, but do not changethe topographic and hydrogeologic set-ting appreciably within the time rangeof climatic stability. Both the physiog-raphy and the vegetation maintain asteady state in dynamic equilibrium.How does the climax forest maintainits overall composition? What factorsand mechanism determine what treewill replace one that falls?

In the mixed mesophytic Appalachianforest, windstorms are the major dis-ruptions renewing succession. Old andinfirm trees blow over, exposing theentire root mat. The opening is rapidlyfilled by a microsuccession of plants inwhich young trees take their placeamong the old. Goodlett's study (33)of the mound microrelief in anAllegheny Plateau forest in Pennsyl-vania showed that white pine is relatedto the incidence of windthrow of oldhemlock trees. Windthrow opens thecanopy, giving light for the germina-tion and early growth of white pine inthe seedbed of loose mineral materialfallen from the upturned mat of roots.Windthrow is also critical in soil de-velopment. Local single-aged stands ofwhite pine are attributed to decimationof large sections of forest by tornadoesor hurricanes that cut across a poly-climax and initiate succession (33).

SCIENCE, VOL. 186

r-r-r

0

Cloudbursts cause debris avalanches onslopes and washouts in ravines (32).

Fire is another natural disturbancethat provides opportunity for regenera-tion in the Appalachians (34). The even-aged stands of white pine mentionedabove have been attributed alternativelyto regrowth after fire. Many pine standsare confined to sandstone ridge crestsand plateaus, where dry soils and windexposure are favorable for fire.The mixed mesophytic forest of

Appalachian hillslopes and hollows,where the forest floor is customarilymoist and winds cannot easily pene-trate, probably rarely burn. Many ofthe hardwood species resprout fromroots. Here windstorms are the moresignificant factors in vegetation disrup-tion. For most of the Southeast, how-ever, the historical record of fire is notadequate for quantitative evaluations ofits importance.

Although the role of fire and wind-throw in Appalachian vegetation devel-opment can be generally inferred,studies of modern forests are hamperedbecause of major disturbances by man.The extent to which features and pro-cesses are natural or artificial cannotoften be determined. Lumbering andagriculture have been extensive in mostof the eastern United States, and fewlarge blocks of undisturbed forest re-main. Even second-growth forests havebeen subject to disturbances such aschestnut blight, decimation of browsingwildlife or of predators that controltheir populations, and sheep and cattlegrazing. Although some New Englandold fields have had 150 years for re-forestation under fairly natural condi-tions (35), the effects of past farmingon soils and forest succession are diffi-cult to evaluate. Goodlett (33) showedthat the frequency of windthrow wasgreatly reduced after clear-cutting, be-cause no high old trees project abovethe regenerated forest canopy to catchthe wind. The modern forest conse-quently contains no white pine.Under man's watch, fires in south-

eastern pine and oak forests have beenless frequent than in the natural re-gime, so even the second-growth forestsafter cutting are still denied a majorecological influence. Natural ecologicalrelations are thus difficult to work outin any detail.

In a study of Catskill Mountain for-ests in New York, McIntosh (36)noted that sugar maple has gained dom-inance over beech and hemlock sincethe area was settled in about 1880, anexpansion he attributed to forest dis-8 NOVEMBER 1974

turbance. Another view is that mapleis the natural potential climax (37).McIntosh (36) cautioned, however,"The assessment of potentiality fromprimeval remnants is now largely pre-cluded over extensive areas of the earth.... As the possibility of reconstructinga primeval climax from undisturbedremnants recedes and vanishes we mustdepend on analyses of current vegeta-tion as it is, not as what it might be-come."

Forests of presettlement structuremay be restored naturally to some de-gree, but widespread removal of seedsources through clear-cutting precludesthis for hundreds of years. In areascleared for agriculture, even root sys-tems have been removed, so thatsprouters, which usually flourish afternatural disturbances, are eliminated. Inany case, the situation is unsatisfactoryfor studies of long-term natural suc-cession. Knowledge lost through thedestruction of natural forests can neverbe regained, yet destruction continues.The areas preserved in the East do notinclude all major forest types and aretoo small for major studies.

