6. E. L. Thorndike, Animal Intelligence (Mac-millan, New York, 1911).

7. E. D. Adrian, The Physical Background ofPerception (Oxford Univ. Press, London,1947).

8. D. H. Raab and H. W. Ades, Amer. J.Psychol. 59, 59 (1946); M. R. Rosenzweig,ibid., p. 127; R. A. Butler, I. T. Diamond,W. D. Neff, J. Neurophysiol. 20, 108 (1957);R. F. Thompson, ibid. 23, 321 (1960).

9. W. D. Neff and I. T. Diamond, in Biologicaland Biochemical Bases of Behavior, H. F.Harlow and C. N. Woolsey, Eds. (Univ. ofWisconsin Press, Madison, 1958); R. B. Mas-terton and I. T. Diamond, J. Neurophysiol.27, 15 (1964).

10. 1. M. Warren, Ann. Rev. Psychol. 16, 95(1965).

11. K. S. Lashley, in Symposia of the Society forExperimental Biology (Academic Press, NewYork, 1950), vol. 4.

12. K. Z. Lorenz, ibid., vol. 4.13. G. Elliot Smith, The Evolution of Man (Ox-

ford Univ. Press, London, 1924); C. J. Her-rick, Brain of the Tiger Salamander (Univ.of Chicago Press, Chicago, 1948).

14. G. G. Simpson, Bull. Amer. Museum Nat.Hist. 85, 1307 (1945); A. S. Romer, VertebratePaleontology (Univ. of Chicago Press, Chi-cago, 1945); J. Z. Young, The Life of Ver-tebrates (Oxford Univ. Press, London, 1962).

15. W. C. Hall and I. T. Diamond, Brain, Be.havior Evolution 1, 181 (1968).

16. 3. Kaas, W. C. Hall, I. T. Diamond, "Dualrepresentation of the retina in the cortex ofthe hedgehog in relation to architectonicboundaries and thalamic projections," in prep-aration.

17. R. A. Lende and K. M. Sadler, Brain Res.5, 390 (1967).

18. P. Abplanalp and W. Nauta, personal com-munication.

19. J. Altman and M. B. Carpenter, J. Comp.Neurol. 116, 157 (1961); D. K. Morest, Anat.Record 151, 390 (1965); E. C. Tarlov andR. Y. Moore, J. Comp. Neurol. 126, 403(1966).

20. H. S. Gasser and J. Erlanger, Amer. J. Physiol.88, 581 (1929).

21. G. H. Bishop, J. Nervous Mental Disease 128,89 (1959).

22. C. J. Herrick, Brains of Rats and Men (Univ.of Chicago Press, Chicago, 1926).

23. R. P. Erickson, W. C. Hall, J. A. Jane, M.Snyder, I. T. Diamond, J. Comp. Neurol. 131,103 (1967).

24. J. E. Rose and C. N. Woolsey, Electroen-cephalog. Clin. Neurophysiol. 1, 391 (1949).

25. Unpublished experiments in our laboratory.26. M. Snyder, W. C. Hall, I. T. Diamond, Psy-

chonomic Sci. 6, 243 (1966); M. Snyder andI. T. Diamond, Brain Behavior Evolution 1,244 (1968).

27. H. Killackey, I. T. Diamond, W. C. Hall, G.Hudgins, Federation Proc. 27, 517 (1968).

28. J. E. Rose and C. N. Woolsey, J. Comp.

Neurol. 91, 441 (1949); -, in Biologicaland Biochemical Bases of Behavior, H. F.Harlow and C. N. Woolsey, Eds. (Univ. ofWisconsin Press, Madison, 1958); L. T. Dia-mond, K. L. Chow, W. D. Neff, J. Comp.Neurol. 109, 349 (1958).

29. L. J. Garey and T. P. S. Powell, Proc. Roy.Soc. London Ser. B 169, 107 (1967); M.Glickstein, R. A. King, J. Miller, M. Berkley,J. Comp. Neurol. 130, 55 (1967).

30. The following abbreviations are used in thefigures and figure legends: A, auditory area ofcortex; CM + Pf, centromedian + parafas-cicular nuclei; GL, lateral geniculate nucleus;GM, medial geniculate nucleus; Hab, habe-nula; Hip, hippocampus; L, lateral group ofnuclei; LP, lateroposterior nucleus; MD, medi-odorsal nucleus; PC, posterior commissure;Po, posterior nucleus; Pul, pulvinar nucleus;R, reticular nucleus; S I, somatic area I;S II, somatic area II; SC, superior colliculus;T, temporal area of cortex; TO, optic tract;V, ventral group; V I, visual area I; V II,visual area II; VGL, ventral lateral geniculate;VL, ventrolateral nucleus; VM, ventromedialnucleus; VP, ventroposterior nucleus; ZI,zona incerta.

31. The research described was supported by grantMH-04849 from the National Institute ofMental Health. For their helpful comments wethank Doctors G. Bishop, J. Kaas, G. Kimble,N. Guttman, J. Jane, R. B. Masterton, andM. Warren.

The principles of ecological succes-sion bear importantly on the relation-ships between man and nature. Theframework of successional theory needsto be examined as a basis for resolvingman's present environmental crisis. Mostideas pertaining to the development ofecological systems are based on descrip-tive data obtained by observing changesin biotic communities over long periods,or on highly theoretical assumptions;very few of the generally accepted hy-potheses have been tested experimental-ly. Some of the confusion, vagueness,and lack of experimental work in thisarea stems from the tendency of ecolo-

The author is director of the Institute of Ecol-ogy, and Alumni Foundation Professor, at theUniversity of Georgia, Athens. This article isbased on a presidential address presented beforethe annual meeting of the Ecological Society ofAmerica at the University of Maryland, August1966.

262

gists to regard "succession" as a singlestraightforward idea; in actual fact, itentails an interacting complex of proc-esses, some of which counteract oneanother.As viewed here, ecological succession

involves the development of ecosystems;it has many parallels in the develop-mental biology of organisms, and alsoin the development of human society.The ecosystem, or ecological system, isconsidered to be a unit of biologicalorganization made up of all of the or-ganisms in a given area (that is, "com-munity") interacting with the physicalenvironment so that a flow of energyleads to characteristic trophic structureand material cycles within the system.It is the purpose of this article to sum-marize, in the form of a tabular model,components and stages of development

at the ecosystem level as a means ofemphasizing those aspects of ecologicalsuccession that can be accepted on thebasis of present knowledge, those thatrequire more study, and those that havespecial relevance to human ecology.

