Resource Partitioning in Ecological Communities

Thomas W. Schoener

Science, New Scrics, Vol. 185, No. 4145 (Jul. 5, 1974), 27-39.

Stable URL:
http://links.jstor.org/sici?sici=0036-8075
% 28 19740705 %293% 3A 185%3A4 145 %3C27%3 ARPIEC %3E2.0.CO%3B2-I

Science is currently published by American Association for the Advancement of Science.

Your use of the JSTOR archive indicates your acceptance of JSTOR’s Terms and Conditions of Use, available at
http://www.jstor.org/about/terms.html. JSTOR’s Terms and Conditions of Use provides, in part, that unless you
have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and
you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/journals/aaas.html.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or
printed page of such transmission.

JSTOR is an independent not-for-profit organization dedicated to creating and preserving a digital archive of
scholarly journals. For more information regarding JSTOR, please contact support @jstor.org.

http://www.jstor.org/
Fri Jan 28 11:44:54 2005
 genetic predisposition, the hypothesis of a print of 1955 publication, Signet, New York, 23. W. Barlow, The Alexander Principle (Gollancz,
purely genetic deviation conflicts with this 1971); G. Thieme, Leben mit unserem autisti- London, 1973),
type of observation, At the same time we schen Kind (Hilfe fiir das Autistische Kind 24. J. Dewey, see, for example, introduction to
know that even when twins grow up in the e.V., Liidenscheid, West Germany, 1971). No F_M. Alexander, The Use of Self (Chaterston,
same family, their experiences can never be two of these seven children received the London, 1932).
identical, same treatment, but on the whole one can 25. A. Huxley, Ends and Means (Chatto and
19, See, for example, L. Wing, Autistic Children say that those who were treated primarily Windus, London, 1937); ‘“End-gaining and
(Constable, London, 1971), p. 8, at the emotional level rather than at the means whereby,” Alexander J. 4, 19 (1965).
20. Although not all the authors of the follow- Ievel of specific skills showed the most 26. G, E. Coghill, “Appreciation: The educational
ing books label their subject as ‘‘autistic,” I striking improvement, methods of F. Mathias Alexander,” in F. M.
mention them because the descriptions of 21. The clearest intreducticn is: F. M. Alexander, Alexander, The Universal Constant in. Living
the initial behavior conform in whole or in The Use of Self (Chaterston, London, 1932); (Dutton, New York, 1941).
part to Kanner’s syndrome; and, as I have but a great deal of interest can also be found 27, R. A. Dart, S. Afr. Med, J. 24, 74 (1947);
said, L consider such descriptions the only in: F. M. Alexander, Man’s Supreme Inherit- An Anatomist’s Tribute to F, M, Alexander
acceptable starting roints: R. d’Ambrosio, ance (Chaterston, London, 1910) and The Uni- (Sheildrake Press, London, 1970).
No Language but a Cry (Cassell, London, versal Constant in Living (Chaterston, London, 28. C. S. Sherrington, The Endeavour of Jean
1971); V. Axline, Dibs—In Search of Self 1942), Fernel (Cambridge Univ. Press, London, 1946):
(reprint of 1964 publication, Penguin Books, 22. The history of medical science is full of such Man on His Nature (Cambridge Univ. Press,
Harmondsworth, England, 1971); J. Copeland examples of breakthroughs due to a reorien- London, 1951),
and J. Hodges, For the Love of Ann (Arrow tation of attention, Compare, for example, 29. E. von Holst and H. Mittelstaedt, Natur-
Original, London, 1973); J. M. Hundley, The Jenner’s discovery that milkmaids did not wissenschaften 37, 464 (1950).
Small Outsider (reprint of 1971 publication, contract smallpox; Goldberger’s observation 30. N, Tinbergen, Proc. Roy. Soc. Lond, Ser. B
Angus and Robertson, Sydney, 1973}; C. C, that the staff of a “lunatic asylum” did not 182, 385 (1973).
Park, The Siege (reprint of 1967 publication, develop pellagra; Fleming’s wondering ubout at. This plea is nowadays heard more often: see
Penguin Books, Harmondsworth, England, empty areas around the Penicillium in his cul- L. Kanner, J. Autism Child Schizophrenia 1,
1972); S. S. Wexler. The Story of Sandy fre- tures. 453 (1971): see also p. 457.

having two quite different explanations

for bill differences in Darwin’s finches
appear simultaneously (6)!
Hutchinson’s (7) reformulation of
the concept of ecological niche pro-
vided a precise language for the descrip-
tion of resource partitioning. In essence,
Resource Partitioning in he proposed that a species’ population
could be characterized by its position
Ecological Communities along each of a set of dimensions
ordering environmental variables such
as ambient temperature, prey size, and
Research on how similar species divide resources helps so on. Ideally, these dimensions would
reveal the natural regulation of species diversity. be few and-independent. Hutchinson
originally conceived of the niche as
comprising intervals of population sur-
Thomas W. Schoener vival along each dimension, Now, how-
ever, many ecologists consider the fre-
quency distribution of utilization or
occurrence along dimensions as the
Biologists have long been intrigued Lotka and Volterra (3). The idea was
niche. :
by differences in morphology and habit What evidence demonstrates that the
supported by experiments of Gause (4)
among closely related species, for to pattern of resource utilization among
and later workers on simple organisms
comprehend the manner and extent of in laboratory containers, in which sim- species results from competition? Mere
such differences is to comprehend much ilar species tended to cause one anoth-
presence of differences is not enough,
for even if niches were arranged ran-
of the natural control of organic diver- er’s extinction. The infusion of models
sity. Ecologists especially have concen-
domly with respect to one another, dif-
and data crystallized into the Gause
trated on differences in the way species principle, one version of which states
ferences would exist. Hence a suffi-
in the same community utilize re- that species cannot coexist for long if
ciently precise search for differences
sources. Studies of this resource parti- they too similarly use the same kinds would be bound to result in their detec-
tion, a state of affairs that led Slobod-
tioning are currently enjoying great of resources. The application of this
popularity. In fact, in the 12 years
kin (8) to rephrase the Gause principle
idea to natural communities, however,
since Hutchinson (/) posed his cele- was begun primarily in the 1940’s dur-
as a “rule of ecological procedure”
brated riddle: “Why are there so many ing the time of the New Systematics.
rather than a verified or even verifiable
proposition.
kinds of animals?” such studies have Although at first interested in differ-
Ecologists now follow two approaches
grown exponentially at a rate four ences between species that might ensure
times that typical of scientific works (2).
in their attempts to make a case for
reproductive isolation, evolutionists
The major purpose of resource-parti- quickly seized upon the idea of reduc- the importance of competition in na-
tioning studies is to analyze the limits ture. One approach is experimental,
ing competition as an alternative ra-
interspecific competition place on the tionale for those differences (5). So
the other is observational.
number of species that can stably coex- rapidly, in fact, did this idea take hold
ist. That such limits exist was sug- that David Lack, eventually its prin- The author is associate professor of biology at
gested by the mathematical models of cipal proselytizer, was placed by a pub- the Biological Laboratories, 16 Divinity Avenue,
Harvard University, Cambridge, Massachusetts
two early 20th-century theoreticians, lication lag in the awkward position of 38.

5 JULY 1974 27
 At its most extreme, the experimental avoid habit-forming predators. In the Overdispersion of Niches
approach argues that a descriptive in- remainder of this article I examine
vestigation of the equilibrium state of such predictions in detail. I attempt to Although if species had no influence
a process reveals little about what that show to what extent they are confirmed on each other’s resource utilization
process is. To prove that a particular by patterns of resource partitioning their niches would still differ, competi-
dynamics operates, one must perturb and how they allow a better under- tion should result in an overdispersion
the system away from equilibrium by standing of those patterns. of niches in niche-space. Where niches
adding or removing individuals. Such The evidence (see Table 1) is drawn are regularly and widely spaced over
experiments, mostly quite recent, have from 81 studies that, explicitly or not, one or more dimensions, the alterna-
demonstrated that compensatory changes bear on resource partitioning in groups tive or “null” hypothesis of randomly
sometimes occur in population num- of three or more species. The studies generated differences must be rejected.
bers and in the condition of individ- heavily favor terrestrial vertebrates, Overdispersion patterns can be arranged
uals (9, 10). When such changes are probably reflecting my own interests under three headings.
accomplished at several places along as well as a real bias in the literature. Regular spacing along a single di-
an environmental gradient, they show Autotrophs are not included, because mension. In 1959 Hutchinson (/)
how competition maintains niche differ- the way in which they partition re- called attention to certain groups whose
ences in nature (J/~13). For example, sources is somewhat different and, in otherwise similar species differed in the
Connell (J7) has shown for barnacles, particular, seldom involves food types size of their feeding apparatus by a
and Grant and others (/2) for rodents, (16). Table 1 shows, for each group, constant factor of 1.2 to 1.4. Using
that habitat differences can result from the number of its species and genera, this morphological indicator, one may
competitive interactions. its location, its trophic position, and infer that species adjacent on a size
Simple perturbation experiments in the dimensions, ranked according to scale differ in mean food size by a
themselves, however, have several short- importance, that separate its species constant ratio and thereby have their
comings. First, while demonstrating the (77-20). Many of these dimensions or- niches regularly spaced. Subsequent
effect of one species on another, they der continuous variables such as food work has often, but not always, sup-
fail to reveal the mechanism of the com- size. Some, however, such as food ported Hutchinson’s generalization (/9,
petition. For example, one species may taxon, are easier to give as nominal 22-28). Certain lizards seem regularly
reduce the abundance of a second in a categories, though these could prob- to show higher ratios (29). Further-
particular habitat by directly depleting ably be ordered along one or more more, insectivorous birds and lizards
its resources, by interfering with its axes. sometimes increase ratios with increas-
ability to obtain those resources, or by Most investigators do not state how ing body size (27, 29).
using up in aggressive encounters en- they delimit the group studied. Usually Why should sizes be a constant or
ergy obtained from those resources. species of some taxonomic category, increasing multiple of one another rath-
Additional information, both observa- especially the genus, are studied in a er than be separated by a constant
tional and experimental, is required to small area. This procedure centers on difference? If populations were close
evaluate these alternatives [for example, theaters of most intense competition, to numbers that would make extinction
see (14)]. Second, they shed little light but it introduces arbitrariness: if a spe- probable, the pattern might result from
on the origin of differences. Indeed, if cies becomes too different in morphol- those species feeding on large items,
species differences have a strong genetic ogy or behavior, it is taxonomically which are often rare (30), having to
component, short-term experiments will defined out of the system, and if it be spaced farther apart on the size axis
not result in much niche expansion, becomes too different in habitat, it is to maintain a viable population size
even though competition may have geographically defined out of the sys- (27). It is likely, however, that the
caused those differences in the first tem. These effects must differ for dif- limit of minimum population size is not
place. Even long-term experiments, ferent kinds of animals and may bias as important as an increased variance
however, may fail to show the evolu- the generalizations made in the follow- in food-size utilization for large spe-
tionary consequences of competition if ing sections. cies: the larger the variance for a
it acts rarely but with intense selective Sometimes it is convenient or neces- given distance separating the niches of
pressure (/5). Third, perturbation is sary to estimate resource differences two species, the greater the overlap
sometimes impractical—it is best when by using species characteristics, usually and, for a certain set of assumptions
generation time is short, populations morphological, that indicate the posi- detailed below, the more intense the
are insular, and only a small number tion of its utilization on the resource competition. Hence, to hold competi-
of populations need to be studied. dimension. The commonest indicator is tion just below a certain intensity, larg-
The observational approach lacks the size of feeding structures, which is er species must space more widely
some of these drawbacks but has correlated usually with mean food size, (31). The increased variance results
others. In exchange for a direct dem- hardness, or depth in some protective from two factors: (i) large individuals
onstration of ongoing competition, it medium. But many other indicators usually eat a greater range of food
attempts to implicate competition indi- (see Table 1) exist, such as body tem- sizes than smaller ones, probably be-
rectly. This can be done by elaboration perature for activity time and hindleg cause their optimal food is relatively
of the predictions of competition mod- length for habitat (27). Indicators not rare; and (ii) species whose adults are
els in such a way as to rule out other only short-cut the compilation of data, large have a greater diversity of sizes
mechanisms that might cause species but they imply strong functional rela- because of their younger, smaller indi-
differences, such as selection for repro- tions between utilization and pheno- viduals. An increased niche variance is
ductive isolation to avoid interbreed- type, relations which constrain resource probably also responsible for lizards
ing or for divergence in appearance to partitioning. having greater ratios than birds. Lizard

