Pattern and Process in Competition

Pattern and Process in Competition

Pattern and Process in Competition RICHARD S . MILLER Department of Biology. University of Saskatchewan. Smkatoon. Canada I. Introduction ...

5MB Sizes 2 Downloads 53 Views

Pattern and Process in Competition RICHARD S



Department of Biology. University of Saskatchewan. Smkatoon. Canada I. Introduction ......................................................... I1. The Nature of Competition ............................................ A The Process of Competition......................................... B. Component Elements of Competition................................ I11 The Ecological Niche ................................................. A . Historical Development of the Niche Concept......................... B Niche Relationships............................................... I V Competition in Nature ................................................ A Distribution Patterns .............................................. B. Community Composition........................................... C Species-Abundance................................................ D Character Displacement ............................................ E Invasions ........................................................ V Conditions of Coexistence.............................................. V I Conditions of Competitive Exclusion.................................... A Mammals......................................................... B Birds ............................................................ C Amphibia........................................................ D Clustacea ........................................................ E Insects ........................................................... V I I . Species Diversity ..................................................... References...............................................................


. .

. .


. .

1 6 7 8 16 16 17

20 21 26

. .


. . . . .

48 49

30 31 33

55 67 68 60 65 69

I. INTRODUCTION The development of competition theory has traditionally included three stages: (1) inferences drawn from observation of natural populations. (2) construction of mathematical models and (3) laboratory experiments designed to test elements of competitive interactions in controlled environments. Unfortunately. even though competition has occupied the minds of scientists in various biological disciplines for many years and is presumed to be a dominant force in biological evolution. remarkably few testable hypotheses have emerged and almost all of our conclusions about the importance of competition in natural systems are still based on speculation and inference. Darwin (1859) assigned a major role to competition between closely related species in the process of natural selection. He wrote. “As the 1



species of the same genus usually have, though by no means invariably, much similarity in habits and constitution, and always in stmcture, the struggle will generally be more severe between them, if they come into competition with each other, than between the species of distinct genera.” He cited examples of the increase of the missel-thrush at the expense of the song-thrush in Scotland, invasions in which one species of rat has displaced another, the spread of the Asiatic cockroach a t the expense of a cogener, and the extermination of the native stingless bee of Australia by the imported hive bee. Darwin also stated one of the central problems of competition theory, which exists today as it did then, when he said “We can dimly see why the competition should be most severe between allied forms, which fill nearly the same place in the economy of nature; but probably in no one case could we precisely say why one species has been victorious over another in the great battle of life. ’’ The well-known mathematical models that were subsequently developed by Volterra, Lotka and Gause from extensions of the logistic theory, to describe population growth in single- and two-species systems, were attempts to state in mathematical terms the events that naturalists such as Darwin had recorded from field observation. I n logistic theory, each individual that is added to a population N reduces the growth capacity of the population by a constant increment. Population growth is thus described by a sigmoid curve in which N approaches an upper asymptote K , which represents the carrying capacity of the environment. If two species with carrying capacities K , and K , comPete for the same resource in the same environment, and if an individual of either species reduces the growth capacities of both populations, the growth of the two species can be represented by the following equation system :

where N , and N , are the numbers of each species, r1 and rz are their respective rates of increase, and K , and K , are carrying capacities or saturation values determined by growing each species alone in this environment. The term B/K2is the inhibitory effect of N , on the growth of N , and a/K, is the reciprocal effect of N , on N,. The outcome of competition will then depend on the inequalities tc > K,/K, and B > K,/Kl.



Volterra (1926) and Lotka (1932) used this equation system to demonstrate that two species using the same resources cannot coexist indefinitely in a limited environment, and Gause (1935) confkmed this conclusion experimentally. The competitive exclusion principle that emerged as a result of this research has become known variously as the “Gause Hypothesis”, “Gause’s Law”, the “Volterra-Gause Principle”, etc. The verbal statement of this principle has also taken several forms. Crombie (1947) states, “. . . it can be demonstrated by simple argument that species with identical needs and habits cannot survive in the same place if they compete for limited resources - a t least if their needs and habits remain identical.” Hardin (1960) presents the case more simply by saying “complete competitors cannot coexist.” He defines the principle in the following terms: (1) if two non-breeding populations occupy the same niche, and (2) if they are sympatric, and (3) if A multiplies faster than B, then ultimately A will replace B and B will become extinct. Hardin notes that in practice we remove the hypothetical character of (3) because we subscribe to the “axiom of inequality” which states that “no two things or processes, in a real world, are precisely equal.” If, for example, two populations are sufficiently distinct morphologically to be recognized as species, they differ to some degree in their genetics, physiology and ecology as well. There is therefore an a priori assumption that no two species are ecologicallyidentical. This introduces a circularity, which has been discussed by Gilbert et al. ‘ (1952), such that the competitive exclusion principle can neither be proved nor disproved. If one species excludes another we say the principle is “proved”, but if they coexist we conclude that they differ ecologically and therefore occupy different niches. What importance can we assign to this axiom as a conceptual model of competition? Can we justifiably claim that the competitive exclusion principle is “perhaps the most important theoretical development in general ecology”, or that it is “one of the chief foundations of modern ecology”? (Hutchinson and Deevey, 1949). As it is usually stated, it cannot be used to either confirm or deny the existence of competition as a natural process and, as Cole (1960) has noted, it is in danger of becoming dogma. He points out that the doctrine of competitive exclusion contains a device which may be used to avoid Hardin’s (1960) admonition that “every instance of coexistence must be accounted for”, for it is easy to dismiss observations of apparent coexistence by merely saying “they obviously have to occupy different niches or they couldn’t exist.” The laboratory experiments that followed the initial research of Volterra, Lotka and Gause have not been particularly instructive insofar as competition in natural communities is concerned. The tendency



has been to reconfirm one case of the Volterra-Gause model, namely that competitive exclusion can be demonstrated, and to ignore the more interesting fact that coexistence for many generations has been repeatedly observed in highly simplified laboratory environments. After showing that Drosophilafunebris and D . melanogaster were able to coexist in population cages where the only noticeable variable was a periodic alteration in the freshness of the food medium, Merrell (1951) stated that these results do not contradict the Gause hypothesis, because the two species were shown to have different responses to the character of the medium. Merrell chose to devalue the important result that, in an otherwise constant environment, a minor change in one variable was sufficient to permit coexistence of the two species for almost two years, after which the experiment was terminated. I n their anxiety to “prove” the competitive exclusion principle, ecologists have repeatedly ignored the important fact that many, if not most, laboratory studies of competition have provided better evidence of coexistence than they have of competitive exclusion. While competition between and within species has unquestionably been an important force in the evolutionary history of natural communities and is presumed to be a major factor in the organization of modern ecosystems, there is still no satisfactory explanation for the tremendous diversity of species that is observed in natural systems. When we consider the relatively low ecological efficiencies of animals, and the fact that a classical food chain is limited for this reason to about 4 or 5 direct links, we can understand why there is “room” for a certain variety of forms. It is surprising, nevertheless, that discrete biotopes such as small ponds or grasslands, with relatively uniform structural characteristics, support the number of speciesl they do, especially large numbers of closely related species. During four years of light trap collections at Rothampsted Experimental Station, 154 to 198 species of lepidoptera were recorded each year (Williams, 1964). Kontkanen (1950) collected 4 288 individuals of 56 species of leafhopper in one day at a locality in North Karelia, Finland. I n three summers he never collected fewer than 37 species in any sample during the months of July and August, and the mean per sample was 2 374 individuals representing 48 species. The H.M.S. Challenger expedition collected 4 248 species of benthic animals, representing 1 438 genera, from 70 stations less than 180 metres deep (Williams, 1964). Mitchell (1964) asks, “Why should several species share a common energy source if one of the species could have become adapted to take all of the energy?” Biotic communities are unquestionably the most highly organized of natural systems. There is a high degree of interrelationship among different parts, self-regulation maintains a more or




less constant ratio of components in a continuous flow of materials, and the system returns to a steady state after a disturbance of stimulus (Bray, 1958). Nevertheless, is the amount of diversity that exists in ecosystems essential to the orderly function of its component parts, and what mechanisms make the diversity possible? It is often observed that older ecosystems such as those in the tropics are richest in structural and species diversity, even though the differences in primary productivity between comparable tropic and temperate systems is not great and may, in fact, be greater in the temperate ecosystem type (Odum, 1959). Bray (1958) advances the hypothesis that re-utilization efficiency may offset the reduced primary productivity that occurs in the succession from simple to more complex communities. Re-utilization efficiency reduces the relative loss that may occur through production of nonavailable energy forms and is a function of the number and kinds of organisms and their ability to create new energy niches and thus increase the number and pathways of energy exchange (Bray, 1958). The diversification of animal food chains is regarded as an example of this function. The most obvious source of species diversity among closely related forms, or within the assemblage of organisms a t one trophic level, is through specialization (Klopfer, 1962). Any natural community contains large numbers of specialists as well as more widely adapted nonspecialists, and it is worth noting that this kind of division seems to be an almost universal phenomenon in every taxonomic group, even within genera. Curiously enough, the more ubiquitous species have asserted themselves to only a limited extent. How then does specialization confer a competitive advantage? Even though we tend to accept the premise that specialization leads to competitive superiority, we lack empirical evidence to show how this occurs. There is no clear reason why the widely adapted wood mouse (Apodemus sylvaticus) has not entirely replaced its cogener, the yellow-necked mouse (Apodemus Jlavicollis), in Britain, or why Peromyscus maniculatus is not the only member of this genus in North America. A. sylvaticus would appear to be able to inhabit every woodland habitat in Great Britain, and there is no evidence to show that A . Jlavicollis is a more efficient or more productive species. I n the absence of competition from its cogeners the least chipmunk (Eutamias minimus) can apparently inhabit any environment in which western chipmunks are found, yet there are 16 species of this genus in North America and this more widely adapted species seems, usually, to be displaced by its cogeners (Sheppard, unpublished). It is obvious that genetic systems allow a rather wide range of adaptation - most taxonomic groups contain a complement of species that are capable of inhabiting a variety of habitats, in addition



to a closely related group of more specialized forms -yet adaptability does not seem necessarily to be correlated with competitive superiority. MacArthur and MacArthur (1961) examine two ways in which species may share the resources of a community: (1) each species may have strict habitat preferences, feeding throughout its particular habitat on all kinds of food, or (2) all species may share the same habitat, each species feeding on a different food or in a different situation within the habitat. The first alternative violates the “jack of all trades - master of none” principle that natural selection favors the increased efficiency of a certain amount of specialization; in the second alternative specialization has proceeded so far that time and energy are wasted travelling from one point to another, where the particular food item or requirement is located (MacArthur and MacArthur, 1961). These authors note that it is difficult to determine where the balance of these opposing requirements is reached. As long as specialization leads to increased efficiency, it will be favored by natural selection, but only up to the point where time and energy are not wasted. While ecologists subscribe to the general proposition that evolution proceeds as a product of diversity and that diversity is accompanied by homeostatic mechanisms that promote stability, considerable development and modification of conventional competition theory will be required before species interactions can be described in terms that can be incorporated into testable hypotheses, and which will lead to a clear understanding of the exact nature of the mechanisms of competition and species diversity. 11. THE NATUREOF COMPETITION In dictionary usage “competition” is defined simply as “a contest between two rivals for the same object.” This definition is not sufficiently precise for ecological use and various attempts have been made to qualify biological competition with respect to ( 1 ) the nature of the contest, (2) the object of the competition and (3) the characters of the rivals. The inconsistencies in the literature regarding the meaning and significance of competition have been reviewed by Crombie (1947), Solomon (1949), Birch (195T), Milne (1961) and others. I n the discussions that follow I will simply employ a modified version of the definition proposed by Clements and Shelford (1939), namely, that “biological competition is the active demand by two or more individuals of the same species population (intraspecies competition) or members of two or more species a t the same trophic level (interspecies competition) for acommonresourceorrequirement that is actually or potentially limiting.” Questions that arise in any consideration of the nature of competition are: What precisely is the form of the competitive process and what