Fire-Dependent Forests and the

Vegetation Climax

Clements' climax concept involvessuccessional stages leading to vegetationthat reproduces in equilibrium with pre-vailing climatic conditions. Successionhas been documented in numerousstudies. Primary succession is relativelyeasy to describe: observations can bemade over a few years, or a complexof fresh land surfaces of differentknown ages can be studied as a group,as in the case of abandoned river bars,landslide scars, or moraines of retreat-ing valley glaciers. Secondary succes-sion can be studied on old fields.The dimensions of these situations

are usually small, and seed sources forsuccessional plants may be close-by.Starting times and conditions for suc-cession can be determined easily, if notby observation then by techniques liketree-ring analysis. In some cases thetime span can be extended by observa-tions in second-generation forest, forexample, by counting rings on downedtrees on which younger trees haverooted (38). However, only evidencefor limited spans of time can be ex-amined.

Succession after fires is complicated.In regions where fires commonly oc-cur, certain plants are adapted so that

their regrowth is immediate and accel-erated, whereas other plants are killedand can be reintroduced only yearslater at the appropriate successionalstage, and then only if seed sources arepresent nearby. Starting times for treesoriginating after a fire can be deter-mined by tree-ring analysis, and thedates can be confirmed by identifyingfire scars in the ring sequences of treesthat survived the fire.

Tree-ring analysis and fire-scar stud-ies were developed principally in ItascaState Park in northwestern Minnesotaby Spurr (39) and Frissell (40). Acomprehensive investigation has beenmade in the virgin forests-more than500,000 acres (200,000 hectares)-ofthe Boundary Waters Canoe Area(BWCA) of northeastern Minnesota.Here a systematic survey by Heinsel-man (41 ) showed the fire dates for theorigins of all pine stands. Fires arerecorded at intervals of 1 to 8 yearssomewhere in the BWCA for 71 of thelast 377 years. Major fires occurred in1910, 1894, 1863-1864, 1824, 1801,1759, 1727, 1692, and 1681. On theaverage, an area equivalent to the totalland area was burned over during each100-year period. This is the natural firerotation. Reburn intervals for particu-lar localities ranged from 30 to 300years.

Because this BWCA study combinesecology and paleoecology and has im-portant implications for concepts ofboth vegetation history and wildernessmanagement, the ecological situation isdescribed in detail (41-43).The topography of the BWCA is di-

versified, with rocky hills 20 to 100 mabove basins containing bogs or stream-connected lakes. The upland vegetationmosaic, however, is controlled by theincidence of fire rather than by topog-raphy (44).An even-aged stand of jack pine

dates to a fire that killed the previousstand but provided for regeneration, be-cause jack pine cones persist on treesuntil they are opened by intense heat.Then seeds fall to the ground, wherefire has burned off competitors, openedan area for sunlight, and convertednutrients to easily soluble form in ash.Stands of red and white pine also tendto be even-aged, because periodic firesclear the underbrush for seeds that fallfrom surviving trees. Even-aged standsof aspen and birch record vigoroussprouting after the canopy is openedby fire. Between fires, pine or hardwoodstands tend to be succeeded by fir,spruce, and cedar, which are not fire-

491

resistant and would presumably formthe climax forest in the absence offire.

Fire rotation controls the distribu-tion of age classes of stands and thesuccession within stands. The resultingdiversity may represent long-range sta-bility, as implied by the paleoecologicalrecord. Suppression of fire may lead toa theoretical climax of spruce and fir,but the succession may be arrested byinsect epidemics, resulting in an unpre-dictable forest composition. The wide-spread infestation of spruce budwormon balsam fir in the BWCA today mayreflect the long interval without fire.The dead trees may fuel a holocaustsomeday-despite protection. Thus, firesuppression prevents the frequent per-turbations under which the forest de-velops and maintains its diversity. Thiselimination of the perturbations may bethe most profound effect of man on thisnatural ecosystem (45).The 370 years covered by tree-ring

studies in the BWCA represent only ashort time in the total life-span of theforest. During those years explorers, furtraders, loggers, fire guards, and tour-ists may have modified the natural fre-quency of fire. To determine the naturalfire frequency for prehistoric time, ourknowledge must be extended backwardby paleoecological studies, which de-pend on finding fossils in a datablestratigraphic sequence. Pollen fromlake sediments records the vegetationhistory, and charcoal concentrations re-veal the fire frequency. Radiocarbondating is possible for the last 40,000years. Analyses of other microfossilsprovide independent evidence of en-vironmental conditions.Such studies were made of Lake of

the Clouds in the heart of the BWCA(Fig. 3). The lake is chemically strati-fied with organic sediments depositedin annual layers.

