Thinking of Biology

On a state factor model of ecosystems

Therefore, speak I to them inparables; because they seeing seenot. (Matthew 13: 13)

The study of terrestrial ecosys­tems is an area of intellectualinquiry that is fundamental to

our very concept of nature. Like allmodern sciences, the ecosystem sci­ences are concerned with a quantita­tive understanding of various prop­erties and processes and the controlson these features. Yet, the facts orknowledge that a science can acquireand the very questions that it asksdepend on the concepts, models,and theories that it collectivelypossesses (Kuhn 1962). From thisperspective, models function as theparables of science.

The concepts and theories of eco­systems have, like those of other natu­ral sciences, evolved greatly since thescientific renaissance of the nine­teenth century and the subsequentcoining of the term "ecosystem" byAlfred Tansley (1935). Many, if notmost, of these ecosystem models haveevolved from the field of ecology(Golley 1993). Yet, there is anotherperspective of ecosystems, and amodel to study them, that evolvedapart from ecology but coincidentwith the development of the scienceof pedology in the late nineteenthand early twentieth centuries. Forthe purposes of this article, we callthis perspective the state factormodel. This model is now being used,either implicitly or explicitly, in agrowing number of ecosystem stud­ies (e.g., Pastor et al. 1984, Schimelet al. 1985). In this article, we reviewthe structure and attributes of thismodel, briefly identify its origin, traceits evolution, and illustrate its useduring the twentieth century, with

by Ronald Amundsonand Hans Jenny

536

the goal of bringing it to a wider audi­ence in the ecological community.

Structure of the statefactor model

The earth's surface, although it var­ies greatly from place to place, isessentially a continuum in both spaceand time (Figure 1). Ecosystems,therefore, are arguably human con­structs that break the continuum intomanageable and differing segmentsfor study. The size of the ecosystemchosen for study is also somewhatarbitrary (Likens 1992), but it willpossess certain characteristics: it isopen with respect to its surround­ings and is capable of exchangingmatter and energy with the surround­ing environment. Moreover, its prop­erties at any time-that is, the stateof the system-depend to a greatdegree on the characteristics of thesurroundings.

What variables determine the stateof an ecosystem at any time, t? Fromcomparisons with other sciences, themain factors can be grouped into thefollowing categories: initial state ofthe system, external conditions, andage of the system. Over a century offield 0 bservation has revealed a setof independent variables that, formost ecosystems, can be shown todefine or control the ecosystem's state(i.e., its properties, or dependent vari­ables) at any given time. Expressedin general mathematical form, theequation is

ecosystem properties = f (climate,organisms, topography, parent ma­terial, time, humans, ... )

All state factors share several gen­eral properties (Jenny 1958). First,they are independent of the systembeing studied (i.e., the system doesnot affect the state factors). This willbe true only if the system being stud-

ied is small in relation to its sur­roundings. For ecological studiesdealing with landscape segments,such as sites or plots, such a restric­tion is unimportant. Second, the statefactors may, in many locations of theearth or for certain periods of geo­logical time, vary independently ofone another. This independence ofone state factor relative to the otherscreates the possibility that throughjudicious system (i.e., site) selection,the influence of a single factor onecosystem properties may be ob­served and quantified in nature.There are, of course, many site com­parisons for which it can be shownthat not all of the state factors areindependent of one another. For ex­ample, it is well known that climatevaries through long expanses of geo­logic time. Consequently, differencesbetween ecosystems of different agesmay in part reflect differences inclimatic histories.

The ecosystem properties equa­tion defines an ecosystem. FollowingJenny (1941), ecosystems can be de­fined as those portions of the earth'sterrestrial surface whose propertiesvary in response to variations in thestate factors. Therefore, a differencein state factor assemblages (no mat­ter how small) between two loca­tions on the earth's surface results intwo differing ecosystems. Theoreti­cally, therefore, the earth is composedof an infinite number of ecosystems.

The terms describing the state fac­tors have been in use for more than acentury, and their meanings havevaried greatly over this time. In thisarticle, we use the current defini­tions of the state factors, which areconsistent with formulations of themodel during the past half century(Amundson and Jenny 1991, Jenny1941, 1958). A discussion of theway in which these definitions andconcepts differ from those of thenineteenth and early twentieth cen-

BioScience Vol. 47 No. 8

Dow

nloaded from https://academ

ic.oup.com/bioscience/article/47/8/536/231390 by guest on 26 August 2022

Depend nt... lion

Tlm = l03yr

In light of this definition, it isapparent that many of the earth'secosystems have been influenced bymultiple constellations of state fac­tors and may be regarded as polyge­ne tic-that is, as reflecting two ormore contrasting periods of forma­tion, each with a different constella­tion of state factors. Pedologists havelong recognized that many soils olderthan the Holocene reflect conditionsof past climates (Hilgard 1887). Thefact that these soils do not entirelyreflect the modern assemblage of statefactors does not invalidate the use ofthe state factor model to investigatethese soils or ecosystems. Instead, itopens the possibility that relict fea­tures in those ecosystems may be usedto reconstruct past climates using cli­mate and ecosystem relationships de­veloped in Holocene environments(Retallack 1990).• The state factor humans could con­ceptually be subsumed under the bi­otic factor (0; e.g., Jenny 1941) be­cause, like all other biota, humanscontain a genetic component (a geno­type). However, unlike many otherspecies, human populations possessa cultural component (Amundsonand Jenny 1991, Jenny 1980) that

Tlm =0 yrFigure 1. Two terrestrialecosystems at differentstages of development.State factors are identi­fied in italic type, andimportant ecosystemcomponents are indicatedby roman type. The sedi­ment in the system on therecent fluvial deposit (t =0; upper left) is defined asthe parent material. Thesystem has had insuffi­cient time for establish­ment of flora and fauna,even though a bioticsource (potential biota)exists . In the ecosystemon the stabilized river ter­race (time = 103 years;upper right), pedogenicprocesses have formed asoil. These two ecosys­tems are but the most re­cem members of a longtemporal continuum ofsystems, many of which are buried in the sediments illustrated in the lower part of thefigure. A rich vegetation, reflecting controls imposed by the system's state factorassemblage, has also evolved. The microclimate within this ecosystem is no longer thesame as that of the regional climate, as in the young ecosystem, bur is instead a functionof the system's age, biotic factors, and other state factors. After Retallack 1983.