are its component elements? What constitutes a common resource or ecological requirement that can legitimately be considered the object of the competitive process? Is competition limited to interactions between individuals and species at the same trophic level, or is any interaction which has a deleterious effect on the existence or increase potential of another individual or species a form of competition? How can the action of competition be detected and evaluated in natural situations, and what constitutes adequate proof of competition? A. THE PROCESS O F COMPETITION Because most of our knowledge of competition has come from laboratory studies designed to test the outcome of a competitive interaction, there has been a tendency to ignore the fact and nature of the process and to view competition almost entirely in terms of its end result of selective elimination. It is perfectly legitimate to design an experiment in which competition is defined as the elimination of one species by another, as in Park’s atudies of competition in Tribolium, but one must also recognize that this design imposes an artificial criterion which is merely a condition of the experiment. Thus, in describing one such series of experiments, Park (1954) writes, “. . . for the purposes of this paper only I shall refer to competition merely as those new events which emerge when two species co-associate and which lead to the persistence of one species and the elimination of the other.” Park cautions, however, that this is not advanced as a general definition of competition, but is merely a sct of ground rules he has selected for a particular purpose. Considerable ambiguity and misunderstanding has resulted from attempts to define the competition process according to its consequences, and through failure to establish a common resource as the object of competition. There may be distinct component elements within the competition process, associated for example with the ecological characteristics of particular taxonomic groups, but these may not be readily discernible in the outcome of the interaction and their considerable theoretical significance may therefore be ignored. For this reason, arbitrary schemes for classifying species interactions in terms of their positive, negative and neutral effects may be quite misleading. According to Burkholder’s (1952) system of classification, predator-prey and parasite-host interactions both have f values and may therefore be classed together. Such a classification ignores their separate ecological characteristics and requires that we view such interactions entirely as abstract energy transformations. There may be valid theoretical reasons for considering the energy flow between all primary and secondary



consumers as a single set of interactions, but this need not disguise the fact that an array of distinctive phenomena, each with unique ecological and evolutionary properties, has produced the observed results. Birch (1957) has reviewed several cases in which various authors have described interactions such as predation and parasitism as forms of competition (cf. Nicholson, 1933), regardless of the fact that the species interaction did not involve a common resource. Thus, Nicholson (1937) states that “. . . competition for food and space, and the interaction of natural enemies and their hosts, can both be represented by the same fundamental formula and its corresponding exponential curve. . .” so that the criterion of competition is merely that survival decreases as density increases. There is a substantial body of literature and theory pertaining strictly to interactions between members of the same species or between species at the same trophic level, in which the interaction is directed toward a resource which is a common requirement for both members of the interaction. There is no conceptual problem in delineating the phenomenon of competition in these terms, nor is there any obvious theoretical advantage in considering competition to be more than an ecologically qualified “contest between two rivals for a common object.”



Elton and Miller (1954), Park (1954) and others distinguish between interference and exploitation as component elements of the competition process. While the general consequences of a competitive interaction may be the same in either case (e.g. elimination of one species from a niche), the process leading to this result may involve one or both of these elements. “Interference” refers to any activity which either directly or indirectly limits a competitor’s access to a necessary resource or requirement. It usually operates in a spatial context, as in the case of territoriality, and assures possession of some minimum requirement by one individual or species. It may involve a structural resource such as a nest site or song post, or it may be a space containing a requisite amount of food or nutrients. The term “interference” has been used in somewhat different ways. It is broadly synonymous with the term “contest” as used by Nicholson (19!55), in that a contest in this sense invariably involves interference. Brian (1956) refers to interference and exploitation as “isolated components of the dual competition concept” and concludes that the Lotka-Volterra model involves competition by interference alone while the Windsor (1934) model involves only exploitation. An objection to the Lotka-Volterra model put forth by Andrewartha and



Birch (1953) also states that it includes no differential exploitation, merely interference between species. I n this connection, they assert that there is no competition unless a common resource is involved and unless one species seeks power over this resource a t the expense of another, and seem therefore to imply that interference is not a sufficient mechanism for this purpose. This objection seems, however, to have no real connection with either the Lotka-Volterra model or the question of interference or exploitation. As long as competition is defined in terms of a limiting resource, which is clearly contained in the term for carrying capacity ( K )in the model, “seeking power” in its broadest sense could be accomplished either by interference or exploitation -in fact interference would appear to be the more effective mechanism for gaining power over a resource at the expense of another species. Some authors have introduced special terms to distinguish between interspecies and intraspecies interference or territoriality (cf. Frank, 1952; Simmons, 1951), but the evidence that is available supports the view that mechanisms such as interspecific territoriality do not differ from those involved in intraspecies relationships (Hinde, 1956) and it is doubtful that such mechanisms will be evolved unless they have first arisen among members of the same species. Indirect interference through chemical repellents may constitute a class of exceptions, although this is by no means certain. Wilson and Bossert (1964) list several examples of territory and home range marking by chemical secretions in different taxonomic groups. They state, “There is no reason to doubt that the essential characteristics of intraspecific communication . . . are also true of interspecific chemical communication.” The evidence they give also suggests that most chemical repellents that act as isolating mechanisms between closely related species may act as interspecies interference mechanisms as well, and have their origins in intraspecific relationships. The spatial organization and requirements of animal populations are usually discussed in relation to territory, but attempts to assign common origins and functions to the territories of different species have led to considerable disagreement and misunderstanding among ecologists (cf. Lack, 1954; Hinde, 1956). Wynne-Edwards (1962) has compiled an impressive number of examples that are presumed to show that animal dispersion is mostly in relation to an optimum food supply and that competition for territory or individual rank and dominance is the proximate agent limiting numbers for this purpose. This is supposed to provide an effective proximate buffer to limit density near the optimum for that species, in relation to the ultimate agent of food. Unfortunately, Wynne-Edwards attempts to categorize all species in the same way, limited proximately by territory and ultimately by food,



and this weakens his argument. He does not define “optimum” and it is therefore impossible to test this critical hypothesis, but a more serious objection is the lack of evidence for food as an ultimate limiting factor in many populations (Andrewartha and Birch, 1954). While we must accept the fact that territories are rendered incompressibleat some point by psychological factors (Klopfer, 1962) and that competition for space is a form of limiting convention, it cannot be assumed that all species require space for the same reasons or that they are ultimately limited by the same factor. There are numerous examples among birds to show that territory is related to factors other than food (cf. Hinde, 1956). The principal area of defense among hole-nesting birds is usually the immediate vicinity of the nest hole, and the density of breeding pairs can be increased in most natural habitats by providing artificial nest boxes (Von Haartman, 1956). Intraspecific and interspecific territorialism exists among closely related species of plovers, but these birds, as do many others, feed outside the limits of their territories. Simmons (1956) suggests that the most important function of territory in these species is the spacing of nests and eggs, as Tinbergen (1956) has shown for other ground-nesting birds. Hartley (1949) found that the only common requirement among the territories of mourning chats (Oemnathe lugens) in their wintering quarters in the African deserts was a measure of shade. This could be provided by broken ground or some artificial counterpart such as a scar in a hill or hutments of a camp. A structural feature which is required for breeding or shelter may be more limiting than food in the habitats of some species, and it is frequently observed that the limits of population in territorial species are well below the ultimate limit of food supply, even in years of food scarcity. However, if we define “territory” simply as any “defended area or space”, regardless of its origin or function, it is evident that territorialism exists both within and between some species, and leads to competitive exclusion from particular habitats and niches. This concept can be accommodated within the broader term “interference”, which may be direct, as in the case of territoriality or overt aggression, or indirect chemical communication. I n each case the effect of interference is to ensure possession of a particular array of resources at the possible expense of other individuals or species. There is an apparent evolutionary tendency toward more formalized mechanisms of territorial behavior, with greater emphasis on display and correspondingly less on physical contact. Wynne-Edwards (1962) uses the term “epideictic display” to denote conventions that have evolved away from the direct, primitive contest and have come to assume a symbolic quality, not even directly implying threat in some cases, but which nevertheless serve to interfere with intrusion into a



territory or defended space. A question pertinent to this discussion is the extent to which such behavior is effective between species, as well as within species, and how interspecific dominance is established and maintained. Direct, physical contests require no special methods of species recognition in order to be effective. The barnacle Balanus balanoides merely overruns its competitor Chthmalus stellatus, crushing or grinding its way into the niche space it seeks to occupy (Connell, 1961). Taxonomic aEnity in such cases functions mainly to bring two species into competition, in that closely related species are more liable to have similar niche requirements. When more elaborate behavioral mechanisms are present, however, taxonomic affinity assumes greater importance. The meadowlarks Sturnella magna and S. neglecta tolerate intrusion into their territories by species outside this genus, but defense and display between these two competing species are as frequent and intense as within species, resulting in complete segregation of their territories (Lanyon, 1956). Where the plovers Charadrius dubius, C. hiaticula and Leucopolius alexandrinus occur together they defend and maintain distinct territories, spaced in such a way that they occupy three contiguous areas (Sluiters, 1954; Simmons, 1956). Territory is advertised by the patrolling behavior of the male as he curves in a “butterfly” display flight, slowly backwards and forwards, singing all the while. These display flights obviously require that visual and song signals be recognized in order to be effective interference mechanisms, and it is evident that relatively elaborate behavior elements that are found in one species are present in, and recognized by, closely related competitors. Moore (1964) also notes that interspecific encounters are more frequent and are more effective between dragonflies that resemble each other, e.g. he found that Anax irnperator invariably drove Aeshnu juncea from small ponds but the presence of other, less similar species did not provoke aggressive attacks. Interestingly enough the motivation for interspecific encounters among dragonflies might be aggressive or sexual, e.g. an attacking male might assume a copulatory position on contact with another male, but the result is the same in that interference is successful and only one male remains within the territory. In each of these and other examples, interference between species almost invariably involves a behavioral element that is present and effective within the species. Thus, if intraspecific mechanisms are lacking, it is unlikely that competitive exclusion will operate through interference between species. Closely related species are not only more likely to have similar ecological requirements and, hence, overlapping or included niches, but are also more able to evolverecognizable, formalized mechanisms of interspecific interference with less dependence on harmful physical contests.