Recent fires known from tree-ringcounts show stratigraphically as max-imums in the charcoal profile in thecurve for lamination thickness (possiblyrepresenting temporary increases in theinwash of soil from the drainage basin),and in pollen curves for shrubs thatsprout after fires (43). The record forthe last 1000 years shows major firesevery 80 years on the average-aboutthe same frequency as during the 18thand 19th centuries. Charcoal analysesof the entire depth of sediment indi-cate periodic fires in this forest eco-system for more than 9000 years.

Pollen profiles generally record re-gional rather than local vegetation.

492

Consequently, even when major firesoccurred in the area, the pollen raininto individual lakes was not signifi-cantly changed. The pollen diagram forLake of the Clouds (46) shows sub-stantial uniformity since the borealspruce forest was transformed to pineforest about 9000 years ago (Fig. 3).Slight changes support subdivision ofthe profiles into pollen zones attributa-ble to climatic change, and independentevidence for this comes from paleo-limnologic studies elsewhere in Minne-sota. Other long-range changes may re-sult from other external factors, suchas progressive soil leaching, or immi-gration of tree species controlled bysuch biologic factors as rate of seeddispersal.The principal result is that the forest

has been reasonably stable for 9000years, despite severe short-term per-turbations. A Clementsian climax isnever reached, because the evolutionof each forest stand is repeatedly inter-rupted by fire, but in the long rangethe forest mosaic as a whole maintainsan equilibrium.

The long-range equilibrium is also ex-pressed by the aquatic components ofthe fire-dependent forest ecosystems.Studies of hydrologic and chemicalbudgets of the Little Sioux fire of 1971in the BWCA indicate that releasednutrients are taken up rapidly by vege-tative regrowth and soil storage, sothat little reaches the lakes to nourishalgal growth (47). Chemical anddiatom analyses of lake sediments indi-cate that past fires have had a negligibleeffect on the quality of the lake water.

The evidence for the dependence ofBWCA forests on fire is convincingbecause it is based on studies of a verylarge area of virgin forest. Where largeareas of virgin forest remain in westernNorth America, all indications point toa similar dependence (42). The west-ern forests include most of the desig-nated wilderness areas in North Amer-ica, for which the stated goal is main-tenance of natural conditions. Knowl-edge of successional relations in suchvirgin forests is essential for intelligentmanagement. Yet where such knowl-edge exists, as for the BWCA, it is dif-ficult to modify traditional fire-protec-tion policies of management.

Management of Wilderness Areas

Most potential wilderness areas inthe eastern United States are foreststhat were cut more than 100 years ago

and reforested with new composition.We do not know how much they differfrom the primeval forests they replacedbecause the remaining virgin areas aretoo small for comparison, and the roleof large-scale factors like fire cannotbe evaluated. Although many second-growth forests serve some of the recre-ational functions of designated wilder-ness areas, they are man-made and can-not take the place of primeval wilder-ness. The wilderness experience dimin-ishes when one encounters a cut stumpor a fence in what superficially resem-bles a virgin forest. Further, scientificinvestigations of natural regimes cannotbe conclusive when data have beenaffected by significant human distur-bance.

Considering the scarcity of true wil-derness, designated wilderness areasshould be managed to reestablish nat-ural regimes, so that visitors can seeprimeval wilderness, scientific researchcan progress, and complete primevalecosystems can persist, whether for thebenefit of man or not.The 1964 Wilderness Act (7) states

that wilderness areas must be managedto maintain their primitive character.Heretofore, management has sup-pressed fire, under the assumption thatfire destroys forests. However, theecological studies recounted above indi-cate that fire is a major factor in main-taining the dynamic equilibrium ofmany natural ecosystems. Fire-suppres-sion policies should therefore bechanged. The situation is critical forthe BWCA, where evidence for fireadaptation is overwhelming, and wherea new management plan (6) has beenadopted. The new plan ignores ecologi-cal principles and wilderness preserva-tion, for it authorizes timber cutting inmore than a quarter of the remainingvirgin forest, rejects fire as a forest-maintenance tool, and introduces in-stead the concept of "administrativecutting" in virgin forest to reduce fuelbuildup.