Definition of Succession

Ecological succession may be definedin terms of the following three param-eters (1). (i) It is an orderly processof community development that is rea-

sonably directional and, therefore, pre-dictable. (ii) It results from modifica-tion of the physical environment by thecommunity; that is, succession is com-

munity-controlled even though the phys-ical environment determines the pattern,the rate of change, and often sets limitsas to how far development can go. (iii)It culminates in a stabilized ecosystemin which maximum biomass (or highinformation content) and symbioticfunction between organisms are main-tained per unit of available energy flow.In a word, the "strategy" of successionas a short-term process is basically thesame as the "strategy" of long-termevolutionary development of the bio-sphere-namely, increased control of,or homeostasis with, the physical en-vironment in the sense of achievingmaximum protection from its pertur-bations. As I illustrate below, the strat-egy of "maximum protection" (that is,trying to achieve maximum support ofcomplex biomass structure) often con-flicts with man's goal of "maximum

SCIENCE, VOL. 164

The Strategy ofEcosystem Development

An understanding of ecological succession providesa basis for resolving man's conflict with nature.

Eugene P. Odum

on

Mar

ch 3

, 201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

production" (trying to obtain the high-est possible yield). Recognition of theecological basis for this conflict is, Ibelieve, a first step in establishingrational land-use policies.The earlier descriptive studies of suc-

cession on sand dunes, grasslands, for-ests, marine shores, or other sites, andmore recent functional considerations,have led to the basic theory contained inthe definition given above. H. T. Odumand Pinkerton (2), building on Lotka's(3) "law of maximum energy in bio-logical systems," were the first to pointout that succession involves a funda-mental shift in energy flows as increas-ing energy is relegated to maintenance.Margalef (4) has recently documentedthis bioenergetic basis for succession andhas extended the concept.

Changes that occur in major structur-al and functional characteristics of adeveloping ecosystem are listed in Table1. Twenty-four attributes of ecologicalsystems are grouped, for convenience ofdiscussion, under six headings. Trendsare emphasized by contrasting the situ-ation in early and late development. Thedegree of absolute change, the rate ofchange, and the time required to reacha steady state may vary not only withdifferent climatic and physiographic sit-uations but also with different ecosystemattributes in the same physical environ-ment. Where good data are available,rate-of-change curves are usually con-vex, with changes occurring most rapid-ly at the beginning, but bimodal or cy-clic patterns may also occur.

Bioenergetics of EcosystemDevelopment

Attributes 1 through 5 in Table 1represent the bioenergetics of the eco-system. In the early stages of ecologicalsuccession, or in "young nature," so tospeak, the rate of primary productionor total (gross) photosynthesis (P) ex-ceeds the rate of community respiration(R), so that the PIR ratio is greaterthan 1. In the special case of organicpollution, the PIR ratio is typically lessthan 1. In both cases, however, thetheory is that PIR approaches 1 as suc-cession occurs. In other words, energyfixed tends to be balanced by the energycost of maintenance (that is, total com-munity respiration) in the mature or"climax" ecosystem. The PIPR ratio,therefore, should be an excellent func-tional index of the relative maturity ofthe system.

So long as P exceeds R, organic mat-IS APRIL 1969

ter and biomass (B) will accumulate inthe system (Table 1, item 6), with theresult that ratio P/B will tend to de-crease or, conversely, the BIP, BIR, orBIE ratios (where E = P + R) willincrease (Table 1, items 2 and 3). Theo-retically, then, the amount of standing-crop biomass supported by the availableenergy flow (E) increases to a maximumin the mature or climax stages (Table 1,item 3). As a consequence, the netcommunity production, or yield, in anannual cycle is large in young natureand small or zero in mature nature(Table 1, item 4).

Comparison of Succession in a

Laboratory Microcosm and a Forest

One can readily observe bioenergeticchanges by initiating succession in ex-perimental laboratory microecosystems.Aquatic microecosystems, derived fromvarious types of outdoor systems, suchas ponds, have been cultured by Beyers(5), and certain of these mixed culturesare easily replicated and maintain them-selves in the climax state indefinitely ondefined media in a flask with only lightinput (6). If samples from the climaxsystem are inoculated into fresh media,succession occurs, the mature systemdeveloping in less than 100 days. InFig. 1 the general pattern of a 100-dayautotrophic succession in a microcosmbased on data of Cooke (7) is comparedwith a hypothetical model of a 100-year forest succession as presented byKira and Shidei (8).

During the first 40 to 60 days in atypical microcosm experiment, daytimenet production (P) exceeds nighttimerespiration (R), so that biomass (B)accumulates in the system (9). After anearly "bloom" at about 30 days, bothrates decline, and they become approxi-mately equal at 60 to 80 days. the B/Pratio, in terms of grams of carbon sup-ported per gram of daily carbon produc-tion, increases from less than 20 to morethan 100 as the steady state is reached.Not only are autotrophic and hetero-trophic metabolism balanced in the cli-max, but a large organic structure issupported by small daily production andrespiratory rates.

While direct projection from the smalllaboratory microecosystem to open na-ture may not be entirely valid, there isevidence that the same basic trends thatare seen in the laboratory are character-istic of succession on land and in largebodies of water. Seasonal successionsalso often follow the same pattern, an

early seasonal bloom characterized byrapid growth of a few dominant speciesbeing followed by the development laterin the season of high B/P ratios, in-creased diversity, and a relatively steady,if temporary, state in terms of P and R(4). Open systems may not experiencea decline, at maturity, in total or grossproductivity, as the space-limited micro-cosms do, but the general pattern of bio-energetic change in the latter seems tomimic nature quite well.

These trends are not, as might at firstseem to be the case, contrary to theclassical limnological teaching whichdescribes lakes as progressing in timefrom the less productive (oligotrophic)to the more productive (eutrophic)state. Table 1, as already emphasized,refers to changes which are broughtabout by biological processes within theecosystem in question. Eutrophication,whether natural or cultural, results whennutrients are imported into the lakefrom outside the lake-that is, from thewatershed. This is equivalent to addingnutrients to the laboratory microecosys-tem or fertilizing a field; the system ispushed back, in successional terms, to a

younger or "bloom" state. Recent stud-ies on lake sediments (10), as well astheoretical considerations (11), have in-dicated that lakes can and do progressto a more oligotrophic condition whenthe nutrient input from the watershedslows or ceases. Thus, there is hope thatthe troublesome cultural eutrophicationof our waters can be reversed if the in-flow of nutrients from the watershedcan be greatly reduced. Most of all,however, this situation emphasizes thatit is the entire drainage or catchmentbasin, not just the lake or stream, thatmust be considered the ecosystem unitif we are to deal successfully with ourwater pollution problems. Ecosystematicstudy of entire landscape catchmentunits is a major goal of the Americanplan for the proposed InternationalBiological Program. Despite the obviouslogic of such a proposal, it is provingsurprisingly difficult to get tradition-bound scientists and granting agenciesto look beyond their specialties towardthe support of functional studies oflarge units of the landscape.