“Exploitation” is the utilization of a resource once access to it has been achieved. Two individuals or species with unlimited access to a common source of food or nutrients will frequently have different abilities to exploit the available supply. This form of competition is referred to by Nicholson (1955) as a “scramble”, in that the most efficient competitor is the one that can obtain and utilize the greatest amount. Elements of both interference and exploitation are undoubtedly present in almost any competitive interaction, but we may inquire whether their relative values in any particular interaction are such that there are distinct theoretical consequences. Interference seems, for example, to be especially well developed among certain of the higher vertebrates, especially birds, and is often so effective that direct exploitation is a relatively unimportant element of competition. Conversely, exploitation seems to be characteristic of simple metazoans in which the populations are more directly controlled by factors in the physical environment apd social interactions are simple. I n other words, exploitation seems to be the dominant form of competition whenever interference mechanisms are poorly developed. If this is indeed the case, it implies that competition by direct exploitation is a primitive and unstable form of interaction, and that interference mechanisms such as territoriality represent evolutionary advances toward more effectivebiotic control of population interactions and numerical stability, as partly suggested by Wynne-Edwards (1962). Wynne-Edwards (1962) states “The substitution of a parcel of ground as the object of competition in place of the actual food it contains . . . is the simplest and most direct kind of convention it is possible to have.” Such an arrangement implies interference with access to a resource, which Wynne-Edwards assumes is a property of all species; buti for many species such conventions have not evolved and direct exploitation of food resources is a characteristic form of competition. Examples of competition through exploitation are found in laboratory studies of insects which spend all or part of their life cycle within their food medium, and there is evidence that similar interactions occur under certain circumstances in natural environments. Such species seem, generally, to be limited by absolute amounts of food or nutrients and there is little evidence of spatial organization in their populations. As the level of population increases relative to the volume of food and space available, there tends to be a non-linear increase in mortality which may ultimately result in the complete extinction of all members of a population. During artificially induced crowding in larval populations of Drosophila (Sokoloff, 1955; Robertson, 1960; Mishima, 1964; Miller, 1964b), moderate increases in density relative to available food and space cause



an increase in the length of the larval period and a corresponding decrease in final body size. Because of the correlation that exists between adult body size and fecundity (Chiang and Hodson, 1950),these physiological adjustments produce a somewhat lower potential rate of increase for the populations, but they also allow the larvae to compensate for the effects of moderate crowding without a significant increase in mortality (Miller, 1964b). Figure 1 shows the changes in larval period +0.3 ,

BODY S I Z E -05











FIG.1. Deviation in larval period and adult body size, with larval crowding in Drosophiln 1964b).

melanogaster (After Miller,

and adult body size that occurred in populations of Drosophila melanogaster when the larval density was increased from 10 to 80 per 5 cc of medium (Miller, 1964b). Over this range of density the larval viability remained the same and there was only a slight increase in pupal mortality. With further increases in larval density these compensatory mechanisms lose their effectiveness and the flies reach what Sang (1949) refers to as “a minimum survival size”, after which further crowding is reflected in greater larval and pupal mortality (Fig. 5). In spite of the extremely high densities that were reached in these experiments, it can be shown that the limits on survival were due primarily t o food quality rather than food volume or space. The larvae were cultured in 8-dram vials containing 5 cc of a synthetic medium which acted as a substrate for a live yeast population. I n later studies it was found that a banana-agar medium enriched with dead yeast allowed much greater survival. The numbers of adults of D. melanogaster



and D. simulans produced from initial larval densities of 20, 40, 130, 120 and 160 per 5 cc of each medium are compared in Table I. Survival was nearly equal at densities of 20 and 40. At a density of 160 larvae per vial survival of D. melanogaster was reduced to 39.2% in Kalmus medium as compared with 72.1% in the banana-agar medium. Likewise, survival of D. simulans was only 18.1% at this density in the Kalmus medium compared with 70.6% in the banana-agar. Bakker (1961) has shown that competitive exploitation of food in D. melanogaster is controlled by subtle differences in (1) rate of feeding, (2) duration of the molting periods, (3) food requirements, (4)initial larval weight and (5) resistance to disoperative effects of crowding. There .is no evidence in these or other studies of larval competition in Drosophila of an allocation of space such that a relatively fixed number of larvae survive to pupate and produce adults. Given the constant physical environment of these experiments and the apparent absence of spaceregulating mechanisms, competition seems instead to be a process of selective elimination through direct exploitation of food resources. TABLEI Production of Adult Drosophila melanogaster and D. simulans in Different Food Media (Kalmus and Banana-agar) Adults Produced

D.n i e l a i t o p q t e r

Kalmii 8

Initial Larval Density


20 40 80 120 160

17.0 33.6 58.8 74.6 62-7


* S.E.) (Mean

i. 0.29

f 0.58 f 2.28 f 6.46 f 6.68

& S.E.)

16.2 Jr 0.34 32.2 f 0.54 66.0 f 0.69 95.9 + 1.14 115.4 3.74





(Mean f S.E.) (Mean & S.E.) 15.9 33.6 54.4 69.0 29.0

& 0.47 f 0.79 2.16 f 3-89 f 2.52

16.5 f 0.32 32.3 f 0.42 70.6 0.62 89.2 j= 1-56 112.0 f 2.37


Studies of competition in daphnids also show competition through a process of selective elimination based on differential rates of exploitation among different species or age classes within the populations. Slobodkin (1954) has shown that growth, reproduction and the size of Daphnia obtusa populations is linearly dependent on food supply, and suggested that the same was true in Pratt’s (1944) studies of Daphnia obtusa. Frank ( 1952) concluded from his experiments with competition in Daphnia pulicaria and Simocephalus vetulus that absolute food shortage



and chemical conditioning by nonvolatile substances were not limiting factors in these populations, and suggested that lack of oxygen or repeated physical contact might depress the feeding rate, or that food quality might be a limiting factor (P. W. Frank, personal communication), but there was no evidence of interference with access to a resource. Ullyett (1950) found no evidence of aggressive activity in competition among larvae of the sheep blowflies Lucilia sericata and Chrysomyia chlorophyga. They feed on the same food and compete only through exploitation of the same resource. Ullyett regards competition in these species in terms of “the provisions of nature rather than in the light of a specific ‘struggle’. The phenomenon which really acts as the controlling agent is basically sheer starvation.” It is in this sense that Nicholson used the term “scramble” to distinguish between this and a form of competition in which “each successful individual lays claim to a supply of requisites sufficient to maintain it and to enable it to produce offspring. The unsuccessful individuals are denied access to the critical requisites by their successful competitors” (Nicholson, 1955). In each of these examples the populations were directly limited by absolute amounts of food, food quality, or by a factor which affected their abilities to exploit their food resources. There is the suggestion, in other words, that food becomes limiting unless (1) some form of interference acts as a regulating mechanism relative to the space that contains a supply of resources, or (2) extrinsic factors such as climate prevent the level of population from reaching a critical value. The crowded populations and variations in body size that are commonly observed in natural populations of some dung, carrion and log-dwelling insects may reflect events similar to those observed in laboratory studies, in which exploitation is the principal form of competition. But there remains the question of how often natural populations are in fact directly limited by a resource such as food, or how frequently they are forced into a competitive interaction dominated by exploitation.

111. THEECOLOGICAL NICHE A. HISTORICAL DEVELOPMENT O F THE NICHE CONCEPT Elton and Miller (1954) have noted that the “ecological niche” is a far more elusive concept than was supposed by either Elton or Grinnell when they introduced the term into the literature. Elton (1927) observed that although there may be a difference in species composition between one community and another, there is usually an obvious similarity in their “ground plans”. We can expect to find groups of herbivores, predators, parasites and scavengers in every animal community, even though these roles may be filled by different species.



Elton proposed the term “niche” to describe this status of an animal in its community and spoke of the niche in the sense of an animal’s “profession” rather than its “address”. A reconnaissance of woodland bird communities in England and Wales (Elton, 1935) showed, for example, that “woodland birds in Britain form a clear-cut and ancient community that is much the same in different woodland types. . . It forms part of a life-zone widely spread over Europe.” Also, in a table of comparisons between comparable plant communities in California, Finland and Britain, Elton (1935)observed that “there is a high degree of resemblance in the structure of such bird communities right around the world.” Elton’s use of the niche concept was therefore primarily one of function, emphasizing that the same niches exist in each community, regardless of what species fill these roles. Udvardy (1859) attempts to establish that Elton and Grinnell used the term “niche” in the same way and that their concepts are therefore compatible, but it is obvious that Grinnell’s emphasis was on the “address” of a species, rather than its functional role. Grinnell (1917a) listed three facts of distribution: (1)that each animal has a habitat or range which is distinctive enough to be included among the characters used to describe a species, ( 2 )that some species range widely and others are restricted, and (3) that groups of species nearly or entirely coincide in range so that realms, zones, faunas, and other units can be easily recognized. These and similar observations laid the basis for his later development of the niche as “the ultimate distributional unit” (Grinnell, 1924, 1928).He also stated that it is axiomatic that no two species in a single fauna can occupy precisely the same niche (Grinnell, 1917b, 1924). Dice (1952) follows Grinnell’s usage to define the niche as all of the features of the ecosystem that a species utilizes and states that “The term does not include, except indirectly, any consideration of the functions that a species serves in the community.” He also emphasizes the distinctness of the niche of each species. It is in this and Gause’s sense that Hutchinson (1944) described the ecological niche as “the sum of all the environmental factors acting on an organism; the niche thus defined is a region of n-dimensional hyper-space, comparable to the phase-space of statistical mechanics.” A difficulty with the Grinnellian concept is that it cannot be used to define the conditions of competition without considerable modification. The broader concept, in which the niche is the functional status of an animal in the trophic system, clearly implies that this function may be filled by any of several species and that competition is not only possible within this context, but is also highly probable. This concept also provides for the possibility of “vacant niches”, in that a functional role



may exist to be occupied by subsequently invading species (cf. Darwin, 1859; Elton, 1958). It is in this connection that Weatherly (1963) defines a niche as the nutritional role of an animal, existing to be occupied in widely separated ecosystems. However, while the broad functional concept of the niche allows us to describe the composition of a community in such a way that particular roles can be occupied by different species, it does not at the same time allow us to define degrees of overlap in the ecological requirements of species competing at the same trophic level. Obviously the value of the niche concept as a theoretical tool is as a means of describing environmental relationships in such a way that the conditions of competition or coexistence can be defined and evaluated. As Crombie (l947), Hutchinson (1957) and others have pointed out, the degree of similarity that leads to competition or, conversely, the ecological differentiation that is necessary for coexistence, can only be defined empirically. Thus we require a conceptual model for this purpose which will accommodate both the functional and distributional concepts of the niche (cf. Lack, 1944). B.


Elton and Miller (1954) recognized certain of the difficulties inherent in using the niche concept to define competitive interactions when they pointed out, “There is no difficulty in seeing the reality of a very broad distinction of ecological function, such as between herbivore and carnivore. The distinctions get a bit blurred if we divide up carnivores into separate niches, since a carnivore may in fact ‘belong’ to more than one consumer layer. At the other end of the scale, each species has a unique ecological niche in the sense that its particular mosaic of abilities and habits is unique. I n between the broad consumer type and the species,. one may create any number of ‘niches’, by choosing some well-marked type of habit: ‘mouse-eater’, ‘conifer-needle feeder’, ‘bark.-beetle’ and so on. What ecologists usually imply by such groupings is that within them interspecific competition occurs or may occur. But when one considers an actual example like the grain-beetle Rhizoperth dominica competing with the grain-moth Sitotroga cerealella (in laboratory culture), it is seen that the similarities of habit (larvae feeding inside wheat grains for example) that bring about competition can be matched by other differences in properties of the species (they will have different parasites; they have different life spans; they presumably have Werent tolerance. . .) We think, therefore, that analysis of communities should pay attention more to tracing the consequences af one species in a key position being replaced by another, rather than trying to classify all the functions of species into a few niches.”