28 SCIENCE, VOL. 185

populations comprise individuals of tion. The Spearman rank correlation another, if resources are to be suffi-
many sizes, whereas birds mature coefficient (r,) of number of species ciently distinct. Such complementari-
quickly relative to their life spans and in a group compared with the number tics illustrate especially well the trouble
reach a monomorphic size. of dimensions segregating the species individuals and their genes seem to take
Increase in number of important di- is .277 (P< .01). Because dimensions to avoid other species’ niches, for they
mensions with increase in species within the broad categories of food, often fly in the face of expected func-
number. Levins (32) and MacArthur space, and time tend to be the most tional relationships between phenotype
(33) argue that in a competitive sys- correlated, to reduce the effect of de- and ecology. For example, small lizards
tem, as the number of species accumu- pendence one can recompute the rela- optimally should eat small insects, and
lates, those species will eventually have tion using only one dimension from they can use small perches. Yet if food
to segregate on more and more dimen- each of the three categories, But the were limiting, this arrangement would
sions to preserve minimal resource over- trend is less decisive (r, = .146, P <.1). result in the spatially most overlapping
lap. This argument assumes an eventual Higher correlations should result from animals competing.
for the most similar
incompressibility of a species niche on a taxonomically more narrowly defined sizes of food. How is this conflict re-
any one dimension, an assumption analysis; our sample includes repre- solved? The dilemma is not imaginary
which must be true because of the en- sentatives from slime molds to lions but real, and we can examine the four
ergy requirements and perceptual con- (34, 35). common lowland Anolis lizards of Ja-
straints of the individuals comprising What is the commonest number of” maica to discover their solution (36).
the species. In contrast, if species were dimensions separating species? For no Curiously, the four show an inverse
thrown together randomly, there should limit on the maximum, three is the correlation between body size and perch
be no tendency for the number of mode, whereas for a maximum of three diameter: the larger the species, the
important dimensions (defined as di- dimensions, two is by far the common- thinner the perches! The situation is
mensions separating some minimal per- est value. Even if one admits the im- complicated by the fact that the species
centage of species pairs) to increase precise identification of important di- break down into classes of different-
with species number. In practice, be- mensions, separation appears generally sized individuals (a class is uniquely
cause of difficulties in ascertaining im- to be multidimensional. defined by age and sex, and adult
portance and determining nonindepen- Separation of species along comple- females are usually considerably small-
dence of dimensions, it will be hard to mentary dimensions. For groups where er than adult males). Moreover, within
disentangle random effects, more than one dimension is important, species, larger individuals appropriate
Table 1 allows a crude investigation similarity of species along one dimen- larger perches. As Fig. 1 shows, the
of the dimensionality of species separa~ sion should imply dissimilarity along arrangement whereby there is a direct

High : Horizontal habitat

separation (%)
a

EK
i
x

yo Ss
f tio
4

\
io
Toes
eS
Thick Thin / Food type
separation (%)
Vertical = /
habitat /100
separation/
(%)

Fig. 1 (left). Spatial arrangement of lowland Jamaican Anolis

lizards. Bottom line of lizards shows how a direct relation be-
tween perch diameter and lizard size within species and an in-
verse relation between species results in those classes of different
species adjacent in perch diameter being of maximally different
size. The top of the figure shows the actual spatial placement
of lizards ‘at one locality [Mona, see (36)], which is one varia-
tion on the theme of the bottom line. Sizes are scaled according
Thick to actual mean sizes. Fully shaded lizards denote species found
mostly in shaded places, unshaded lizards denote species found
mostly in sunny places. Fig. 2 (above). Cody’s grassland bird
communities, showing how total separation in horizontal habitat,
vertical habitat, and food type is constant, but the relative
proportion of the three types of separation varies, Squares are
g SM South American communities, circles are North American.
Opalinus Grahami Lineatopus Valencienni {From Cody (37); courtesy of the University of Chicago Press]
3 JULY 1974 29
Table 1. Ecological differences between similar species. Numbers in columns “Macrohabitat” through “Time” denote importance rank of
that decision. Abbreviations are as follows. Rank: A, thought to be strictly an aspect of separation on another dimension; b, breeding
time; f, feeding time; i, indicator used; n, thought not to be necessarily related to resource partitioning; X, known not to be important
dimension. Macrohabitat: Alt, altitude; Aqu, aquatic-terrestrial gradient; Dis, foraging distance from land: For, marine formation; Geo,
geographic; Hor, aquatic horizontal zone; Lat, latitude; Soi, soil; Str, stream size or part of stream; Veg, vegetation type; Wat, size
or type of water body, Microhabitat: s, used as shelter; structural habitat, food or perch substrate in vegetation, Food type: B, artifi-
cial baits; F, feeding type; H, hardness; S, size; T, taxonomic catcgory [letters following ‘I refer to species of food (S), higher taxonomic cate-
gory (H), life stage of host (L), part of an individual prey (P)]. Indicators: 1, body size; 2 bill size; 3, head size; 4, ovipositor length; 5, bill
shape; 6, body temperature; 7, mouth shape; 8, body form; 9, hindleg length; 10, hair quantity; 11, group size; 12, nutrient-utilization ability.

Consumers Rank and description of resource dimensions

No. No. Time
Group, location, and reference of of Food Macro- Microhabitat Food
spe- gen- habitat type D
: ay Year
cies era

Spnple organisms
Slime molds, forest, eastern 4 2 Bacteria x 1-TS
North America (93)
Paramecium, near Ann Arbor, Mich. (94) 5 1 Organic minutia 1-Wat 1
Triclads, shallow littoral zone of lakes, 4 3 Invertebrates 2 (X)-Depth 1-TH; X-S
Britain (95)
Nematodes, psammolittoral, 46 ? Invertebrates, plants 1-Hor 1-Depth 3-F
Gulf of Mexico (96)
Rotifers, small lake, central Sweden (97) 3 } Flagellates t-Depth li-S*
Tubiticid oligochaetes, Toronto Harbor (98) 3 3 Bacteria x 1-TS?
Polychactes, soft bottom, Beaufort, N.C. (99) 5 3 Deposit feeder 1-Sediment type
2-Vertical zone
Chaetognaths, Agulhas Current, 18 4 Mostly copepods 2-Hor 2-Depth 1i-S
Indian Ocean (7/00)
Mollusks
Gastropods, shallow water, Florida (47) 8 6 Invertebrates 1-TS 2
Corus, Hawati (761) 25 1 Polychaetes, fish, 3-For 2-Substrate 1-TSH 4
gastropods
Conus, Pacific atolis (702) 17 1 Mostly polychaetes 2-Substrate 1-TSH
Crustacea
Crabs, intertidal bench, Tasmania (103) 1! 9 Algae, inverte- 1-Hor 1-As macrohabitat® 3-TH 3h
brates, detritus 2-Cover
5-Vertical zone
1-Shell shape*®
2-Shell weight*®
Hermit crabs, intertidal, 3 1 Detritus 3-Bed and tidepool
San Juan Islands, Wash, (6/) type®"
Diaptomus copepods, Clarke Lake, 3 1 Plant, animal 1-Depth 1S x 1
Ontario (44) particles
Diaptomus copepods, Saskatchewan 7 1 Plant, animal 1-Geo 4. (X)-Depth 21-8 2
ponds (/04) particles
Amphipods, marine sand beaches, 5 5 Mostly detritus, 1-Hor 2-Depth in sand 2-8! Qe
Georgia (44) algae, protozoa 5-TH
Crustaceans, cave streams, 4 3 Leaves, 1*-Riffles or type Xx
West Virginia (/05) microorganisms pool
Insects
Grasshoppers, prairie, northeastern 14 1 Grasses, forbs 1-TS; 1-TS 2
Colorado (106)
Melanoplus grasshoppers, grasslands, 3 1 Mosily grasses 2-Veg 1-TS 2
Boulder, Colo. (707)
Termites, Savannah-woodland, 5 1 Grasses 1"-Veg 4-TS; 4-S 3 2
West Africa (108)
9 5 Bark algac, fungi 4-Alt 2-Twig condition 1-THP 3
Psocids, larch trees, Britain (109)
12 7 Decaying fruit X-Microclimate X-B has
Butterflies, lowland rain forest, .
Costa Rica (10)
8 2 Mainly scavengers 1-Veg ; A-TH; A-S 2 5
Carabid beetles, fen, England (1/7)
3 1 Predators, 1-Lat 1-Lake size 3(X)
Whirligig beetles, Michigan (J/2)
scavengers . : \
(7/3) 19 1 Nectar 2-Veg 3'-Microclimate rs x
Euglossa bees, Panama
Ants, Colorado (114) 4 3. Animals 1-Veg _-3-Type log or cover 28
5 2 Seeds 2-Veg 3-Type log or cover 1S x
Ants, Colorado (1/4) 2
3 1 Parasitoids 1!-Depth of food’ x
Megarhyssu wasps, beech-maple,
Michigan (1/5) X-Leaf type’ LTL: 148
Wasps, Neodiprion, Quebec (19) W 9 Parasitoids 4-Veg 1!-Depth of host “Ts .