Timber cutting (8) in the largestarea of virgin forest east of the RockyMountains will reduce an area whoselarge size is one of its major scientificvalues. Large size is vital for large-scaleecological studies such as determiningthe distribution and extent of past firesand studying the movements and ter-ritories of mammals like moose, bear,and wolf, and for preserving the geneticvariation in natural populations. Cut-ting also removes attractive virgin for-est from the most heavily visited areaof the Federal Wilderness System, at a

SCIENCE, VOL. 186

time when use is increasing at about10 percent per year.Although the virgin stands of jack

pine and black spruce constitute thelast major source of long-fibered pulp-wood in the Great Lakes region, theirlong-term value for wilderness recrea-tion and scientific study far exceeds theshort-term return from timber sales,especially when the cost of preparationfor sale, administration, and replantingis counted. Modern logging practicesof clear-cutting, rock-raking, and road-building destroy the natural landscape.Recovery takes hundreds or even thou-sands of years. Nutrients are removedthrough increased runoff and soil ero-sion (48), breakdown of the nitrogencycle in the soil (49), and removal ofbiomass that ordinarily decomposesand releases its nutrients for vegetativeregrowth (50). Streams are affected byinflow of excess nutrients (49). Thestratigraphic records for lakes affectedeven by nonmechanized logging indi-cate that abrupt changes in water qual-ity follow cutting (51).The decision not to use fire to restore

the virgin forest is based largely on fearthat visitors will be endangered andthat fire might spread to Canada or onto private property. It is based on thetraditional view that virgin forest isdestroyed by fire, and on a fear ofadverse public reaction after so manyyears of Smokey Bear warnings. How-ever, lakes and streams in the BWCAprovide excellent firebreaks for limit-ing prescribed fires, and experiencewith free-running lightning fires atSequoia-Kings Canyon National Parkindicates that the public will accept achange in management based on soundscientific principles (52).With rejection of fire as a manage-

ment tool in fire-dependent virgin for-est, the manager must either continuefire suppression, with the resulting fuelbuildup and insect infestation until anuncontrollable fire inevitably occurs, orinitiate "administrative cutting," whichpresumably means the felling of deadand diseased trees, removal of under-brush, and perhaps prescribed burning.Although administrative cutting is pref-erable to fire suppression, it is not away to preserve the virgin characterof the wilderness. The time has comefor serious experimentation with nat-ural and prescribed fires under con-trolled conditions, so that virgin forestcan be renewed rather than destroyed,and so that true wilderness areas willbe available for future generations ofvisitors.8 NOVEMBER 1974

Conclusions

Both the landforms and the vegeta-tion of the earth develop to states thatare maintained in dynamic equilibrium.Short-term equilibrium of a hillslopeor river valley results from intersectionbetween erosional and depositionaltendencies, controlled by gravitationalforce and the efficiency of the trans-porting medium. Long-term equilibriumof major landforms depends on crustaluplift and the resistance of the rock toweathering. In most parts of the worldlandscape evolves toward a peneplain,but the reduction rate approaches zeroas the cycle progresses, and the counter-acting force of crustal uplift intercedesbefore the end form is reached.

Davis described this theoreticalmodel in elegant terms. Leopold andHack ha e provided a new and quanti-tative understanding of short-range geo-morphic interactions that tend to dis-credit the Davisian model in the eyesof many. However, the substitute mod-els of quasi-equilibrium or dynamicequilibrium merely describe short-rangesituations in which this or that Davisianstage is maintained despite uplift ordownwasting. Given crustal stabilityand an unchanging climate, landformswould presumably still evolve throughDavisian stages. However, the Davismodel cannot be tested, for despitetremendous inventions in geochronol-ogy and impressive advances in strati-graphic knowledge, we cannot yetestablish the rates or even the fact ofcrustal uplift in most areas. We areleft with an unresolvable problem, forthe sedimentary records of erosionalhistory are largely inaccessible, undat-able, and indecipherable, at least in thedetail necessary to describe long-termevolution of the landscape.We know more about the evolution

and maintenance of vegetation assem-blages than about landform evolution,for even long-term vegetation sequencesare within the scope of radiocarbondating, and the biostratigraphic recordis detailed. Even here, however, distinc-tions between short-term and long-termsituations must be made, so that Cle-ments' grand scheme of vegetationalclimax-created soon after Davis'smodel of landform development-canbe evaluated in terms of modernknowledge. Disillusion with the climaxmodel paralleled disillusion with Davis'smodel in the 1950's, but the climaxmodel can be tested, because the rec-ord of vegetational history is accessible,datable, and decipherable.