• The state factor parent material (p)is defined as the initial state (at t =0)of the sediment, rock, or soil compo­nent of the ecosystem. For an ecosys­tem developing on fresh alluvium oron a recent volcanic flow, the parentmaterial is the geological materialitself. For an ecosystem that is re­forming on a site that has been burnedor clear-cut, the parent material isthe soil that is present at the time atwhich the new flora begin to reinha bitthe site.• The state factor time (t) is definedas the elapsed time since the ecosys­tem began forming or since the assem­blage of state factors of an ecosystemchanged. For some ecosystems, thisstarting point, or t = 0, begins after adepositional event, such as fluvialsediment or volcanic ash deposition.In others, t = 0 might be consideredthe time at which a major distur­bance or change in the configurationof the remaining state factors oc­curred. Climatic change, biotic in­flux, or human disturbance can allbe of such magnitude that, from aninvestigative standpoint, one mustconsider them capable of resettingthe "clock"-that is, starting the de­velopment of a new ecosystem.

rur ies can be found in Jenny (1941,1989).

• The state factor climate (cl) is theclimate (e.g., rainfall, temperature,and humidity) surrounding the eco ­system. It is often consistent with theconcept of regional climate. The cli­mate within an ecosystem (which is aproperty of that ecosystem) is re­ferred to as rnicroclimate or depen­dent climate (cl'). This dependentclimate is related to the regional cli­mate but is also conditioned by theother state factors that are affectingthe ecosystem.• The biotic factor, sometimes calledorganisms (0), is defined as the po­tential biota of the system. From apractical standpoint, the biotic fac­tor is the microbial, plant, and ani­mal gene flux that enters the systemfrom the surroundings. The vegeta­tion (v) and animals (a) that actuallysurvive or reproduce in the systemare dependent ecosystem properties.These living organisms may not di­rectly reflect the potential biota be­cause they depend on the constella­tion of all the state factors. Forexample, plants that grow at a sitecan be a function of climate (Arnund­son et al. 1989), topography (Elga­baly 1953), parent material (Pastoret al. 1984), and ecosystem age(Burges 1960, Van Clcve et al. 1991).Given the ease with which seeds andanimals are dispersed, it is difficultto compare natural ecosystems thatdiffer only in their biotic factors.Sites separated by biotic barriers,such as mountain ranges andoceans, may be used for these com­parisons. In addition, the invasionof introduced species into other­wise undisturbed island ecosystemsoffers an opportunity to observethe initial phases of a change in thebiotic factor (Vitousek and Walker1989).• The state factor topography (r) ismade up of a number of subvariablesthat correspond to the physical con­figuration of an ecosystem at t = 0(the beginning of the system's devel­opment or the beginning of an obser­vation period). These subvariablesinclude the site's topographic posi­tion on a hills lope complex, the as­pect of the slope, and the proximityof the system to a shallow ground­water table.

September 1997 537

Dow

nloaded from https://academ

ic.oup.com/bioscience/article/47/8/536/231390 by guest on 26 August 2022

varies greatly from society to societyirrespective of genotype. Culture isdefined as the assemblage of tech­nologies, ideas, and philosophies pos­sessed by a group of individuals.Thus, to distinguish the purely physi­ological characteristics of humansfrom their sociological aspects, thestate factor cultural inheritance (c;the cultural assemblage of a popula­tion in a system at t = 0) has beendefined as a distinctive state factorof ecosystem formation (Amundsonand Jenny 1991).1

It should be recognized that thecollective culture of a human popu­lation at any time t may vary greatlyfrom that at t = O. These changes inthe resulting culture (i.e., in the de­pendent variable Cl) reflect the influ­ence of the constellation of the statefactors of the ecosystem. This depen­dent variable varies in response tostate factors. For example, the socialstructure of early nineteenth-centuryGreat Plains indigenous populationsvaried with climate (Bamforth 1988),and the agricultural practices (andassociated technologies) of migrat­ing Polynesian populations varied inresponse to the characteristics of theislands being settled (Kirch 1982).

This list of state factors is notexhaustive, and the equation on page536 is written to allow the additionof other factors of local importance,as the ellipses indicate. Some addi­tional factors include fire, coastalsalt spray, and dust (Jenny 1980).

The key to using the state factormodel quantitatively lies in beingable to numerically represent theproperties of both the ecosystem andthe state factors. If numerical rela­tionships can be constructed, thenvarious univariant functions describ­ing the relationship between ecosys­tem properties (e) and sta te factors canbe derived from field observations:

climofunctions: e =f (cl) e.r.p.t;«, ...

biofunctions: e = f (o)cl,r,p,t,c, ...

"The full human state factor (h) therefore con­tains genetic (0h) and cultural components: h =(0h, c). For most ecosystem studies, the role ofculture is of interest, and it is stressed here. In thisarticle, cultural inheritance is designated simplyas c (in contrast to our earlier use of the symbolc), and dependent culture is designated as c'(compared with our earlier use of c; Amundsonand Jenny 1991).

538

topofunctions: e = f (r) cl,0,p,t,C,...

lithofunctions: e = f (p) cl.e.r.t,c, ...

chronofunctions: e = f (t) cl.e.r.p.t,«, ..

anthropofunctions: e = f (c)cl,o,r,p,t, .