Other arthropods
7 7 Leaf litter, 1-Position in log, K-F
Millipedes, maple-oak forest,
central Illinois (116) decaying wood litter \ 4
7,9 1 Invertebrates 1-Veg 1-Depth in soil 1i-S
Mites, deciduous forest, central
Maryland (117) 1-TS; 2-P 4
20 1 Parasites 3-Wat
Water ‘nites, ponds, central
New York (1/8)
Fish
2-Str 2-Depth 1-TH 3
Stream fish, dry season, Panama, 12 12 Animals, plants
moist tropics (42)
River fish, River Endrick, Scotland (1/9) 5 4 — Arthropods, algac 1-Str <-Depth® ee 3 3
17 15 Animals, plants 1-Hor* Dept Segue ,
Lake fish, eastern Ontario (43)
13. «12 Invertebrates, algae 1-Hor 1-As macrohabitat 2-TH
Intertidal fish, Brittany (720)
SCIENCE, VOL. 185
30
 Consumers Rank and description of resource dimensions
No. No. Time
Group, location, and reference of of Food Macro- Microhabitat Food
spe- gen- habitat type Day Year
cies era

Salamanders
Desmognathwus, Appalachians (727) 5 2 Arthropods 1-Aqu 2-St x
2-Alt A-TH
Triturus, ponds, England (1/22) 3 1 Invertebrates 22-Alt 2-Temperature 1-S, 4-TH
Frogs
Tropical Rana, streamsides, rain forest, 3 { Small animals 12-Wat {-Distance from 3-8 x
Borneo (/23) stream 3-TH

Temperate Rana, northeastern 6 I Smail animals 2-Aqu BLS 1»

North America (/24) 3-Lat

Lizards
Ameiva teids, Osa, Costa Rica (40) 3 i Arthropods, fruit i-Veg 2-§' 2!
Ctenotus skinks, desert, Australia (22) 7 1 Arthropods 4-Plant cover®® 3-S*; 1-TH 58 2
Cnemidophorus whiptails, Trans-Pecos (725) 5 I Arthropods 1-Veg 2-TH
Cuemidophorus, south and central 4 1 Arthropods 1-Veg 2-TH; 2-8! x 4"
New Mexico (126)
Anolis, Bimini (20) 4 1 Arthropods 5-Veg 1-Structural habitat 2-S*; 2-TH 2
Anolis, Jamaica (35) 6 I Arthropods, fruit 3-Veg 1-Structural habitat 1S 4
Anolis, Puerto Rico (1/27) 10 | Arthropods, fruit 2-Veg 1-Structural habitat 3i-S 4?
Phyllodactylus geckos, Sechura desert, 4 1 Mostly 2-Alt 3-Plant species 1-TH x
Peru (728) arthropods 3-Soi 5-Foraging substrate 6-5

Other reptiles
Sternothaerus turtles, southeastern 4 1 Mollusks, i-Wat 3 (X)-TH !
United States (129) arthropods
Garter snakes, Michigan (730) 3 t Animals 2-Veg 1-TH 3
Birds
Alcids, Olympic Peninsula (49) 6 6 Fish, invertebrates 1-Dis 2-Feeding depth 2-TH x»
Alcids, St. Lawrence Island, Alaska (1317) 3 2 Invertebrates 4-Dis X-Feeding depth 1-S? 3 xe
1-TH®
Terns, Christmas Islands (23) 5 4 Fish, invertebrates 2-Dis 1-S13 2
A-TH
Sandpipers, tundra, Alaska (232) 4 I Insects 1-Veg i-TS
3-S
Herons, Lake Alice, Fla. (133) 4 4 Animals 2-Feeding place 1-THS; 3-S
Ducks, Medway Island, Britain (24) 7 3 Plants, animals 3-Veg 2-Feeding method 1-THP
Hummingbirds, Arima Valley, Trinidad (38) 9 8 Nectar, insects 2-Veg 1-Kind of plant 2-$12
Flycatchers, deciduous forest, 5 5 Insects {-Veg 4-Feeding height 1'-S33
south West Virginia (134) 3-Vegetation A-TH
density
Flycatchers, deciduous forest, eastern 5 4 Insects 1-Veg 2-§)
United States (39)
X-TH
Titmice, broadleaved woods, Britain (78) 5 1 Insects, seeds 1-Veg 2-Structural habitat 2S
Vireos, New World (/35) 17 1 Insects 2-Alt 2-Foliage layer 4g
1-Geo
Warblers, boreal forests, Vermont (136) 5 i Insects x 1-Part of tree X-TH xX 3
2-Feeding style
Icterids, channeled scabland, 4 4 Insects 1-Veg 1-As macrohabitat 2-TH
Washington (/37) (for nestlings)
Tanagers, Trinidad (238) 10 5 Fruit, insects 2-Veg 1-Structural habitat 2-THS
Honeycreepers, Trinidad (138) 5 4 Fruits, insects 3-Veg 1-Structural habitat 2-THS
Finches, southeastern United States (139) 5 5 Seeds, insects 1-Veg 1-Height 3 (A)-S
Finches, near Oxford, England (140) 10 5 Seeds, insects, 3-Veg 4-Ground, air, i-TSH*? 8
buds or foliage 1-S*
Geospiza finches, central Galapagos 5 1 Seeds, fruit, 2-Veg 1-8, H
Islands (78) buds, insects
Camarhynchus finches, central 4 i Insects 3-Veg 1-Structural habitat 3-5
Galdpagos Islands (78) 1-Foraging method
Grassland birds, ten sites, New World (37) 2-4 2-4 Seeds, insects 2-Veg 3-Vertical 1'-S x x
Foliage gleaners, oak woods, 5 3 Insects 3-Substrate 1-8
California (25) 4-Foraging layer 1-TH
Upltand birds, broadleaved woods, 22 #177 ~~ Insects 2-Veg 1-Structural habitat 2-TH A
Britain (50) A-holes*
Mammals
Pocket gophers, Colorado (7/7) 4 3 Plants 1-Soi X-FT
Chipmunks, Sierra Nevada, Calif. (/4/) 4 1 Seeds, fruits 1-Alt
Rodents, deserts, North America (142) 10 3 Seeds ret 1-Foliage height X?-S x
-Soi
Peromyscus, Ozarks, Mo. (143) 3 1 Insects, seeds 1-Veg X-THP
Giant rats, western Malaysia (144) 4 3 Plants, insects 1-Alt 2-Height in trees 3-THP
3-Veg
Carnivores, Serengeti, Africa (145) 7 5 Large animals 2-Veg 1-Set 3
Bats, central Iowa (146) 8 6 Insects 2-Veg 1
Bats, Central American lowlands (52) 31021 Animals, plants 2-Veg 1-§;2-TH

3 JULY 1974 31
 intraspecific relation between size and io eat different foods is perhaps the mond’s numerous examples (28) of
perch diameter, and an inverse inter- commonest combination. In addition altitudinal separation for New Guinea
specific relation, results in maximizing to having been reported for birds (23, birds, the situation is similar to those
the size difference between adjacent 37-39) and lizards (20, 36, 40, 41), just described. However, other cases of
classes from different species. Thus it has also been reported for fish (42, geographic separation do not appear to
sizes are staggered in space so as to 43) and crustaceans (22, 44). The covary with any dimension and are
minimize resource overlap between spe- Anolis lizards of Bimini provide a pre- particularly common in archipelagos,
cies, The way the lizards solved the cise example of this combination. In where terrain is fragmented (0, 28,
awkward problem of balancing heavy Fig. 3, overlap in food size is plotted 45).
animals on thin perches was to evolve against overlap in structural habitat 2) Food type and time: Temporal
elongated body form and short-femured (perch height and diameter) for all separation can be on a daily or a yearly
limbs. pairwise combinations of classes of basis and both sorts can complement
More complex examples are provided Bimini lizards. For pairs from different separation by food type. For example,
by the grassland bird communities species, where structural habitat is sim- two similarly sized terns feed at differ-
studied by Cody (37): in each com- ilar, food size is not, and vice versa. ent times of the day (23), and similar-
munity, the species separate according For pairs from the same species, sim~- ly sized lizards (46), crustaceans (22,
to various mixtures of differences in ilarity in one dimension implies sim- 44), and gastropods (47). reach peak
vertical habitat, horizontal habitat, and ilarity in another. This indicates the abundance at different times of the year.
food type. The proportional importance relative effectiveness of competition in 3) Habitat and time: Lizards in habi-
of the three dimensions varies from spacing similarly sized individuals, de- tats where climatic factors vary sub-
community to community, but the total pending upon whether those individuals stantially during the day can show
separation is remarkably constant (Fig. are In the same gene pool and repro- nonsynchronous spatial overlap (36,
2). ductively attractive to one another. 40, 48).
Cases of complementarity involve Many otherwise similar species are 4) Habitat and habitat: In addition
nearly all possible combinations of separated, often with extraordinarily to Cody’s birds, other species similar
kinds of dimensions: small overlap, in geographic range. in horizontal habitat often differ in ver-
1) Food type and habitat: The tend- When that separation parallels changes tical habitat—seabirds partition depth
ency for species that overlap in habitat in habitat variables, such as in Dia- under water (49), while woodland pas-
serines partition foraging height (50).
5) Food type and food type: Mc-
09% Nab constructed a matrix whose col-
interspecific pairs © umns index food size and whose rows
Intraspecific pairs © index food taxon; members of the
0.8 e Trinidad bat fauna fill this matrix fairly
evenly (57, 52).
Despite there being so many cases
Cd

of complementarity, there are more

e
24
T

cases where similarity of species along

Overlap in prey size by volume

one dimension implies similarity along

another. This is because the dimensions
2°
ao
oe

that ecologists recognize are rarely in-

°
e

dependent: for example, moisture and

food size can be correlated (30). The
on

6
T
2

point is not to compare numbers of

examples. Rather, given correlations
between environmental variables and
°
aS
7
°

functional relations between consumer

and resource characteristics, one must
explain why examples of complemen-
2ur
t

tarity should exist at all.