Food Chains and Food Webs

As the ecosystem develops, subtlechanges in the network pattern of foodchains may be expected. The manner inwhich organisms are linked togetherthrough food tends to be relatively simn

263

on

Mar

ch 3

, 201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

0 20 40 60 80 100Days

Fig. 1. Comparison of the energetics of succession in a forest and a laboratory inicro-cosm. PG, gross produiction; PN, net production; R, total community respiration; B, total

ple and linear in the very early stagesof succession, as a consequence of lowdiversity. Furthermore, heterotrophicutilization of net production occurs pre-dominantly by way of grazing foodchains-that is, plant-herbivore-carni-vore sequences. In contrast, food chainsbecome complex webs in mature stages,with the bull of biological energy flowfollowing detritus pathways (Table 1,item 5). In a mature forest, for ex-ample, less than 10 percent of annualnet production is consumed (that is,grazed) in the living state (12); most isutilized as dead matter (detritus)through delayed and complex pathwaysinvolving as yet little understood animal-microorganism interactions. The timeinvolved in an uninterrupted successionallows for increasingly intimate associa-tions and reciprocal adaptations betweenplants and animals, which lead to thedevelopment of many mechanisms thatreduce grazing-such as the develop-ment of indigestible supporting tissues(cellulose, lignin, and so on), feedbackcontrol between plants and herbivores(13), and increasing predatory pressureon herbivores (14). Such mechanismsenable the biological community tomaintain the large and complex organicstructure that mitigates perturbations ofthe physical environment. Severe stressor rapid changes brought about by out-side forces can, of course, rob the sys-

264

tem of these protective mechanisms andallow irruptive, cancerous growths ofcertain species to occur, as man toooften finds to his sorrow. An exampleof a stress-induced pest irruption oc-curred at Brookhaven National Labora-tory, where oaks became vulnerable toaphids when translocation of sugarsand amino acids was impaired by con-tinuing gamma irradiation (15).

Radionuclide tracers are providing ameans of charting food chains in theintact outdoor ecosystem to a degreethat will permit analysis within the con-cepts of network or matrix algebra. Forexample, we have recently been able tomap, by use of a radiophosphorustracer, the open, relatively linear foodlinkage between plants and insects in anearly old-field successional stage (16).

Diversity and Succession

Perhaps the most controversial of thesuccessional trends pertain to the com-plex and much discussed subject ofdiversity (17). It is important to distin-guish between different kinds of diversi-ty indices, since they may not followparallel trends in the same gradient ordevelopmental series. Four componentsof diversity are listed in Table 1, items8 through 11.The variety of species, expressed as

a species-number ratio or a species-area ratio, tends to increase during theearly stages of community development.A second component of species diver-sity is what has been called equitability,or evenness (18), in the apportionmentof individuals among the species. Forexample, two systems each containing10 species and 100 individuals have thesame diversity in terms of species-num-ber ratio but could have widely differ-ent equitabilities depending on the ap-portionment of the 100 individualsamong the 10 species-for example,91-1-1-1-1-1-1-1-1-1 at one extreme or10 individuals per species at the other.The Shannon formula,

ni ni-2 N log2-

where ni is the number of individualsin each species and N is the total num-ber of individuals, is widely used as a di-versity index because it combines thevariety and equitability components inone approximation. But, like all suchlumping parameters, Shannon's formulamay obscure the behavior of these tworather different aspects of diversity. Forexample, in our most recent field ex-periments, an acute stress from insecti-cide reduced the number of species ofinsects relative to the number of individ-uals but increased the evenness in therelative abundances of the survivingspecies (19). Thus, in this case the"variety" and "evenness" componentswould tend to cancel each other inShannon's formula.

While an increase in the variety ofspecies together with reduced domi-nance by any one species or small groupof species (that is, increased evenness)can be accepted as a general probabilityduring succession (20), there are othercommunity changes that may workagainst these trends. An increase in thesize of organisms, an increase in thelength and complexity of life histories,and an increase in interspecific compe-tition that may result in competitiveexclusion of species (Table 1, items 12-14) are trends that may reduce thenumber of species that can live in agiven area. In the bloom stage of suc-cession organisms tend to be small andto have simple life histories and rapidrates of reproduction. Changes in sizeappear to be a consequence of, or anadaptation to, a shift in nutrients frominorganic to organic (Table 1, item 7).In a mineral nutrient-rich environment,small size is of selective advantage, es-pecially to autotrophs, because of thegreater surface-to-volume ratio. As the

SCIENCE, VOL. 164

biomass. on

Mar

ch 3

, 201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

ecosystemrr develops, however, inorganicnutrients tend to become more andmore tied up in the biomass (that is, tobecome intrabiotic), so that the selec-tive advantage shifts to larger orga-nisms (either larger individuals of thesame species or larger species, or both)which have greater storage capacitiesand more complex life histories, thusare adapted to exploiting seasonal orperiodic releases of nutrients or otherresources. The question of whether theseemingly direct relationship betweenorganism size and stability is the resultof positive feedback or is merely for-tuitous remains unanswered (21).

Thus, whether or not species diversitycontinues to increase during successionwill depend on whether the increase inpotential niches resulting from increasedbiomass, stratification (Table 1, item 9),and other consequences of biological or-ganization exceeds the countereffects ofincreasing size and competition. No onehas yet been able to catalogue all thespecies in any sizable area, much lessfollow total species diversity in a suc-cessional series. Data are so far availableonly for segments of the community(trees, birds, and so on). Margalef (4)postulates that diversity will tend topeak during the early or middle stagesof succession and then decline in theclimax. In a study of bird populationsalong a successional gradient we founda bimodal pattern (22); the number ofspecies increased during the early stagesof old-field succession, declined duringthe early forest stages, and then in-creased again in the mature forest.

Species variety, equitability, and strat-ification are only three aspects of di-versity which change during succession.Perhaps an even more important trendis an increase in the diversity of organiccompounds, not only of those withinthe biomass but also of those excretedand secreted into the media (air, soil,water) as by-products of the increas-ing community metabolism. An increasein such "biochemical diversity" (Table1, item 10) is illustrated by the increasein the variety of plant pigments along asuccessional gradient in aquatic situa-tions, as described by Margalef (4, 23).Biochemical diversity within popula-tions, or within systems as a whole, hasnot yet been systematically studied tothe degree the subject of species diver-sity has been. Consequently, few gen-eralizations can be made, except thatit seems safe to say that, as succes-sion progresses, organic extrametabo-lites probably serve increasingly impor-tant functions as regulators which stabi-18 APRIL 1969

lize the growth and composition of theecosystem. Such metabolites may, infact, be extremely important in prevent-ing populations from overshooting theequilibrial density, thus in reducing os-cillations as the system develops sta-bility.The cause-and-effect relationship be-

tween diversity and stability is not clearand needs to be investigated from manyangles. If it can be shown that bioticdiversity does indeed enhance physicalstability in the ecosystem, or is the re-sult of it, then we would have an im-portant guide for conservation practice.Preservation of hedgerows, woodlots,noneconomic species, noneutrophicatedwaters, and other biotic variety in man'slandscape could then be justified onscientific as well as esthestic grounds,even though such preservation oftenmust result in some reduction in theproduction of food or other immediateconsumer needs. In other words, is va-riety only the spice of life, or is it anecessity for the long life of the totalecosystem comprising man and nature?