The design and quantification ofunivariant functions have been theprimary field application of the statefactor model for studying soils andecosystems (Jenny 1941, 1961,1980). The effect of a given factor onan ecosystem property can be evalu­ated by holding the other factorsinvariant between sites. This prin­ciple can be best understood by con­sidering the differential form of theequation that describes the differ­ence in properties between ecosys­tems (Jenny 1941):

de =(ae/acl)dclo,r,p,t,c, ... + (ae/)d0et,r,p,t,c, ...+ (ae/ar)drcl,0,p,t,C, ...+(ae/ap )dpcl,0,r,t,c, +(ae/at)dtcl,0,r,p,c, ... + (ae/ac)dCet,0,r,p,t,...+ .

The total change in propertiesbetween ecosystems (de) can be evalu­ated with respect to a single factor inone of two ways. First, the range, ortotal variation, in the other factorsbetween systems can be kept negli­gible through clever site selection.For a study in which the effect ofclimate is to be evaluated, for ex­ample, do, dr, dp, dt, and de all equal0, or nearly so. A second way to holdother factors constant emerges fromconsidering the second equation.Even if total variation in some factoris not held constant between sites(dfactor *- 0), then the effect of thefactor may still be unimportant, solong as fJe/fJfactor = 0 for the sites ofinterest. For example, in soil studies,the effect of climate on soil organiccarbon storage can be evaluated bychoosing sites that vary greatly inclimate but have similar slopes, par­ent materials, and biotic factors.However, the age of the soil (i.e., theage of the deposits on which the soilsform) may be difficult to hold con­stant across a broad climatic gradi­ent. Although the age of the soilsmay vary considerably, if they are allmore than several thousand yearsold they may be near steady statewith respect to organic carbon stor­age (Schlesinger 1990), making varia­tions in carbon storage between sites

due to variations in the time factornegligible. The existence of steadystate (or, more generally, the insensi­tivity of an ecosystem property toany factor) cannot be assumed; itmust be verified through empiricalobservation for each ecosystem prop­erty of interest.

Development of the statefactor modelThe state factor model is not wellknown in ecology (Golley 1993), al­though methodological approachessuch as comparative ecosystem analy­sis (Duarte 1991, Peters et al. 1991)appear to be closely related to it. Therelative obscurity of the state factormodel in ecology is probably relatedto a historical lack of communica­tion between the field in which itdeveloped (the earth sciences) andthe field in which most ecosystemsresearchers are trained (the biologi­cal sciences; Golley 1993). Golley(1993) noted that important con­cepts related to ecosystem studieswere developed by Russian earth sci­entists but concluded that these con­cepts had little impact on presentecosystem thought.

Although the early Russian ideasremained unknown to the English­speaking world for some time, theseconcepts did eventually reach theWest through a circuitous path­way and have become part of main­stream pedological thought formuch of the second half of thetwentieth century. These ideas, asmodified and redefined by others(Jenny 1941, 1958, 1961), serve asthe initial framework for the statefactor model described here.

The origin of the state factor modelis commonly traced to VasiliDokuchaev, who is widely recog­nized as being among the origina­tors, if not the founder, of modernpedology (Krupenikov 1993, Vuci­nich 1970). Dokuchaev was born inRussia in 1846, the son of a priest.On beginning a fellowship at theTheological Seminary in St. Peters­burg, his attentions turned to natu­ral science, and he eventually ob­tained his doctorate in geology,becoming curator of the geologicallaboratory at St. Petersburg Univer­sity in 1872 (Joffe 1936). An impor­tant event for Dokuchaev was his

BioScience Vol. 47 No. 8

Dow

nloaded from https://academ

ic.oup.com/bioscience/article/47/8/536/231390 by guest on 26 August 2022

selection, in 1876, by the Free Eco­nomic Society to organize the firstsystematic study of the Russian eh er­nozem (prairie soil), with specificinstructions to explain its structure,origin, and evolution (Dokuchaev1883, Krupenikov 1993, Vil'yams1936). The participants in Doku­chaev's study sought advice fromscientists of a broad range of disci­plines' using the most advancedchemical and physical analyses avail­able to them for their investigations.

Within a short time, Dokuchaevperceived soil to be an "independentnatural body" (Vil'yams 1936) thatis amenable to explanation by natu­rallaws. As early as 1880, he wrote(Dokuchaev 1880):

Like any other individual body innature, any other organism, soilhas also its specific origin, itschemical composition and physi­cal attributes, its individual struc­ture, its habitude, its specificity ingeographical distribution. As hasbeen explained in other papers,every normally located surface soilbearing vegetation has to be re­garded as a function of the:

(a) Local maternal rock variety

(b) Age of the land (in particularthe time since it became surfaceland)

(c) Climate

(d) Vegetation

... By issuing from these assump­tions, the distribution of soil can­not any more be regarded as anabsolutely casual phenomenon.The geography of soils, just likethe distribution of other organ­isms, conforms to definite laws ....

Vil'yams (1936) stated that thesefactors were recognized by Doku­chaev's predecessors, but he creditsDokuchaev with considering theircombined effect and evaluating eachindependently. Dokuchaev clearlyrecognized that his theory of soilformation had implications for thenatural world (i.e., terrestrial ecosys­tems) as a whole (Dokuchaev 1898):

It is my undaunted opinion, thatthe concepts of modern soil knowl­edge, as developed in Russia,should be placed in the center ofthe basis on which we construct

September 1997

our new understanding of natureas a whole. I believe that our soilknowledge should become the veryspringboard from which our teach­ings on the relationship between"living" and "dead" nature derivetheir impetus for the grasping ofthe idea of the relationship be­tween man and the rest of theworld, i.e., its organic, as well asmineral part.

