PATTERN AND THE BALANCE OF NATURE 31

Leigh Van ValenDepartment of BiologyUniversi ty of Chlcago

Chicago, I l l inols 60637

@.!: Some maJor ecological pat terns indicate that the herbivore t rophicl-evel (as def ined) is regulaEed ln the Long run by food. Three Levels of fac-tors of populat ion control" must be clear ly dlst lnguished and have di f ferentevolut ionary consequences. The remarkably regular evo1ut lon of mammaLlanbody size on isLands needs study. A considerat ion gf var ious possibl l i t iesdoes not cLearLy resolve the Enigma of Balance: Hori can i t be that somespecies regulated even uLt i .mateLy by food do not per lodical ly great ly re-duce their food supply by overeat ing?

J

Introduct ion

The subjeet of th ls paper ls one that is easy to gl-oss over. I t has aremarkable propensity for encouraging superf ic ial thought, and so is oftenregarded as t r iv ia l . Yet i t is one of the most important wel l -def lned prob-lems in ecology. I t has unexpectedly deep roots, and their ramlf icat ionsunder ly apparent ly qui te unrelated matters. Many anslrers seem pLausibleunt i l they are put lnto a broader context . My argument is necessar l ly con-voluted and branched; most of the notes are an integral part of the argu-ment but are separated for ease of readlng. I have found the problem lntel-Lectual ly more di f f icuLt than any other I have considered, including thatof the preceding paper (Van Valen, 1"973b).

Tn their wel l -known paper ' Hairston, Smith and Slobodkin (1950i = HSS)proposed, among other points, that in terrestr ial communit ies the dominantherbivores are predator- l imi ted whi le dominant predators are food- l iml ted.Murdoch (1966) and Ehr l ich and Birch (L967) cr i t ic ized HSS; most of thelrarguments were adequately met by Slobodkin, Smith, and Haireton (L967; = SSH).I wiLl present evidence that seems more compeLLing than that of HSS, thatleads to a reverse concLusion, but that does not destroy HSS's argument. Thelatter nol^r apPears as a Paradox.

To avoid i r re levant wrangles we can def ine the rrdominant herbivoresrt ofa cornrnunity as those herbivores thaL joint ly use m()st of the energy or re-duced carbon from plants that is used by al l herbivores (cf . Van Valen, L973a) .HSSts concl"usion can be rephrased as saying that, of the energy !n l iv ingplant mater ia l that is eaten, most is eaten by species whose Local popuLa-t ions are reguLated by predat ion. The over lap of t rophic levels can bepart i t ioned ln the same way. However, HSS's argument acLual ly leads to asl ight ly di f ferent conclusion: that the rate of energy f l "ow through theherbivore t rophic LeveL is reguLated by predat ion. In other words, i f theamount of predat ion on herbivores is changed and the system remal-ns stabLeand biot ical ly seLf-contained, the rate of energy f low from plants to al l eaters ofpl"ants wi l l change to a di f ferent equl l ibr ium vaLue. Removal of a l l pre-

dators would be an appropr iate i f d i f f icul t procedure.SSH dlst ingulshed between seed-eaters and other herblvoree. Janzen

(1969) has pointed out that seeds are young pLants and that their l -oss to

herbivores can have an important ef fect on the plant popul"at lon; I there-fore provis ional ly incLude seed-eaters among herblvores (1). Predators

incl-ude parasi tes here (U,

Evol. Theory L:31-49 (.rulY 1973)

32 L. VAN VALEN

Janzenrs point is important ln a more general context . I t is t rue thati f herbivores ate the same proport ion of the leaves of a plant as they do of

i ts seeds, the ef fect woul-d be much greater and usual" ly lethal , thereby pre-

cluding future seed product ion as weLl- . The existence of some Levet of seed-eat lng which strains the energy budget of the plant is al l that is necessaryto regard seed-eaters and leaf-eaters as comparable, though, wi th referenceto a s ingle plant species. The greater proport ion of seeds than leavesthat can be eaEen wlthout ser ious harm to the plant may explain, v ia food-l lml tat ion of herbivores, the observat ion that a greater proport ion ofseeds than leaves are in fact eaten. SSH used the lat ter Pattern to suPPortthe unlmportance of food- l funi tat ion for leaf-eaters, but i t is consistentwith ei ther al ternat lve. The mechani .sm poses a probLem which I consider1ater.

HSS proposed predator regulat ion for herbivores for three reasons:green plants are usualLy not appreclabl"y depleted by herbivores, weather isusual ly impl"ausible as a reguLator, and reguLat ion by predators seemed theonly widely appl lcabl .e al ternat ive. HSS proposed food regulat ion forpredators slmply because of consequences of the argument that the predatorsregulate their own food. SSH did not substant ia l ly modify any of these argu-ments. The last three arguments may be accepted in the form given and thef i rst seems descr ipt iveLy unexcePt ionable (3 ' 4)

t 'The world ls green" (Slobodkin in Murdoch, L966). This is the fact ( i fI t is a fact) to be expLained. A di f ferent way of looking at most of thesame issues is to ask how i t ls that more than one monophagous species (oftenseveral) of herblvores coexlst Local ly on a s lngle specles of p lant. Thisls related to the nature of a resource (Haigh and Maynard Smlth, L972)(5).

Levels of ControL

There are three Levdls of populat lon control that must be clearLydlst lngulshed. The dl f ference of levels ls of fundamental importance inevolut lonary ecoLogy, and confusion among them seems to have been one of themaJor dt f f icul t ies wi th theoret ical ecology in the past decade or more (g).Domlnant factors can be def ined as those that cause the greatest amount ofmortal i ty, suppresslon of reproduct ion, and other ef fects that lower theactual rate of lncrease of a populat lon beLow rm (Z). The other two l"eveLsoperate ln the regulat lon of popuLat ion densi ty. Proximal lv regulat ing fac-torg are those that are densi ty-dependent and suff ic ient ly important to keeppopulat lon densi ty wi th in some equl l ibr ium range of values. The third leve1operates more or less on the evolut ionary t ime scale and may be cal led thatof u l t imately regulat lng factors (8). These factors determine the densi tythat can be supported lndef in i te ly by the environment ( i .e. how much of theresource space is used) and thereby the leve1 at which proximal ly regulat ingfactors act. They may operate only ln very bad years,

An example where al I three levels di f fer sharpLy wouLd be a case wheremost mortal i ty and variat ton in reproduct l-on were caused over the whole pop-uLat lon by weather, apart f rom the ef fect of any densi ty-dependent refuges;the usual- reguLat lon of densi ty wae by r ig id ter t t tor ia l i ty (or densi ty-pro-duced stress, etc) ; and the slze of terr l tor ies (or the densi ty at whichstress occurs) was determlned genet icalLy, the threshoLds having been or ig in-al ly selected ln relat ion to food supply. In the exper imental popuLat ions ofthe weevi l Calandra orytae that Ehrl ich and Birch (L967) used as an i l lustra-t ion, proximal regul.at lon is by egg cannibal lsm and uLt lmate regulat ion isby food. Str ix @, the tawny owl, seems to be a species where alL threeLevels ere separate (review by Lack, L956).

PATTERN AND THE BALANCE OF NATURE

The three leve1s do not, of course, have to have di f ferent factors; one

facror (e.g. food Ltmltat ton) may operate at each level . In fact , not a l l

three levels need to occur t rn every case. Dominant factors must aLways occur 'by def ln i r lon and blol-ogical real t ty. ColonlzLng (ot , wi th dl f ferent emphases

and referents, fugl t ive, nonequt l lbr ium, r-seLected, opportunlst ic) speciesmay often, perhaps, have even no proximal- ly reguLatory factors, which is tosay they may never reach densi t ies where such factors wouLd operate. Therelat ive frequency of such species is unknown even beyond an order of mag-

ni tude and depends on the freguency of ext inct ion of l -ocal populat ions (9).

I f local ext inct ion is caused by t rophic compet i t ion, as i t normal ly isdur ing succession, then regulatory factors must occur. Species or popula-

t lons regulated proximal ly by predat ion ( including Parasi t ism) may have noul t imate regulat ion that acts speci f ical ly on them (10), a l though at someLevel of compLexi ty the predator-prey systems of which they are members wi l lbe subject to ut t imate regul-at lon (cf . Van Valen and Sloan, L966, and below).