Nutrient Cycling

An important trend in successionaldevelopment is the closing or "tighten-ing" of the biogeochemical cycling ofmajor nutrients, such as nitrogen, phos-phorus, and calcium (Table 1, items 15-17). Mature systems, as compared to de-veloping ones, have a greater capacityto entrap and hold nutrients for cyclingwithin the system. For example, Bor-mann and Likens (24) have estimatedthat only 8 kilograms per hectare outof a total pool of exchangeable calciumof 365 kilograms per hectare is lost peryear in stream outflow from a NorthTemperate watershed covered with amature forest. Of this, about 3 kilo-grams per hectare is replaced by rain-fall, leaving only 5 kilograms to beobtained from weathering of the under-lying rocks in order for the system tomaintain mineral balance. Reducing thevolume of the vegetation, or otherwisesetting the succession back to a youngerstate, results in increased water yieldby way of stream outflow (25), but this

Table 1. A tabultr model of ecological succession: trends to be expected in the developmentof ecosystems.

Ecosystem attributes Developmental Mature stagesstages

Community energetics1. Gross production/community Greater or less Approaches I

respiration (PIR ratio) than I2. Gross production/standing crop High Low

biomass (P/B ratio)3. Biomass supported/unit energy Low High

flow (BIE ratio)4. Net community production (yield) High Low5. Food chains Linear, predom- Weblike, predom-

inantly grazing inantly detritusCommunity structure

6. Total organic matter Small Large7. Inorganic nutrients Extrabiotic Intrabiotic8. Species diversity-variety component Low High9. Species diversity-equitability Low High

component10. Biochemical diversity Low High11. Stratification and spatial Poorly organized Well-organized

heterogeneity (pattern diversity)Life history

12. Niche specialization Broad Narrow13. Size of organism Small Large14. Life cycles Short, simple Long, complex

Nutrient cycling15. Mineral cycles Open Closed16. Nutrient exchange rate, between Rapid Slow

organisms and environment17. Role of detritus in nutrient Unimportant Important

regenerationSelection pressure

18. Growth form For rapid growth For feedback control("r-selection") ("K-selection")

19. Production Quantity QualityOverall homeostasis

20. Internal symbiosis Undeveloped Developed21. Nutrient conservation Poor Good22. Stability (resistance to external Poor Good

perturbations)23. Entropy High Low24. Information Low High

265

on

Mar

ch 3

, 201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

greater outflow is accompanied bygreater losses of nutrients, which mayalso produce downstream eutrophica-tion. Unless there is a compensating in-crease in the rate of weathering, theexchangeable pool of nutrients suffersgradual depletion (not to mention pos-sible effects on soil structure resultingfrom erosion). High fertility in "youngsystems" which have open nutrientcycles cannot be maintained withoutcompensating inputs of new nutrients;examples of such practice are the con-tinuous-flow culture of algae, or inten-sive agriculture where large amounts offertilizer are imported into the systemeach year.

Because rates of leaching increase ina latitudinal gradient from the poles tothe equator, the role of the biotic com-munity in nutrient retention is especiallyimportant in the high-rainfall areas ofthe subtropical and tropical latitudes,including not only land areas but alsoestuaries. Theoretically, as one goesequatorward, a larger percentage of theavailable nutrient pool is tied up in thebionfass and a correspondingly lowerpercentage is in the soil or sediment.This theory, however, needs testing,since data to show such a geographicaltrend are incomplete. It is perhaps sig-nificant that conventional North Tem-perate row-type agriculture, whichrepresents a very youthful type of eco-system, is successful in the humidtropics only if carried out in a systemof "shifting agriculture" in which thecrops alternate with periods of naturalvegetative redevelopment. Tree cultureand the semiaquatic culture of rice pro-vide much better nutrient retention andconsequently have a longer life expect-ancy on a given site in these warmerlatitudes.

Selection Pressure:

Quantity versus Quality

MacArthur and Wilson (26) have re-viewed stages of colonization of islandswhich provide direct parallels withstages in ecological succession on conti-nents. Species with high rates of repro-duction and growth, they find, are morelikely to survive in the early uncrowdedstages of island colonization. In contrast,selection pressure favors species withlower growth potential but better capa-bilities for competitive survival underthe equilibrium density of late stages.Using the terminology of growth equa-tions, where r is the intrinsic rate of

266

increase and K is the upper asymptoteor equilibrium population size, we maysay that "r selection" predominates inearly colonization, with "K selection"prevailing as more and more species andindividuals attempt to colonize (Table 1,item 18). The same sort of thing is evenseen within the species in certain "cy-clic" northern insects in which "active"genetic strains found at low densitiesare replaced at high densities by "slug-gish" strains that are adapted tocrowding (27).

Genetic changes involving the wholebiota may be presumed to accompanythe successional gradient, since, as de-scribed above, quantity production char-acterizes the young ecosystem whilequality production and feedback controlare the trademarks of the mature system(Table 1, item 19). Selection at the eco-system level may be primarily interspe-cific, since species replacement is acharacteristic of successional series orseres. However, in most well-studiedseres there seem to be a few early suc-cessional species that are able to persistthrough to late stages. Whether geneticchanges contribute to adaptation in suchspecies has not been determined, so faras I know, but studies on populationgenetics of Drosophila suggest thatchanges in genetic composition couldbe important in population regulation(28). Certainly, the human population,if it survives beyond its present rapidgrowth stage, is destined to be moreand more affected by such selectionpressures as adaptation to crowding be-comes essential.

Overall Homeostasis

This brief review of ecosystem de-velopment emphasizes the complex na-ture of processes that interact. Whileone may well question whether all thetrends described are characteristic ofall types of ecosystems, there can belittle doubt that the net. result of com-munity actions is symbiosis, nutrientconservation, stability, a decrease inentropy, and an increase in information(Table 1, items 20-24). The overallstrategy is, as I stated at the beginningof this article, directed toward achievingas large and diverse an organic struc-ture as is possible within the limits setby the available energy input and theprevailing physical conditions of exist-ence (soil, water, climate, and so on).As studies of biotic communities be-come more functional and sophisticated,

one is impressed with the importance ofmutualism, parasitism, predation, com-mensalism, and other forms of symbi-osis. Partnership between unrelated spe-cies is often noteworthy (for example,that between coral coelenterates andalgae, or between mycorrhizae andtrees). In many cases, at least, bioticcontrol of grazing, population density,and nutrient cycling provide the chiefpositive-feedback mechanisms that con-tribute to stability in the mature systemby preventing overshoots and destruc-tive oscillations. The intriguing questionis, Do mature ecosystems age, as orga-nisms do? In other words, after a longperiod of relative stability or "adult-hood," do ecosystems again develop un-balanced metabolism and become morevulnerable to diseases and other per-turbations?