According to Krupenikov (1993),Dokuchaev commented, in the fol­lowing year, that the "time is not farwhen in its own right and because ofits great importance for humanity, it(i.e., the science of the study of ter­restrial systems as a whole) will oc­cupy an independent and fully re­spected place."

The dissemination of Russian ped­ological thought to the westernworld, particularly the United States,was hindered by a language barrier.In 1908, the Russian scientist N. M.Tulaikov wrote, in a review of soilscience in the United States (Tulaikov1908):

I wish to express my regret that thesuccess in the understanding andthe application of fundamentalprinciples of scientific "soil-knowl­edge" as emphasized in the contri­butions of Russian soil scientists,remain still to a great extent un­known to the foreign reader inter­ested in this realm of science. Wehave not yet gained such a posi­tion among the circle of Europeanand other scientists, to be read inour own native tongue, therefore,we remain completely unknown,notwithstanding that in manyfields we should deserve a differ­ent attitude toward us.

Eugene W. Hilgard, the eminentAmerican soil scientist, urged theRussians to publish abstracts in ei­ther French or German, which would"assist to the distribution of Russiancontributions from the border ofRussia to the Bay of San Francisco"(Tulaikov 1908). It was the eventualGerman publication of Russian ped­ological ideas that led to the passageof Russian pedological thought toAmerican science. With help fromthe German scientist HermannStreme, Konstantin D. Glinka, a stu­dent of Dokuchaev and an impor­tant contributor to Russian pedol­ogy in his own right, wrote Die Typen

der Bodenbildungin 1914 (Simonson1997, Tandarich and Sprecher 1994).By 1917, this book was translatedinto English by the American soilscientist Curtis F. Marbut, the direc­tor of the Division of Soil Survey, USDepartment of Agriculture (USDA).According to Joffe (1936), Marbut'sintroduction to Russian pedologicaltheory inspired him to transformmany aspects of the US soil classifi­cation system and the way in whichsoils were studied in the United States.

The infusion of the Russian con­cepts of soils and nature into ecologybegan as early as 1930, with thepublication of Charles Shaw's dis­cussion of factors of soil formation(Shaw 1930). However, the primarybreakthrough was Hans Jenny's re­formulation and further quantifica­tion of the state factor concept andthe publication of these integratedideas as a book in 1941 (Jenny 1941).As noted elsewhere (Amundson1994b), this publication provided adetailed definition of both soil and the"larger system," as well as a methodto quantitatively and numerically linksoil and larger system properties tostate factors.

The concept of the larger systemimmediately suggested ecosystem toJenny and his students. In 1951,JackMajor published the first paper (Ma­jor 1951) explicitly linking the fac­torial model to ecosystem analysisby focusing on the relationship ofstate factors to vegetation. A subse­quent paper by Jenny (1958) furtherdiscussed the significance and inter­pretation of the biotic factor, and inlater publications he explicitly ex­panded the model to ecosystemwideapplications (Jenny 1961,1980). Thepresent status of the factorial modelin ecology and ecosystem science hasbeen discussed recently by Phillips(1989) and Vitousek (1994).

The state factor model as ahistorical scienceThe state factor model of ecosystemsshares elements of concepts used byother branches of the historical sci­ences, of which geology is a notableexample (Frodeman 1995). A his­torical science is one in which theobject of understanding-in this case,ecosystems-involves a knowledgeof both place and time (Frodeman

539

Dow

nloaded from https://academ

ic.oup.com/bioscience/article/47/8/536/231390 by guest on 26 August 2022

25 10

Figure 2. A schema tic rep rese nta tion of glo ba l so il carbo n in relat ion to meanann ua l temperature (MAT) and precipitation (MA P). Th e th ree-dimensional sur­face was gene ra ted by inte rpo la ting between th e carbo n sto rage to cl imate relation­ships given in Figure 1 of Post et al. 1982.

60

50N

~ 40~

~30e*20o

10

1995) . One challenge common to allhistorical sciences is to define th eobject of study, to ident ify th e set ofcharacteristics th at define it, and todet ermine how mu ch cha nge ca nocc ur befor e it is considered a newentity (Fro deman 1995 ).

In th e hi storical sciences, asStephen J. Gould has noted, "the re­sults of history lie strewn aro und us,but we ca nnot, in princip le, observeth e processes that produced th em .H ow then can we be scientific aboutth e pa st?" (Gould 1983). It is com­mon in the earth sciences to assem blethe results of a given process in itsd ifferent sta ges to determine th e rat eand direction in which th e processpro ceeds (Gould 19 83). The sta tefactor mod el of ecosys tems sharesth is premise but expands on it bysugges ting that it is a lso possibl e toexamine portions of th e ea rth, all a tth e same stage of development but ina varie ty of di fferent environments,to decipher qu antitative relat ionshipsbetween ecosystem pr operties andth e ind ependent varia bles kn own assta te fac to rs . Th e sta te fact or mod elis, the refore, one of severa l likelyapproaches to "compa ra t ive ecosys­tem analysis " (Co le et a l. 1991 ), andit s use in thi s context ha s been dis­cussed and illustrated in a few rece nt

540

publica tions (e.g., Vi tousek 1994,Vitousek and Matson 1991).

The sta te factor mod el (Jenny1941 ), like many comparat ive eco­system models (Peters et al. 1991 ), isnonmech ani st ic. Strictly speaki ng,th e sta te fact or mod el tells us noth­ing ab out how a system acq uires itpr operties-it is phenomenological(Amundso n an d Jenny 1991 ). Toaddress processes or mechanisms, onemust resort to different types ofmodels that address other aspects ofthe ecosystem.The dichotomy betweenstate factor models (nonmechanistic)and process models (mechanistic) hasbeen much discussed in the pedologi­cal literature (e.g., I-Ioosb eek andBryant 1992 ) and in the history ofscience literature (e.g., Ga le 1984),with th e recognit ion that the tw o ap­proa ches are, in many cases, comple­mentary and essent ial to a fuller un­derstanding of natural phenom ena(Amundso n 1994 a).