In the short term of a few decades,successional vegetation stages occur ina variety of situations, as confirmed byobservation or by techniques such astree-ring analysis. The successional veg-etation stages are reactions to nutrients,weather, competition, and consumption.Such succession implies long-term dis-equilibrium, or at least unidirectionaldevelopment.The long-term controlling factor in

Clements' model of vegetation devel-opment is climate. With climatic stabil-ity the succession will proceed to aclimax. In the Appalachian Mountains,geomorphic, microclimatic, and edaphicconditions limit climax development,producing a polyclimax, which is gen-erally sustained by the dominance ofthese factors. Death and regenerationof single forest trees is controlledmostly by windstorms. The distribu-tional pattern may be locally transectedby lightning fires, major windstorms, orwashouts. However, the long-term sta-bility of Appalachian forests is demon-strated by pollen stratigraphy.

Although we can infer the long-termstability of Appalachian forests, thetrends and mechanics of short-termvegetational succession are not fullyunderstood, because lack of sizableareas of virgin forest limits investiga-tions of natural conditions. In this re-spect, the eastern United States is al-ready much like western Europe, whereclimatic and disturbance factors in veg-etational history cannot be disentangled.

In the Great Lakes region, a largearea of virgin forest exists in theBWCA of northeastern Minnesota.Here short- and long-term studies showthat for at least 9000 years the prin-cipal stabilizing factor has been thefrequent occurrence of fire. Major firesoccur so often that the vegetation pat-tern is a record of fire history. All ele-ments in the forest mosaic are in vari-ous stages of postfire succession, withonly a few approaching climax. Fireinterrupts the successful sequence to-ward climax. Geomorphic and edaphicfactors in vegetational distribution arelargely submerged by the fire regime,except for bog and other lowland vege-tation. Fire recycles nutrients and re-news succession. Nevertheless, despitethe fire regime, the resulting long-termequilibrium of the forest mosaic, char-acterized by severe and irregular fluctu-ations of individual elements, reflectsregional climate.

In the BWCA and the western moun-tains, large virgin forests can be pre-served for study and wilderness recrea-

493

tion. These wilderness areas must bemanaged to return them to the naturalequilibrium which has been disturbedby 50 to 70 years of fire suppression.The goal should be to maintain virginforests as primeval wilderness. Thiscan be done by management that per-mits fire and other natural processesto determine the forest mosaic. Mecha-nized tree-felling and other human dis-turbances should be kept to an absoluteminimum.

Natural landforms also should bepreserved for study and for certain non-destructive recreational activities. Itis somewhat late for the Colorado Riverand other rivers of the West, becausenatural balances are upset by drainage-basin disturbances. Modification ofplant cover on hillslopes changes in-filtration and erosion rates and thus thestream discharge and sediment load, sothe stream balance is altered fromprimeval conditions. Scenic Rivers leg-islation should thus be used to restorecertain river systems and their drainagebasins.

Mountain meadows, badlands, desertplains, and patterned permafrost terrainare extremely fragile and sensitive. In-tricate stream and weathering processesleave patterns easily obliterated bymechanized vehicles. Tire tracks canlast for decades or centuries. The min-eral patina or lichen cover on desert oralpine rocks are records of long stabil-ity, and slight differences in their de-velopment record the relative ages oflandforms, to the year in the case oflichens. Delicate color differences in atalus slope or desert fan show long-term effects just as does the arborealvegetation mosaic in another climaticsetting.

Preservation of virgin wilderness forstudy is viewed by some as a selfishgoal of scientists, to be achieved at theexpense of commercial and recreationaldevelopment. However, scientific studyand nonmechanized recreational usesare compatible in wilderness areas.Furthermore, the public does appreci-ate intellectual stimulation from naturalhistory, as witnessed by massive sup-port for conservation, the WildernessAct, and a dozen magazines like Na-tionial Geographic. Finally, no knowl-edgeable American today is unawarethat ecological insights are necessary topreserve the national heritage. Westerndust bowls, deforested slopes, gulliXedfields, silted rivers, strip mine wast -

lands, and the like might have beenavoided had long-term problems beenbalanced against short-te4m profits.