Relevance of Ecosystem DevelopmentTheory to Human Ecology

Figure 1 depicts a basic conflict be-tween the strategies of man and of na-ture. The, "bloom-type" relationships, asexhibited by the 30-day microcosm orthe 30-year forest, illustrate man's pres-ent idea of how nature should be di-rected. For example, the goal of agri-culture or intensive forestry, as nowgenerally practiced, is to achieve highrates of production of readily harvest-able products with little standing cropleft to accumulate on the landscape-inother words, a high P/B efficiency. Na-ture's strategy, on the other hand, asseen in the outcome of the successionalprocess, is directed toward the reverseefficiency-a high B/P ratio, as is de-picted by the relationship at the right inFig. 1. Man has generally been preoc-cupied with obtaining as much "pro-duction" from the landscape as possible,by developing and maintaining earlysuccessional types of ecosystems, usu-ally monocultures. But, of course, mandoes not live by food and fiber alone;he also needs a balanced C02-02 at-mosphere, the climatic buffer providedby oceans and masses of vegetation,and clean (that is, unproductive) waterfor cultural and industrial uses. Manyessential life-cycle resources, not tomention recreational and esthetic needs,are best provided man by the less "pro-ductive" landscapes. In other words, thelandscape is not just a supply depot butis also the oikos-the home-in whichwe must live. Until recently mankindhas more or less taken for granted the

SCIENCE, VOL. 164

on

Mar

ch 3

, 201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

gas-exchange, water-purification, nutri-ent-cycling, and other protective func-tions of self-maintaining ecosystems,chiefly because neither his numbers norhis environmental manipulations havebeen great enough to affect regional andglobal balances. Now, of course, it ispainfully evident that such balances arebeing affected, often detrimentally. The"one problem, one solution approach"is no longer adequate and must be re-placed by some form of ecosystem anal-ysis that considers man as a part of, notapart from, the environment.The most pleasant and certainly the

safest landscape to live in is one con-taining a variety of crops, forests, lakes,streams, roadsides, marshes, seashores,and "waste places"-in other words, amixture of communities of differentecological ages. As individuals we moreor less instinctively surround our houseswith protective, nonedible cover (trees,shrubs, grass) at the same time that westrive to coax extra bushels from ourcornfield. We all consider the cornfield a"good thing," of course, but most of uswould not want to live there, and itwould certainly be suicidal to cover thewhole land area of the biosphere withcornfields, since the boom and bust os-cillation in such a situation would besevere.The basic problem facing organized

society today boils down to determiningin some objective manner when we aregetting "too much of a good thing."This is a completely new challenge tomankind because, up until now, he hashad to be concerned largely with toolittle rather than too much. Thus, con-crete is a "good thing," but not if halfthe world is covered with it. Insecticidesare "good things," but not when used,as they now are, in an indiscriminateand wholesale manner. Likewise, waterimpoundments have proved to be veryuseful man-made additions to the land-scape, but obviously we don't want thewhole country inundated! Vast man-made lakes solve some problems, at leasttemporarily, but yield comparative littlefood or fiber, and, because of highevaporative losses, they may not even bethe best device for storing water; itmight better be stored in the watershed,or underground in aquafers. Also, thecost of building large dams is a drain onalready overtaxed revenues. Althoughas individuals we readily recognize thatwe can have too many dams or otherlarge-scale environmental changes, gov-ernments are so fragmented and lackingin systems-analysis capabilities that there18 APRIL 1969

Table 2. Contrasting characteristics of youngand mature-type ecosystems.

Young Mature

Production ProtectionGrowth StabilityQuantity Quality

is no effective mechanism wherebynegative feedback signals can be re-ceived and acted on before there hasbeen a serious overshoot. Thus, todaythere are governmental agencies, spur-red on by popular and political en-thusiasm for dams, that are putting onthe drawing boards plans for dammingevery river and stream in NorthAmerica!

Society needs, and must find asquickly as possible, a way to deal withthe landscape as a whole, so that ma-nipulative skills (that is, technology)will not run too far ahead of our under-standing of the impact of change. Re-cently a national ecological center out-side of government and a coalition ofgovernmental agencies have been pro-posed as two possible steps in theestablishment of a political controlmechanism for dealing with majorenvironmental questions. The soil con-servation movement in America is anexcellent example of a program dedi-cated to the consideration of the wholefarm or the whole watershed as an eco-logical unit. Soil conservation is wellunderstood and supported by the public.However, soil conservation organiza-tions have remained too exclusivelyfarm-oriented, and have not yet risen tothe challenge of the urban-rural land-scape, where lie today's most seriousproblems. We do, then, have potentialmechanisms in American society thatcould speak for the ecosystem as awhole, but none of them are reallyoperational (29).The general relevance of ecosystem

development theory to landscape plan-ning can, perhaps, be emphasized by the"mini-model" of Table 2, which con-trasts the characteristics of young andmature-type ecosystems in more generalterms than those provided by Table 1.It is mathematically impossible to ob-tain a maximum for more than onething at a time, so one cannot haveboth extremes at the same time andplace. Since all six characteristics listedin Table 2 are desirable in the aggre-gate, two possible solutions to the di-lemma immediately suggest themselves.We can compromise so as to provide

moderate quality and moderate yield onall the landscape, or we can deliberatelyplan to compartmentalize the landscapeso as to simultaneously maintain highlyproductive and predominantly protec-tive types as separate units subject todifferent management strategies (strate-gies ranging, for example, from inten-sive cropping on the one hand to wilder-ness management on the other). Ifecosystem development theory is validand applicable to planning, then theso-called multiple-use strategy, aboutwhich we hear so much, will work onlythrough one or both of these ap-proaches, because, in most cases, theprojected multiple uses conflict withone another. It is appropriate, then, toexamine some examples of the com-

promise and the compartmental strat-egies.

Pulse Stability

A more or less reguLlar btut acutephysical perturbation imposed fromwithout can maintain an ecosystem atsome intermediate point in the develop-mental sequence, resulting in, so tospeak, a compromise between youth andmaturity. What I would term "fluctuat-ing water level ecosystems" are goodexamples. Estuaries, and intertidal zonesin general, are maintained in an early,relatively fertile stage by the tides, whichprovide the energy for rapid nutrientcycling. Likewise, freshwater marshes,such as the Florida Everglades, areheld at an early successional stage bythe seasonal fluctuations in water levels.The dry-season drawdown speeds upaerobic decomposition of accumulatedorganic matter, releasing nutrients that,on reflooding, support a wet-seasonbloom in productivity. The life historiesof many organisms are intimately cou-pled to this periodicity. The wood stork,for example, breeds when the waterlevels are falling and the small fish onwhich it feeds become concentrated andeasy to catch in the drying pools. If thewater level remains high during theusual dry season or fails to rise in thewet season, the stork will not nest (30).Stabilizing water levels in the Evergladesby means of dikes, locks, and impound-ments, as is now advocated by some,would, in my opinion, destroy ratherthan preserve the Everglades as we nowknow them just as surely as completedrainage would. Without periodic draw-downs and fires, the shallow basinswould fill up with organic matter and

267

on

Mar

ch 3

, 201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

succession would proceed from the pres-ent pond-and-prairie condition towarda scrub or swamp forest.