As long as the phenomenon of competition is described only by its outcome and in the abstract language of the Lotka-Volterra equations, we are liable to overlook the important fact that any interaction between two individuals or populations is an event which has dimensions of both space and time. Neither the broad, functional classification nor the total ecological characterization into mutually exclusive niches is capable, without modification, of expressing these properties of competition. A partial solution was offered by Elton and Miller (1954)when they proposed the term “arena” to describe a locus “within which some temporary relation between the species of animals is arrived at, by a combination of mutual opposition and various degrees of symbiosis”. This was an attempt to define species interactions in such a way that the total ecological characterizations of species express potentials which may overlap in time and space, leading under appropriate conditions to interspecies competition. A more satisfactory, formal description of the relationships between the ecological niches of potential competitors has been developed by Hutchinson (1957)using conventional notations of set theory. This approach allows us to define a niche in terms of all variables relative to the species, regardless of their exact nature or quality, and we may at the same time follow the usual set-theoretic practice of representing an infinite number of variables in a restricted number of dimensions. Hutchinson’s formalization also includes both the fundamental niche that might be expressed in the absence of competitive interactions and the realized niche that is occupied when competition restricts the expression of the total species potential. If we consider the total array of variables which limit the survival of a species S,, we may define an n-dimensional hypervolume N, which is the fundamental niche of that species. The fundamental niche, as so defined, may be regarded as a set of points in an abstract N space which will completely define the ecological properties of S,. Secondly, if B is a limited volume of physical space comprising the biotope of a collection of species S,, S,, . . . S,, the biotope is complete relative to S, if all the points in N, are represented in B. If N, and N, are two fundamental niches they may have no points in common and are therefore separate, or they may have points in common and are said to intersect. N,.N, is the subset of points common to N, and N, and is their intersection subset. The following restrictions apply to this mode of expression: (1) It is assumed that all points in each funda.menta1 niche imply equal probability of survival, and all points outside the niche imply zero survival of the relevant species. (2) It is also assumed that all environmental variables can be linearly



ordered, even though this is obviously not possible with our present state of knowledge. (3) The model refers to a single instant in time. (4) Only a few species can be considered at once, even though all other species in the community are regarded as part of the coordinate system. As Hutchinson (1957) points out and earlier discussions have also shown, some of the confusion surrounding the “Volterra-Gause Principle” has arisen from the concept of two species not being able to co-occur when they occupy identical niches. This problem was dealt with in detail by Gilbert et al. (1952). According to the set-theoretic formulation of niches, identity of fundamental niche would imply N, = N,, that is every point of N, is a member of N, and vice versa. It is axiomatic that this is impossible (Hardin, 1960; Cole, 1960). Omitting this quasi-tautological case, Hutchinson ( 1957) distinguished two cases of intersection between fundamental niches: (1) N, is a proper subset of N, (N, is “inside” N, and is therefore a smaller niche) and (2) N,-N, is a proper subset of both N, and N,. These relationships are illustrated by Euler diagrams in Fig. 2 for two independent variables



FIQ.2. Relationships between fundamental niohes (N, and N,) defined by two variables (z and y) in a biotope ( B ) .

x and y which can be measured along ordinary rectangular coordinates. In the first case N, coincides with the intersection subset N,.N, and is included within N,, while in the second case the intersection subset N,.N, is formed by the “overlap”. between the two niches.



The case of an “included niche” is particularly interesting, as it imposes strict conditions for the outcome of competition and, especially, for the survival of S,. I n the simplified, two-dimensional relationship shown in Fig. 2 (Case l), the niche N, is described by the coordinates x, - x3 and y, - y3 and is included within the total niche N,, which is in turn described by the coordinates x, - xpand y, - ya. The following outcomes of competition are possible under the conditions of Case 1: ( a ) competition proceeds in favor of S , in all the elements of B corresponding to N,.N, and, given adequate time, only S , survives; or ( b ) S , survives in all elements of B corresponding to some part of the intersection subset and both species survive. The first alternative (a) implies that S , is the superior competitor and, with its greater ecological amplitude, will eliminate S , from all parts of the fundamental niche N,. Coexistence of S, and S, is impossible under the conditions of an included niche when S , is the superior competitor. The second alternative ( b ) requires that S , be the superior competitor, in spite of its narrower ecological tolerance. Under these conditions, S, will be excluded from N, but will survive in the parts of N, that lie outside the intersection subset. The realized niche of S, will therefore be the difference subset N, - N, (N, - N, N,). As long as the niche space N, - N, is sufficient to allow the survival of S, there will be coexistence of both species within the total biotope B . The conditions described by alternative ( b ) of an included niche would seem, at first glance, to contradict conventional theories of competition. We do not ordinarily expect a species with a relatively narrow range of tolerance to be superior in competition with a closely related species with greater ecological amplitude, yet later examples will show that the approximate conditions of an included niche are relatively common in species relationships, and that Case l ( b ) may in fact explain some of the species diversity that exists in natural communities.


IV. COMPETITION IN NATURE It is difficult to identify the actual process of competition in nature, either as an active or historical factor, although certain patterns of geographical or habitat distribution and niche relationships between species are often thought to reflect competitive exclusion. When a species occupies a broader niche in the absence of a closely rela.ted species than it does when the two species are sympatric, competitive exclusion is often inferred (Lack, 1944); contiguous allopatry can only be explained by a consistent environmental discontinuity or by a competitive interaction which maintains the pattern along a line of contact (Miller, 1964a); character displacement is usually explained as a product



of competition (Brown and Wilson, 1956; Hutchinson, 1959;Mayr, 1963); the species composition of natural communities may reveal fewer repre sentatives of a genus than would otherwise be expected in the absence of competition (Elton, 1946); and MacArthur’s (1957) model of nonoverlapping niches has been advanced as a test for competitive exclusion (Hutchinson, 1957; Slobodkin, 1961). I n each case we are left, however, with the question of whether the observed phenomena are in fact due to competitive exclusion, or whether they cannot be explained instead by ecological differences which have arisen through some other agent of natural selection (Elton and Miller, 1954).

A. DISTRIBUTION PATTERNS Figure 3 shows the relationships that may exist in the spatial distributions and fundamental niches of two populations. For simplicity SPATIAL DISTRIBUTION



A llopat ry

B N1


C N1

Contiguous Allopatry

FIU.3. Distribution patterns of two species (S, and S,) and their possible niche relationships (N, and N2).

the niches are represented by single variables. They illustrate (A) non-intersection; (B) intersection corresponding t o Case 2, Fig. 2; (C) intersection corresponding to Case 1, Fig. 2; and (D) contiguous nonintersection. Even though it is unlikely that the critical variables in B



two niches would be precisely contiguous and non-intersecting (D), this is a theoretical possibility which must at least be considered as a possible cause of observed cases of contiguous allopatry in spatial distributions. The terms allopatry, sympatry and contiguous allopatry are conventionally used to describe geographic distributions, but analogous patterns occur at all levels of distribution and it is not always possible to distinguish between ecological events which affect geographic, habitat or microhabitat distributions (Miller, 1964a). The related questions of spatial distribution and coexistence are relative terms, depending on the level of spatial distribution chosen as a criterion and the sensitivity of sampling procedures in relation to this criterion. For this reason, it seems advisable to consider the general phenomenon of spatial distribution as a single event which can be qualified according to specific circumstances. When it is stated, according to the Gause axiom, that no two species can coexist indefinitely in the same locality, we beg the question of what is meant by “locality”. Populations which appear to be sympatric in their geographic distributions may diverge ecologically so that they are effectively allopatric in their local distributions, even in the absence of a competitive interaction. We know that closely related species frequently coexist in the same habitats, but there is usually some level of spatial distribution at which they are separate. What is required is a critical comparison between the spatial distributions of two supposed competitors and the amount of intersection in their fundamental niches. Unless it can be shown that their spatial distributions, at any level, reflect realized niches which are less than what would be expected from an unrestricted intersection of their fundamental niches, there is no reason to infer that competition is a contributing factor. Although there is no opportunity for competition to occur between allopatric populations, this situation illustrates the difficulty of drawing conclusions about niche relationships from observed distribution patterns. Allopatry may be due to an intervening barrier to dispersal or habitat occupation, in spite of intersection between two fundamental niches (Fig. 3 B); or the allopatric distribution may reflect a difference in fundamental niches (Fig. 3 A) such that N, and N, are separate sets, or at least do not intersect in the niche elements that affect spatial distribution. The Grst of these relationships is shown in the distribution of two races of song sparrow (Passerella melalia) in the San Francisco Bay area. One race occupies tidal marshes and the other is confined to nearby fresh-water habitats (Marshall, 1948). The intervention of unsuitable environments forces them to behave like allopatric species, but there is hybridization between them wherever there is contact between






the two habitats. The fundamental niches of the two populations intersect but their realized niches are expressed in allopatric distributions because of ecological barriers. In the absence of barriers to dispersal and habitat occupancy, intersection in the spatial elements of two fundamental niches will be expressed in various dimensions of sympatry, depending on the nature of the competitive interaction that results from intersection and the spatial requirements of the species concerned. There is considerable evidence showing that competitive exclusion at the level of broad habitat or geographic distributions is invariably accompanied by competitive interference. Species which compete primarily through exploitation are more often controlled by physical factlors which affect their distribution and abundance and are less likely to be spatially exclusive, except at the level of mxrohabitat distribution. They are also more likely to show complete differentiation in some critical element of their fundamental niches. Townes and Townes (1960) remarked on the apparent coexistence of three species of ichneumon wasp (Negarhyssa atrata, M . macrurus and M . greenei), all of which parasitize the same host, the wood-boring larva of the pigeon tremex (Tremex columba). A female Megarhyssa detects a host larva or pupa in a dead log or stump and inserts her ovipositor full length into the wood to deposit an egg on the host. The ovipositor is directed a t a right angle to the surface of the wood and all ovipositions require the complete insertion of the ovipositor (Heatwole and Davis, 1965). Tremex larvae occur a t various depths in the wood, but once established an individual larva tends to remain a t a constant depth. Thus the larva to be parasitized must be at a specified depth corresponding to the length of the extended ovipositor of the female wasp. Figure 4 shows a comparison of the ovipositor lengths of these three species of Megarhyssa. There is no overlap in the range of ovipositor lengths for M . atruta and M . macrurus and only a slight overlap between M . macrurus and M . greenei. There is a significant statistical difference in their means and standard deviations (Heatwole and Davis, 1965) with respect to this structural feature, and it is evident that they do not compete for the same larvae but parasitize different segments of the total host population. I n spite of their apparent sympatry, even to the extent of the adults using the same foods, resting places and oviposition sites and the larvae using the same resource in the same habitat, there is no intersection with respect to the critical factor of ovipositor length. The relationship between the fundamental niches of pairs of these three species is described by Fig. 3 A or D. It is possible that competition w-as a historical factor in the ecological differentiation that now exists anioiig these three species, but its action



is now negligible if it occurs at all. If the tremex larvae are uniformly or randomly distributed at different depths in the logs and do not alter their distributions as a result of the relative effects of the three species of parasite upon their numbers, each species of Megurhyssa will continue to utilize a separate portion of the total resource regardless of the activities or effects of the other two species, and changes in the abundance of tremex larvae at different depths will not lead to inter-specific interactions between the wasps. Thus, three apparently sympatric species which use the same habitat in the same ways are in fact allopatric at the level of their larval distributions and therefore do not compete.










FIG.4. Ovipositor lengths of Megarhyssa atrata, M . macrurus and M . greenei. Horizontal line, mean length; vertical line, range; rectangle, fS.D. (After Heat,wole and Davis, 19661.