494

Many economic questions cannot beanswered intelligently without detailedknowledge of extensive virgin ecosys-tems. Long-term values are enhancedby those uses of natural resources thatare compatible with the preservationof natural ecosystems.

Esthetically, virgin wilderness pro-duced by nature is comparable to anoriginal work of art produced by man.One deserves preservation as much asthe other, and a copy of nature has aslittle value to the scientist or discerninglayman as a reproduction of a paintinghas to an art scholar or an art collector.Nature deserves its own display, notjust in tiny refuges but in major land-scapes. Man is only one of literallycountless species on the earth. Man de-veloped for a million years in a worldecosystem that he is now in danger ofdestroying for short-term benefits. Forhis long-term survival and as an expres-sion of his rationality and morality, heshould nurture natural ecosystems.Some people believe that human loveof nature is self-protective. For manyit is the basis of natural religion.The opposition of many Americans

to the Alaska pipeline is a manifestationof almost religious feeling; most neverexpect to see the Alaskan wilderness,but they are heartened to realize thatit exists and is protected. The same canbe said of those who contribute tosave the redwoods in California. Herecost analysis fails to account for theenormous value people place on natureand on the idea of nature as contrastedto the private gain of a few developers.Americans admire European preserva-tion of works of art. Europeans admireAmerican foresight in setting aside na-tional parks. However, the distributionof protected natural areas in Americais uneven and inadequate, and vastareas continue to be developed or badlymanaged despite widespread newknowledge about long-term human in-terest in wilderness preservation.

Darwin turned nature study into thestudy of natural history. He could ob-serve natural features in vast undis-turbed areas with no thought that hu-man interference had been a factor intheir development. Today such naturallandscapes have practically vanished.Those that remain should be preservedas extensively as possible, and man-aged with scientific knowledge of thenatural processes that brought them tobeing. At the present accelerating rateof exploitation, massive disturbance,and unscientific management, soon nona ural areas will be left for research

or wilderness recreation. Some say thatscientific curiosity and the ability forrecreation define man. This is reasonenough for wilderness preservation.However, a more ominous conclusionis that the survival of man may dependon what can be learned from the studyof extensive natural ecosystems.

References and Notes

1. C. Lyell, Principles of Geology (Murray,London, 1930).

2. C. Darwin, On the Origin of Species (Murray,London, 1859).

3. W. M. Davis, Natl. Geogr. Mag. 1, 81 (1889);reprinted in D. W. Johnson, Ed., GeographicalEssays (Dover, New York, 1954), pp. 485-513.

4. H. C. Cowles, Bot. Gaz. 31, 73 (1901).5. F. E. Clements, Research Methods in Ecology

(University Publishing, Lincoln, Neb., 1905).6. Boundary Waters Canoe Area Land Use Plan

(U.S. Forest Service, Superior National Forest,Duluth, Minn. 1974).

7. Wilderness Act (PL 88-577, 3 September1964).

8. The Wilderness Act, in a special section de-voted to the BWCA, states that managementof the BWCA shall have "the general purposeof maintaining, without unnecessary restric-tions on other uses, including that of timber,the primitive character of the area." TheSecretary of Agriculture thereupon authorizedthat the BWCA be subdivided into two zones,and that timber cutting be permitted in thePortal Zone. This was challenged in a lawsuit[Minnesota Public Interest Research Groupv. Butz et at., Suppl. No. 4-72 Civil 598 (Dis-trict Court, Minnesota, 1973)] seeking atemporary injunction against cutting underthe terms of the National EnvironmentalProtection Act. In complying, the courtstated: "The language used makes it clearthat the Secretary of Agriculture is to enun-ciate and enforce any and all restrictionswhich are necessary to maintain the primi-tive character of the BWCA. It is only if arestriction is not necessary to fulfill thispurpose that it can be challenged as 'un-necessary.' Where there is a conflict be-tween maintaining the primitive characterof the BWCA and allowing logging or otheruses, the former must be supreme." None-theless, the new BWCA management planretains the policy of logging in the PortalZone.

9. W. M. Davis, Anm. Geol. 23, 207 (1899); re-printed in D. W. Johnson, Ed., GeographicalEssays (Dover, New York, 1954), pp. 350-380.

10. , J. Geol. 13, 381 (1905); reprinted inD. W. Johnson, Ed., Geographical Essays(Dover, New York, 1954), pp. 296-322.