Fin ally, th e sta te fac tor model dif­fer s from some other conce pts ofecosystems in th at it explici tly in­cludes soil, wa ter, and th e atmos­phere-all of which are subject tochan ges in respon se to any cha ngesin the indepen dent sta te fac tors-asintegral co mponen ts of ecos ystems .In this sense, th ese component s are

much mor e than abiotic factors thataHect th e living organisms of th esystem- they are dynamic ecosys­tem fea tures that vary spatially andtemp or ally across the globe.

The state factor model in thestudy of terrestrial ecosystems

The array of concepts and modelsfor describing ecosys tems is diverse(Likens 1992), and communicationbetween divergent points of view hasnot been effective historically (Golley19 93, Pimm 1994). This Balkaniza­tion may be due to several factors,among th em th e tendency of practic­ing scientists as a group to be rela­tively unconcerned with the ori ginand implications of the models andth eories th at they use (Gale 1984,Kuhn 1962 ). But as many scientistshave ar gued (Frodeman 1995, Gale1984) , a better aw ar ene ss of th e na­ture and origin of scientific rea son­ing will benefit not only the indi ­vidua l scientis t, but also the scientificcommunity as a whole.

Th e state factor model is a un iqu ean d useful mea ns of studyin g nature:both conceptually and qu antitati vely.It is particularl y useful for exper i­mental design in comparative eco ­system studies, becau se it for ces th einvestigator to explicitly determinethe variation in the complete arrayof fac tors that affect the systemsbeing exa mined. For many scientists,th e usefulness of a model is the mostcritical cr iterion for general accep­ta nce. Nearly 75 years of data colle c­tion and analysis attest to the appli­cability of th e state factor model toth e study of soils. As an example ofits util ity in thi s application, we sho whow it ca n be applied to an ecos ys­tem probl em of immediate concern:th e ro le o f terrestrial ecosystems inth e global carbon cycle.

It is well documented th at th e larg­est terrestrial reservoir of carbon re­sides in soi l organic matter (Schlesin­ger 1991 ). There ha ve been twoapproaches to quantifying and orga ­nizing the glo bal soil carbon pool:taxonom ic (e.g. , Bohn 1976 ) andeco logica l (i.e ., based on a sta te fac ­to r appro ach; Post et al. 1982) . Bothgive similar estima tes of soil carbo npools (Kern 1994 ), but the taxon omicapproa ch ha s no predictive value,wh ereas the ecological approach has

BioScience Vo/. 47 N o. 8

Dow

nloaded from https://academ

ic.oup.com/bioscience/article/47/8/536/231390 by guest on 26 August 2022

a 4 r---:---.....-----...---., Figure 3. Th e effect of state factors, otherthan climate , on the organic carbon con­tent of soils. (a) The relationship betweenorganic carbon (in the upper 20 cm of soil )and mean annual temperature in the grass­land-forest boundary in the Great Plain s(data from Jenn y 1930 ). The difference incarbon betw een the grass land and forestsites approxima tes a biosequence becau sethe vegeta tive difference is most likely afunction of fire (Brow n 1985), which maybe considered a state factor in its ownright, rather than of differences in poten­tial vegetat ion. (h) The relationship be­tween soil organic carbon and hillslopeposition and parent materia l in the north­ern Great Plains (data from Aguilar et a!.1988 ). (c) The relationship between soilorganic carbon and soil age (Californ iadata from Harden 1987; New Zealanddata from Syers et al. 1970; Baffin Islanddata from Evans and Cameron 19 79). (d)The relationship between length of cultiva­tion and remaining soil carbon in somemajor biotic zones of the world (da ta com­piled bySchlesinger [1986J). For data wherenitrogen rath er than carbon content wasreported, a C:N ratio of 12 was assumed inthe calculat ion of the carbon content. In allgraphs except (d), explicit attempt s weremade to control variations in all state fac­tor s other than tho se being examined.

clim ate, because both soil organiccarbon and plant communities ar efun cti ons of regional climate. Figure2 uses data from Post et al. (1982 ) toillustrate the relationship of soil car­bon to mean annual temperature andprecipitation that can be derived froman "ecological life zone" or "Hol­dridge life zone" organization ofcarbon storage. This figure illus ­trates, at a global sca le, wha t is wellknown from smaller, regional stud­ies: Soil carbon inc reases with meanannual precipitation and decreaseswith mean annual temperature(Burke et a!. 1989, Jenny 19 30,1941 ). Most important, as noted byPost et a1. (1982) , this analysis re­veals the climatic regions that ar emo st suscept ible to large changes inso il carbon sto r age with sm allchanges in climate (i.e., Figure 2shows areas where carbon ch an gesgrea tly with sma ll ch anges in eithertemperature or precipitation ). Fig­ure 2 do es not illustrate th e lar gevariability of soil carbon in any givenclimatic zone (generally at least 50 %of the mean value in either direction;Post et al. 1982) . Presumably, thisvariabi lity re sults from the fact that

• • •! ]j 2

L. ndsclpe pollUon

C )0

N

ElO

~

~ 10....l 0

0

SOlI "lII (log yro)

d

100 •0 . . 0

-i .• •.., c SO • • •• •.. - • 0i CS I. • 060 • • •

30 •0- -. 40

0 _ .'--u 200 20 40 60 SO 100

L.nglh 01 cultrvlhon (yro)

b

10 5 20In Innu lernperolUf, (OC)

~ 0 0 0 0 0 00 0 •

l Oooo

N

E 151'---=-.....~

20

effective means to reconstruct past(or future) soil carbon storage basedon ecological changes. For example,changes in global terrestrial carbonstorage from the last glacial maxi­mum to the pre sent ha ve been calcu­lated by reconstructing glac ial maxi­mum vegetation patterns (either frompaleoecological data [Adarns et a1.1990) or from mod eling [Prentice andFung 1990]).ln these appro aches, soilcarbon storage at theglacia l maximumwas calcu lated using the relationshipsbetween present -day soil carbon andecosystem vegetation established byPost et al. (1982) and others.