Each level of populat ion control has a di f ferent evolut ionary s igni f i -cance. Dominant factors are of course those most imPortant in natural selec-t ion, at l -east insofar as naturaL select ion operaLes on the intr insic rateof natural increase (11). This ls not to say that natural select ion canttoperate important ly on other aspects of the 1- i fe cycle, such as deveLopment

rate or meiot ic segregat ion, that are not dlrect ly factors of populat ion

control . (Development rate is an important part of the intr insic rate ofnatural increase and so i ts determinat ion couLd perhaps be considered anaspect of populat ion control . )

Proximal ly regul-at ing factors, on the other hand, are those most impor-tant for the K select ion of MacArthur and Inl i lson (L967). T use the term inl ts or ig lnal and useful sense, select ion to increase the equi l ibr ium densi ty

K. Thls select lon operates near equi l " ibr ium densi ty and concerns t rophicef f ic lency and the l l -ke. Interference compet i t ion and defense against Pre-dat ion have a s imi lar re lat lon to proxi"mal regul-at ion but are related toother theoret ical- parameters, Because proximalLy regulat ing factors are

those most important in keeping the populat lon near equi l - ibr ium densi ty, they

wiLl" be most important in the sel-ect ion based on this densi ty.Ul t imately regutat ing factors are the basls for success or fat lure in

trophlc compet l t ion. Var l .ey (1949) seems to have been the f i rst to see this.

They determine what part of the resource space a species uses. Trophiccompet i t ion occurs onLy when two species use the same resource, normaLly an

aspect of food or space in the broadest senses, and then only when thisresource is at Least ul t lmately regulatory for at least one of the species@)

11 select ion is of ten regarded as the basis for success in t rophic com-pet l tTon. I bel leve this is l r rong in general , for the reasons given above,al though i t can be true when proximal and ul t imately regulat ing factors arethe same. In th ls case sel .ect ion to improve compet i t ion Iwhich by analogycould be cal led C select ion (13)] would coincide with K select ion. Further-more, the same adaptat ions are undoubtedly somet imes ef fect ive for d i f ferentproximal and ul t imate factors. A Larger body size may permit a longer period

without food and also be useful in int imidat ion.I f u l t lmate regul-at ion happens suff ic ient ly rarely, there may be cycl ic

microevol"ut ion. As an example, mean Lerr l tory s ize may decrease becauselndlv iduals wl th smal ler terr i tor ies can get enough food, unt i l there is anextreme food shortage, which would remove such individuaLs.

Even l f each species of herbivore (or green pl-ant) is proximal ly regu-lated by the next higher t rophic Level , i t doesn' t fo l low that the t rophicLevel 1n quest ion is so regui .ated, because the number of species coul-d be

ul t imately regulated by food. Unused food permits immigrat ion of species touse l t , however these specles are regulated (10). For th is reason' and more

33

34 L. VAN VALEN

important ly because the leve1 of regulat ion muet i tsel f be determined for theent ire predator-prey system, as discussed below, the argument of HSS canLead to conclusions onLy about ul t imate regulat ion al though they state theargument part ly in terms of proximal regulat ion. Predator and prey mayeach regulate the ot,her proximal ly, but the joint equl l ibr ium range requiresul t lmate regulat ion (14). The fact that the cLassical predator-preyequat ions are sel f-contained, at the Level- of proximal regul-at lon, is anargument against their adequacy rather than against the biol"ogy.

For the sequel i t is necessary to def ine a nehl Parameter ' ^tr@, .g,1!g,of increase (r*) . Thls is the rate of increase of a popul"at ion at opt imal-density in ar$teaL habitat , but with competi t ion and predat ion removed. Theremoval ur ight be direct l -y done experlmental l -y but is probably more easi lydone indirectLy by determining the effects of competl t ion and predat lon. I iobviously depends on the popul-at ions's phenotype and habltat , and could begeneral ized to depend on density. The effect ive environment here, the nlchein a cormon usage, is the physical environment ( including nonblologlcal inter-act ions of a l l t rophic levels) , the t rophic levels below that of the speciesconsidered, decomposers ( for the environment of non-decomposer species), andany commensal or mutual lst ic ef fects. L ike r*r r t must be posi t lve toavoid ext inct ion.

Patterns

The argtrment of HSS starts on the level of ul t imate regulat lon, wlthan observat ion of an apparentLy under-used resource. There are, ho&tever,several observat ions of comparably general appLicabi l i ty that can be made onthe same level and that lead to the reverse concl-usion (Van Valen and Sloan,L966). These observat ions are of major patterns of the distr ibut ion ofspecies in the resource space, analogous to those of El ton (1927) and Jordan(1971), and may be l is ted as foLlows:

L. The maxlmum equiLibr ium density of Large herbivores is much less perspecies than that for smal1 ones. Compare deer, mice, and aphide ln the sarneforest . This does not of course deny the posslbiLi ty of rare smat l species.The same pattern occurs for obl igate leaf-eaters.

2. The maximum number of sympatr ic species of large herblvores ls muchsmaller than that for smaLl ones. The same example appl ies. Observat ions1 and 2 can be combined, as Elton (1927) did for ent i re communit ies ratherthan single t rophic Levels in histrpyramld of numbers.rr Such a combinat lonis stronger than et ther observat ion separately and is what is direct ly re-lated to the avaiLabi l i ty of resources.

3. The densi ty wi th in comparabLe herblvorous species is usuaLly lessin habirats of low product iv i ty than in habitats of high Product lv iLy. Iknow of no data speci f icalLy on this point for terrestr ia l herbivores beyondthose revlewed by Lack (L954), but i ts t ruth is further supPorted by the great

decrease in insect density in ar id regions as one goes from a stream-border conrnunity to the dry divides that have l i t tLe vegetat ion. How thls

decrease is part i t ioned between species dlversi ty and single-species densl ty

is conjectural , but ei ther suff ices for the argurnent and I suspect bothoccur. The fact that here, as with other points, formal studies are ln-

adequate doesnft detract f rom the value of a strong quaLl tat ive observat lon

which is cornmon knowl"edge and which T have personaLly observed. I t is the

qual i tat ive observat ion i tsel f which is relevant (15).

4. The nurnber of sympatr ic specles of comparable herbivores is usual ly

less in habitats of l -ow product iv i ty than ln habttats of hlgh product iv i ty.

The evidence for this point is the same as for the previous one but the argu-ment from species number ls weaker, s ince terrestr ial predictabi l i ty and

PATTERN AND THE BALANCE OF NATURE

perhaps habitat diversi ty are highLy correlated with product iv l ty and thesemay be the important var iabtes (cf . Sanders, L968, for benthic marine comparl-sons). Observat ions 3 and 4 can also be combined.

I submit that none of these patterns can be expl"ained unless the densltyof the dominant herbivores in any habitat is regulated, at the uLt imate level,by food. Proof of such a statement requires eLiminat ion of a l l concelvablealternat ives and so is l imited by our imaginat lon. Posslble al ternat iveinterpretat ions wiLl . be evaluated later. The patterns seem to be majorstructuraL features of the l iv ing wor ld (16) and, L1ke any natural pat terns,need an explanat ion.

The Island Rule tn Marrcnal-s

As I have remarked elsewhere (Van Valen, 1970) with another emphasis,the reguLar evol-ut ion of marmnal ian body size on is lands is an extraordinaryphenomenon which seems to have fewer except lons than any other ecotypic rulein animals. Smal l herbivorous mammal"s increase ln size, whi le carnivores andungulates, and apparent l"y aLso lnsect ivores, become smalLer. Adul teLephants can be a meter ta1l . The cause of thls Patt ,ern is unknown and evenan adequate descr ipt ion of the pattern i tseLf is unavaLlabl"e.

Despi te some progress, as by Rosenzweig (1968) for carnivores, we dontt

understand why mammals have the sizes they do. The isLand rule br ings thisinto focus. Perhaps predat ion is a component that tends to make herbivores

targer; i t is commonly bel ieved that th is is t rue (Rensch, 1960). Removalof th is vector on many is lands would then lead to smalLer s lze. But rodents

and lagomorphs become larger. t r{e might compLicate the hypothesls by saying

that predators prefer middLe-sized mammal"s. This would help somewhat; i tpredicts the obsenlat ion (Foster, 1965) that Pergmyscus manlculalu.s 1s

smal ler on the tvro Queen Charlot te Is l"ands that have carnlvores ( these is lands

are also the largest, though) than on the other isLands. But lnsect lvores

thenbecomeembarrassing,and@agreSqisisas1argeontheScott ishis lands with carnivores as on those without. I t would also be desirable to

accorrnodare the ShetLand pony. Grant (1965) suggested plaustbLy that is land

birds and rodents tend to become Larger in order to use a wider var lety of

food in an area with fewer competl tors than on the mainland,One class of explanat ions for the is land rul"e involves di f ferent sortg

of reguLat ions of herbivores and predators. I f on the mainland most herbi-

vores are regulated proximal ly by predat ion and most predators by food, then

removal of each group to is lands wi lL have dl f ferent ef fects. I f there are

no carnivores on the isLands, the herbivores wlLl- have Lo be regulated by

something else, l ike food, at a perhaps higher densi ty. Kleln (1964) has

an exampLe of th is, wi th appropr iate phenotypic resul ts, in is land deer.