0.2
° Importance of Particular Dimensions

i £ I i [ J } 1 } in Resource Partitioning
02 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Overlap in structural habitat The data just reviewed raise the ques-
tions of which dimensions are important
Fig. 3. Similarity in prey size (weighted by prey volume) plotted against similarity for which groups and why some kinds
in structural habitat for all pairwise combinations of classes of Bimini Avolis lizards.
of animals show greater dimensionality
Overlap is expressed as percentages. For interspecific pairs, similarity in habitat im-
plies dissimilarity in prey size and vice versa, whereas the reverse is true for intra- than others. To approach these ques-
specific pairs. More exactly, divide the plot into four sections bounded by the axes and tions, we need to answer two others:
two lines corresponding to 0.5 overlap for habitat and 0.5 for food. Then 25 inter- (i) What is the absolute heterogeneity
specific pairs fall into the top left or lower right sector, whereas no pairs fall into
pairs are 0 and 8 of resources and, in particular, what
the top right sector. In contrast, the same figures for intraspecific
(P < .001, in either separate or combined binomial tests). resource kinds renew as separate popu-
SCIENCE, VOL, 185
32
lations, and (ii) to what degree is it once they are entered, but does not is simultaneously deciding instead to
adaptive or necessary for consumers to affect the per-unit-time value of eating forage in or travel between other habi-
distinguish that heterogeneity? types of food once they are found (54). tat types; he is weighing one positive
Phrased another way, we need a Thus, at first, species will partition energy gain against what is nearly al-
theory of the feasibility of resource habitats, However, evolution should ways another. But in deciding to omit
partitioning as it relates to particular eventually redistribute phenotypes of certain time periods, the consumer is
dimensions. This theory is now in a the consumers, those phenotypes better usually trading something—a lowered
formative stage, in which the qualita- at extracting energy from the more but positive yield in the time period
tive sorting out of biologically impor- abundant food types becoming more frequented by competitors—for noth-
tant effects is critical, To obtain sug- prevalent. Unless the competing species ing, no yield at all. Only where ability
gestions about the directions that such eliminate the possibility of such evolu- to process food is limited relative to
a theory might take, we can examine tion by their separating into nonover- risk of being eaten during feeding
the data of Table 1. Five major gener- lapping habitats and quickly becoming should temporal specialization be mark-
alizations emerge. genetically fixed in them, the result ed. A similar argument can be made
Habitat dimensions are important can be a narrowing of diet with con- for seasonal activity, where what is at
more often than food-type dimensions, sequent specialization. Hence, the Mac- stake is some reproduction as opposed
which are important more often than Arthur-Pianka argument may not be to no reproduction. No wonder tem-
temporal dimensions. There are two sufficient to explain the preponderance poral partitioning is relatively rare (55).
ways to show the considerable magni- of separation by habitat. The importance of seasonal differ-
tude of these differences. First, we can Perhaps a more critical factor re- ences between species decreases with
give the percentage of groups where lates to the contrasting ways in which increasing species number (r,= .377,
the most important dimension falls into habitat and food are distributed in P <.05). So does the importance of
one of the three categories. The result space and encountered by consumers, diel differences (r,
= .313). Although
is as follows: 55 percent of the groups An individual must encounter many the latter is not quite statistically sig-
fall into the habitat category, 40 percent food items (and to a lesser extent mi- nificant, both of the temporal correla-
into the food category, and S percent crohabitats) in its foraging and if it tions are substantially greater than those
into the time category. Second, we can skips too many it will lose too much for other dimensions. Hence it seems
give the percentage of groups where energy and time in searching relative that the possibility for temporal parti-
each kind of dimension is known to to its energy input. However, because tioning is rapidly exhausted as species
separate species. The result is: in 90 macrohabitat patches are more con- are added to an assemblage.
percent of the groups the species are tinuously distributed and are large rela- Predators separate more often by
separated by habitat; in 78 percent, tive to the need and ability of indi- being active at different times of the
the species are separated by food; and viduals to range through them, it is day than do other groups. Of the 63
in 41 percent, the species are separated feasible for an individual to spend groups for whose species no differences
by time. most of its time in a single macrohabi- in daily activity were noted, 49 percent
We can begin to assess the likeli- tat. Thus, when it becomes impossible are primarily predators, 27 percent are
hood of partitioning by habitat as com- to specialize further on food types, primarily herbivores or scavengers, and
pared to food type with MacArthur species can continue to partition macro- 24 percent are primarily omnivores. Of
and Pianka’s (53) foraging-strategy habitats. the 17 groups showing partitioning by
scheme. Picture a set of habitat patches Despite these arguments, there is no daily activity, 82 percent are predators
between which individuals of a given statistically significant overall trend for and 12 percent herbivores [P < .025
species travel and within which they habitat rank to increase or food-type (56)1.
forage, and suppose that individuals of rank to decrease with increasing spe- The most likely hypothesis for this
a second species invade and reduce cies number in an assemblage. Certain pattern gives a second reason for the
food density in certain kinds of patches, kinds of animals, such as lizards [for rarity of temporal partitioning: re-
Then it may no longer be worth the their microhabitats, see (35)}, do seem sources cannot often be separated into
while of the given species to visit to show an overall trend, but others, independently renewing populations by
patches of that habitat kind, since the such as terrestrial mammals, do not, their own daily activity. This is obvi-
time spent there might better be spent Indeed, in the latter group, separation ously true for plants and most plant
traveling to and feeding in less depleted by macrohabitat is more important than products—a leaf not eaten during the
places, However, picture now the situa- it is in any other group and species day can be eaten at night. It is least
tion within such a habitat patch, where often seem to be separated only in Jikely to be true for animal foods,
a competing species reduces the density this way. This is in contrast to groups which themselves often show peaks of
of certain kinds of items. The yield such as birds, which are able to parti- activity. Therefore, predators will be
per unit time spent there for individuals tion space vertically and which often more likely to partition resources by
of the original species must go down, show greater horizontal overlap (13). being active at different times of the
but if those individuals still find it Should the partitioning of time be day than will other animals.
optimal to feed in that kind of habitat more like habitat or food type? As for Terrestrial poikilotherms relatively
patch, they should at least eat all the habitats, competitors reduce the value often partition food by being active at
types of food that they ate before of foraging in certain time periods by different times of the day. Of the 23
competition and should probably eat lowering the density of available food. such groups, 43 percent segregate by
other kinds of food as well. In short, However, there is a fundamental dif- having different periods of daily activ-
competition changes the value, per unit ference. In deciding not to forage in a ity, whereas only 12 percent of other
time, of feeding in kinds of patches particular type of habitat, a consumer animals do [P < .005 (56)]. Those ani-
5 JULY 1974
33
 @
mals most sensitive to diel climatic K Limiting Similarity
variation are poikilotherms, because MN. saddlee) No
(unstabl * Stable
they are less buffered against external Although niche overdispersion falsi-
temperature change. And such changes Ny Ny fies the hypothesis of randomly placed
are greater on land than in water. niches, it fails to rule out alternatives
Therefore, both consumer-perceived to competition, such as predation and
and absolute temporal heterogeneity is Kg Ko reproductive isolation, that can cause
No N
high for terrestrial poikilotherms. , © 2 species differences. Analysis of the
Vertebrates segregate less by seasonal Stable particular dimensions used by particu-
activity than do lower animals. Forty- node lar kinds of animals may to some ex-
Ny
four percent of invertebrates but only Kho No tent enable us to discriminate among
13 percent of vertebrates differ in sca- my the hypotheses. For example, instead of
sonal activity or breeding (P
< .0i). the divergence in time of activity or
The obvious explanation for this trend Ka in time of reproduction that would be
is that animals whose generation times © Ne expected from competition, predation
Stable Ny Wins
are relatively long cannot partition the K Ny node K might sometimes result in these times
1 e 1 Ni
year as finely as those that mature in becoming synchronized so as to satu-
shorter time periods. What is more, even Ni} N; Ny rate the predators, as in the case of
when there is unusual opportunity for —— No _. periodical cicada species (62). How-
vertebrates to stagger breeding, as for Ko Ko ever, an entirely different approach is
birds in tropical climates, the species No No through a theory specifying quantita-
simply expand their breeding periods so Fig. 4. (a) The two kinds of equilibria tively just how similar species could be
that overlap is no fess than for tem- for Lotka-Volterra competition equations. and yet coexist. This number, called
perate species (57). (b) Equilibrium in a competition model the limiting similarity, could then be
Segregation by food type is more where each species has exclusive resources checked against differences in hypothet-
(Bq. 3). (c) The two possible outcomes
important for animals feeding on food ically equilibrial communities. The
of competition in an “included niche”
that is large in relation to their own model, where one species has exclusive theory, only begun during the last
size than it is for animals feeding on resources but the other does not. Either several years, links properties of the
relatively small food items. This trend the first species wins. or there is stable niche to dynamical models that specify
holds separately for predators and for coexistence. the outcome of competition between
other trophic types. For the groups populations.
whose food items are relatively large, The dynamical system that has served
the combined totals are as follows: 71 tistically significant (P< .025). The as the basis for models of limiting
percent separate by food type and 33 pattern supports the idea that specializ- similarity is that of Lotka-Volterra for
percent separate by habitat; for groups ing on food size is difficult for con- n competing species:
whose food items are small, 28 percent sumer specics whose populations at any
dNifdt=
separate by food type and 72 percent one time comprise many sizes because, Hn