Contiguous allopatry is a particularly interesting phenomenon which has received very little attention from ecologists, in spite of the fact that it seems by its nature to indicate a strong element of competition. It is unlikely that the fundamental niches of two species would be entirely separate but contiguous (Fig. 3 D) over any appreciable area. Nevertheless, such a'distribution pattern could be produced by a sharp environmental discontinuity, and there is nothing in the pattern itself which immediately reveals whether it is controlled by environmental factors or competitive exclusion. The geographic and habitat distributions of the four species of pocket gopher (Beomyidae)that occur in Colorado show a pattern of contiguous



allopatry between the ranges of adjacent species (Miller, 1964a). There are a t least three possible explanations for this rather striking distribution pattern. Each species could be so highly adapted t o some critical factor such as soil type that the relationship between their fundamental niches is that shown in Fig. 3 D; their fundamental niches may intersect but some intervening barrier, such as a discontinuity between soil types or a physical barrier such as a river, might prevent the intersection from being expressed in their realized niches; or, in the absence of effective geographical or habitat barriers, the intersection in their fundamental niches might be controlled by competition so that only one of any species pair can occupy the niche space corresponding to the intersection subset. This example will be discussed in R later section when the evidence for competition will be examined in more detail, but suffice to say it can be shown that there is intersection in the fundamental niches of all four species. I n fact, the relationship between their fundamental niches is essentially a system of included niches as illustrated in Fig. 3 C , and competitive exclusion is the most reasonable explanation of the distribution patterns shown by these species. A similar relationship is illustrated in Fig. 5, which shows the spatial distributions of two species of triclnd when they occur separately and together in highland brooks and streams (Beauchamp and UllyBtt, 1932). The distributions of fresh-water triclads are governed primarily by the temperature and rate of flow of springs and streams. Both of these species were found to exist alone in several springs where the water temperature was 6.5 to 8.5"C, but Planaria montenegrina can only tolerate temperatures up t o 16 or 17°C while P. gonocephala inhabits waters as warm as 23°C. Their fundamental niches therefore intersect (Fig. 3 C) in the range ofnpproximately 6.5 t o 1 G or 17°C with respect to this critical factor. \4heu the two species are sympatric in the sense that they both occur in the same stream, P. gonocephala is excluded from the spring head t o a point where the temperature reaches 13 to 14"C, below which P. nronteticgrina is absent and P . gonocephala is the only species present. It is interesting to note that competitive exclusion does not occur a t the extreme of the fundamental niche of P . montenegrina, but that there is a shift of 3 or 4 degrees to the exclusion point of 13 to 14°C. Andrewartha and Birch (1954) contend that this is not necessarily an example of Competitive exclusion, and that other selection processes could have produced the same result. They offer no alternative t o what seems a clear case of exclusive realized niches between two species whose fundamental niches intersect; the most reasonable explanation of this phenomenon is that competitive exclusion operates predominaiitly in favor of 1'. rnontenegrina, the more specialized species.



B. C O M M U N I T Y C O M P O S I T I O N Elton (1946) analysed published ecological surveys of 55 animal communities and 27 plant communities and concluded that the number of species present in any given community is less than would be expected from chance. He found a constantly high percentage of genera represented by only one species (86% for animal and 84% for plant communities) and average numbers of species per genus of 1.38 in STREAM FLOW




I 6.5-8.5


D! c



-I 23’









11 14”




FIQ. 5. Distributions of Planaria montenegrina and P . gonocepknla when they occur separately (allopatric) and together (sympatrir) in streams (Data froin Beauchamp and Ullyett. 1932).

animal communities and 1.22 in plant communities. These figures were found to differ considerably from those of faunal lists for large regions. For example, only 50% of the genera in 11 large groups of British insects contain one species and the average number of species per genus for all British insects is 4.23. Elton (1946) attributed the difference in species/genus frequencies between ecological surveys of relatively small habitats and those from faunal lists for large regions to “existing or historical effects of competition between species of the same genus,



resulting in a strong tendency for the species of any genus to be distributed as ecckypes in different habitats, or if not, t o be unable to coexist permanently on the same area of the same habitat.” Williams (1947) argues that it is insufficient to show that the species per genus is smaller in smaller communities than in larger, as this is a mathematical result of taking a smaller sample from a larger group. It is necessary to show that the species/genus frequency in the smaller communities is different from expectation in a randomized sample of the same number of species, selected without reference to generic relationships from the smaller or larger fauna. Using an index of diversity Williams (1947) conc~ludcdthat Hlton’s data could have been obtained from a random sample and that the cvidcnce in fact indicates natural selection in favor of more rather than fewer species in the same genus in small communities. Williams ( 1951) reiterated his earlier conclusions after analysing Moreau’s (1948) records of East African bird communities. He showed an excess of‘congeneric groups with two or more species above what would be expected by selection without reference t o generic relationships, and further that the excess seems to increase as the number of congeneric species increases. Williams ( 1951) concluded from this analysis that (1) competition between closely related species is probably on the average greater than between less closely related species, and (2) that closely related species are probably more suited t o similar physical environments and, therefore, similar extra-generic competition. Bagenal (1951) examined the evidence presented by Elton and Williams and concluded that many of the communities analysed by Elton (1946) were unsuitable Serause they were often very heterogeneous, were sampled by a variety of methods and the community was often sampled over a long period of time. Hutchinson’s (1957) distinction between homogeneously dicerse and heterogeneously diverse environments implies that we cannot expect species-abundance distributions to behave in the same way under both conditions, a fact which is relevant t o community surveys. As this classification is related to the free paths of animals in relation to the mosaic of structural and microenvironmental features of a habitat, it follows that comparisons between species with different powers of dispersal and daily activity will also be suspect. Elton and Williams gave different meanings to “habitat” and the term is used especially broadly by Williams. Bageiial (1951) states “From this standpoint &Iton’s conclusions appear t o be correct, though based on unsuitable data, and Williams, far from contradicting Gause and Elton, has provided the corollary that related species are more likely t o occur in similar, though not identical, habitats than are unrelated ones.”



What these analyses do not, and cannot, show is any actual relationship between fundamental and realized niches, and therefore empirical evidence of either competitive exclusion or coexistence at the level of species interactions. I n order to verify or deny the possibility that competition affects community composition, we must first know how many species are geographically and ecologically available to the community and whether there is intersection in their fundamental niches. As Elton and Miller (1954) have noted, there is very little empirical evidence that gives satisfactory proof of interspecific competition in natural communities, “most of that adduced being equally explicable by selective processes acting through other density-dependent population pressures.” This distinction was not clearly realized by Elton (1946) in his analysis. We have not, therefore, answered the question of whether competition is a daily occurrence in the lives of many animals, or whether it has occurred in their evolutionary histories, even though laboratory evidence and some competition theory “is so strong and persuasive that it must be taken into account in theories about community ecology” (Elton and Miller, 1954).


When the MacArthur (1957) model of species-abundance is used to demonstrate competitive exclusion, it suffers from the same inadequacies as other inferences predicated on the a priori assumption that mutually exclusive distributions are invariably the product of competition. Of the three alternative situations considered in this model, the one based on non-overlapping niches most nearly fits the conditions of competitive exclusion. This case assumes that the Gause hypothesis is universally valid and that all species in the community are mutually exclusive, i.e. that each species in the community “utilizes the environment in some way so as to make it completely unavailable to the other species in the community” (Slobodkin, 1961). If we consider the total community environment to be a line of finite length, we may divide it into n random parts by throwing (n - 1) random points onto the line. The abundance of each species in the community will have the same distribution as the length of parts of the line. Under these conditions the expected distribution of the rth rarest species is given by:

in which there are S species and N individuals. There is a limitation on this theory such that the ratio of total numbers of individuals to



total number of species must remain constant. Hutchinson (1957) notes that this is likely in a homogeneously diverse environment, in which the mosaic of structural features (e.g. logs, stones, bushes, etc.) is small compared with the free paths of the organisms concerned, but is less likely in a heterogeneously diverse environment where, for instance, stands of woodland may be separated by areas of grassland. Two examples of a correspondence between observed species-abundance distributions and the predictions of this case of the model have been cited as evidence of competitive exclusion. MacArthur (1957,1958) obtained a reasonable fit t o the model when he plotted data for birds in the Quaker Run Valley of New York from component habitats (pasture, orchards, mature oak-hickory forest) which were homogeneously diverse; and Kohn (1959) found a remarkably good correspondence for some of his samples of snails (Conus)in Hawaii. However, these studies would appear to involve two quite different kinds of interaction. The birds studied by MacArthur are mostly territorial, they are relatively uninfluenced by minor variations in environmental factors, particularly microenvironment, and interfere with each other sufficiently to suggest that competitive exclusion is a t least a distinct possibility. Kohn (1959), however, found that the adult ecological niches of each species of Conus differ significantly with respect to a t least two of the following characteristics: nature of the food, nature of and relation to the substratum, and zonation of the marine environment. Kohn (1959) stated, “These differences are concluded to be the primary factors by which the ecological niches of species of Conus are differentiated. This is the mechanism which enables the maintenance of populations of large numbers of closely related, sympatric species of Conus in tropical regions.” I n other words, the fundamental niches of birds are large and, for the most part, independent of close environmental control, and the results obtained by MacArthur (1957) may reflect a series of nonoverlapping realized niches determined by competitive exclusion. But there is no evidence of territoriality or interference among different species of Conus (A. J. Kohn, personal communication). They are so sensitive to variations in the microenvironment that their fit to the MacArthur model probably reflects a condition of specialization with respect to food and substrate and mutually exclusive distributions that are controlled by environmental discontinuities and not competition. Correspondence with the predictions of the MacArthur model of nonoverlapping niches can be produced by (1) separate realized niches due to competitive exclusion of one species from the intersection subset of two intersecting fundamental niches (Fig. 3 B or C), (2) separate realized niches due to environmental discontinuities that are greater than the amount of intersection between fundamental niches, or (3) B*



non-intersecting fundamental niches (Fig. 3 A). Only the first of these alternatives requires the action of interspecies competition.



Brown and Wilson (1956) proposed the term “character displacement” to describe a situation in which, when two species become sympatric, the differences between them are accentuated in the zone of overlap and are weakened or lost entirely in the parts of the range outside this zone. Any of a variety of morphological, ethological, ecological or physiological characters may diverge in this manner and they are assumed to have a genetic basis. While the most obvious function of character displacement may be to reinforce existing isolating mechanisms, it is usually thought to reflect a competitive interaction which has led also to niche differentiation (Brown and Wilson, 1956; Kohn and Orians, 1962; Mayr, 1963). Once a secondary contact between two cognate populations has been established, the species may interact in two ways to augment their initial divergence: (1) if inbreeding occurs, hybrid sterility or non-viability will lead to the reinforcement of reproductive barriers, and natural selection will favor a reduction in “gamete wastage” and further ethological or genetic divergence, and ( 2 )ecological displacement and a reduction in competition will also be favored by natural selection if the ecological characteristics have a genetic basis (Brown and Wilson, 1956). According to this theory interspecies competition leads initially to ecological divergence in the physical space of the intersection between the two fundamental niches. Ecological divergence is accompanied by character divergence, reduced competition, and reinforcement of ethological and ecological isolation. This theory describes a positive feedback mechanism which will continue to increase diversity, as long as it has selective value and hybridization does not intervene at an early stage disrupting the process. The adaptive differences between two sympatric species of nuthatches (Sitta) have been cited as a particularly clear example of character displacement (Brown and Wilson, 1956; Mayr, 1963). Xitta neumayer and X. tephronota largely replace one another in eastern and western Eurasia but overlap broadly in Iran. Vaurie (1951) showed that the two species are nearly identical in areas outside the zone of overlap, and in fact can only be distinguished by an experienced taxonomist; but where the two species occur in more or less equal numbers in the zone of sympatry in Iran, S. neumayer shows a marked reduction in bill length and overall bill size, and in the width, size and distinctness of the facial stripe. Conversely, S. tephronota shows a positive augmentation of all these characters in specimens from the zone of sympatry.