11. W. Penck, Morphological Analysis of Land-form.s, H. Czech and K. C. Boswell, Transl.(St. Martin's, New York, 1953).

12. J. H. Mackin, Geol. Soc. Am. Butll. 59, 463(1948).

13. L. B. Leopold and T. Maddock, Jr., U.S.Geol. Suirv. Prof. Pap. No. 252 (1953).

14. J. T. Hack, Am. J. Sci. 258A, 80 (1960).15. C. D. Holmes, ibid. 262, 436 (1964).16. S. A. Schumm and R. W. Lichty, ibid. 263,

110 (1965).17. M. A. Carson and M. J. Kirby, Hillslope

Form and Process (Cambridge Univ. Press,Cambridge, Mass., 1972).

18. H. C. Cowles, Bot. Gaz. 51, 161 (1911).19. H. A. Gleason, Bull. Torrey Bot. Cilub 53, 7

(1926).20. R. H. Whittaker, Ecol. Monogr. 23, 41 (1953).21. A. G. Tansley, Ecology 16, 284 (1935).22. E. L. Braun, Decidutouis Forests of Easterni

North A merica (Blakiston, Philadelphia, 1950).23. E. S. Deevey, Geol. Soc. An,. Bultl. 60, 1315

(1949).24. E. L. Braun, Ohio J. Sci. 51, 1939 (1951).25. W. A. Watts, in Proceedin1gs of the Interna-

tional Geobotany Conference (Univ. of Ten-nessee Press, Knoxville, in press).

26. Ecology 51, 17 (1970).27. , Quat. Res. 3, 257 (1973); D. R.

Whitehead, Ecol. Monogr. 42, 301 (1972).28. J. A. Maxwell and M. B. Davis, Quiat. Res.

2, 506 (1972); W. A. Watts, Geol. Soc. Ant.Buill., in press.

29. A. J. Craig, Geol. Soc. AmtI. Spec. Pap. No.

SCIENCE, VOL. 186

123 (1969), p. 283; W. A. Watts, Geol. Soc.Am. Bull. 80, 631 (1969).

30. W. A. Watts, Ecology 52, 676 (1971).31. M. B. Davis, in The Quaternary of the UInited

States, H. E. Wright, Jr., and D. G. Frey,Eds. (Princeton Univ. Press, Princeton, N.J.,1965), pp. 377-402.

32. J. T. Hack and J. C. Goodlett, U.S. Geol.Surv. Prof. Pap. No. 347 (1960).

33. J. C. Goodlett, Harv. For. Bull. No. 25(1954).

34. K. H. Garren, Bot. Rev. 9, 617 (1943).35. S. H. Spurr, Ecol. Monogr. 26, 245 (1956).36. R. P. McIntosh, ibid. 42, 143 (1972).37. A. L. Langford and M. F. Buell, Adv. Ecol.

Res. 6, 84 (1969).38. R. S. Sigafoos and E. L. Hendricks, U.S.

Geol. Surv. Prof. Pap. No. 387A (1961);D. B. Lawrence, Geogr. Rev. 40, 191 (1950).

39. S. H. Spurr, Ecology 35, 21 (1954).40. S. S. Frissell, Jr., Quat. Res. 3, 397 (1973).41. M. L. Heinselman, ibid., p. 329.

42. H. E. Wright, Jr., and M. L. Heinselman,ibid., p. 319.

43. A. M. Swain, ibid., p. 383.44. L. F. Ohmann and R. R. Ream, U.S. For.

Serv. Res. Pap. NC-63 (1971).45. 0. F. Loucks, Am. Zool. 10, 17 (1970).46. A. J. Craig, Ecology 53, 46 (1972).47. R. F. Wright, Univ. Minn. Limnol. Res.

Cent. Interim Rep. No. 10 (1974); J. G.McColl and D. F. Grigal, Agron. Abstr. Am.Soc. Agron. (1973); J. P. Bradbury, J. C. B.Waddington, S. J. Tarapchak, R. F. Wright,in Proceedings of the 19th Congress of theInternational Association of Limnology (Win-nipeg, 1974).

48. L. B. Leopold, paper presented at the NewZealand Symposium on Experimental andRepresentative Research Basins, Wellington(1970); R. R. Curry, Assoc. Southeast. Biol.Bull. 18, 117 (1971).