Moreover, using the ecosystem ap­proach to tabulate soi l carbon pro­vides an opportunity to exp licitlylink global soil storage directly to

been the basis for estimates of theeffects of both past and future cli­mate changes on the global stora geof terrestrial soil (and plant) carbon.

The taxonomic approach relies onthe determination of the areal extentand average carbon content of themajor soil taxonomic gro ups of theworld (e.g., soil orders or suborders ofthe USDA system or mapping units ofthe Food and Agricultural Organiza­tion world soil map). The USDA soilclassification scheme is deliberatelynonscientific in the sense that taxa areestablished for practical purposes ofland management and are not definedor determined by their direct relation­ship to sta te facto rs or to soil pro ­cesses. The concepts used in the defini­tion of soil taxa in the USDA systemare outlined by Smith (1965, p. 19):"Properties selected as differentiaeshould be soil properties, The use oftheo ries act s to limit the possibility ofacquiring new knowledge. Th e use ofnon-soil properties, such as the bed­rock or the climate, tend s to focus ourattention on the climate or th e geologyrather than on th e soil." Althoughpost-factum arguments ha vc beenraised to highl ight the relationship ofdefined taxa to state factors or pro ­cesses (Cline and Johnson 196 3), itremains inherently impossible to pre­dict what new soil taxa and, th erefore,soil carbon might develop in responseto a climate or environmental changebecause, as Smith (1965, p. 22 ) notes,"At least one step of reasoning is nec­essary to develop genetic interpreta­tions from the definitions of the (soiltaxonomic) classes."

The ecosystem approach providesan en tirel y different basis for esti ­mating and organizing global soilcarbon reser vo irs. The basi s for thi smethod is the observation th at broadeco log ica l life zones (Ho ldridge194 7) are distributed in relation tovariations in climate, a state factor.Based on the general observation th atplant communities and so il organicca rbon covary in response to climatevariations, Post et al. (1982 ) tabu­lated the areal extent of the wo rld' smajor life zones and calculated theaverage soil organic carbon content ofeach life zone to assemble what mightbe the most widely qu oted est imateof the global soil carbon pool.

This ecosystem approach has sev­eral advantages. Firs t, it provides an

September 1997 541

Dow

nloaded from https://academ

ic.oup.com/bioscience/article/47/8/536/231390 by guest on 26 August 2022

carbon storage does not depend onclimate alone but results from theentire series of state factors (Figure 3):Variations in biota, topography andparent material, soil age, and hu­mans within any given climatic re­gion all probably contribute to thereported "noise" in the data (Post etal. 1982).

These applications of the statefactor model in its broadest senseare, like all historical sciences andrelated methodologies, limited intheir predictive value. To paraphraseFrodeman (1995), the present worldmay be too narrow a window to viewthe past or the future, and our mod­ern analogues may not directly applybackward or forward in time. Nev­ertheless, historical approaches may,in many cases, represent our best,or even our only, means of scien­tifically addressing many ecosys­tem problems of immediate globalconcern. Not only their limitations,but also their strengths, should beboth acknowledged and explored.

The state factor model of ecosys­tems possesses an elegance, outwardsimplicity, and potential breadth ofapplication that makes it an appeal­ing means of studying nature. Yet inactual practice, the selection of eco­systems to study the effect of onestate factor or another is seldom asimple matter, and rigorous experi­mental design commonly involves theexpertise and interaction of geornor­phologists, ecologists, pedologists,.climatologists, and many other spe­cialists. This need for a multidisci­plinary approach strikes at the heartof what ecosystems are conceived tobe (Tansley 1935) and provides anopportunity for greater communica­tion among disparate sciences thatare now confronted with urgentprob­lems and questions of a global andhuman dimension.

Acknowledgments

I thank Oliver Chadwick and T. N.Narasimhan for discussions pertain­ing to this article. Rebecca Chasan,Peter Groffman, and an unidentifiedreviewer provided numerous com­ments that improved this article. TroyBaisden's expertise and assistance ingenerating Figure 2 is greatly appre­ciated. HopeJahren and Pam Matsonreviewed an early version of this ar-

542

ticle, providing useful commentsand criticisms. This research wassupported by the University of Califor­nia Agricultural Experiment Station.

References citedAdams JM, Faure H, Faure-Denard L,

McGlade JM, Woodward FI. 1990. In­creases in terrestrial carbon storage fromthe last Glacial Maximum to the present.Nature 348: 711-714.

Aguilar R, Kelly EF, Heil RD. 1988. Effectsof cultivation on soils in northern GreatPlains rangeland. Soil Science Society ofAmerica Journal 52: 1081-1085.

Amundson R. 1994a. Towards the quantita­tive modeling of pedogenesis-a review­Comment. Functional vs. mechanistictheories: the paradox of paradigms.Geoderma 63: 299-302.

__. 1994b. Foreward. Pages vii-xiv inJenny H. Factors of soil formation: a sys­tem of quantitative pedology. New York:Dover Publications.

Amundson R, Jenny H. 1991. The place ofhumans in the state factor theory of eco­systems and their soils. Soil Science 151:99-109.

Amundson RG, Chadwick OA, Sowers JM.1989. A comparison of soil climate andbiological activity along an elevation gra­dient in the eastern Mojave Desert.Oecologia 80: 395-400.

Bamforth DB. 1988. Ecology and human or­ganization on the Great Plains. New York:Plenum Press.