by habitat [P < .05; P< .01 (58)]. for a given range of available food (riNi/ Ki) (Ki = Ni = So aN)
sizes, overlap must be greater among
The reasons for this pattern may be
several. First, unless some kind of such species.
i
ix d,...,2 (1)
small food is abundant, a necessary There is another rough tendency
condition for individual specialization shown by the data in Table 1. Habitat where WN, is the population size of the
is that the animal cat only a few items, is less often the most important dimen- ith competitor, r; is its intrinsic rate of
as do those animals feeding on rela- sion in aquatic animals than it is in increase, K; is its carrying capacity
tively large foods. Many herbivorots terrestrial animals (43 percent as op- (the number of individuals at equilib-
insects, for example, spend most of posed to 65 percent). A striking excep- rium with no competing species), and
their lives on a single “item.” Second, tion to this pattern is terrestrial arthro- a;; is a competition coefficient (the ef-
animals that cat relatively large prey pods. If real, the pattern is hard to fect of an individual of competitor j on
are often pursuers, and several models explain except on the basis of a the growth of competitor i relative to
of foraging strategies argue that pur- lower absolute habitat heterogeneity in the effect of an individual of competi-
suers should specialize on food of a aquatic systems. As a terrestrial ecolo- tor i). Depending upon the K’s, the
limited size range (53, 59). gist, I fear this hypothesis may be near- a’s, and (for n > 2) the r’s, Eq. 1 can
In addition to these five patterns, sighted. Yet there is agreement among give mathematical equilibrium with
there are others which are not quite ecologists working on benthic marine some nonpositive N,, in which case
statistically significant but which may systems (60) that spatial heterogeneity those species are excluded from the
be real. there is often low. In addition, because community (63). When equilibrium oc-
For example, animals that mature in of the high specific heat of water com- curs with all positive XR. it can be
size quickly relative to their life span, pared to air, spatial climatic variation stable or unstable. Figure 4a illustrates
or whose adults are ecologically dis- in aquatic systems should often be less the two possible kinds of equilibria for
similar to their young, partition food severe. Finally, because of greater re- {wo competitors: a stable node (64)
by size more often than do others (59 source mobility, spatial dimensions may and a saddle (the “unstable equilibri-
percent as opposed to 39 percent). A characterize resource kinds less well um” of some texts}. Both positivity
striking exception is mammals, and if in aquatic than in other systems [for and stability are necessary for coex-
these are deleted, the result is sta- example, see (6/)]. istence.
SCIENCE, VOL. 185
34
 Conditions for coexistence were in- cave rather than bell shaped. Most of
vestigated with deterministic models these new and exciting ideas have yet
first by MacArthur and Levins (65) to be tested in detail.
and then by Roughgarden (66) and
May (67). These investigators assumed
that competitors separate along one di- Other Mathematical Approaches
mension and obey Eq. 1. To incor- to Resource Partitioning
porate niches into the differential equa-
tions, they calculated competition coef- While providing the gateway to the
ficients as current theory of limiting similarity,
the Lotka-Volterra equations (Eq. 1)
ety =o pap! pus (2) are less than ideal for modeling re-
k Kk
Fig. 5. Conditions of coexistence for three source utilization. This is because they
or the continuous analog. Here, pj, is Species arrayed along a resource dimen- fail to incorporate explicitly the mech-
sion. The K’s of peripheral species (1 and
the percentage of utilization of resource anism of that utilization. Indeed, inter-
3) are equal; species with d/w and K2/K:
k by competitor i. Thus the p’s specify within the limiting curves will coexist; A, preted literally they seem better to
the niche of competitor i along the the eigenvalue that determines stability in model competition by direct interfer-
dimension indexed by k, They also as- May’s stochastic model and sets the rate ence (69, 70), though if considered a
sumed that species have niches with of return to equilibrium in the determinis- Taylor series, they can approximate a
tic model, is plotted as a third dimension
the same shape and variance (w?) and (contours spaced 0.2 d/w-units apart). variety of more complex equations.
with means differing from those of Note the rapid rise of A past d/w= 1. Alternatives to Eq. 1 now exist for
adjacent niches by a constant amount, Dashed line shows A’s for which all equi- both one and several trophic levels,
d. Not only does this last assumption librium populations are equal. [From May and I shall review three examples,
(1/47); courtesy of the American Mathe-
collapse the algebra, but it also sets the
matical Society]
Without much explicit justification,
stage for predictions about how varia- Eq. 1 is sometimes said to model the
tion in d, w, and niche shape affects a situation where each competing species
and thereby affects coexistence. has an exclusive set of resources, this
MacArthur and Levins dealt with the increasing species diversity of an as- resulting in the stable node of Fig. 4a,
survival of a species trying to wedge semblage. However, if competition were effected
its niche between the niches of two A second approach to limiting sim- purely through exploitation of re-
species that were already established, ilarity, developed by May and Mac- sources, it would be difficult to imagine
For bell-shaped niches and equal K’s, Arthur (3/) and May (67), incorporates how, as Eq. 1. implies, one species
limiting d/w for invasion is 1.56, and stochastic variation. For competition could reduce the other’s abundance to
this decreases as the invader’s K in- coefficients given by Eq, 2, equilibria zero, much less how this reduction
creases. Roughgarden dealt with the involving all positive N, are neces- could be linear. A model (70) for one
survival of the peripheral as well as sarily stable for the deterministic ver- trophic level that incorporates resource
the “sandwiched” species and deter- sion of Eq. 1, Hence, as far as stability partitioning explicitly is
mined how variation in niche kurtosis is concerned, there is no limiting sim-
dN. _ tian To
{a shape
similarity.
parameter) affects limiting
As utilization curves with
ilarity, so that the deterministic ap-
proaches just discussed deal with posi-
“= PN Re aca 7)
constant w vary from normal to back- tivity. However, the situation is (3)
to-back exponentials, limiting similarity different for a stochastic analog. By in- where R, is the number of individuals
decreases. He speculated that concave corporating environmental fluctuations of consumer 1 produced from a unit
niches may characterize species whose (white noise) into the competitors’ K’s, of processed energy, C, is the cost in
resources show little long-term depend- May showed that over a wide range of energy of the death and maintenance
ability. He also predicted a bimodality the fluctuations’ amplitude, limiting d/w of an individual of competitor 1, Ip; is
in the distribution of d/w in nature. is close to one (Fig. 5). Again, the the energy extractable from competitor
Species with d/w larger than the upper more species along a resource dimen- I’s exclusive resources per unit time,
bound of a certain interval can always sion, the less tightly they can be packed, Ty, is the energy that individuals of
coexist. Below the lower bound, suc- although this effect diminishes rapidly competitor 1 can extract from re-
cessful invasion can often result in one for 2 >4, sources used by both species per unit
of the peripheral species becoming There are many cases in nature of time, and @ is the relative likelihood
extinct, giving a small d/w between the species separating on one dimension of an individual of competitor 2 get-
two survivors, May cxamined limiting having d/w 21. Examples are birds ting a unit of the overlapping resource
similarity for arbitrary numbers of spe- separating by prey size and feeding relative to an individual of competitor
cies and showed that the more species height, and parasitic wasps separating 1.
along the resource dimension, the less by depth of prey in wood. There are The equation for competitor 2 is
tightly they can be packed. It is inter- also cases in which limiting similarity similar to Eq. 3 but has a 8 multiply-
esting that Pianka (47) has evidence appears to be much less than 1, for ing the second term; the format easily
from desert lizards for this “diffuse example, tropical frugivorous birds (66, generalizes to # species, Equation 3 is
competition,” that is, the aggregate 68). According to the theory just pre- not appropriate where the Rs are small
competition from many species rather sented, such exceptions may occur in relative to the J’s but is better close to
than just from neighboring ones. He extremely stable environments or equilibrium. Figure 4b shows the equi-
finds that niche overlap decreases with among species whose niches are con- librial solutions for two competing spe-
5 JULY 1974
35
cies, neither of which can be reduced A second two-level system (73) has redistributing its phenotypes (78).
below the level of that population size consumers the same as in Eq. 4 but Niches that change in tandem with
supportable on its exclusive resources. resources growing according to abundance of consumers and resources
When two competitors are mimicked could lead to highly complicated sys-
by equations such as Eq. 3, there is dF ./dt = Sx — Fad aiN: tems behavior, including multiple equi-
always a single stable equilibrium. By i
ko l,....m (6) libria and periodic rather than point
including more resources, the equations equilibria, Nonetheless, in situations of
can be made more complicated and where §,, is the number of resource k tightest specics-packing the niches them-
less tractable. However, they can also entering the system per unit time (/5, selves may approach some limiting
be simplified so as to model coexistence 69, 70). At equilibrium, Eq. 6 is struc- form.
or extinction for Miller’s (71) “in- turally the same as models of the type The most critical problem in apply-
cluded niche” phenomenon. Here one of Eq. 3. ing the models is identification of re-
species has its resource kinds contained Two-level models show explicitly how sources in nature. Theorems have
entirely within those of the other, many key variables affect competition. appeared for various model-systems
whereas the other has resource kinds For example, Eq. 5 says that the larger stating that a necessary condition for
not used by the first. The appropriate a resource kind’s net caloric value to coexistence of n consumers is that
model for the broad-niched species is consumers and the smaller its intrinsic there be at least n resources (79). In
Eq. 3, and for the included species, rate of increase (a measure of recov- what may well become a classic paper,
AN,/ dt = RyNyA[8/(Ny + BN.) — C2} ery), the greater that resource’s con- Haigh and Maynard-Smith (80) ex-
is the appropriate model (70). The tribution to the magnitude of the com- amine the question of when population
broad-niched species always persists, petition coefficient, a. When resources units can act as resources in the
but the narrow-niched species survives have a one-to-one correspondence with theorems. They conclude that, to act
if and only if K,Co<1I,8 (Fig. 4c). habitats, Eq. 5 shows that the more as resources, populations need only not
Among other things, this condition re- similar the habitat preferences (reflect- be functionally dependent (81); for
quires that individuals of the narrow- ed in the a’s), the greater a. This in- example, different parts of the same
niched species be sufficiently better terpretation of resources provides an plant and different life stages of an
than those of the broad-niched species approximate justification for the form insect species could be separate re-
at appropriating the overlapping re- of Eq. 2, since the a’s are proportional sources even if their abundances were
sources, and that those resources be to the p’s. When dietary utilizations are highly correlated. One way to determine
sufficiently abundant and calorically known, a;, in Eq. 5 must be replaced such units is to examine demographic
worthwhile. This model thus conditions by 1,,/f,, where ny, is the number of properties of the prospective resources
coexistence on parameters more directly items of resource & an individual of themselves. Another is to demonstrate
related to resource utilization than K competitor i eats per unit time, and f, or assume resource competition, then
and a. is the standing relative abundance of employ statistical techniques such as
Instead of fluxes through the food resource k (74). Then the more sim- discriminant or multiway-contingency-
processing machinery of competitors, ilar the diets, the greater o; the smaller table analysis (36, 82) to determine
resources can be modeled as reproduc- the standing frequency of a resource what axes and categories best separate
ing populations. MacArthur’s (72) two- kind relative to its utilization, the great- consumer niches. Incidentally, the hoop-
level system is the best-known cxample: er its contribution to a. la over resource identification has ob-
While yielding intriguing results, to scured what, from a resource’s view-
AN
if dt = RiNi >) aixbioF's — Cs) achieve tractability the mathematical point, is an equivalent problem, and
i theory of resource partitioning has had that is consumer identification. Indeed,
i= I, wee lt
dP.
/ dt = nF, — (Fi?/ Ki) _ so far to ignore major kinds of bio- in certain pollination systems, plants
. F, SrauN: kK=1,...,m
logical variability. Most of its restric- and insects seasonally reverse the roles
i tive assumptions are too technical to of competitors and resources (83)!
(4) give here. Rather, we can condense the Finally, an entirely different approach
where F,, is the number of resource k, major ones by stating that the theory to interspecific habitat separation is to
4, is the consumption rate of resource is one of species with static niches or consider the habitat kinds as units
k by competitor i, by, is the net energy abilities. That is, the utilizations (p’s), within which species come to competi-
per item of resource & extractable by consumption rates (a’s), and efficiencies tive equilibrium separately. This as-
an individual of competitor i, r; is the (6’s) vary neither with abundance of sumption would certainly be true were
intrinsic rate of increase of resource k, the resources nor with time. Yet often the habitat patches completely isolated
K,, is the carrying capacity of resource in reality individuals saturate with too from one another. But if individuals
k, and other symbols are as before. At much food (75), restrict the kinds of distributed their utilizations among hab-
equilibrium, Eqs. 1 and 4 are structur- resources eaten when food is abundant itats so as to balance their habitat-
ally equivalent, whence (59), and apparently change their abil- specific caloric benefits and costs, the
ity to perceive (76) and digest (77) assumption could be approximately met.
wy | » ews | |
resource kinds, depending on abun- Then spatial niches would not be static
dances. Aside from these adjustments in population-dynamics time. The a’s
k=1 x
of individuals, natural selection, with could be estimated by regression, with
| > othe | (5) some lag, should change a_ species’ habitat-specific K’s and N’s being used
k=1 7 utilization and efficiency parameters by as variables. Complementary occur-
36 SCIENCE, VOL. 185
rences of species among the habitats come be extinction or coexistence. For studies of resource partitioning, ques-
then could indicate the outcome of a wide range of parameters, however, tions this article has related to the
strong competition rather than the weak such predation has no effect (86). (ii) theory in a preliminary way.
competition that Eq. 2 implies (70). Predation in which prey-specific con- 1) What is the mechanism of com-
sumption rate (a) monotonically de- petition? What is the relative impor-
creases as that prey’s relative abun- tance of predation? Are differences
Cross-Community Comparisons dance decreases and goes to zero as that likely to be caused by pressures toward
prey vanishes allows prey species to reproductive isolation?
Another way to assess the importance coexist with greater a’s and correspond- 2) Are niches (utilizations) regularly
of competition is to examine changes ingly lesser ecological differences (87). spaced along a single dimension?
in properties of single species across (iit) Prey when rare may evolve dif- 3) How many dimensions are impor-
communities. When such changes cor- ferences in appearance so as to escape tant, and is there a tendency for more
relate with presence or absence of espe- predators whose prey-specific consump- dimensions to be added as species
cially similar species, one may infer a tion rates vary with that prey’s abun- number increases?
causal relationship. When the proper- dance (88). (iv) The premium placed 4) Is dimensional separation com-
ties are morphological, the changes are on the consumer gaining energy quick- plementary?
labeled character displacement (84); ly while feeding is set in part by the 5) Which dimensions are utilized,
when they are ecological, the changes likelihood of the consumer being eaten how do they rank in importance, and
are labeled ecological shift (85). while feeding (59). Hence predators why? How do particular dimensions
Unfortunately, such simple compari- may reinforce selection for feeding effi- change in rank as species number in-
sons fail to discriminate between ciency, including resource specializa- creases?
changes resulting from competition and tion, rather than obviate it. 6) What is the relation of dimen-
changes resulting from other mecha- To disentangle the possible effects of sional separation to difference in phe-
nisms, such as reproductive isolation. predation and to show how these effects notypic indicators? To what extent does
Again, elaboration of predictions are interact with competition is now one the functional relation of phenotype to
required. For example, one can exam- of the major challenges of ecology. To resource characteristics constrain parti-
ine the modality of character change: discriminate between the effects de- tioning?
characteristics especially effective for scribed in (ii) and (iii), for example, 7) What is the distance between
reproductive isolation, such as voice again requires an examination of the mean position of niches, what is the
and plumage for birds, should differ modality of differences: differences re- niche standard deviation, and what is
from those most facilitating resource lating to resource partitioning should the ratio of the two? What is the niche
partitioning, such as bill and body size. decrease, whereas those relating to shape?
Or one can ask whether the various predator avoidance should increase.
mechanisms should always produce dis- And whether competition or predation References and Notes