For predators, on the other hand, food may wel l be scarcer or less predlctabLe

than on the mainland. Smal ler individuals need to eat l "ess and so might be

selected for, a l - though i f food is proximal ly regulatory in both regions i t

is ent i reLy unclear why such selecLlon would oPerate more strongly on is lands.

Starvat ion is starvat ion at any densi ty. Tf food is relat ively scarcer for

large herbivores than for smal l ones, the same implausible mechanism would

apply to them as to predatorg. Thus the explanat ion is rather general and

is ' l imi ted only by i ts impLausibi l i ty , the Lack of evidence, and the trouble-

some appl icat ion of the is land rule to some herbivores on isLands that have

carnivores. ObviousLy, the phenomenon needs study'

35

36 L. VAN VALEN

Outl ine of the Major Argument

Some ubiqui tous patterns apply to terrestr ia l , herbivores as wel l as toother animaLs and strongly suggest that the ul t imate regulat ion of th istrophic 1evel as a whole, l ike that of other t rophic levels, is by resources,in th is case food. The patterns are that there are more smaLl animal indiv i -duals than large ones, and that communit ies with greater pr imary product iv i tyhave more animals. There shouLd also be evoLut ion to use avai lable resourcesthat aren' t ueed. A moderate proport ion of herbivores are even proximal lyregulated by food.

The general conclusion on regul-at ion of herbivores then makes i t para-doxical that the wor ld is green, Several ways exist to escape from the para-dox, but none seem especial ly pLausibLe as a general resolut ion:

1. Perhaps the wor l -d isnrt green. But i t seems to be.2. Perhaps predat ion is ul t imately regulatory af ter a l l , wi th a reser-

voir of food in the more abundant decomposer trophic l"eveL. But not enoughpredators seem to act th is way.

3. Perhaps predat ion is ul t imately regulatory af ter a1L, wi th eachpredator-prey subsystem reguLat ing i tseLf. But other species should theninrnigrate to use the avai l -abl"e food.

4. Perhaps the patterns donrt require regulat ion by food. But al terna-t ives are implausible, and the jo int herbivore-predator system i tsel f needsul t imate regulat ion.

5. Perhaps some aspect of space is regul"atory. But there is so muchspace in a forest for animals, even of restr icted kinds. We woul"d thenhave the problern that the world is open.

6. Perhaps greenness persists because herbivores eat as much as theplants can tolerate. But a sui table mechanism is di f f icul . t to f ind; theherbivores donrt know in advance that they are overeat ing.

7, Perhaps the trophic dist inct ion between l iv ing and dead plants isarbi t rary, so i t doesnrt matter to the argument i f the wor ld is green orbrown. But eaters of l ive food get there f i rst and shouLdnrt leave avai lablemorseLs to the decomposers f rom the goodness of their hearts.

8. Perhaps much that is green is inedibLe. But every plant has i tsherbivores, each of which has a posi t ive t rophic rate of increase and mustsomehow be regulated beLow that rate,

9. Perhaps the degree of palatabi l i ty depends on the densi ty of theherbivore. But th is seems implausible for vertebrate herbivores, whateverval- id i ty i t may have for arthropods.

10. Perhaps.

Other Evidence

Control of most herblvores at the ul t imate Level by resources is al-sosuggested by another argument, somewhat weaker but st i l l important. I fresource densi ty is i r re levant to herbivore densi ty, as wouLd be true byHSS, then i t is hard to understand why other species have not evolved to moreor Less f i l l up the interst ices in the resource space. At one level" theproblem is easiLy resolved, because predator- l imi ted species would evolverandomly wi th respect to resources and so some parts of the resource spacewould be empty, a l though accessible evolut ionar i ly , at any t ime. The problemreappears, however, when we consider the total" system of predators and prey(Van Valen and Sloan, L966). This system is necessar i ly at or below the den-si ty set by ul t imate regulat ion, and such a densi ty would not exist i f therehrere no ul t imate regulat ion. We do not see very dense populat ions thatl i teral- ly f i l -1 the air , as would somet imes happen without ul t imate regulat ion.

PATTERN AND THE BAT,ANCE OF NATURE

(There would then be a random walk of levels of proximal regulat ion, s incethere wouLd be no counterbalance to any force changing any current LeveL ofproximaL regulat ion. I f the forces in each direct ion are more or Less equal ,but appl ied randomLy, the expected densi ty of surviv ing populat ions increaseswithout l imi t . ) As a reLated point , species seem to be packed wlthout majorgaps in the resource space; cf .MacArthur and Wilson (L967), Raven (L967r,Cody (1968) and CuLver (1970, L972). I t foLLows that there is evoLur ionarypressure to use unoccupied or under-occupied parts of the reE ource space (andto herbivory i f HSS are r ight) , insofar as these are evolut ionar iLy accessible.

The dist inct ion between l iv ing and dead organisms ign' t a lways important.hle could th ink that what herbivores donrt eat , decomposers wl11, so i t doesnrtmatter whether the worLd is green or brown, The dist tnct ion between thesetwo trophic cLasses is probabl"y best thought of in terrns of the reproduct lveval"ues (Fisher, 1930) of the food organigms, or an analogous parameter thathas a constant maximum value unt i l reproduct ion starts. Such parametersshould be evaluated for the present purpose in the absence of predat ion andperhaps mult ip l ied by the energy avai labLe in the prey. Then surplus orsenescent indiv iduals that woul"d die anyway wi l .1 be trophicaLly equivalentto dead individuals, and predators or herbivores compete with scavengers anddecomposers. However, any herbivores (and parasi t ic fungi are such herbi-vores) that eat potent ia l ly reproduct ive plants wiLl" have an advantage be-cause they get there f i rst ; th is Part of the resource space ls open to suchorganisms wlthout interference by decomposers. Therefore the problem remains.

The evolut ion of p lants in response (against palatabl l i ty , etc) maywel l be unimportant to the argument al though important to the plants andherbivores, because the existence of any posi t ive t rophic rate of lncreasecan Lead to any equi l ibr ium densi ty. Exgept ions to th is ef fect of a posi t ivert exist and may be signi f icant i Dixon and McKay (1970) found what can beff i terpreted as a chemical defense by a tree r .r i th an effect on an aphid thatdepends on the densi ty of the aphid. We have here a model explanat lon ofregul"at ion by food that does not depLete the food. But how general thlsexplanat ion is, especiaLl"y for larger herbivores, may be quest ionabl .e. I tdoes give a causaL path for the di f ference between leaf-eaters and seed-eaters, because seeds lack the polrer of responding to the intensi ty of pre-dat ion on them after they are formed. I t is crucial , but insuff ic ient fora conclusion of the unimportance of a l le lopathy at th ls 1eveL, that , as SSHpointed out in another connect ion, every plant has i ts herbivores. This isbecause each herbivore must have a posit ive trophic rate of increase and mustsomehor^r be regul"ated below this level (L7).

Some patterns simi lar to those in a previous sect ion give further evi-dence on regulat ion by food; e.8. , the deer-wol f system is l -ess dense thanthe aphid- ladybedt l-e system ( including or excl-uding the other components ofboth). Since predator-prey systems themselves therefore fa lL under ul t lmateregulat ion, there wi lL be some amount of compet i t lon for resources in mosthabi tats, at Least occasional ly.

I f predators ul t imately regulate herbivores and are ul t imately reguLatedby them, i t is d i f f icul t to see what regulates the jo int system. HSSpointed out thatr faLl- organisms taken together are l imi ted by the amount ofenergy f ixedr ' and that decomposers (and also pr imary producers) are resource-l- imited. I t fol lows by subract ion that the tsro remaining classes, herbivoresand predators, are also resource- l imi ted wtten considered as a s ingl .e systemor are ul t imately reguLated below this level , But why should ul t lmate regu-l"at ion, i f not i tsel f due to relat lve ocarci ty of resources, have any rela-t ion whatever to the amount of resources? There are many communlt les onearth, of wldely di f fer ing k inds, snd l f the level of regulat ion of theirherbivore-carnivore systems were independent of resources r^re woul-d expect

37

38 L. VAN VALEN

many or most of them to be proximal ly regulated, on occasion, above this1eveL, and even many orders of magnitude above i t . Of course such cornrnuni-t ies would soon exper ience more or less catastrophic decLines, but s lncethese decl ines would be due to resource scarci ty i t fo l lows that ul t imateregulat ion by resources occurs even for the hypothet ical ly exempt comrnuni-t ies. I t therefore occurs for al l conrnunlt ies. Because l-n a herblvore-carnivore system i t is the herbivores which are trophical ly in contact withthe rest of the community, the conclusion that ul t imate regul"at ion occurgfor th is system must also apply to the herbivores alone, especial ly elncethe large major i ty of the energy or reduced carbon in the system occurs inthe herbivores.