It is strange that man does not readilyrecognize the importance of recurrentchanges in water level in a natural situ-ation such as the Everglades when simi-lar pulses are the basis for some of hismost enduring food culture systems(31). Alternate filling and draining ofponds has been a standard procedure infish culture for centuries in Europe andthe Orient. The flooding, draining, andsoil-aeration procedure in rice culture isanother example. The rice paddy is thusthe cultivated analogue of the naturalmarsh or the intertidal ecosystem.

Fire is another physical factor whoseperiodicity has been of vital importanceto man and nature over the centuries.Whole biotas, such as those of theAfrican grasslands and the Californiachaparral, have become adapted to peri-odic fires producing what ecologistsoften call "fire climaxes" (32). Man usesfire deliberately to maintain such cli-maxes or to set back succession to somedesired point. In the southeasterncoastal plain, for example, light fires ofmoderate frequency can maintain a pineforest against the encroachment of oldersuccessional stages which, at the presenttime at least, are considered economi-cally less desirable. The fire-controlledforest yields less wood than a tree farmdoes (that is, young trees, all of aboutthe same age, planted in rows and har-vested on a short rotation schedule), butit provides a greater protective cover forthe landscape, wood of higher quality,and a home for game birds (quail, wildturkey, and so on) which could not sur-vive in a tree farm. The fire climax,then, is an example of a compromisebetween production simplicity and pro-tection diversity.

It should be emphasized that pulsestability works only if there is a com-plete community (including not onlyplants but animals and microorganisms)adapted to the particular intensity andfrequency of the perturbation. Adapta-tion-operation of the selection process-requires times measurable on the evo-lutionary scale. Most physical stressesintroduced by man are too sudden, tooviolent, or too arrhythmic for adaptationto occur at the ecosystem level, so severeoscillation rather than stability results.In many cases, at least, modification ofnaturally adapted ecosystems for cul-tural purposes would seem preferable to

complete redesign.268

Prospects for a Detritus Agriculture

As indicated above, heterotrophicutilization of primary production in ma-ture ecosystems involves largely a de-layed consumption of detritus. There isno reason why man cannot make greateruse of detritus and thus obtain food orother products from the more protectivetype of ecosystem. Again, this wouldrepresent a compromise, since the short-term yield could not be as great as theyield obtained by direct exploitation ofthe grazing food chain. A detritus agri-culture, however, would have some com-pensating advantages. Present agricul-tural strategy is based on selection forrapid growth and edibility in foodplants, which, of course, make themvulnerable to attack by insects and dis-ease. Consequently, the more we selectfor succulence and growth, the moreeffort we must invest in the chemicalcontrol of pests; this effort, in turn, in-creases the likelihood of our poisoninguseful organisms, not to mention our-selves. Why not also practice the reversestrategy-that is, select plants which areessentially unpalatable, or which pro-duce their own systemic insecticideswhile they are growing, and then con-vert the net production into edible prod-ucts by microbial and chemical enrich-ment in food factories? We could thendevote our biochemical genius to theenrichment process instead of fouling upour living space with chemical poisons!The production of silage by fermenta-tion of low-grade fodder is an exampleof such a procedure already in wide-spread use. The cultivation of detritus-eating fishes in the Orient is anotherexample.By tapping the detritus food chain

man can also obtain an appreciable har-vest from many natural systems withoutgreatly modifying them or destroyingtheir protective and esthetic value. Oys-ter culture in estuaries is a good exam-ple. In Japan, raft and long-line cultureof oysters has proved to be a very prac-tical way to harvest the natural micro-bial products of estuaries and shallowbays. Furukawa (33) reports that theyield of cultured oysters in the Hiro-shima Prefecture has increased tenfoldsince 1950, and that the yield of oysters(some 240,000 tons of meat) from thisone district alone in 1965 was ten timesthe yield of natural oysters from theentire country. Such oyster culture isfeasible along the entire Atlantic andGulf coasts of the United States. A large

investment in the culture of oysters andother seafoods would also provide thebest possible deterrent against pollution,since the first threat of damage to thepollution-sensitive oyster industry wouldbe immediately translated into politicalaction!

The Compartment Model

Successful though they often are,compromise systems are not suitablenor desirable for the whole landscape.More emphasis needs to be placed oncompartmentalization, so that growth-type, steady-state, and intermediate-typeecosystems can be linked with urbanand industrial areas for mutual benefit.Knowing the transfer coefficients thatdefine the flow of energy and the move-ment of materials and organisms (in-cluding man) between compartments, itshould be possible to determine, throughanalog-computer manipulation, rationallimits for the size and capacity of eachcompartment. We might start, for ex-ample, with a simplified model, shownin Fig. 2, consisting of four compart-ments of equal area, partitioned accord-ing to the basic biotic-function criterion-that is, according to whether thearea is (i) productive, (ii) protective, (iii)a compromise between (i) and (ii) or (iv),urban-industrial. By continually refiningthe transfer coefficients on the basis ofreal world situations, and by increasingand decreasing the size and capacity ofeach compartment through compu-ter simulation, it would be possibleto determine objectively the limitsthat must eventually be imposed oneach compartment in order to maintainregional and global balances in the ex-change of vital energy and of materials.A systems-analysis procedure providesat least one approach to the solution ofthe basic dilemma posed by the question"How do we determine when we aregetting too much of a good thing?" Alsoit provides a means of evaluating theenergy drains imposed on ecosystems bypollution, radiation, harvest, and otherstresses (34).

Implementing any kind of compart-mentalization plan, of course, would re-quire procedures for zoning the land-scape and restricting the use of someland and water areas. While the princi-ple of zoning in cities is universallyaccepted, the procedures now followeddo not work very well because zoningrestrictions are too easily overturned by

SCIENCE, VOL. 164

on

Mar

ch 3

, 201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

short-term economic and populationpressures. Zoning the landscape wouldrequire a whole new order of thinking.Greater use of legal measures providingfor tax relief, restrictions on use,scenic easements, and public owner-ship will be required if appreciable landand water areas are to be held in the"protective" categories. Several states(for example, New Jersey and Califor-nia), where pollution and populationpressure are beginning to hurt, havemade a start in this direction by enact-ing "open space" legislation designed toget as much unoccupied land as possi-ble into a "protective" status so thatfuture uses can be planned on a rationaland scientific basis. The United Statesas a whole is fortunate in that largeareas of the country are in nationalforests, parks, wildlife refuges, and soon. The fact that such areas, as well asthe bordering oceans, are not quicklyexploitable gives us time for the accel-erated ecological study and program-ming needed to determine what propor-tions of different types of landscapeprovide a safe balance between man andnature. The open oceans, for example,should forever be allowed to remainprotective rather than productive terri-tory, if Alfred Redfield's (35) assump-tions are correct. Redfield views theoceans, the major part of the hydro-sphere, as the biosphere's governor,which slows down and controls the rateof decomposition and nutrient regenera-tion, thereby creating and maintainingthe highly aerobic terrestrial environ-ment to which the higher forms of life,such as man, are adapted. Eutrophica-tion of the ocean in a last-ditch effort tofeed the populations of the land couldwell have an adverse effect on the oxy-gen reservoir in the atmosphere.