Vaurie (1951) surmised that differences in bill size and shapr are correlated with different food habits and constitute a basis upon which the two species can avoid competition where they are sympat,ric, although he presented no evidence that would substantiate this conclusion. Unfortunate]y character divergence has received very little serious attention as an ccological phenomenon and there is only indirect evidence to support the view that it is related t o competition, attractive as this theory may be. Mayr (1963) has correctly emphasized the need for more quantitative analysis in the evaluation of character divergence. Sonie of the examplcs cited by Brown and Wilson (l95G) may illustrate hybridization rather than character displacement. There are cases in which species differences arc reduced in a region of sympatry, and there are also cases where the ovcrlap seems to have no effect on the phenotypes of the sympatric populations (e.g. sibling species). As Mayr (1963) points out, “Only a statistical analysis can bring out the facts needed for valid generalizations.”

E. I N V A S I O N S Elton (1958) has documented examples of changes in distribution resulting from the invasions of species into the geographic ranges of their cogeners, or of species which in some way require similar resources and become displaced in the resulting interaction. Thus the persistent spread of the ubiquitous European starling (Sturnus vulgaris) in North America has occurred at tlie expense of species such as the bluebird (Sialia sialis) and the flicker (Colaytes nztmtus) which have been forced to compete for a limited supply of the nest holes which they require for breeding. An overwhelming amount of evidence exists to show that when plants or animals invade a new area, they seldom do so without disrupting the ecological balance that existed before they arrived (Elton, 1958). The effect of competition in nature is perhaps best demonstrated by such invasions. Island faunas are especially vulnerable to invasion by new competitors (Mayr, 1963) and there are many well-documented examples. Greenwap (1958) has shown that most of the species of birds that have become extinct during the past 200 years have been island birds. However, as Mayr (1963) points gut, sweeping generalizations that attempt to predict the outcome of invasions from one fauna to another are not entirely reliable. There are exceptions, both in the fact that species native t o the larger more diversified areas are not necessarily superior competitors, and t o the assumption that such invasions inevitably lead to competitive exclusion. Moreover, there is indirect



evidence that suggests that many of the birds that became extinct in New Zealand and the Hawaiian Islands were the victims of diseases introduced by the invaders (Mayr, 1963). Again, it is necessary that successful invasions and the extinction of members of the native fauna be carefully documented with supporting evidence of niche relationships before they are classed as evidence of competition. I n summary we may assign the Werent types of evidence for competition in nature to four principal categories, all of which contain elements of spatial relationships: 1. Mutually exclusive spatial distributions without supporting evi-

dence of a competitive interaction. 2. Mutually exclusive spatial distributions with supporting evidence of a competitive interaction. 3. Observed or inferred ecological displacement (usually correlated with character divergence) in sympatric populations. 4. Induced changes in distribution pattern. Udvardy (1952) and Andrewartha and Birch (1954) are critical of the use of competition to explain any aspect of variation and distribution of animals. After a review of examples from Lack’s (1944) article on species formation in passerine birds, Andrewartha and Birch (1954) concede that closely related birds either seem to live in different places or use different foods, but they note that “this is true, only more so, of distantly related species; but no one seriously suggests ‘competition’ as a cause for this.” They ask “Why then should it be necessary t o invoke competition to explain the same phenomenon among closely related species, especially when there is no empirical evidence for it?” Brown and Wilson (1956) agree that the evidence for competition in nature is scanty indeed, but they also suggest that Andrewartha and Birch (1954) have failed to appreciate the amount of evidence that does exist. The tendency for closely related species to inhabit different areas or to exploit different niches may in many cases originate from causes quite different from competition (Elton and Miller, 1954), and careful analysis of the critical variables in fundamental and realized niches is an obvious requirement in evaluating the role of competition in species interactions, as the preceding examples have shown. Statistical tests such as those for specieslgenus frequency, index of diversity, or species abundance distributions provide estimates which are more quantitative but are not necessarily more informative than direct observation, and they cannot be used to either prove or disprove the existence of competition in nature. Mutually exclusive distributions do not necessarily depend on competition, even when this seems to be the most immediate explanation. It is unfortunate, in this connection, that Mayr (1963)



has chosen to use the term “exclusion” as though it were synonymous with competitive exclusion. It is obviously necessary to distinguish between observed patterns of mutual exclusion in spatial distributions and the cause of the distribution, which may or may not be competition. The correct alternative is not usually evident in the distribution pattern alone, but requires supporting evidence which will reveal the relationship between the fundamental niches of potential competitors. Nevertheless, there is also a growing body of evidence that establishes competition as an active force in species relationships in nature. When ecological displacement occurs in the realized niches of sympatric species that occupy broader niches outside the zone of sympatry, competition is clearly implied. If these observations are supported by accurate descriptions of the intersection of their fundamental niches with respect to critical factors which control their survival and habitat occupancy, a stronger case for competition is established. This is especially true when there is a strong element of interference in the interaction between the two species, as later discussions will show. It would be more convincing to be able to show, with appropriate controls, that the experimental addition or removal of a species affects the realized niche distribution of another. This has seldom been attempted, in spite of the potential value of such experiments.


It was stated earlier that the outcome of competition according to equations (1) and (2) depends on the inequalities a > K , / K , and fl > K,/K,. By reversing the aigns of the inequalities one by one it can be shown that there are four possible outcomes of competition in this model. Gause and Witt (1935) analysed the properties of this equation system to consider the essential types of competition that might exist between species. Two of the cases included in their analysis are of particular interest: Case 1 : a > K J K , and j? Case 2 : a < K , / K , and j?

> K2/Kl < K,/K,


(5) These cases are illustrated diagramatically in Fig. 6. In the diagrams, neither species can increase above its saturation line K,/a, K , or K , , RI/fland below its saturation line each species will tend to increase. I n the first case either N , or N , will be the sole survivor, depending on the initial concentrations of each species. I n the second case competition will lead to a stable, mixed-species population, regardless of the initial concentration of each species. An important feature of the second case is that there is homeostatic regulation of the composition of the stable



two-species population; disturbance of the balance in numbers will. lead automatically to re-establishment of the stable combination. Lotka (1932) and Windsor (1934) also recognized the theoretical possibility of infinite survival of both species in a mixed population. Windsor put the conditions in the form: up < 1. This requires that a singular point exists which is a “knot” on the surface of N , N , . This is guaranteed by the conditions u < K , / K , and p < K , / K , . The stable equilibrium will inevitably arrive a t the “knot” represented by point E (Fig. 6, Case 2) regardless of the initial concentration of the two species.

Care 1

Core 2

FIQ.6. Casen 1 and 2 for the outcoine of competition in the TAotke-Volterramodel (After Gause and Witt, 1935).

Gause and Witt (1935) assumed that the conditions of Case 2 would apply only when there was slight mutual depression of species, which might occur when two species belong to different niches in the same microcosm. They cited the example of Gause’s (1935) experimental demonstration of coexistence of stable, mixed-species populations of protozoans. Paramecium caudatum and P. aurelia are effective consumers of bacterial components suspended in the upper surface layer of a medium, while P. bursaria feeds on yeast cells sedimenting on the bottom of the experimental microcosm. Combinations of P . caudntum and P. bursaria, or P. aurelia and P . bursaria, can coexist because of this niche differentiation. Hutchinson and Deevey (1949) dismiss the possibility of coexistence in this case because they feel, as Gause and Witt (1935) assumed, that i t “implies that the ecological niches of the two species do not overlap completely.” There is nothing in the logistic theory that requires or supports this assumption, although it is presumably a corollary of the



axiom of inequality. Kostitzin (1939), who also dismisses the case of coexistence, reaches the further conclusion that intraspecies competition should be less violent than competition between two allied species. This statement introduces a curious mystique whereby speciation involves not only the evolution of normal isolating mechanisms, but also the sudden emergence of a competitive antagonism. Interspecies competition is assigned an arbitrary force which is presumed not t o exist within species. Cole (1960) finds no basis for these pronouncements and notes that Darwin not only referred to the fact that competition between closely related species will be more severe than between more distant relations, but that “the struggle will almost invariably be most severe between the individuals of the same species.” Cole (1960) states, “If Darwin was right the Volterra-Lotka analyses predict not competitive exclusion but coexistence.” Let us accept the proposition that some degree of ecological differentiation is one condition that will permit coexistence, provided the differentiation is great enough. The species diversity of any self-sustaining ecosystem, no matter how small, is sufficient t o substantiate this conclusion. We generally assume that the different species in such an ecosystem have distinct adaptations and requirements and are therefore compatible to this extent. It is also evident, at the other end of the scale of species difference, that members of the same species are able to coexist, even tbough their individual requirements are nearly identical. As they will also exhibit approximately identical responses to factors affecting their distribution and abundance, competitive advantage should be least pronounced among members of the same species. Or, stated in another way, the individual interaction values that comprise u or /3 should be nearly equal. Following this line of reasoning further, the degree of similarity between two species should also be reflected in the species interaction values u and /3. It is quite conceivable, in fact, that less competitive advantage ( u - /3) might exist between two sibling species than between the individuals of a highly polytypic species. Competitive exclusion, in such a case, would depend on the extent to which critical resources were limiting (e.g. the size of the total two-species population that the environment could support) and the constancy of the environment. Miller (1964b) examined the case of competition between larvae of the sibling species Drosophila melanogaster and D . sirnulam. These species are sympatric and apparently coexist in similar habitats throughout most of the temperate and tropic regions of the world. An initial assumption in the experiments was that these sibling species, because of the genetic similarity that underlies “their unusual morphological similarity (Moore, 1952; Miller, 1964b), are not only likely to



have similar geographic distributions and occupy similar niches in nature, but are also more likely than most species to have competitive interactions which are comparable. We might logically expect that the more nearly the ecological requirements of two species are alike and the greater the amount of intersection in their fundamental niches, the higher the probability their species interaction values will be the same. The experiments consisted of placing early first instar larvae in vials containing 5 cc of a standard medium to which was added 1/20 cc of a 15% yeast solution (Miller, 1964b).The cultures were allowed to develop a t 25OC and the production of pupae and larvae was recorded. In singlespecies cultures the larvae responded to increased crowding by extending the larval period and by progressive reductions in the final adult body size without significant increases in mortality. As noted earlier, this compensatory mechanism seems to occur in species which live in their food medium and compete mainly through direct exploitation of their food resources. There is, however, a minimum survival size which marks the limit of compensation for the effects of crowding, and both species showed a pronounced increase in mortality between formation of the pupa and emergence of the adult when the initial larval densities were raised above 120 per 5 cc of medium (Fig. 7). As maximum numbers of adults were produced when the initial larval density was 120 in single-species populations, this was defined as the “saturation density”. A t densities up to 120 larvae there was no significant difference in IS0

0 0



D.mdonogartbr D.rimulans





5 0



0 I





FIQ.I. Mean numbers of adult Drosophila melanogaster and D . sirnulam produced at different initial larval densities (After Miller, 1964b).



the numbers of adults or pupae produced by the two species, although D. melanogaster had a higher production than D. simulans when the initial larval density was increased above this value. During interspecies competition between equal numbers of the two species at combined densities of 10, 20, 40, 80 and 120 larvae there was also no significant difference in the numbers of adults of each species produced. If we allow K , and K , to stand for the maximum production of adults of D. melanogaster and D . simuluns respectively at each experimental density up to saturation in single-species cultures, and consider N , and N , to be the number of adults of each from mixed populations, we can make the comparison shown in Table I1 and can test for the values of the interactions between them at each of the five density levels. TABLEI1 Numbers of Adult D. melanogaster (K, and N,) and D. simulans (K, and N,) in Single-Speciesand Two-SpeciesCultures (From Miller, 19643) Initial