placement. Interestingly, the answer for is dominant will depend on size and 1, G. E. Hutchinson, Am. Nat, 93, 145 (1959),
2. Based on data of Table 1, using studies from
competition is no. Several models using trophic position: herbivores (89) and 1959 through 1972. Growth of papers during
population-dynamics equations like Eqs. small animals (90) are more likely to that period is exponential (NV, = Ne"), with
yo 0.25 year-!, In contrast, D. J. de Solla
i and 4 predict convergence for gen- be regulated by predation than are sec- Price [Little Science, Big Science (Columbia
eralized feeders or homogeneous re- ondary consumers and large animals. Univ, Press, New York, 1963)] finds that
exponentially growing scientific publications
sources (33, 65). Another model, based Most of the classical examples of slight have r~ 0.05 to 0.07, a doubling time of
about four times as long. Although the small
on an optimal-feeding argument, pre- ecological differences, such as Ross’ interval of time (14 years) and the tendency
dicts that a generalist species should leafhoppers (97), do seem to fit one to select newer studies for a review place
the magnitude of the difference in question,
shrink in size when contacting another, or both of the two former categories. the relation is quite regular and the true
whether that other is larger or smaller. And large predatory animals, such as difference probably still substantial.
3. A. J. Lotka, J. Wash. Acad. Sci, 22, 461
Thus, convergence should be associated Accipiter hawks, Kingfishers, and mus- (1932); V. Volterra, Mem. R. Accad. Lincei
with and only with shrinking, a predic- telids, provide classical cases of re- Ser. 6, Ti (1926),
. G. F. Gause, Struggle for Existence (Hafner,
tion verified for Anolis lizards (see source partitioning (75, 27, 92), But New York, 1934).
we