Until we can determine more pre-cisely how far we may safely go in ex-panding intensive agriculture and urbansprawl at the expense of the protectivelandscape, it will be good insurance tohold inviolate as much of the latter aspossible. Thus, the preservation of na-tural areas is not a peripheral luxury forsociety but a capital investment fromwhich we expect to draw interest. Also,it may well be that restrictions in theuse of land and water are our onlypractical means of avoiding overpopula-tion or too great an exploitation of re-sources, or both. Interestingly enough,restriction of land use is the analogue ofa natural behavioral control mechanismknown as "territoriality" by which18 APRIL 1969

Protective// (mature systems) \

. 2 C rtm ent

Productive - 'Compromise(growth systems) - (multiple-use systems)

environment renviro nment

partitioned according to ecosystem develop-ment and life-cycle resource criteria.

many species of animals avoid crowd-ing and social stress (36).

Since the legal and economic prob-lems pertaining to zoning and compart-mentalization are likely to be thorny,I urge law schools to establish depart-ments, or institutes, of "landscape law"and to start training "landscape lawyers"who will be capable not only of clari-fying existing procedures but also ofdrawing up new enabling legislation forconsideration by state and national gov-erning bodies. At present, society isconcerned-and rightly so-with hu-man rights, but environmental rightsare equally vital. The "one man onevote" idea is important, but so also is a"one man one hectare" proposition.

Education, as always, must play arole in increasing man's awareness ofhis dependence on the natural environ-ment. Perhaps we need to start teachingthe principles of ecosystem in the thirdgrade. A grammar school primer onman and his environment could logicallyconsist of four chapters, one for eachof the four essential kinds of environ-ment, shown diagrammatically in Fig. 2.Of the many books and articles that

are being written these days aboutman's environmental crisis, I would liketo cite two that go beyond "crying outin alarm" to suggestions for bringingabout a reorientation of the goals ofsociety. Garrett Hardin, in a recent arti-cle in Science (37), points out that, sincethe optimum population density is lessthan the maximum, there is no strictlytechnical solution to the problem ofpollution caused by overpopulation; asolution, he suggests, can only beachieved through moral and legalmeans of "mutual coercion, mutuallyagreed upon by the majority of people."

Earl F. Murphy, in a book entitledGoverning Nature (38), emphasizes thatthe regulatory approach alone is notenough to protect life-cycle resources,such as air and water, that cannot beallowed to deteriorate. He discusses per-mit systems, effluent charges, receptorlevies, assessment, and cost-internaliz-ing procedures as economic incentivesfor achieving Hardin's "mutually agreedupon coercion."

It goes without saying that the tabu-lar model for ecosystem developmentwhich I have presented here has manyparallels in the development of humansociety itself. In the pioneer society, asin the pioneer ecosystem, high birthrates, rapid growth, high economicprofits, and exploitation of accessibleand unused resources are advantageous,but, as the saturation level is approached,these drives must be shifted to consid-erations of symbiosis (that is, "civilrights," "law and order," "education,"and "culture"), birth control, and therecycling of resources. A balance be-tween youth and maturity in the socio-environmental system is, therefore, thereally basic goal that must be achievedif man as a species is to successfullypass through the present rapid-growthstage, to which he is clearly welladapted, to the ultimate equilibrium-density stage, of which he as yet showslittle understanding and to which henow shows little tendency to adapt.

References and Notes

1. E. P. Odum, Ecology (Holt, Rinehart &Winston, New York, 1963), chap. 6.

2. H. T. Odum and R. C. Pinkerton, Amer.Scientist 43, 331 (1955).

3. A. J. Lotka, Elements of Physical Biology(Williams and Wilkins, Baltimore, 1925).

4. R. Margalef, Advan. Frontiers Plant Sci. 2,137 (1963); Amer. Naturalist 97, 357 (1963).

5. R. J. Beyers, Ecol. Monographs 33, 281(1963).

6. The systems so far used to test ecologicalprinciples have been derived from sewageand farm ponds and are cultured in half-strength No. 36 Taub and Dollar medium[Limnol. Oceanog. 9, 61 (1964)]. They areclosed to organic imput or output but areopen to the atmosphere through the cottonplug in the neck of the flask. Typically,liter-sized microecosystems contain two orthree species of nonflagellated algae and oneto three species each of flagellated protozoans,ciliated protozoans, rotifers, nematodes, andostracods; a system derived from a sewagepond contained at least three species of fungiand 13 bacterial isolates [R. Gordon, thesis,University of Georgia (1967)]. These cul-tures are thus a kind of minimum ecosystemcontaining those small species originallyfound in the ancestral pond that are able tofunction together as a self-contained unitunder the restricted conditions of the labora-tory flask and the controlled environment ofa growth chamber [temperature, 650 to 75°F(18' to 24'C); photoperiod, 12 hours; illumi-nation, 100 to 1000 footcandles].

7. G. D. Cooke, BioScience 17, 717 (1967).8. T. Kira and T. Shidei, Japan. J. Ecol. 17, 70

(1967).9. The metabolism of the microcosms was

269

on

Mar

ch 3

, 201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

monitored by measuring diurnal pH changes,and the biomass (in terms of total organicmatter and total carbon) was determinedby periodic harvesting of replicate systems.

10. P. J. H. Mackereth, Proc. Roy. Soc. LondonSer. B 161, 295 (1965); U. M. Cowgill and0. E. Hutchinson, Proc. Intern. Limnol. Ass.15, 644 (1964); A. D. Harrison, Trans. Roy.Soc. S. Africa 36, 213 (1962).

11. R. Margalef, Proc. Intern. Limnol. Ass. 15,169 (1964).

12. J. R. Bray, Oikos 12, 70 (1961).13. D. Pimentel, Amer. Naturalist 95, 65 (1961).14. R. T. Paine, ibid. 100, 65 (1966).15. G. M. Woodwell, Brookhaven Nat. Lab.