10 20 40 80 120

7 -8 17-0 33.6 58.8 74.6

7.9 16.9 33.6 54.4 69.0

4.3 8 -7 15.8 28.2 32.3

4.1 7 *7 14.3 28.2 37-2

Analysis of variance for the values of K,, K , , N , and N , at the five levels of density permitted the following conclusions: (1) K , N K , : the conditions of the experimental environment had the same survival value for both species at the density levels tested. (2) N , NN,:the total resources of the environment were shared equally between the two species. (3) ( K , - N , ) N , : the effect of N , on N , was the same as the effect of N , on members of its own species and therefore a = 1. (4) ( K , - N , ) N , : the effect of N , on N , was the same as the effect of N , on members of its own species and therefore /3 = 1. (6) a /3: the effect of N , on N , was the same as the effect of N , on N,. To test for differences in the values of a and /3 the logarithms of ( K , - N , ) , ( K , - N , ) and the observed values for N , and N , were andysed for variance. The comparison log a log /3 given by: N



Clog ( K , - N , ) - 1%

N 2 1



[log (K2- N , )

- 1%

N 1 1




was seen to have an insignificant mean square, so that the values of a and fl may be considered equal. A considerable amount of ambiguity exists in interpretations of the results of interspecies competition. It is necessary to distinguish between two considerations: (1) maximum exploitation of the available resources by one species, or (2) equitable utilization of the resources and possible coexistence in a mixed species system. As most of the emphasis in competition experiments has been on the extent to which a population N , reduces the resources available to its competitor N , , there has been a tendency to ignore the far more interesting fact that the resources may be shared in such a way as to support two species. Terms such as “degree of interference” or “intensity of competition” may be completely misleading in this respect (cf. Merrell, 1951; Moore, 1952; Sokoloff, 1955). It is usually assumed that when the ecological requirements of two species are similar or nearly identical, the “intensity of competition” will be high and coexistence impossible; when their requirements are dissimilar they will compete less with each other and, therefore, may coexist. As noted earlier, this was assumed by Gause and Witt (1935) in their case for a stable equilibrium in mixed populations. The relationship between the sibling species Drosophila persimilis and D. pseudoobscura is remarkably like that between D. melanogaster and D. simulans. Their ecological requirements are almost identical and competitive superiority is only shown at extremely high densities. Sokoloff (1955) concluded that when these two species comPete with each other there is intense “interference”, even though adult survival is nearly the same for both species. When they compete with D. miranda, a more distantly related species, interference with the stronger competitor is not so great, even though the weaker competitor may be eliminated entirely. These results pose an apparent contradiction to the Gause principle, in that coexistence is usually more successful between closely related species than between more distant relatives. The expression ( K , - N , - aN,) merely implies that two species which compete with each other cannot each survive at the same density levels that are possible when only one species is present. Obviously N , cannot attain its maximum density K , when it shares the resources with N,. This refers to the exploitation of the total resources by two species, but does not exclude the possibility of coexistence, which depends on the relative values of cc and 8. Analysis of the results of the experiments with D. melanogaster and D. simulans showed that the values of K , and K , were not significantly different and we may therefore assume that K , = K , . It was also shown that a = 1 and fl = 1 and a and fl were not significantly different, and we may conclude that the competitive interactions between species were the same as those



within species and a = jl = 1. This means that K , = K,/a and K , = Rs/Band the isoclines in Fig. 6 (Case 1) coincide. We may now state

the general case for this relationship as follows: N , and N , will coexist at equilibrium densities which depend on initial concentration when : a = K,/K, and /3 = K , / K ,


These experiments were designed so that the initial concentrations of the larvae were always equal. The predictability of this model for different concentrations of the two species was not tested. However, equilibria occurred at points along the line 0s which corresponded to the density levels in the design of the experiments, and to this extent the data for N , = N , confirm the prediction of the general case for a two-species equilibrium. As this study only considered the case of competition during the larval stages of the Drosophila life cycle, there remained the possibility that factors affecting fecundity and fertility and the selection of oviposition sites will alter the course of competition by varying the initial concentrationsof the two species in this system. A series of experiments waa therefore designed to test competition between these two species for one complete life cycle (Miller, 1964~).Controls consisted of two pairs of adults of one species in each vial containing the same food medium used in the larval experiments; the experimental vials contained one pair of each species. The adult flies were allowed to lay eggs for 6 days and were then removed and the subsequent production of F, adults recorded. A total of 3 705 flies was produced from 111 competition replicates. The ratio of D. melanogaster to D. simulans F, adults was 51.1 to 48.9, which was not a significant difference. However, each replicate was a self-contained population in which the course of events had no direct relation to events in other replicates, and it was noted that there was marked dominance of one species or the other in individual populations. As competition was confined to a single generation and the total elimination of one species or the other was not a criterion of the experiments, there were two measurable categories of species dominance or competitive superiority: (1)the number of replicates in which a species waa the more successful (between-replicate dominance) and (2) the numerical dominance of one species over the other in individual replicates (within-replicate dominance). The lack of competitive advantage shown in the total numbers of each species produced was also reflected in the relative numbers of replicates in which each species was the more successful. D. melanogaster produced greater numbers in 51-4% of the 111 Competition replicates and D . simulans was dominant in 46.8%



with the numbers of each species equal in only 1 4 % . Table I11 shows the within-replicate dominance that existed in the competition cultures and Table I V shows the numbers of each species produced in control cultures. Three important facts emerge from these data: (1) the total numbers of adults produced in both the experimental and control replicates was well below the numbers that would be expected if the initial numbers of eggs and larvae were at the saturation densities established previously (Miller, 1964b); (2) total numbers of F, adults were not significantly different in the control and experimental populations; and (3) there was a distinct tendency for one species or the other to establish strong numerical dominance in each mixed-species population. A quantitative estimate of the species interactions within replicate populations was obtained by calculating the numbers produced by one species at each density level of the other. This relationship showed a general decrease of X2 ( D . simulans) with each increase in N , ( D . melanogaster)and vice versa. The rate of change of N , with each change in N , and the rate of change of N , with each change in N , was as follows:

With these and other data (see Miller, 1964c for details) it is now possible to construct a revised model of competition for one generation between D. melanogaster and D. simulans. This model (Fig. 8) shows that the numbers of F, adults that would be expected from the saturation densities established for larval populations of these species (Miller, 1964b) are not obtained when production depends on eggs oviposited by adult females. Secondly, when within-replicate dominance strongly favors one species over the other, the total numbers produced are higher than would be expected by chance; when the numbers of each species are nearly equal, the total numbers in each replicate are less than expected. I n what way can these data be used to explain the possible consequences of interactions in natural populations? There are two stages in the Drosophila life cycle that might be affected by competition: (1) the larval period when food quantity and quality might be critical and (2) the adult stage which is responsible for dispersal, habitat selection and reproduction. Larval competition, when it occurs, seems to consist of direct exploitation of the available food resources and will only be a


Within-Replicate Dominance in Competition Cultures (FromMiller, 196aC) ~




Mean & S.E. (Range)

Mean f S.E. (Range)

16.8 f 0-8 (5-30) 16.6 f 0.9 (5-35)

32.1 & 1.5 (12-53) 34.8 f 1.6 (13-60)

Males Dominant Species

Number of Replicates

D. melanogaster D . aimuluna

57 52

D. sirnulana

D . melanogaster

Mean f S.E. (Range) Mean f S.E. (Range) 13.7 f 0.8 (3-26) 2.9 f 0.5 (0-13)

1.6 f 0.4 (0-12) 15-3 f 0.8 (4-27)

T A B L EIV Number of Flies Produced in Control Cultures (From Miller, 1964-c) Males

Females Mean f S.E. (Range)

Mean & S.E. (Range)

13.8 f 1.4 (0-30) 16.5 f 1.7 (0-28)

27.5 f 2.6 (0-48) 30.9 f 3.1 (0-52)


Number of Replicates

Mean f S.E. (Range)

D. melanogaster D. aimulana

36 29

13.7 1.3 (0-24) 14.5 f 1.6 (0-27)





factor when the population reaches a critical density in relation to the amount and quality of food. It is evident that species interactions at this stage need not lead to the selective elimination of one of the competing species. The interactions values ( a = fi = 1) may in fact allow complete coexistence of both species, except at extremely high densities. The results obtained by Sokoloff (1955) for larval competition between the sibling species D . pseudoobsczsra and D. persirnilis were not analysed in this way but suggest an identical relationship. It would appear therefore that the outcome of competition in this system depends primarily on genetic and environmental influences affecting fecundity and fertility of the parental generation.


K1.5 a PIO.8. Model of competition between Drosophila melamqaater and D . tiimulans for one generation (After Miller, 1 9 6 4 ~ ) . Nc,


Competition could occur a t the adult stage between females that require access to a limited number of oviposition sites, or through interactions between larval and adult populations when the larvae alter the suitability of the medium as an oviposition surface (Sang, 1950; Miller, 196413). Interference apparently does occur among adults, or between adults and larvae, as shown by the exponential change of N , with changes in N , in the model in Fig. 8. Nevertheless, the potential fecundity of the adult females is not realized and the larval population never reaches saturation in this system. Chiang and Hodson (1950) have shown that the character of the surface of the medium changes rapidly as the numbers of larvae increase. These changes lead to marked



reductions in realized fecundity and fertility after the first few days of cultures, so that the conditions suitable for a high rate of production are restricted to a very short period (Robertson and Sang, 1944). Natural breeding sites also deteriorate rapidly in relation to their suitability as oviposition surfaces (Carson and Stalker, 1951) and seldom contain more than a few eggs or larvae (cf. Gordon, 1942; Birch and Battaglia, 1957; Sokoloff, 1957). Harrison (1964) studied the factors affecting the abundance of four species of Lepidoptera that spend their life cycles on banana plants, and concluded that they are not affected by intra- or interspecies competition because they normally lay so few eggs per unit area that their population levels never reach critical densities in relation to the food that is available. Adult female Ceramidia butleri will lay far more eggs on banana plants in cages in an insectary than they will under natural conditions in plantations, and the larval densities required to produce a significant decrease in pupal weight and an increase in mortality in the laboratory are much higher than densities recorded in the field. Population control through limitations on realized fecundity is so effective that competitive exploitation does not occur (Harrison, 1964). An absolute shortage of food is probably a rare event in the lives of animals such as insects and other invertebrates that would normally compete through exploitation. I n spite of the tremendous increase potential these animals possess, their populations appear to be regulated quite strongly by climatic factors that tend to reduce fecundity, fertility and adult longevity t o values that are only fractions of those reached under optimum conditions (Birch, 1948). Birch (1945, 1963) has shown, for example, that the beetles Rhizopertha dominica and Calandra oryzae are extremely sensitive to temperature and moisture in their rates of development and in population characteristics affecting their intrinsic rates of increase. Data for other poikilotherms are less precise than those provided by Birch’s elegant studies, but it is well known that their population characteristics are affected in much the same ways by relatively slight changes in physical factors. If competition for food occurs in such populations, it is likely to be a transient phenomenon that occurs only infrequently. The fact that population irruptions or “outbreaks” are especially characteristic of certain insect populations and are seldom observed in homiotherms lends support to the view that the populations of most terrestrial insects are normally under climatic control. They only express rates of increase approaching their full potential when an unusual combination of optimum conditions occurs. I n species such as dung and carrion insects, on the other hand, in which the oviposition site and larval environment are less subject to the effects of physical factors and over-population is more liable to