49. G. E. Likens et al., Ecol. Monogr. 40, 23(1970).

50. G. F. Weetman and B. Webber, Can. J.For. Res. 2, 351 (1972); D. W. Cole, S. P.Gessel, S. F. Dice, in Proceedings of Sym-posium on Primary Productivity and MineralCycling in Natural Ecosystems, H. E. Young,Ed. (Univ. of Maine Press, Orono, 1968), pp.197-232.

51. D. M. Stark, thesis, University of Minnesota(1971); J. P. Bradbury and J. C. B. Wad-dington, in Quaternary Plant Ecology, H. J.B. Birks and R. G. West, Eds. (CambridgeUniv. Press, Cambridge, 1973), pp. 289-308.

52. B. Kilgore, Quat. Res. 3, 496 (1973).53. Discussions leading to this article started

with M. L. Heinselman and E. J. Cushingaround a winter campfire in the BWCA, andI thank them and numerous students forcontinued incentive to evaluate the time factorin landscape evolution and to demonstratea scientific rationale for wilderness preserva-tion. I also appreciate the editorial and thesubstantive critique of R. A. Watson.

From Mars with Love

Missions to return a Mars surface sample are feasible,but pose potential back-contamination problems.

Richard S. Young and Donald L. DeVincenzi

Soil samples can now be returned tothe earth from another planet by usingunmanned spacecraft. One possibleconsequence of this ability is the poten-tial for returning a viable extraterres-trial organism that may interact withterrestrial life forms.The bioscience community is polar-

ized on the issue of back-contaminationof the earth; some believe the risk isvirtually nonexistent while others be-lieve it is high. There is no questionthat scientific interest in exploring thesurface material of another planet isgreat. The problem arises in determin-ing the method of study that is the mostproductive and cost-effective, that is,in situ on the planetary surface withautomated landers or by direct study ofreturned samples in terrestrial labora-tories.The purposes of this article are to

outline the mission possibilities for re-turning surface samples from Mars, toreview experience gained from returnedlunar samples, and to discuss the scien-tific value of sterilized samples com-pared to unsterilized samples. This isnot a statement of NASA policy, and8 NOVEMBER 1974

there is no intention to take sides onthe issue of back-contamination. Ouraim is to describe the mission optionsand the potential back-contaminationproblems they pose and thus to stimu-late thought that we hope will generatesolutions.

Historical Considerations

In July 1964, a conference on thepotential hazards of back-contamina-tion of the earth by returned extrater-restrial samples was held under theauspices of the Space Science Board ofthe National Academy of Sciences (1).The participants included representa-tives of the Space Science Board's LifeScience Committee, the Department ofAgriculture, the National Institutes ofHealth and Public Health Service, andNASA, as well as selected scientistswith backgrounds in public health andpathology from various universities.The committee considered the ques-tion of returning samples from themoon and the .planets, the potentialhazards to terrestrial life, and the need

for action to assure the safety of lifeon the earth.The committee stated its belief that

the existence of life on the moon or onthe planets could not be precluded, andthat the likelihood of life on the moonwas much less than on Mars. It re-viewed the history of the harmfulspread of biological agents on the earth(for example, tuberculosis, smallpox,and measles), where disease agents wereinadvertently introduced into humanpopulations that had not previouslybeen exposed to the disease and there-fore had not evolved protective mecha-nisms. The committee also reviewed thehistory of nonhuman epidemics (suchas the Irish potato famine, in which afungus infection literally destroyed thepotato crop on which Ireland de-pended). The group pointed out thatorganisms harmless to man but patho-genic to plants or animals might beas deleterious to man as those whichaffect him directly. It was felt that theintroduction of a completely new(extraterrestrial) organism must beconsidered a potential catastrophe sinceterrestrial forms of life would have hadno previous history of exposure andtherefore no opportunity to have de-veloped natural immunities or artificialvaccines. The committee concluded thatextraterrestrial life, and the concomi-tant possibility of back-contamination,must be presumed to exist, and thatany "policies of defense against back-contamination must be based on theproposition that if infection of theearth by extraterrestrial organisms ispossible, it will occur."The committee therefore strongly

Dr. Young is chief of the Planetary BiologyProgram, National Aeronautics and Space Ad-ministration (NASA) Headquarters, Washington,D.C. 20546. Dr. DeVincenzi is assistant chief,Planetary Biology Division, NASA Ames Re-search Center, Moffett Field, California 94035.

495