I f even every species at one trophlc leve1 is opportunist ic or predator-control ted, the patterns of food avai l -abi l i ty would seem to lndicate a con-trol of species number by resources. Addi t ional species wouLd then lmmigrateto use the uneaten food. What is t rue for each may not be true for al ljo int ly. But the s ize patterns seem more restr ict ive on the regulat ion ofindiv idual specles.

HSS state that plants which are overgrazed would be replaced by otherswhich are not. Clear ly th is occurs somet imes, as l -n the replacement ofelephant-eaten trees by grass in East Afr lca (Wing and Buss, 1970). I t doesnot always occur, however (nood, I97A; Bartholomett ' 1970), and the existence

at ecological equi l ibr iurn of such cases of bare ground caused by overgrazlngmakes part icularLy acute the problem of why overgrazing is not more common.

In the intert idal zone and oceanic plankton, aLgae are conmonly more or Leseel iminated by herbivores (Sutherland, L970; Beers and Stewart, L97L). I f tynot on Land? I , Ihy is i t that the grasses which replace the trees are not

themselves overgrazed, by something other than elephants? Again i t 18 impor-

tant to note that every p1"ant has i ts herbivores.SmaLl animals tend to have a larger trophic rate of increase ( i ) than

do large animaLs, in a causat feedback l r t i th higher mortal i ty, but thrs ls

i r re levant to average densi ty for equiLibr ium species. For colonizers

i t would give the observed resul t , Yet i t is inadeguate, Aphids ere more

numerous than voLes, and both have gome attr ibutes of colonizers. Clear ly,i f aphids had as smal l an x-L as voLes, expressed in unLts of absolute t ime'they would be much less numerous and perhaps ext inct. hle see, though, thati f voles did become as numerous as aphids there would be a rapid deplettonof sui table vegetat ion. r r can evoLve in part independent ly of body size.

The amount of energy avai l3ble sets a f . imit to the amount that can pass

through herbivores, however the lat ter may be part i t ioned in body size and

species. (Some pret iminary resul ts of mine indicate that the part i t ioning

is var iable but rather predictabLe.) Without ul t imate regulat ion, ordinar i lyby food, there seems to be no reason why even each coLonizing species should

not of ten be, sgy, a miLLion t imes more numerous than i t is (18). In fact ,

i f there were extreme coloniz ing species and so rg and dispersal were al l

that ever mattered, a species wi th as low an r t F a vole shoul"dnft exist .

Species that do not sometimes experience regulat ion may therefore be rare.

Body size has other ef fects. A given environment tends to be of a

larger grain s ize for sma1l animals, i .e. their smal l" s ize in l tseLf helpe

them dist inguish among aspects of the environment that would be more unlform

to a larger animal. . They shoul-d, therefore, of ten have e competl t ive advan-tage; cf . Levins (1968). Thts and other disadvantagee of large eLze areoften of fset by advantages (Rensch, 1960). I t ls therefore unclear whygrain s ize should have a di f ferent status f rom other adapt lve aspects ofbody size. I t is not an imposed constraint , as is resource avai labl l " i ty .perhaps there are more species of smal l animals because they can Part i t ionthe environment more f inely. This l touLd be related to ul t imate regulat ion

PATTERN AND THE BALANCE OF NATT]RE

by resources because i t is determined by t rophic compet i t ion, so i t does notaffect the argument.

Large terrestr ial species have fewer refuges from predat ion and somight be considered more suscept ib le to predat ion. This reverses the usual ,and I bel" ieve wel"L-founded, statement of the reLat ionship. There are manysmaLl predators, and Larger body size is a protect ion against them; targepredators of ten eat much smal ler prey (woLves eat mice, skunks eat lnsects,etc. ) (19).

1f Leaf-eat ing herbivores compete for food they should, ln appropr latecases, exempl" i fy the contrnon phenomena of character displacement and ecoLogicalreLease in feeding-related structures and behavior, This hrould then impLyul t imate regulat ion by food. T know of no evidence on this point (20).

Darwin (1859, PP. 67'68), Paine (L966, L97L), and others have foundthat a predator of ten (not a1-ways: Harper, 1-969) permits coexistence of morespecies than could coexist in i ts absence. This is commonly bel ieved toindicate relat ive unimportance of resource part i t ioning in coexistence. Thereverse, however, seems to be the case. lJhen adequate analysis is made ofsuch cases l t is the compet i t ively domlnant specles that is predominant lyharmed by the predator. Th{s reduces i ts compet i t lve advantage ln relat tvelymarginal s i tuat ions and forces i t into a smal ler n iche, where i t is st iLLsuper ior and from which i t can overf low into other niches i f predat londecreases (Van VaLen, in press).

Conc 1us ion

I interpret the total argument above, in conjunct ion wi th HSS and SSH,to indicate that dominant herbivores, L ike green plants, decomposers, andcarnivores, are ul t imately regulated by food. This is a somewhat odd concl"u-sion because i t requires each trophic leveL to be regul"ated by the one belowwithout in turn regulat ing i t . The dist inct ion between proximal and ul t imateregulat ion may help, because the regulat ion from bel"ow can be ul t imate whi l .ethat f rom above cannot be, The large overlap in trophic levels may alsohe 1p.

There seems to be evidence, therefore, that decomposers, green pLants,herbivores, and predators al l exper ience ul t imate reguLat ion of populat iondensity to en appreciabLe extent. What then are l i re to do with the observat ionof an apparent ly highly under-used food resource, palatable l iv ing leaves?I 'The world is green.r ' I f u l t imate regulaEion occurs onLy very rarely,this would suff ice, but known mechanisms seem to require i t to happen toofrequent ly to be invis ibLe. Perhaps one of the observat ions (such as palat-abi l i ty) is wrong, al though this wouLd need to be shown. In areas dominatedby annual"s, perennial herbs that die down in the winter or dry seaaon, anddeciduous woody plants, the probLem is not acute because herbivore regulat ioncould be in the unfavorable season. Diapause of ten al leviates th is seasonal

severi ty, however, and the abundance of conmunit ies with leaves presentthroughout the year makes i t i r reLevant to the generaL problem. Many insectsfeed on f lowers, which are only seasonal ly avai labLe for most species (Evansand Murdoch, 1968). Hughes and Gi lbert (1968) suggest that there ie not infact much under-ut iL izat ion of green plants, because they need thelr Sreenparts to survive and may need about as much as ls lef t to them by the herbl-vores. Rafes (1970), Batzl i and Pi teLka (1970), l { lng and Buea (1970) andothers imply the sarne conclusion.

The probl"em nevertheLess remains. With unused food in the plants,however necessary to the pLanE populat ion' there should be intraspeci f icselect ion in herbivores for use of th is food i f the popuLat ion is proximal ly

seLf-regu|ated. Greater use rnight el iminate the food pLant (Wlegert and

39

40 L. VAN VALEN

Owen, L97L), but the herbivore doesn't know this before i t happens. Inrmediateeffects of overuse wi l l of ten be distr ibuted among the ent i re herbivore popu-lat ion, not just one fami ly, so k in select , ion for seLf-regulat ion seemsimpl"ausibl"e as a general mechanism. Group select ion is too weak to offsetindividual select ion (Maynard Smith, 1964), a conclusion which also appl iesto sel .ect ion among ecosystems. This statement is not contradicted by recentuse of group select ion in special contexts (Levins, L97I; Van Val .en,L97L) (31). Wynne-Edwards (1962) has discussed the general subject f rom adi f ferent v iewpoint . Interference compet i t ion, such as terr i tor ia l i ty ora1- leLopathy, is re l -evant onl"y to proximal reguLat ion, s ince i tg leveI- isdetermined by ul- t imate reguLat ion,

An al ternat ive explanat ion for the strong relat ionship between insectabundance and pLant densi ty in dry areas is that the plants provide a sui t -abte microcLimate, especial ly in relat ion to conservat ion of moisture. Thenplant product iv i ty controls insect abundance but by a component, of epacerather than of food, This seems plausibLe, al though I know of no studiesthat dist inguish the al ternat ives other than proximal ly. I t apparent lydoesnrt explain the reguLat ion of insect abundance in moist areas, even r lverval leys in ar id reglons. There may be a correlat ion between product iv l ty andthe abundance and diversi ty of refuges from cl imate and predat ion. Sat is-factory refuges di f fer extremely among di f ferent k inds of species, though,and i t must be shown that these vary together or that one kind is predominant lylmportant, Plant mater ia l a lso can provide cover f rom predators, so l tsdensi ty may be direct ly related to suscept ib iL i ty of an animal" to predat ion.Ul t imate regulat ion, however, wouLd explain why the predator-prey system lsat i ts actual LeveL rather than at some very di f ferent leveL, and unlessthere is a suf f ic ient reservoir of prey in the decomposer t rophic level (22)i t is c lear that predat ion cannot ul t imately reguLate the system. I f pre-dators survived by eat ing pLant mater ia l when animal prey was insuff ic ient ,their u l t imate regulat ion woul"d then be by the abundance of pLant mater la lfor food, which would thereby regulate the system.