59). the tendency has yet to he investigated . J. Huxley, Evolution, the Modern Synthesis
(Allen & Unwin, London, 1942); E. Mayr,
in detail. Systematics and the Origin of Species (Dover,
New York, 1942),
6, E, Mayr, Ibis 115, 432 (1973).
Interaction of Predation 7. G, E, Hutchinson, Cold Spring Harbor Symp,
Quant. Biol. 22, 415 (1957),
and Competition Conclusion 8. L. B, Slobodkin, Growth and Regulation of
Animal Populations (Holt, Rinehart &
Winston, New York, 1962).
The theory of how predation affects To understand resource partitioning, 9, R. G. Jaeger, Ecology 52, 632 (1971); B. A.
Menge, ibid, 53, 635 (1972).
species differences and diversity is much essentially a community phenomenon, 10. BE. Nevo, G. Gorman, M. Soulé, S. Y. Yang,
less well-developed than is the theory we require a holistic theory that draws R. Clover, K. Javonovié, Oecologia (Beri.)
10, 183 (1972).
for competition. Several major ele- upon models at the individual and pop- il, J. H. Connell, Ecology 42, 710 (1961).
ments for the former have already ap- ulation level. Yet some investigators 12. P. R. Grant, Annu. Rev. Ecol. Syst. 3, 79
(1972); R. F. Inger and B. Greenberg, Ecol-
peared, however. (i) Predation in are still content mainly to document ogy 47, 746 (1966); D. H. Sheppard, ibid. 52,
which the consumption rates per unit 320 (1971); W. Sheppe, Can. Field Nat. 81,
differences between species, a proce- 81 (1967).
prey (the a’s in Eqs. 4 and 6) are con- dure of only limited interest. There- 13. J. R. Koplin and.R, 8. Hoffmann, 4m. Mid.
stant can reverse the outcome of com- fore, it may be useful to conclude with Nat, 80, 494 (1968).
14, J. L, Menge and B. A. Menge, Ecol. Monogr,
petition among prey, whether that out- a list of questions appropriate for in press,
3 JULY 1974 37
 15, R. H. MacArthur, Geographical Ecology 49. M. L. Cody, ibid. 54, 31 (1973). S. A. Levin, dm. Nat. 104, 413 (1970); A.
(Harper & Row, New York, 1972). 50, J. M. Edington and M. A. Edington, J. Rescigno and I, W. Richardson, Bull. Math.
16. J. L. Harper [Brookhaven Symp. Biol. 22, 48 Anim, Ecol. 41, 331 (1972). Biophys, 27, 85 (1965); A. Rescigno, ibid,
(1969)] writes ‘the essentially similar require- 51. B. K, McNab, Ecology 52, 352 (1971). How- 30, 291 (1968).
ments of all green plants—solar radiation, ever, Fleming et al. (52) found for Central 80. J. Haigh and J. Maynard-Smith, Theor. Pop.
water, carbon dioxide, and a basic set of America that this procedure crowds cells Biol. 3, 290 (1973),
mineral nutrients—provide little opportunity corresponding to small species, perhaps be- 81. In the sense that there exists a function ¢
for diversification in relation to food supply.” cause of greater importance of habitat or such that g(x,, x,)=0, where x: and x,
Discussions of resource partitioning in plants food-species partitioning on the mainland. are the abundances of the two resources.
are given by R. H. Whittaker [Science 147, 52. T. H. Fleming, E. T. Hooper, D. E, Wilson, 82. R. H, Green, Ecology 52, 543 (1971); S. E.
250 (1965)] and F. D. Putwain and J. L. Ecology 53, 555 (1972), Fienberg, ibid. 51, 419 (1970),
Harper [/. Ecol. 58, 251 (1970)]. 53, R. H. MacArthur and E, Pianka, Am. Nat. 83. T. Mosquin, Oikos 22, 398 (1971).
17. Because it is of central interest to this article 100, 603 (1966). 84, W. L. Brown and E. O, Wilson, Syst. Zool.
to rank dimensions in order of importance 54. This argument is put into precise form as 5, 49 (1956); P. R. Grant, Biol. J. Linn.
for separating species-pairs within groups, follows, Contingency models of optimal feed- Soc. 4, 39 (1972).
Table 1 excludes groups of two species, in ing specify that resource kind x should be 85. R. H. MacArthur and E, O. Wilson, Theory
which evaluation can be made for only one added to a diet not including x, if and only of Island Biogeography (Princeton Univ.
pair. J rank dimensions as follows. Usually, I if Press, Princeton, N.J., 1967); J. Diamond,
prepare a matrix comparing all possible pairs Proc, Natl. Acad. Sci. U.S.A 67, 529 (1970);
of the species of interest, then ask for each
pair whether they differ “significantly” (that
Crfts) > (SZ espa —CsTsV(Z tips + Ts) A.
Keast, Biotropica 2, 61 (1970).
diet diet 86. N. F. Cramer and R. M. May [J. Theor.
usually being a matter of the author’s opinion Biol. 34, 289 (1971)], trying to model R. T,
or particular comparative technique) in each where p, is the environmental frequency of Paine’s results on predator removal in the
of the dimensions investigated, Then for each food type i, e, is the net energy gained from intertidal zone [dm. Nat. 100, 65 (1966)].
dimension I total the number of pairs that an item of type i once found, f, is the time 87, J. Roughgarden and M. Feldman, in prepara-
differ significantly in that dimension, and it takes to eat an item of food type i, tion.
rank the dimensions on the basis of that T, is the search time between two items 88. B. Clarke, Syst. Assoc. Publ. 4 (Taxonomy
number. The method is similar to that of of encountered food, C, is the caloric cost and Geography), 393 (1962); G, B. Moment,
Lack (78), except he only counts that dimen- per unit search time, and sumunations are Science 136, 262 (1962); A. S. Rand, Atas
sion most important in separating each pair, over those item kinds actually eaten. Since do Simpésio sobre a Biota Amazénica 5,
whereas I count all important dimensions for the inequality does not contain pz (nor in (Zoology), 73 (1967).
each pair. If only morphological indicators, rearranged form does it contain the absolute 89. N . G, Hairston, F. E. Smith, L. B. Slobod-
such as size, are available, I total the num- abundance of type x), competitors that reduce kin, Am. Nat. 94, 421 (1960).
ber of pairs differing by at least a ratio of item x’s abundance do not affect choice. 90, D. S. Wilson, in preparation,
1.33 [this procedure is a conservative one But if i indexes types of habitat patches, 91. H, H. Ross, Evolution 11, 113 (1957), But
for estimating significant food differences (19, the situation is different. Then competitors by see D. E, Breedlove and P. R, Ehrlich,
20)}. If numerical overlap values are available eating in particular patches can change the Oecologia (Berl. 10, 99 (1972); D. G. Dodds,
or can be calculated, I rank dimensions ac- value of the e’s and ?’s and thereby change J. Wildl. Manage, 24, 52 (1960); P, J. Jarman,
cording to their average overlap values, the the inequality’s direction. Oecologia (Berl.) 8, 157 (1971).
average being taken over all possible pairs. 55, This subject is discussed in greater detail 92. B, K. McNab, Ecology 52, 845 (1971); R.
All dimensions are treated as if they were in T. W. Schoener, Proc. Natl Acad. Sci. W. Storer, Auk 83, 423 (1966).
independent, not because they all are inde- U.S.A., in press. 93, E, G. Horn, Ecology 52, 475 (1971).
pendent, but because dependence is hard to 56, Significance determined by x2 or exact tests. . N. G. Hairston, Evolution 12, 440 (1958).
ascertain and to correct for in most studies. 57, R. E. Ricklefs, Evolution 20, 235 (1966); 95. T. B, Reynoldson and R. W. Davies, J.
Grouping of dimensions under major head- R. H. MacArthur, Am, Nat, 98, 387 (1964). Anim. Ecol. 39, 599 (1970).
ings was usually not difficult except for dis- 58. Exact data on mean sizes of resources and 96, C. E, King, Ecology 43, 515 (1962),
tinguishing macro- from microhabitat. This consumers are generally not given in the 97, G. E. Hutchinson [The Ecological Theater
was especially hard in aquatic groups and is references. Therefore, my classification is and the Evolutionary Play (Yale Univ. Press,
somewhat arbitrary, somewhat imprecise, and others would prob- New Haven, Conn,, 1965)] discussing B.
18, D. Lack, Ecological Isolation in Birds OHar- ably differ in the way they classified the Berzins’ analysis,
vard Univ. Press, Cambridge, Mass., 1972). data, but the trend is so strong that it would 98. R. O. Brinkhurst and K. E. Chia, /. Fish.
19. P. W. Price, Ecology 51, 445 (1970); ibid. 53, probably not be qualitatively affected. Res. Board, Can. 26, 2659 (1969),
190 (1972). 59. T. W. Schoener, Am. Nat. 103, 277 (1969); 99, Cc, P. Magnum, Limnol. Oceanogr. 9, 12
20. T. W. Schoener, ibid. 49, 704 (1968). Annu, Rey, Ecol. Syst, 2, 369 (1971). (1964).
21, E. R, Pianka, ibid. 50, 1012 (1969). 60, H. L. Sanders and R. R. Hessler, Science 100. J. H. Stone, Ecol Monogr. 39, 433 (1969).
22. R. A. Croker, Ecol. Monogr. 37, 173 (1957). 163, 1419 (1969); P. K. Dayton and R, R. 101, A. J. Kohn, ibid. 29, 47 (1959); Limnol.
23. N. P. Ashmole, Syst. Zool. 17, 292 (1968). Hessler, Deep-Sea Res. 19, 199 (1972), Oceanogr. 16, 332 (1971).
24. R. T. Holmes and F, A. Pitelka, ibid., p. 305. 61. R, R. Vance, Ecology 53, 1062 (1972), 102. , Ecology 49, 1046 (1968).
25. R. B. Root, Ecol. Monogr. 37, 317 (1967). 62. M. Lloyd and H. S. Dybas, Evolution 20, 103. D. J. G. Griffin, J. Anim. Ecol, 40, 597
26. E, J. Maly and M. P. Maly, in preparation. 466 (1966). (1971),
27. T. W, Schoener, Evolution 19, 189 (1965). 63. C. Strobeck, Ecology 54, 650 (1973). 104, V. T. Hammer and W. W. Sawchyn, Limmol.
28. J. M. Diamond, Science 179, 759 (1973), 64. The terminology is that of nonlinear dif- Oceanogr. 13, 476 (1968).
29. T. W. Schoener, Am, Nat. 104, 155 (1970). ferential equations; an introduction very 105, D. C, Culver, Ecology 51, 949 (1970).
30. and D. H. Janzen, ibid. 101, 207 suitable for biologists is given in N. Keyfitz, 106. D. N, Ueckert and R. M. Hansen, Oecologia
(1968). Introduction to the Mathematics of Popula- (Berl.) 8, 276 (1971).
31. R. M. May and R, H, MacArthur, Proc, Nail. tion (Addison-Wesley, Reading, Mass., 1968). 107. E. B, Caplan, Ecology 47, 1074 (1966).
Acad, Sci. U.S.A. 69, 1109 (1972). 65, R. H. MacArthur and R, Levins, Am. Nat. 108. W. A, Sands, J. Anim, Ecol. 34, 557 (1965);
32. R, Levins, Evolution in Changing Environ- 101, 377 (1967). Entomol, Exp. Appl. 4, 277 (1961).
ments (Princeton Univ. Press, Princeton, N.J., 66, J. Roughgarden, Theor. Pep. Biol., in press, 109, E. Broadhead, J, Anim, Ecol. 27, 217 (1958),
1968). 67, R. M. May, Stability and Complexity in 116. A.M. Young, Am. Midl. Nat. 87, 146 (1972).
33. R. H. MacArthur, Biol. Rev. (Camb.) 40, 510 Model Ecosystems (Princeton Univ. Press, 111. N. Dawson, J. Anim, Ecol. 34, 299 (1965),
(1965). Princeton, N.J., 1973), 112. C. A. Istock, Evolution 20, 211 (1966),
34. In fact, correlation coefficients for lizards 113, R. E. Ricklefs, R. M. Adams, R. L, Dress-
68, J. Terborgh and J. M. Diamond, Wilson
are 400 and .536 (35), but these are less ler, Ecology 50, 713 (1969).
Bull, 82, 29 (1970).
likely to be different from zero because of 114. D. C. Culver, ibid, 53, 126 (1972).
the smaller sample size. 69. T. W. Schoener, Theor. Pop. Biol. 4, 56
115. H. Heatwole and D. M. Davis, ibid. 46, 140
35. T. W. Schoener, in Biology of the Reptilia, (1973).
(1965),
C, Gans, Ed. (Academic Press, New York, 0, , ibid., in press,
116. R. V. O'Neill, ibid. 48, 983 (1967).
1974). 71. R. S, Miller, Ecology 45, 256 (1964); Adv.
117, H. W. Hurlbutt, Syst. Zool. 17, 261 (1968).
and A. Schoener, Breviora 368, 1 Ecol. Res. 4, 1 (1967).
118. C. A. Lanciani, Ecology 51, 338 (1970).
T. W. Schoener, Ecology 51, 408 72. R. H. MacArthur, in Population Biology and 119. P. S. Maitland, J. Anim, Ecol, 34, 109 (1965),
Evolution, R. €, Lewontin, Ed. (Syracuse 120, R. N Gibson, ibid. 41, 189 (1972),
37. M, 1. Cody, Am. Nat. 102, 107 (1966). Univ, Press, Syracuse, N.¥., 1968). 121, N. G. Hairston, Ecol. Monogr. 19, 47 (1949).
38. B. K. Snow and D. W. Snow, J. Anim. Ecol, 73. Independently devised by MacArthur @3) 122. R. A. Avery, Oikos 19, 408 (1968).
41, 471 (1972). and Schoener (69, 70), 123. R. F. Inger and B, Greenberg, Ecology 47,
39, H. A. Hespenheide, Ibis 113, 59 (1971). 74. A detailed discussion of this modificd compe~- 746 (1966).
40. P. EF, Hillman, Ecology 50, 476 (1969). tition coefficient is given in T. W. Schoener, 124, J. A. Moore, in Genetics, Paleontology, and
41, E, R. Pianka, paper presented at symposium Am, Nat., in press, Evolution, G. L. Jepsen, E. Mayr, G. G.
for R. H. MacArthur, Princeton, N.J. 75, C, S. Holling, Can, Entomol. 91, 385 (1959); Simpson, Eds. (Princeton Univ, Press, Prince-
(1973). Mem. Entomol. Soc. Can, 48, 1 (1966). ton, N.J., 1949), p. 315.
42.T. M. Zaret and A. S. Rand, Ecology 52, 76. L. Tinbergen, Arch, Neerl. Zool, 13, 265 125, W. W. Milstead, Tex. J. Sci. 9, 410 (1957),
336 (1971). (1960), 126. P. A. Medica, Bull, South Calif, Acad. Sci.
43, A, Keast, in Marine Food Chains, J. H. 77. R. T. Paine, personal communication. 66, 251 (1967),
Steele, Ed. (Oliver & Boyd, Edinburgh, 1970),
78. J. Roughgarden, Am, Nat. 106, 683 (1972). 127. A. S. Rand, Ecology 45, 745 (1964); T. W.
. 377. Roughgarden (unpublished manuscript) has Schoener and A. Schoener, Breviora M.C.Z.
44, &. A, Sandercock, Limnol, Oceanogr, 12, 375, 1 (1971).
recently modeled faunal buildup in which
97 (1967). 128. R. B. Huey, thesis, University of Texas
niche distance (d) undergoes evolutionary
45. T. W. Schoener, in preparation, change between invasions, Then, the greater (1969).
46, D, G. Broadley, Zool. Afr. 3, 1 (1967), the number of species, the smaller d—the 129, D. W, Tinkle, Tulane Stud. Zool. 6, 3 (1958),
47. R. T. Paine, Ecology 44, 63 (1963). opposite of what is expected for species 130. Cc. C, Carpenter, Ecol. Monogr. 22, 235
48. R. F. Inger, ibid. 40, 127 (1959), packed at limiting d/w. (1952),
38 SCIENCE, VOL. 185
131, J. Bédard, Can. J. Zool. 47, 1025 (1969). 139. H. R. Pulliam and F, Enders, Ecology 52, 147. R. M. May, in Some Mathematical Prob-
132. D. A. Jenni, Ecol. Monogr. 39, 245 (1969). 557 (1971), lems in Biology (American Mathematical
133. P. 3. S. Olmey, Trans. Congr. Int. Union 140. 1. Newton, Ibis 109, 33 (1967). Society, Providence, R.1., 1974), vol. 5.
Game Biol. 6, 309 (1963). 141. H. C. Heller and D. M. Gates, Ecology 52, 148. I thank P, B. Hertz, R. B. Huey, J. Rough-
424 (1971). garden, A. Schoener, R. L. Trivers, and E. E.
134. D. W. Johnston, duk 88, 796 (1971). Williams for criticism of an earlier draft
142. M. L. Rosenzweig and J, Winakur, ibid. 56,
135, T. H. Hamilton, Condor 64, 40 (1962). 538 (1969). of this article and R. B. Huey, A. Keast,
136. R. H. MacArthur, Ecology 39, 599 (1958), 143. L. N. Brown, J. Mammal, 45, 189 (1964), E. J. Maly, R. M. May, B. A. Menge,
137. G. H. Orians and H. S, Horn, ibid. 50, 930 144, B. Lim, ibid, 51, 730 (1970). J. L. Menge, E. R, Pianka, J, Roughgarden,
(1969). 145, G. B. Schaller, The Serengeti Lion (Univ. of and D, S. Wilson for access to unpublished
138. B. K. Snow and D. W. Snow, Auk 88, 291 Chicago Press, Chicago, 1972). material. Supported by NSF grant GB-
(1971). 146. T. H. Kunz, J. Mammal. 54, 14 (1973). 37731X.