Bohn HL. 1976. Estimate of organic carbonin world soils. Soil Science Society ofAmerica Journal 40: 468-470.

Brown L. 1985. Grasslands. Knopf (NY):Chanticleer Press.

Burges A. 1960. Time and size as factors inecology. Journal of Ecology 48: 273-285.

Burke IC, Younker CM, Parton WJ, Cole CV,Flach K, Schimel DS. 1989. Texture, cli­mate, and cultivation effects on soil or­ganic matter content in U.S. grasslandsoils. Soil Science Society of America Jour­nal 53: 800-805.

Cline AJ, Johnson DD. 1963. Threads ofgenesis in the seventh approximation. SoilScience Society of America Proceedings27: 220-222.

Co le J, Lovett G, Findlay S. 1991. Compara­tive analyses of ecosystems: patterns,mechanisms, and theories. New York:Springer-Verlag.

Dokuchaev VV. 1880. Protocol of the meet­ing of the branch of geology and mineral­ogy of the St. Petersburg Society of Natu­ralists. [Translated by the Department ofSoils and Plant Nutrition, University ofCalifornia, Berkeley].Transactions of theSt. Petersburg Society of Naturalists XII:65-97.

__. [1883] 1967. Russian chernozem:selected works of V. V. Dokuchaev. Vol.1. [Translated by the Israel Program forScientific Translations, Jerusalem]. Jerusa­lem: Israel Program for Scientific Transla­tions.

__.1898. Plenary session of the membersof the Caucacus division of the ImperialRussian Geographical Society, September

26, 1898. [Translated by the Departmentof Soils and Plant Nutrition, University ofCalifornia, Berkeley].

Duarte CM. 1991. Variance and the descrip­tion of nature. Pages 301-318 in Cole J,Lovett G, Findlay S, eds. Comparativeanalyses of ecosystems: patterns, mecha­nisms, and theories. New York: Springer­Verlag.

Elgabaly MM. 1953. The soils, water supply,and agriculture in northeastern Sinai.Pages 125-152 in Proceedings of theUNESCO Symposium on Scientific Prob­lems of Land Use in Arid Regions; Cairo,Egypt; Dec 1953. Heliopolis (Egypt): Bou­levard Sultan Hussien.

Evans LJ, Cameron H. 1979. A chronose­quence of soil developed from graniticmorainal material, Baffin Island, N.W.T.Canadian Journal of Soil Science 59: 203­210.

Frodeman R. 1995. Geology as an interpre­tive and historical science. GeologicalSociety of America Bulletin 107: 960­968.

Gale G. 1984. Science and the philosophers.Nature 312: 491-495.

Golley FB. 1993. A history of the ecosystemconcept in ecology: more than the sum ofthe parts. New Haven (CT): Yale Univer­sity Press.

Gould S]. 1983. Hen's teeth and horse's toes.New York: Norton.

Harden JW. 1987. Soils developed on gra­nitic alluvium near Merced, California.US Geological Survey Bulletin or 1590-A.Available from: US Government PrintingOffice, Washington, DC.

Hilgard EW. 1887. The processes of soilformation from the Northwestern basalts.Proceedings of the Society for the Promo­tion of Agricultural Science 8: 51-58.

Holdridge LR. 1947. Determination of worldplant formations from simple climaticdata. Science 105: 367-368.

Hoosbeek MR, Bryant RB. 1992. Towardsthe quantitative modeling of pedogenesis­a review. Geoderma 55: 183-210.

Jenny H. 1930. A study of the influence ofclimate upon the nitrogen and organicmatter content of the soil. Missouri Agri­cultural Research Station Bulletin or 152.Columbia (MO): Missouri AgriculturalExperiment Station.

__. 1941. Factors of soil formation: a sys­tem of quantitative pedology. Republishedin 1994. New York: Dover Publications.

__. 1958. Role of the plant factor in thepedogenic functions. Ecology 39: 5-16.

__.1961. Derivation of state factor equa­tions of soils and ecosystems. Soil ScienceSociety of America Proceedings 25: 385­388.

__. 1980. The soil resource, origin andbehavior. New York: Springer-Verlag.

__. 1989. Hans Jenny: soil scientist,teacher, and scholar. [Interviewed by D.Maher and K. Stuart], University HistorySeries, Regional Oral History Office, TheBancroft Library, University of Califor­nia, Berkeley.

Joffe JS. 1936. Pedology. New Brunswick(NJ): Rutgers University Press.

Kern JS. 1994. Spatial patterns of soil organiccarbon in the contiguous United States.Soil Science Society of America Journal

BioScience Vol. 47 No. 8

Dow

nloaded from https://academ

ic.oup.com/bioscience/article/47/8/536/231390 by guest on 26 August 2022

58: 439-455 .Kirch PV. 1982. Ecology and the adaptation

of Polynesian agricultural systems. Ar­chaeology in Oceania 17: 1-6.

Krupenikov lA. 1993. History of soil science:from its inception to the present . RussianTranslation Series 98. Brookfield (VT) : A.A. Balkerna Publishers.

Kuhn TS. 1962. The structure of scientificrevolutions . Chicago: The University ofChicago Press .

Likens GE. 1992. The ecosystem approach:its use and abuse. Oldendorf (Germ any) :Ecology Institute.

Major .1 . 1951. A functional, factorial ap­proach to plant ecology. Ecology 32 : 392 ­412 .

Pastor .1, Aber JD, McClaugherry CA, MelilloJM. 1984. Aboveground producrion andNand P cycling along a nitrogen mineral­ization gradient on Blackhawk Island,Wisconsin . Ecology 65: 256-268.

Peters RH, er al. 1991. On the relevance ofcomparative ecology to rhe larger field ofecology. Pages 46-63 in Cole .1, Lovett G,Findlay S, eds. Comparative analyses ofeco system s: patterns, mechanisms, andtheories. New York: Springer-Verlag.