Space is ul t imateLy regulatory for many mot i le as wel l as sessi le specles,and l t may be that this somehorir provides a vray around the di l -euuna. However,the di f ference between seed-eaters and leaf-eaters 1s then unexplalned; thepath is st iLl foggy and the l ight may be an ignis fatuus. Moreover, manyspecies seem to lack restr ict ive space requirements and shoul"d immigrate orevolve to use the avai labl-e food.

Regulat ion of terrestr ia l" herbivores by pLants, and probably ul t imateregulat ion by food, can be seen in the fer t iL izat ion exper iment of Hurd,Mel- l inger, Wolf , and McNaughton (1971), which was done for another purposebut in which arthropod herbivores consistent ly increased ln the fer t l l izedareas with more plant product lv i ty. Vertebrate herbivores, however, r { rerenot considered. Elaborat ion of th is approach and those of microcl i .mate anddensi ty-dependent a1Lelopathy, considerat ion of rar i ty of a pLant relat iveto the dispersal of i ts special ized herbivores, measurement of the ef fect ofremoval of appropriate predators from natural communit les, and study ofs impler natural systems that are t rophlcal ly more or less sel f -contained,may be the best means of at tack.

The egest ion of sugars by sap-feeding lnsects suggests that energyavai labi l i ty is not di rect ly l iml t lng to them even ul t tmately, but 1f u l t t -mate regulat ion ls by, say, n i t rogen avalLabl l t ty the ef fect on the plant lethe same because the sugars are lost whether used by the aphld, en ant, ora decomposer. I f proximal reguLat ion of herbivores ls ugual ly by predat lon,sui table space, or other external- means, th is reguLat ion Limlts the scope ofselect ion for use of more food but does not el iminate the problem becaugeanother species could then iuunigrate and use the remaining food. This

PATTERN AND THE BALANCE OF NATURE 4L

species is presumably in fact absent because i t is outcompeted by the speciesalready present, but compet i t ion for food requires ul t imate regulat ion byfood and therefore a relat ive shortage. In extreme environments whlch onlya few species can tolerate, iumigrat ion may be unimportant unLess the environ-ment lasts long enough for evolut ion to occur in other species, but suchenvironments are rather rare and so do not affect the problem.

The problem is suff ic ientLy acute that I give i t a name, the Enigrna ofBalance: How can i t be that some species regutated even ul t lmately by fooddo not per iodical- ly great ly reduce their food by overeat ing?

Acknowledgments

This paper rdas once a third i ts present l "ength and straightfornrard, buti t t reated many things too concisely. Perhaps i t st iL l does. I am gratefuLto the fol- lowing people for comments on versions of this paper over severalyears: H, G, Baker, J. I^I . Curts inger, B. DalzelL, J. F. Fox, N. G. Halrston,G.E. Hutchinson, D.E. Janzen, H,W. Kerster, C.E. King, P. Lane, R. Levins,R.C. Lewont in, A. Zomnicki , K.A. MacKay, D.B. Mertz, W.tr{ . Murdoch, P.H. Raven,T.J.M. gchopf, and J.W. Wi lson. The t i t le of the paper is s imi lar to thatof a book by Wi l l iams (1964) on another topic.

Notes

(t) . Seed-eaters may never af fect the total amount of p lant mater la l ln anarea, or more precisely may never af fect the total pr lmary product iv l ty.I f Leaf-eaters of ten do (and they do sometimes, as discussed later) ,then i t rn ight seem.that seed-eaters are i r re levant to the problem afteral l . SSH agree that seed-eaters are regulated by food. For the pur-poses of the present paper, i t doesn' t matter whether one consldersseed-eaters to be herbivores; the arguments apply to both cases.

Q). The usual depl .ct ion of a food pyramid is incomplete; i t should end withthe highest level of parasi tes, not the highest level of predators,when these are (as usual) d ist inguished. Lions are food for f leas.

G). Pul l iam, Odum, and Barret t (1968) and Perkins (1970) th ink that thehigher equi tabi l i ty they found for predaceous insects and f ish thanfor herbivorous ones supports resource l imitat ion for the predatorsand not for the herbivores, but the work of Cohen (1967, 1958) andSanders (1968) in di f ferent ways makes this argument suspect in thatequi tabiLi ty can have diverse causes. This does not, of course,affect the argument of BSS.

G). I t has been suggested that a mathematical t reatment of the subjectwould be preferabl"e. I am hardly ant i -mathematical ( I teach a shortcourse in biometr ics), but I don' t see how at present a useful . mathe-mat. icaL version couLd be ei ther more r igorous or anywhere nearly asgeneral . I ts persuasiveness would be spur ious because the mathematlcalcontext is inadequate. The comrnon prejudice against non-mathematicaltheory is cur ious for a subject l ike ecoLogy whose mathematicalfoundat ions are of shi f t ing sand.

C5). I f a predator has al ternate food, i t can regulate an indef in i te numberof preferred prey species. Man in fact does eo, part ly by th is mecha-nism. Such verbal (and tr iv ia l ly quant l f iable) counter-examples to amathematicaL theory require gett ing outside the theory to thlnk of lhemor even consider their val id i tY.

LO. The f i rst person to dist lnguish between proximal and ul t imate regulat ionseems to have been Baker (1"938). Lack (1954) fo l lowed him ln th is, and

Van Valen and Sloan (1966) mede the same dlst inct ion I t i th di f ferent

l+2 L. VAN VALEN

terms. Nevertheless, the dist inct ion is almost universaLLy ignored,perhaps in part because i t hss never been incorporated into the mathe-mat ical theory. Such incorporat ion can be done directLy but ls on thenext level" of abstract ion from the received theory. A danger ln mathe-matical theory in biology is that the assumptions tend to demarcaEe thedevelopment of subject . This is part icular ly ser ious when, as wi thecology, the assumptions are known to be nrong al though usefuL. Theyare s impl i f icat ions rather than ideaLlzat ions in that the resul ts aresometimes importantLy wrong too.

A). By rm I mean the maximum rate of increase of a populat ion with a stabLeage distr ibut ion; the condi t ions under which i t occurs def lne the opt l -maL environment and density of the populat ion.

(S). Ul t imate regulat ion can i tsel f have more than one leveL. For instance,the amount of food may be ul t imatel"y regulat ing but a greater range offood might permit a greater controL of energy. Then the second levelof uLt imate regulat j .on ls whatever l t ls that keeps such e greaterrange of food from belng achieved. A recent faeclnat lng account byIsaev and Khlebopros (1973) of the beet le Monochamus uruesovi , d i f fereln that a great expanston of d let occurs regular ly at h lgh denslElee.presumably the added klnds of food are subopt imal and contr ibute toproximal regulat ion, but th is is not c lear f rom the rePort .

€) . I t may be possible to est imate the maximum frequency of such species lnanimals most easi ly, aLthough very roughly, by seeing what proport lonof species can be included in a set to which the patterns descr ibed inthe next sect ion do not apPly. For plants, at least easl ly v ls ib leones, and for s irni larLy inunobiLe animals, direct observat ion ls posslbleand shoul-d be attempted, al though for perenniaLs a very Long t lme isclear ly necessary. For some species histor ical records may be appro-pr iate, aLthough these are most l ikely to be avai lable in disturbedareas, and for some cladocerans (cf . Deevey, 1969) and perhaps a fewother cases the fossiL record may be directLy appl icable.

(Lo. I t is appropr iate here to note Janzen's theory of the promot ion of t reediversi ty in t ropicaL forests by the act ion of species-speci f ic seedpredators that eat every seed that fa l"Ls near the parent (Janzen, L970).This produces an ent i re ly seLf-contained system and there would be muchbare ground i f onLy one tree-herbivore pair existed. Of course othersthen inrnigrate, so the mechanism isnrt d i rect ly relevant to the regula-t ion of the herbivore trophic l"evel. Haldane (L949) had proposed astructural- ly s imi lar reguLat ion based on disease.

(U). The symbol r has several re lated meanings, at least two of which areoften confused (cf . MacArthur and Wilson, L967, pp. 78 and L49) z theintr insic rate of increase and the rate of change in actual populat ionsize. Andrewartha and Birch (1954) and others have used separate syrn-bols. Because of the apparent l .y unresolvabLe ambiguity of the symbol"rr I propose that the act_ual rat€ of change of a populat ion's sLze or: .densi ty be cal led gs.

( I2). There can of course be no competi t ion for a resource that is alwayssuperabundant reLat ive to the needs of both species and that they haveno di f f icul ty f i -nding, l ike oxygen for air -breathers, and compet i t iondoes not exist when the l imit tng resource is everywhere reduced by anextraneous catrse that has a much greater effect than the organisms, aefor oxygen ln a reducing hypol imnlon. The l -at ter ef fect ie equlvatentto that of other harsh habi tats but le of ten regarded ae rrophlc co"mpe-t i t ion.