NEWS AND COMMENT The committee, however, leaves no

Killian Committee: Report Urges
Advisory Council in White House
The scientific community has reacted
to the abolition of the science advisory
machinery in the White House rather
like an amputee whose phantom foot
continues to hurt long after the leg is
gone. No doubt propinquity to the
President has a heavy symbolism for
scientists, but many are also convinced
of the merits of the case for
room at the top for a science adviser
and his staff. And in recent months
increasing number of voices calling for
restoration of a science adviser to the
White House have been heard, although
not, it should be noted, from the direc-
tion of the White House.
More on the issue is sure to be said
in hearings before the House Astronau-
tics and Space Committee scheduled
run well into July. These hearings on
federal policy, planning, and organiza-
tion of science and technology began
on 20 June with an appearance by Sen-
ator Edward M. Kennedy (D-Mass.),
who is the senator most strategically
placed to influence science policy issues
on that side of Capitol Hill. The Ken-
nedy appearance was essentially a cour-
tesy call, although his opinion
politely solicited on several matters.
The committee should get down to
more specific cases on 26 June when
it is scheduled to hear former presi-
dential science adviser James R. Kil-
lian, Jr., discuss a recently completed
report by a blue-ribbon committee he
chaired* (Science, 8 February).
Killian committee was formed at the
behest of the council of the National
* Science and Technology in Presidential Policy-
making: A Proposal, Individual copies ate avail-
able from the Printing and Publishing Office, Na-
tional Academy of Sciences, 2101 Constitution
Avenue, NW, Washington, D.C. 20418,
5 JULY 1974
Academy of Sciences “to look into the
question of scientific and technical ad-
vice to the government, including the
advisory and coordinating functions
previously carried out by the White
House science advisory complex.”
The report is likely to have consid-
erable impact, not only because of the
making prestige of the committee members and
the academy’s imprimatur, but also
an because the attention it will get in the
hearings is likely to make the report a
bench mark in future discussions about
science policy arrangements. (This is-
sue of Science went to press before the
report was scheduled to be discussed
at the hearings on 26 June, and this
article is based on a conversation be-
to tween Killian and reporters the previ-
ous week.)
The committee’s principal recom-
mendation is that a “Council on Sci-
ence and Technology” be created in the
Executive Office of the President along
the lines of the Council of Economic
Advisers and be designed to interact
effectively with the other staff units in
the White House and to provide close
was links with the scientific community.
Killian and his committee were acute-
ly aware that they might be accused of
special pleading in behalf of science,
and he says they “didn’t want to cry
over spilt milk or try to reconstitute
PSAC” (the President’s Science Ad-
visory Committee, which was based in
The the Executive Office until PSAC was
abolished in the reorganization of a
year ago). The report concentrates on
what science can do for government
rather than what the government can
do for science in terms of funding and
otherwise.
doubts about its position, beginning the
outline and summary of the report with
the flat statement, “The committee
concludes that the office of the Presi-
dent could benefit from a scientific and
technological presence.” This presence
the committee sees quite clearly in the
form of a Council for Science and
Technology in the Executive Office of
the President. The report describes the
council only in general terms, recom-
mending that it have the following
major features. The council should have
at least three, perhaps more, full-time
members drawn from science, engineer-
ing, and related areas. The council
members, one of whom would serve as
chairman, would be appointed by the
President subject to the advice and con-
sent of the Senate and would serve at
the President’s pleasure. The commit-
tee would prefer to see the council
established by legislative action but
leaves the matter open. A staff of 25 to
30 is suggested as appropriate.
Effective working relations between
such a council and the major White
House staff offices are given heavy
emphasis in the report. The committee
thinks that the council chairman should
sit as a member of the Domestic Coun-
cil in the White House and that the
council should participate actively in
the workings of the National Security
Council and cooperate closely with the
Office of Management and Budget
(OMB).
Stress is also put on the role the
council would play in areas of foreign
policy strongly affected by scientific
and technological considerations. It is
recommended that the Council for Sci-
ence and Technology make an annual
report to the President and through
him to the Congress. The presumable
model is the annual report of the Coun-
cil of Economic Advisers, but Killian
hopes that such a report would not be
simply a survey of activity in science
and technology, but would be devoted
to the analysis of trends which repre-
sent major opportunities or problems
39