Pub. 924(T-381) (1965), pp. 1-15.16. R. G. Wiezert, E. P. Odum, J. H. Schnell,

Ecology 48, 75 (1967).17. For selected general discussions of patterns

of species diversity, see E. H. Simpson, Na-ture 163, 688 (1949): C. B. Williams, J.Animal Ecol. 22, 14 (1953); G. E. Hutchin-son, Amer. Naturalist 93, 145 (1959); R.Margalef, Gen. Systems 3, 36 (1958); R.MacArthur and J. MacArthur, Ecology 42,594 (1961); N. G. Hairston, ibid. 40, 404(1959); B. C. Patten, J. Marine Res. (SearsFound. Marine Res.) 20, 57 (1960); E. G.Leigh, Proc. Nat. Acad. Sci. U.S. 55, 777(1965); E. R. Pianka, Amer. Naturalist 100,33 (1966); E. C. Pielou, J. Theoret. Biol.10, 370 (1966).

18. M. Lloyd and R. J. Ghelardi, J. Animal Ecol.33, 217 (1964); E. C. Pielou, J. Theoret.Biol. 13, 131 (1966).

19. G. W. Barrett, Ecology 49, 1019 (1969).20. In our studies of natural succession following

grain culture, both the species-to-numbers andthe equitability indices increased for alltrophic levels but esoecially for predatorsand parasites. Only 44 percent of the speciesin the natural ecosystem were phytophagous,as compared to 77 percent in the grain field.

21. J. T. Bonner, Size and Cycle (PrincetonUniv. Press, Princeton, N.J., 1963); P. Frank,Ecology 49, 355 (1968).

22. D. W. Johnston and E. P. Odum, Ecology37, 50 (1956).

23. R. Margalef, Oceanog. Marine Biol. Annu.Rev. 5, 257 (1967).

24. F. H. Bormann and G. E. Likens, Science155, 424 (1967).

25. Increased water yield following reduction ofvegetative cover has been frequently demon-strated in experimental watersheds throughoutthe world [see A. R. Hibbert, in InternationalSymposium on Forest Hydrology (PergamonPress, New York, 1967), pp. 527-543]. Dataon the long-term hydrologic budget (rainfallinput relative to stream outflow) are avail-able at many of these sites, but mineralbudgets have yet to be systematically studied.Again, this is a prime objective in the "eco-system analysis" phase of the InternationalBiological Program.

26. R. H. MacArthur and E. 0. Wilson, Theoryof Island Biogeography (Princeton Univ. Press,Princeton, N.J., 1967).

27. Examples are the tent caterpillar [see W. G.Wellington, Can. J. Zool. 35, 293 (1957)]and the larch budworm [see W. Baltens-weiler, Can. Entomologist 96. 792 (1964)].

28. F. J. Ayala, Science 162, 1453 (1968).29. Ira Rubinoff, in discussing the proposed sea-

level canal joining the Atlantic and Pacificoceans [Science 161, 857 (1968)1, calls fora "control commission for environmentalmanipulation" with "broad powers of ap-proving, disapproving, or modifying al ma-jor alterations of the marine or terrestrialenvironments. . . ."

30. See M. P. Kahl, Ecol. Monographs 34, 97(1964).

31. The late Aldo Leopold remarked long ago[Symposium on Hydrobiology (Univ. ofWisconsin Press, Madison, 1941), p. 17] thatman does not perceive organic behavior insystems unless he has built them himself.Let us hope it will not be necessary to re-build the entire biosphere before we recog-nize the worth of natural systems!

32. See C. F. Cooper, Sci. Amer. 204, 150 (April1961).

33. See "Proceedings Oyster Culture Workshop,Marine Fisheries Division, Georgia Game andFish Commission, Brunswick" (1968), pp. 49-61.

34. See H. T. Odum, in Symposium ont PrimaryProductivity and Mineral Cycling in NaturalEcosystems, H. E. Young, Ed. (Univ. ofMaine Press, Orono, 1967), p. 81; , inPollution and Marine Ecology (Wiley, NewYork, 1967), p. 99; K. E. F. Watt, Ecologyand Resource Management (McGraw-Hill,New York, 1968).

35. A. C. Redfield, Amer. Scientist 46, 205 (1958).36. R. Ardrey, The Territorial Imperative (Athe-

neum, New York, 1967).37. G. Hardin, Science 162, 1243 (1968).38. E. F. Murphy, Governing Nature (Quad-

rangle Books, Chicago, 1967).

Visual Receptors andRetinal Interaction

Haldan Keffer Hartline

The neuron is the functional as wellas the structural unit of the nervoussystem. Neurophysiology received animpetus of far-reaching effect in the1920's, when Adrian and his colleaguesdeveloped and exploited methods forrecording the activity of single neuronsand sensory receptors. Adrian andBronk were the first to analyze motorfunction by recording the activity ofsingle fibers dissected from a nervetrunk and Adrian and Zotterman the

Copyright © by the Nobel Foundation.The author is professor of biophysics at the

Rockefeller University in New York City. Thisarticle is the lecture he delivered in Stockholm,Sweden, 12 December 1967, when, with RegnarArthur Granit and George Wald, he receivedthe Nobel Prize in medicine or physiology. Itis published here with the permission of theNobel Foundation and also is included in thecomplete volume of Les Prix Nobel en 1967, aswell as in the series Nobel Lectures (in Eng-lish), published by the Elsevier Publishing Com-pany, Amsterdam and New York.

270

first to elucidate properties of singlesensory receptors (1). These studieslaid the foundations for the unitaryanalysis of nervous function.My early interest in vision was

spurred by another contribution fromAdrian's laboratory: his study, withR. Matthews, of the massed dischargeof nerve impulses in the eel's opticnerve (2). I aspired to the obvious ex-tension of this study: application ofunitary analysis to the receptors andneurons of the visual system.

Oscillograms of the action potentialsin a single nerve fiber are now common-place. The three shown in Fig. 1 arefrom an optic nerve fiber whose retinalreceptor was stimulated by light, therelative values of which are given at theleft of each record. One of the earliestresults of -unitary analysis was to show

that higher intensities are signaled byhigher frequencies of discharge of uni-form nerve impulses.

In 1931, when C. H. Graham and Isought to apply to an optic nerve thetechnique developed by Adrian andBronk for isolating a single fiber, wemade a fortunate choice of experi-mental animal (3). The xiphosuranarachnoid, Limulus polyphemus, com-monly called "Horseshoe crab," aboundson the eastern coast of North Amer-ica (4). These "living fossils" havelateral compound eyes that are coarselyfaceted and connected to the brain bylong optic nerves. The optic nerve inthe adults can be frayed into thinbundles which are easy to split untiljust one active fiber remains. The rec-ords in Fig. 1 were obtained from sucha preparation.The sensory structures in the eye of

Limulus from which the optic nervefibers arise are clusters of receptorcells, arranged radially around the den-dritic process of a bipolar neuron (ec-centric cell) (5). Each cluster lies be-hind its corneal facet and crystallinecone, which give it its own, small visualfield (Fig. 2). Each such ommatidium,though not as simple as I once thought,seems to act as a functional receptorunit. Restriction of the stimulating lightto one facet elicits discharge in onefiber-the axon of the bipolar neuron

SCIENCE, VOL. 164

on

Mar

ch 3

, 201

1w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from