(13). In rhe received theory cu, refers to both trophlc (and other reeource) andinterference compet i t ion, which af fect d i f ferent levels of regulat lon,

PATTERN AND THE BALANCE OF NATURE

It would be useful to separate these two processes,(L{) . I t is impossible for a popuLat ion to be regulated proximal ly at any one

t ime by more than one factor, whether these factors are at one or moretrophic 1eveI"s, unless the weaker factor increases the sensl t lv i ty ofthe populat ion to the stronger factor. Thi .s happens, ln other words,i f ( for resources) an increase in ei ther lets the organism survive ona smal ler amount of the other per indiv idual- . More precisely, agiven strengthening of a factor A (decrease i f a resource, increase l fa physical stress or predat ion) wl thout change in factor B decreasesequi l ibr ium populat ion densi ty in a given environment f rom a level uby some proport ion x, which provides a comparat ive measurement scale.A given strengthening of factor B without change in factor A decreagesequi l" ibr ium density from u by some factor_;g4x. The same strengtheningas before of both factors s imuLtaneously, decreases equi l ibr ium densi tyfrom u by a proport ion ax. The value of ig, whlch may be a funct ion ofthe changes in A and B, measures the interact ion. I f and onLy i f a )1can the populat ion be said to be regul"ated proxlmal ly by both factorssimultaneously, This occurred, for example, ln an exper imentwtth !gg@!g (Slobodkin, 1961), seems to be common for photosyntheeis,and was perhaps f i rst not iced by Darwin (1859, pp, 67-68). Proxlmalregulat lon at appreciably di f ferent t imes or pl .aces by any set ofdi f ferent factors is of course not excluded. The regulat ion is of oneparameter (we can think of i t as densi ty or rg); separat ion lnto twoparameters, as is somet lmes done to give an appearance of doubl .e regula-t ion, is therefore misleading even though double reguLat ion does some-t imes exist .

( f$. Faut in (L949-I957) has given data on cr icet id mice ( largeLy herblvorous;Onvchomvs is absent) for al t i tudinal ly separated communit ies in Wyoming,but the product iv i ty of these communit ies is unknown. I have includedonl"y 3-day standard censuses and used unweighted meens of aLl years foreach local i ty; most local i t ies l rere sampled for several years. Forseven local- i t ies in the dr ier and Lower habi tats, sagebrush and grass-Land, the mean number of mice per year per local i ty was 1l- and the meannumber of species r tas 1-.6. For 3 local i t ies of mountain mahoganychaparral , the values were 53 and 2.4. For 4 local" i t ies of coni ferousforest , the values were L8 and 3.2. The onLy, or the onLy other, valueapparentLy odd is f rom 3 local i t ies in aspen, which had values of 8 andL.6. A11 comparisons except aspen and grassL4nd are pairwise s igni f i -cant. Most of the mice are largely granivores rather than herbivoresin the sense of SSH. The example is weak. Rosenzweig and Winakur(1969) compared numbers of species and indiv idual"s of largely herbivorousmice on Ll- to 15 plots in ar id regions of Ar izona in 2 succegsive years.They have a measure of total fo l iage for each plot , which is herepresumably related but far f rom ident ical to product iv i ty. Their datagive correLat ions between fol iage and number of species of 0.10 and0.72 (only the lat ter s igni f icant) and correlat ions between fol iage andnumber of indiv idual"s of -0.1-6 and 0.54 (nei ther s igni f icant) . Forthe ocean, Blackburn (1966) and previous workers whom he ci tes havefound a posi t ive correlat ion, among stat ions, between the densi ty ofherbivores (zooplankton) and pl"ants (phytoplankton). Wynne-Edwards(1962, pp. 2-3) reviewed simi lar data for marine birds and phytoplankton,as did Lack (1954) for other aquat lc birds. The eeasonal correlat lonof abundance of edible plant$ and herbivores, reviewed by l lutchlnson(1966) for the l imnoptankton, may also be relevant.

(1-6). The ambigui ty of the word "btospheretr is amuslngly evldent f rom theSeptember, 1970, issue of Scient i f ic American, which had thls wordas i ts theme.

43

PATTERN AND THE BALANCE OF NATURE

Vesey-Fi tzGerald, D,F, 1960. Grazing successjon among East Afr ican gameanimals. Jour. Mammal- . 4IzL6l-172.

l latson, T. F. L964. Inf tuence of host pl -ant condi t ion on populat ionincreases of Tetranychus telar ius (Linnaeus) (Acar ina: TeLranychidae.)Hj lgardia 35:273-322.

Way, M.. I . 1968. Intra-specl f ic mechanisms with special reference to aphidpoprr lat ions. Symp. Roy. Entomol. Soc. London 4z18-36.

h/elr l r , S.D. 1969. Ext inct ion-or ig inat ion equi l ibr ia in late Cenozoic landmanmals of North America. Evol-ut ion 232688-702.

Inl t r l taker, J.O. , Jr . 1967, Habi tat re l -at ionships of four species of micein Vigo County, Indiana. EcoLogy 482867-870.

Wiegert , R.G,, and D.F. Owen. I97L. Trophic structure, avai labLe resourcesand populat ion densi ty in terrestr ia l vs. aquat ic ecosystems. Jour.Theor. Bio1" 30:69-81.

Wil l iaros, C.B. 1964. Patterns in the Balance of Nature. London: AcademicPress. 324 pp.

I^I i l l iams, C.E., and A.L. Caskey. L965, Soi l fer t i l i ty and cottontai lfecundi ty in southeastern Missour i . Amer.Midland Nat. 74z2LL-224.

Wlng, L.D., and I .O. Buss. L970. El"ephants and forests. Wi ld l i fe Monog.L9 zL-92.

Wynne-Edwards, t r { .C. L962. Animal Dispersion in Relat ion to Social Behaviour.Edinburgh: Ol iver and Boyd. 653 pp.

49

46 L. VAN VALEN

Iivans, F.S. and I^/ .W. Murdoch. L968. Taxonomic composi t ion, t rophic structureand seasonal occurrence in a grassland insect community. Jour. Anim.l ico1. 37:259-273.

I . 'atr t in, l t .W. 1949-1957. (Wyoming censuses. ) Iq J.B. Calhoun and A.A.Arata (eds.) , North American Census of Sma1l Mammals. No. 2, mimeo-graplred. Nos. 3-4, Bar Harbor, Maine: Jackson Laboratory. Nos. 5-9,l lethesda, Maryland: U.S. Publ ic Heal th Service.

Findley, J.S., and C.J. Jones, L962. Distr ibut ion and var iat ion of thegenus Microtus in New Mexico and adjacent areas, Jour. MammaL.43 : I54'166 .

Foster, J.B. 1965. The evolut ion of the rnanrnals of the Queen Charlot teIs lands, Br i t ish Columbia. Occ. Pap. Br i t ish Columbia Prov. Mus.14: 1- 130.

Glasgow, J.P. 1963. The Distr ibut ion and Abundance of Tsetse. Oxford:Pergamon Press, 24L pp.

Gwynne, M,D., and R.H.V. 8e11. 1968. Select ion of vegetat ion components bygrazing ungut l -ates in the Serenget i Nat ional park. Nature 220:390-393.

Haigh, J. , and J. Maynard Smith. 1972. Can there be more predators thanprey? Theor. Pop. Bio1. 3:290-299.

Hairston, N.G., F.E. Smith, and L.B. Slobodkin, 1960. Community structure,populat ion control , and compet i t ion. Amer. Nat. 94242L-425.

I ta ldane, J. B. s. 1949. Disease and evolut ion. Ricerca Sci . 19(suppl . ) :68-7 6.

I larper, J.L. L969. The role of predat ion in vegetat ionaL diversi ty.Brookhaven Symp. Bio1, 22248-62.

l1orn, l i .s , , and R.H. MacArthur. I972, compet i t ion among fugi t ive speciesin a har lequin environment. Ecology 532749-752.

l luf faker, C.B. 1957, Fundamentals of b io logical controL of weeds. Hi lgardia27zlO1-157.

Hughes, R.D., and N. Gi lbert . 1968. A model of an aphid popuLat ion - ageneral statement. Jour. Anim. Ecol" . 37. .553-563.

l lurd, L.E. , M. v. Mel l inger, L. L. I , r Io l f , and s. J. McNaughton. r971. Stabi l i tyand diversi ty at three trophic levels in terrestr ia l successionalecosystems. Science 173:L134-1136.

Hutchinson, G.E. 1966. A Treat ise on Limnology. Vol . 2. New york: wi l "ey.1115 pp.

fsaev, A.S., and R.G. Khlebopros. I973. Pr ints ip stabi l 'nost i v dinamikechis lennost i lesnykh nasekomykh. Dokl , . Akad. Nauk. S.S.S.R.2O82225-227.