Phillips .ID. 1989. An eva lua tio n of the statefactor model of soil ecosystems . Ecologi­cal Modelling 45: 165-1 77.

Pimm SL. 1994. An American talc. Nature370 : 188 .

Post WM, Emanucl WR, Zinke PJ, Stan­genberger AG. 1982. Soil carbon andwnrld life zones. Nature 298: 156-159.

Prcnrice KC, Fung IY. 1990. Th e sensitivityof terrestrial carbon storage to climatechange. Na ture 346: 48-51.

Retallack GJ. 1983 . Late Eocene and Oli­gocene paleosols from Badlands NationalPark, South Dakota. Special Paper nr 193.Boulder (CO): Geological Society ofAmerica.

_ _ . 1990. Soils of the past: an introduc­tion to paleopedology. London : Unwin­Hyman.

Schimel D, Stil1well MA , Woodmansee RG .1985. Biogeochemistry of C, N, and P ina soil ea ten a of the shorrgrass steppe.Ecology 66 : 276-282 .

Schlesinger WH o 1986. Changes in soil car­bon storage and associated properties withdisturbance and recovery . Page s 194-220in JR, Raichc DE, eds . The ch anging car­bon cycle: a global analysis. New York:Springer-Verlag.

__. 1990. Evidence from ch ronosequencestudies for a low carbon-storage potentia]of soils . Nature 348 : 232-234.

_ _ . 1991. Biogeochemistry: an analysisof global change. San Diego: AcademicPres s.

Shaw CF. 1930. Potent factors in soil forma­tion . Ecology XI: 239-245.

Simonson RW . 1997 . Early teaching in USAof Dokuchae v factors of soil formation.

September 1997

Soil Science Society of America Journal61: 11-16.

Smith GD . 1965. Lectures on soi l classifica ­tion. l'edologie 1965, Special issue nr 4.Rozier 6. Ghcnr (Belgium ): Belgi an SoilScience Society.

Syers .JK, Adams j A, Walker. 1970. Accumu­lation of organic matter in a chronose­quence of soils developed on windblownsand in New Zealand. Journal of SoilScience 21: 146-15 3.

Tandarich JP, Sprecher SW. 1994. The intel­lecrual background for th e factors of soilformation. Pages 1-13 in Amundson R,Harden .1, Singer M, eds. Factors of soilformation: a fiftieth anniversary retro­spect ive. Special Publication nr 33 . Madi­son (WI): Soil Science Society of Am erica .

Tansley AG . 1935. The use and abuse ofvegetational concepts and terms. Ecology16: 284-307.

Tulaikov NM. 1908 . Soil research in theUnited States. [Translated by the Depart­ment of Soils and Plant Nutrition, Univer­sity of California, Berkeley.) l'ochvove­dcnic 10: 293-322.

Van Cleve K, Chapin FS Ill, Dyrncs C'I',Viereck LA. 1991 . Element cycling in raigaforesr: state-factor COntrol. BioScience 41 :78-88 .

Vi lyams VR. [1936] 1967. V. V. Do­ku cha ev's role in th e development of soilscience: introduction to the Russian Cher­nozcm . [Translated by the Israel Programfor Scientific Translations, Jerusalem].Jerusalem: Israel Program for ScientificTranslations .

Vitousek PM. 1994. Factors controlling eco­syst em structure and function. Pages 87­97 in Amundson R, Harden .1, Singer M,cds, Factors of soil formation: a fiftiethanniversary retrospective . Publication nr33. Madison (WI): Soil Science Society ofAmerica .

Virousck PM, Matson PA. 1991. Gradientanalysis of ecosystems. Pages 28 7-298 inCole .T, Lovetr G, Findlay S, ed s. Com­parative analyses of ecosysrems: patterns,mechanisms, and theories . New Yo rk :Springer-Verlag.

Vitousek PM, Walker LR. 1989. Biologicalinvasion by Myrica faya in H awaii : plantdemography, nitrogen fixation, and eco­system effects. Ecological Monographs 59 :247-265.

Vucinicb A. 1970. Science in Russian culture1861-1917. Stanford (CA): Sranford Uni ­versity Press.

Ronald Amundson (e-ntail :earthy@nature.berkeley.edu) is a professor, and the lateHans Jenny was a professor emeritus, inthe Division of Ecosystem Sciences, 151Hil-gard Hall, University of California,Berkeley, California 94720. © 1997 Ameri­can Institute of Biological Sciences.

ithout an colo lcal found tion,economic policy i blind nd

un u t inable.

ithoul n economic found lion,cologic I policy I impr ctic I.

That'. 'hy th lea Inth fj Id and many

oth rs h v oin d I EE.

ITER TIO LOCIET FOR

ECOLOGIC LECO 0 IC

POIK)FA «,

Inlnnt1

Nature and Human SocietyThe Quest for a Sustainable World

National Academy of SciencesOctober 28-30, 1997

Washington, D.C .

Convenors:National Research Council

Smithsonian InstitutionAAAS

Library of Congress

The goal of this forum is to improveunderstanding about the ways in whichbiodiversity supports us and itspotential to further enhance our livesand livelihood. TIle intended audienceis leaders in science, industry, publicpolicy, the media, as well as thegeneral public. Topics will include:

Day I: The Biological Basis ofSustainabilityThreats to Sustainahility

Day 2: The Measure of BiodiversityAddressingthe Problem

Day 3: Social and Economic DimensionsofSustaining Biodiversity

For more information:(202)334-2251biodiver@nas.edu

http://www2.nas.edu/ biodiv

543

Dow

nloaded from https://academ

ic.oup.com/bioscience/article/47/8/536/231390 by guest on 26 August 2022