Janzen, D.H. 1966. Coevolut ion of mutuaLism between ants and acacias inCentraL America. EvoLut ion 20:249-275,

. L967a. Fire, vegetat ion structure, and the ant x acacia inter-act ion in Central America. Ecology 48226-35.

. 1967b. Interact ion of the bul-1rs-horn acacia (Acacia cornigeraL. ) wi th an ant inhabi tant (Pseudomyrmex ferruginea f . .S* i t f i l ineastern Mexico. Univ. Kansas Sci . Bu11. 47:3I5-559.

L969. seed-eaters versus seed size, number, toxic i ty and dis-persal . Evolut ion 23 zl-27.

. l97O' Herbivores and the number of t ree species in t ropicalforests. Amer. Nat. 104:501-528.

Jarman, P.J. L971, Diets of large mammal,s in the woodLands around LakeKariba, Rhodesia. Oecologia B:157-178.

Jordan, c.F. L97L. A world pattern in pLant energet ics. Amer. sci . 592425-433.

Ki tching, J.A., and F.J. Ebl ing. 1961-. The ecology of Lough rne. xr .The contrgl o{ a lgae_by Paracentrotus l iv idus (nchinoidea). Jour.Anim. EcoL. 30:373-383.

PATTERN AND THE BALANCE OF NATUIIE

Klein, D,R. T964. Range-related di f ferences in gror^r th of deer ref lectedin skeletal rat ios. Jour. Manrmal. 452226-235.

Kopl in, J.R., and R.S. Hoffmann. 1968. Habi tat over l -ap and compet i t iveexclusion in voles (Microtus). Amer. Midland Nat. BO:494-507.

Lack, D. 1954, The Natural Regulat ion of Animal Numbers. Oxford: ClarendonPress. 343 pp.

. 1966. Populat ion studies of Birds. oxford: c larendon press.341 PP.

Lamprey, H.F. 1963. Ecological separat j -on of the large mammal. species inthe Tarangire Game Reserve, Tanganyika. East Afr ican Wild Li fe Jour.I :63-92.

Levins, R. 1968. Evolut ion in Changing Enrr i ronments. Pr inceton: Pr incetonUniv. Press. 120 pp.

. L97L. Ext inct ion. f3 Some Mathematical- Quest ions in Biology(M. Gerstenhaber, ed.) , pp. 75-L07. Provldence: American MathematicalSoci ety.

Zomnicki , A. (MS). Evolut ion of the herbivore-plant, predator-prey and para-si te-host systems: a theoret j .cal mode1.

MacArthur, R.H. L971. Patterns of terrestr iaL bird conrmunit ies $ AvianBioLogy (D.S. Farner and J.R. King, eds.) , vol . L, pp. LB9-22L. NewYork: Academic Press.

. 1972. Geographical Ecol-ogy. New york: Harper and Row. 269 pp,and E.O. InI i l "son. L967. The Theory of Is land Biogeography.

Pr inceton: Pr inceton Univ. Press. 203 pp.Maynard smith, J. L964, Group select j .on and kin seLect ion. Nature

2}lzLl45-1L47 .Muidoch, W.I{ . 1966. rrCommunity structure, populat ion control , and compet i -

t ionrr- a cr i t iqrre. Amer. Nat. L00:2L9-226.Murton, R.K,, N.J. Westwood, and A,J. Isaacson. L964. A prel iminary

study of the factors regulat ing populat ion s ize in the woodpigeonColuJnba paluqb-lrs. Ibis L06:482-507 .

Nicholson, A.J. 1-958. Dynamics of insect popuLat ions. Ann. Rev. Entomol- .3: IO7 -136.

Odum, E.P. I97I . Fundamental-s of EcoLogy. 3rd ed. Phi ladelphia: Saunders\74 pp.

Paine, R.T. L966. Food web complexi ty and cornmunity stabiLi ty. Amer Nat.L00:65-75.

. 1 '97L. A short- term exper imental invest igat ion of resource part i -t ioning in a New Zealand rocky intert idal habi tat . EcoLogy 52:1096-1106.

Perkins, P.L. L970. Equi tabi l i ty and trophic leveLs in an Eocene f ishpopulat ion. Lethaia 3:301-310.

Pi te lka, F.A. 1959. Populat ion studies of l "emmings and lemming predatorsin northern Alaska. Proc. I -5 Inr . Cong. ZooL. ( for 1958)2757-759.

Pul l iam, H.R., E.P. odum, and G.w. Barret t . I "968. Equi tabi l i ty and resourcel imitat i on. Ecology 492772-774.

Rafes, P.M. 1970. Est imat i ,on of the ef fects of phytophagous insects onforest product ion. In Analysis of Temperate Forest Ecosystems(D.8. Reichle, ed.) , pp. 100- l -06. New york: Spr inger-Ver l_ag.

Raun, G. G. , and B. J. Ia l i lks , L964. Natural h istory of Baigmys tavlor iin southern Texas and competi t ion with Sigmodon hispidus in a mixedpopulat ion. Texas Jour. Sci . L6:28-49.

Raven, P.H. 1967. The f lor ist ics of the Cal i fornia is lands. Ig proceedingsof the Symposium on the Biology of the cal i fornia ts landi(n.n.Phi lbr ick, ed.) , pp. 57-67. santa Barbara: santa Barbara BotanicGarden.

47

48 L. VAN VALEN

Reader, R.J. , and J.M. Stewart . L972. The relat ionship between net pr imaryproduct ion and accumulat ion for a peat land in southeastern Manitoba.Ecology 53zLO24-L037.

Rensch, B. 1960. Evolut ion Above the Speeies Level . New York: ColumbiaUniv. Press. 4L9 pp.

Robtnette, W.L., O. JuLander, J,S. Gashwi ler , and J.G. Smith. L952.I^/ inter mortal i ty of mule deer in Utah in rel-at ion to range condi t ions.Jour. Wi ld l - t fe Mgnt , 16 z 289-299 ,

Rood, J.P. 1970. Ecology and social behavior of the desert cavy (Microcaviaaustral is) . Amer. Midland Nat. 83z415-454.

Rosenzweig, M.L. 1968. The strategy of body size in mammal ian carnivores.Amer. MidLand Nat. 802299-3L5,

and J. t r r l inakur. 1969, Populat ion ecology of desert rodentcommunit ies: habi tats and environmental complexi ty. Ecology50z558-572.

Sanders, l l .L. 1968. Marine benthic diversi ty: a comparat jve study. Amer.Nat. I022243-282.

Slobodkin, L.B. L96L. Growth and Regul-at ion of Animal Populat ions. NewYork: Hol t , Rinehart , and Winston, 184 pp.

F.E. Srni th, and N.G. Hairston. 1967. Regulat ion in terrestr ia lecosystems, and the impLied ba1ance of nature. Amer. Nat. L01:IO9-L24.

Stuss, R.R. L967. Populat ion dynamics of the walnut aphid, Chromaphisiuglandicola ( fa l t . ) in northern Cal i fornia. Ecol-ogy 48:41--58.

Stewart , D,R,M., and J. Stewait . 1970, Food preference data by faecatanalysis for Afr ican pLains ungulates, ZooL. Afr icana 5zlL5-127.

Suther land, J.P. 1970, Dynamics of h igh and l -ow populat ions of the l impet,Acmaea scabra (Goul-d). EcoL. Monog. 40:169-18B.

Talbot, L.M. L962. Food preferences of some East Afr ican wi ld ungulates.East Afr ican Agr ic. Forest . Jour. 27:131"-138.

. L963. The biological product iv i ty of the t ropical savannaecosystem. Internat. Union Conserv. Nat. Tech. Meet. 9zB8-97.

and M.H. Talbot, L963. The high biomass of wi ld ungul-ates onEast Afr ican savanna, Trans. North A,merican Wil-dLi fe Nat. Res. Conf.28:465-47 6.

Tast, J. 1968. Inf luence of the root vole, Microtus oeconomus (paL1as),upon the habi tat sel-ect ion of the f ieLd vol-e, Microtus agrest is (L.) ,in northern Finland, Ann. Acad. Sci . Fennicaerser. A. IV Biologica,L36zI-23.

Teer, J.G. , J.W. Thomas, and E.A. I , r la lker, 1965. Ecology and management ofwhi te- ta i led deer in the Dano Basin of Texas. Wi ld l i fe Monog. L5:L-62.

Van Valen, L. L970. Late Pleistocene ext inct ions. Proc. North AmericanPaleont. Conv., pp. 469-485. Lawrence: A1len press.

. I97I . Group select ion and the evol"ut ion of d ispersal" . EvoLut ion25259L-598.

. 1973a. Body size and the numbers of pl"ants and animals.Evolut ion 27 227-35.

. L973b. A new evoLut ionary law. Evol- . Theory l" :L-30.

. ( In press). Predat ion and species diversi ty. Jour. Theor. Bio1.

. (Submit ted), Group select ion, sex, and fossi l "s.and R.E. Sl"oan. 1966. The ext inct ion of the mult i tuberculates.

Syst. ZooI. 15z26l-278.Van ZyI, J 'H.M. 1"965. the vegetat ion of the S.A. Lombard Nature Reserve

and i ts ut i l izat ion by certain antelope. Zool . Afr icana Lt55-7L.Var ley, G.C. L949. Popul"at ion changes in German forest pests. Jour. Anim.

Ecol . L8: IL7-L22.

PATTERN AND THE BALANCE OF NATURE

because decomposers and herbivores are for the most part suf f ic ient lydi f ferent that d i f ferent predator species special ize on each class.

Li terature Ci ted

Andrewartha, H.G., and L.C. Birch. L954. The Distr ibut ion and Abundance ofAnimals. Chicago: Universi ty of Chicago Press. 782 pp, :

Ashby, K.R. 1959. Var iat ions in the number of mice and voles (Apode*"s,Clethr ionomvq, and Mlcrotus) in woodland near Durham and an analysisof their ef fect on the regenerat ion of forest . Proc. L5 lnt . Cong.looL. ( f or 1.958) : 7 57 -7 59 .

Baker, J.R. 1938. The evolut ion of breeding seasons. In Essays onAspects of EvoLut ion Presented to Professor E.S. Goodrich (G.R. deBeer,ed.) , pp. L6l- I77. Oxford: Clarendon Press.

Bartholomew, B. 1970. Bare zone between Cal i fornia shrub and grasslandcommunit ies: the role of animals. Science LTOzl2lO-L2L2,

Batzl i , G.O., and F.A. Pi te lka. 1970, Inf luence of meadow mouse populat ionson Cal i fornla grassland. Ecology 5L:.1.O27-L039.

Beers, J.R., and G.L. Stewart . L97L. Micro-zooplankters in the pl"anktoncommunit ies of the upper r^raters of the eastern t ropical Paci f ic .Deep-Sea Res. 18:861-883.

Bel1, R.H.V. 1970, The use of the herb l -ayer by grazjng ungul-ates in theSerenget i - . In Animal Popul"at ions in Relat ion to Their Food Resources(A. I^Iatson, ed.) , pp. LIL-124, Oxford:Blackwel-1.

Bendel1, J.F. 1959. Food as a control of a popul-at ion of whi te- footed mice,Peromvscus leucopus noveboracensis ( f ischer) . Canadian Jour. ZooL.37 zL73-209.

Blackburn, M. L966, Relat ionships between standing crops at three successivetrophic Levels in the eastern t ropical" Paci f ic . Paci f ic Sci , 20:36-59

Caldwe11, L.D. 1964. An invest igat ion of compet i t ion in natural populat ionsof mice. Jour. Marmnal. 452L2-3A,

Cameron, A.W. 1963. ManrnaLlan zoogeography of the MagdaLen IsLandArchipelago, Quebec. Jour. Mammal, 432505-514.

, 1965. Competi t ive exclusion between the rodent genera l l icrotusand Clethr ionomys. Evolut ion, 18 ( tot L964):630'634,

Cody, M.L. 1968. On the methods of resource div is i .on in grassLand birdcommunit ies. Amer. Nat. lO2zL07-L47.

Cohen, J. E. L967. A model of s impl"e compet i t ion. Ann. Computat . Lab. HarvardUniv. 4L ( for 1"966) : 1- 138.

. 1968. Al ternate der ivat ions of a species-abundance reLat ion.Amer. Nat. LO2zL65-L72.

Culver, D.C. L970, Analysis of s impLe cave communit ies: n iche separat ionand species packing. Ecol-ogy 51:949-958.

. 1972. A niche analysis of Colorado anrs. Ecology 532I26-L31.Darwin, C. 1859. On the Origin of Species. ls t ed. London: John Murray.

.502 pp. '

Deevey, E.S. L969. Speci f ic d iversi ty in fossi l assemblages. BrookhavenSymp. Bio1. 22:224-24L.

Dixon, A.F.G., and S. McKay. L970. Aggregat ion in the sycamore aphidDrepanosiphum platanoides (Schr.) (Hemiptera:Aphididae) and i ts reLe-vance to the regulat ion of popul"at ion growth, Jour. Anim. Ecol .39 2439-454.

Ehr l ich, P.R. , and L.c. Birch. L967. The I 'bal"ance of naturet t and t tpopula-t ion contro1." Amer. Nat. l }L:97-L07.

El ton, C. L927. Animal Ecology, London: Sidgwick and Jackson. 207 pp.

45

44 L. VAN VALEN

GZ). There can sometimes (Zomnicki , MS) be a regulated bal-ance between herbi-

vores and a poLymorphism for palatabi l i ty in the plant. The extentof condj- t ions that wi l l g ive such a batance is unknown but seems' on

i ts face, to be rather narron and so inappl" icable to the general prob-

lem of the t roPhic level .(18). Horn and MacArthur (L972) have recent ly discussed a probl-em equivalent

to the ul t imate regulat ion of fugi t ive species.(19). I t may be that an average species of Large animal has a greater proba-

bi l i ty of ext inct ion than an average sma1l animal" . For mammals th isis not t rue but for foramini ferans i t is (Van Valen, submit ted); how-ever, even rats are much larger than most herbivorous animals. A1-though phylettc size increase is more common than a decrease, the large

maJori ty of large animal species are der ived from other large species.

Therefore, i f they become ext inct more readi ly, they must also speciatemore readi- ]"y, for there to be an approximate steady state. So this

approach does not help in explaining the relat ive abundance of l -argeand sma11 animaLs.

(20). There is, as a subsidiary point , d i rect evidence for even the proximal

reguLat ion of some indiv idual species of green plants by herbivoresunder natural condi t ions, and of herbivores by their food, Somespecies of A.cacia grow only when the major herbivores are removed or

discouragea Uy special ized ants (Janzen, L966, L967 a,b). Undoubtedlythere is also a resource l imi tat ion for these species of Acacia, but

i t i .s not known whether they approach i t . An interact ion in regulat ionof Acacia is of course aLso possible. Other references to the pheno-

menon among terrestr ial herbivores are for a mite (Watson, L964);var ious insects (Huffaker, L957); a moth (Cactoblast is; review by

NichoLson, 1958); aphids (Sluss, 1967 i t r r lay, 1968, wi th addi t ional ref-

erences; and Hughes and Gi lbert ,1969); a pigeon (Murton, Westwood and

Isaacson, L964)1 mice ( includj-ng voles) (Ashby, 1959; Pi te l -ka, 1959;

BendeLl, 1959; Findley and Jones,1962; Cameron,1963, L965; Raun and

Wilks , Lg64; Caldwel"1, L964; I , r rh i taker, L967; Tast, L968; and Kopl in and

Hoffmann, 1968); rabbl ts (Wi1Liams and Caskey, 1965); deer (Robinette,

Julander, Gashwiler, and Smith, 19521 and Teer, Thomas and l^Ialker,1965); and antelope and other Afr ican ungul-ates (Vesey-Fi tzGerald, 1960:

Talbot, 1962, 1963; Talbot and Talbot, 1963; Lamprey, 1963; Gwynneand Bel_1, 1968; 8e11, l97Oi Stewart and StewarE, L97O; Jarman, L971-;but not Van ZYL' 1965).

(31t. I 'here may be a s iml lar s i tuat ion, on the next t roPhic Ievel , for the

tsetse f ly, which Glasgow (1963) has ctaimed takes about as much blood

from i ts ungulate hosts as they can stand, There is, however, a

densi ty-dependent defensive response by the ungulates' whjch react more

strongly when there is a greater rat io of tsetse to ungulate. Thismay r{reL1 remove Glasgowrs mystery here, for the ungulates determine

their own react ion and coul"d govern i t , proximal ly or ul- t imateLy' by

Eheir loss of b lood.(22>. On land 80 or 90 per cent of the energy in green pLants, somet imes even

more, goes Lo decomposers, not herbivores (Odum, 1971). Because l i t t le

reduced carbon is bur ied (probabl-y less than 10 per cent of net pr i -

mary product iv i ty even in peat bogs: Reader and Stewart, L972), Lt

fol- lows deduct ivel-y that at equi l - ibr ium much more energy goes through

decomposers than through herbivores. Much of this energy is avai lableto predators, so we might th ink that predators could retreat to de-composers when herbivores became too scarce and thereby give an uLt i-mare regulat ion of the herbivore trophic level. Any herbivore that re-

appeared woul-d be snapped up. Obviousl-y this doesnrt happen, presumabl-y