occur, there is often a strong interference component that limits the number of eggs laid, or a sequence of rapid sera1 changes in the oviposition site that restrict egg laying to a relatively short period. For example, several burying beetles (Necrophorus)may collaborate in the burial of a carcass, but aggressive interference ensures that only one pair of adults will finally remain in possession of it (Wynne-Edwards, 1962). Such interactions are not confined to intraspecific encounters, as I have witnessed the same thing among different species of Necrophorus. I n this example, competitive interference limits the size of the F, generation and consequently reduces the probability of heavy exploitation among the larvae. The ecological advantages of this arrangement are fairly obvious. Laboratory studies of competition have demonstrated that direct exploitation in crowded populations may lead to extremely high mortality of larvae and severe reductions in the size of the F, generation. If climatic or other environmental controls do not prevent the occurrence of potentially harmful rates of exploitation, there will be a selective advantage in the development of effective interference mechanisms within and between species populations. It was noted earlier that the traditional emphasis on selective elimination in laboratory studies of competition has tended to obscure the more interesting fact of extended coexistence that is frequently observed in these populations, even when the conditions of the experiment are designed to achieve maximum competition. The experiments with D. melanogaster and D. simulans adults and larvae (Miller, 1964c) established the fact that interactions between sibling species in a uniform environment can have equal value, but this series of experiments involved events in only one generation. Conventional competition theory would argue with some justification that a competitive advantage, however slight, is bound to accrue to one species or the other in a stochastic projection of a mixed-species system through several generations. If we grant this objection to the models developed earlier, we must then ask "What degree of species difference or environmental fluctuation would permit indefinite coexistence of the two species?" Moore (1952)found that D . melanogaster eliminates D. simuluns when they compete in population cages kept a t a temperature of 25"C, but the competitive advantage is reversed at 15°C. However, the progress of competition and its outcome depended partly on the age and quality of the food, which was a function of the rate at which the food medium was renewed. Conceivably coexistence could have been maintained indefinitely either by keeping the populations at some intermediate temperature between 15" and 25"C, by periodic fluctuations in temperature, or by selecting an appropriate program of food renewal. Even with the experimental design that was followed, D. melanogaster was



only reduced to 45% of the total population in one cage after 402 days at 15OC. I n a similar study of competition between D. melanogaster and D. funebris in population cages, it was found that fresh medium favored the production of D. melanogaster while D. funebris was more successful than D. melanogaster in older food (Merrell, 1951). I n an otherwise constant environment, periodic renewal of the food cups introduced environmental fluctuations that were sufficient to allow coexistence of these species for almost 2 years, after which the experiment was terminated. Table V shows the time required for extinction of either Tribolium castaneum or T . confusum in mixed-species populations in different volumes of flour a t 29°C and 65 to 70% relative humidity (Park, 1948). The populations were started with an equal number of adults of each species. The minimum time of 270 days required for extinction of T . castaneum in 8 grams of flour is equivalent to approximately 8 generations at 29”. The maximum time of l 470 days in 80 grams of medium is roughly equivalent to 42 genwitions of coexistence in a uniform environment. TABLEV Period Required for Extinction of Tribolium castaneum or T. confusum at 29°C and 65-70% R.H. (Datafrom Park, 1948) Initial Adult Population

Amoiint of Medium (grams)

8 40 80

8 40 80


Days to Extinction Minimum Maximum Mean ~


0 2

270 300 840


780 1020 1470






548 513 1155

There are numerous examples in the literature showing that selective elimination of species through exploitation of food may require an extremely long time, even when food and space are artificially limited and the environment is kept as uniform as possible. The remarkable fact in these results is not the traditional observation that “there is some ecological difference between the two species that permits coexistence” -this much is axiomatic -but that coexistence even in crowded populations seems t o require such slight alterations in a single factor, especially in relation to event,s in natural environments. The restrictions mentioned previously on expressing niche relationships in terms of set theory included the assumption that the probability of survival is equal at all point,s within the fundamental niche. This, as





Hutchinsori ( 1 ! ~ 7 )points out, is highly unlikely and we ordinarily expect that there will be an optimal part of the niche with suboptimal conditions near its boundaries. This is reflected, of course, in the law of tolerance and is seen in the outcome of competition between T . cnstaneum and T . confusum under different conditions of temperature and moisture (Park, 1954). Table VI shows the percentage of replicates for each treatment in which T . castaneum or T . confusum eventually persisted and its competitor was eliminated (extinction of one species was a condition of the experiment). At high temperature and moisture T . castaneum persisted and T . confusum was eliminated in all replicates; a t low temperature and humidity the reverse occurred. I n the latter situation the failure of T . cnstaneum to survive was not due entirely to competition. In single-species controls T . castaneum had a mean life-duration of 350.0 f 34.1 days, but this was reduced to 27.15 f 18-1 days when T . confusum was present, so that the effect of the mixedspecies interaction was to hasten the elimination of T . castaneum which would have occurred in any event. The most interesting feature of these results is, however, the “indeterminate” outcome of competition when temperature and moisture were between these extremes. Thus, although T . confusum eliminated T . castaneum. in 71% of 28 replicates a t 24°C and 70% relative humidity, T . castaneum won in 29% of the replicates. The results were therefore “probabilistic” rather than “deterministic” (Park, 1054). TABLE


Selective Eliminotion .f Triboliuin cavtaneum and T. confusuin in Relation to Temperature and Moisture. (Data front Park, 1954) Treatment Tempera,ture Moisture

34 34 29 29 24 24

Percent of Replicates in which Species Persists

(Percent R.H.)


70 30 70 30 70 30

100 10 86 13

29 0

T.coi&wrn 0 90 14 87 71 100

This interaction has been analysed in detail by Neyman, Park and Scott (1968) and is discussed in terms of the Gause cases by Slobodkin (1961). The patterns of competition for each set of conditions (Table VI) can be represented by a series of empirical diagrams that effectively



replace the diagram of Case 1 (Fig. 6). The diagram for 24OC and 70% RH is shown in Fig. 9. The line 0s (Fig. 6) becomes an indeterminate zone of possible initial numerical combinations of T . castaneum and T. confusum (Fig. 9) in which either species may eliminate the other. Outside this zone the outcome of competition is deterministic. Different combinations of temperature and moisture alter both the shape and the position of the indeterminate zone. Thus, a t the highest temperature and humidity T . castaneum invariably persists and T. confusum is eliminated in mixed-species populations, even though these conditions are within the fundamental niche of both species. I n other words, at high temperature and humidity the zone of indeterminacy disappears ; the lowest temperature and humidity created conditions outside the fundamental niche of T . castaneum.

T confusum

FIG. 9. Diagram of the outcome of coniprtition between Triboliirm cnslatiruni nnd 7'. confwum at 24OC and 70% H.H. (After Neyniaii, Park and Rrott. 1958).

Inasmuch as different parts of tho fundamental niche have different survival values, the same is true for points within the intersection subset of competing species. Slobodkin (1961) points out that if there is also differential survival of the two species in the zone of indeterminacy, as the analysis by Neyman, Park and Scott (1958) suggests, their diagrams become probability of outcome surfaces, introducing stochastic processes into the Gause case for this interaction. The indeterminate model for Triboliurn and the analogous model for Drosophila imply that a position may exist within the intersection subset of two fundamental niches corresponding to equal or nearly equal probability of survival of both species. Such a conditioii would only be I~elevitntto direct or



indirect exploitation of the resources contained within the intersection subset of their niches, and would not usually apply in the case of competitive interference with access to the resource. Chance oscillations in the environmental variables controlling the values of tc and /Imight pass back and forth through the equilibrium point for the two species, continually reversing the direction of competition and the probability of survival of one or the other of the competitors (Hutchinson, 1948). The theoretical possibility of a stable equilibrium is instructive, even though it could not be sustained in a naturally changing environment. Its principal value lies in the fact that it allows us to recognize conditions which would permit the coexistence of two species and to examine these criteria with respect to events in natural ecosystems. Thus, these analyses support Hutchinson’s (1948) observation that the rule of competitive exclusion need-not apply when (1) external factors act to rarefy the mixed-species population so that the required resources are not heavily exploited, or ( 2 ) when the values for tc and @ are under environmental control and chance oscillations prevent the establishment of a permanent equilibrium.

EXCLUSION VI. CONDITIONSOF COMPETITIVE The conditions of temperature and moisture in Park’s experiments with Tribolium confusum and T . castaneum may be viewed as a set of points in the intersection subset between two fundamental niches (Fig. 2, Case 2). It is evident from the data for control replicates that all points in the fundamental niche spaces N, or N, do not have equal survival value for the species, which in turn implies that the outcome of competition between two species with intersecting niches will not be the same at all points in their intersection subset N,.N,. This will be the case especially in interactions where the dominant form of competition is exploitation and the outcome of the interaction depends on (1) the ability of each species to survive, in the absence of interspecies competition, a t different points in the niche space of the intersection subset, and (2) their different exploitation and survival rates during interspecies competition. Park’s experiments provide an especially good illustration of the change in interaction values occurring at different points in the intersection subset, and of the fact that an indeterminate zone exists where the outcome of competition cannot be predicted with certainty. Rather than the strict system of fundamental and realized niches referred to earlier with reference to Case 1 and Case 2 (Fig. 2), the indeterminate zone is an area of probabilities in which both species may occur with changing frequency. However, when competition is chiefly through interference, there tends to be a more definite exclusion point corresponding to the



combination of factors which represents the limits of niche space within which species S, is able to exert its competitive superiority and completely exclude S,, or vice versa. As long as interference prevents access to the resources contained within a niche space, the realized niches of the two species can be described in spatial terms relative to the environmental variables that control the interaction. The conditions of competitive exclusion through interference are particularly well-illustrated in the case of an “included niche”, in which the niche (N,) of species 8,is a proper subset of and is included within the niche (N,) of species S,. As noted earlier, this situation immediately imposes certain strict conditions on the interaction. I n order for S , to survive in an included niche, it must be competitively superior to S,, otherwise it would be completely eliminated from this biotope. This with its smaller niche space, is a more relationship also implies that AS,, specialized species with respect to the variables determining N, and N,. A system of included niches therefore offers an opportunity for coexistence within the total biotope B by means of specialization of S, and the ability of S, to survive within the difference subset N, - N,. It is of considerable interest that several examples of this relationship exist among different taxonomic groups. These examples will be reviewed in some detail in order to define their common properties.



It has often been noted that pocket gophers (Geomyidae) do not form mixed-species populations and are mutually exclusive, even in local habitats. Their geographic distributions reflect this intolerance, in that the ranges of two species may meet in contiguous allopatry but do not become truly sympatric. An investigation of the ecological relationships among four species of pocket gopher which are at or near limits of their continental distributions in Colorado showed that their habitat and geographic distributions are essentially governed by three factors: soil depth, soil texture and competitive exclusion (Miller, 1964a). All four species seem to prefer deep, sandy soils but, as illustrated in Fig. 10, they differ in their abilities to burrow in shallower and coarser soils. For example, Geomys bursarius is mostly confined to deep sand or sandy loam soils, while indurate soils such as compacted clays or coarse gravels are barriers to its habitat and geographic distributions. At the other extreme, Thomomys tdpoides also prefers deep, sandy soils but can burrow in extremely coarse soils and occurs throughout a great variety of soil types and habitats. Although the four species differ slightly in food habits and perhaps in other respects as well, these differences do not appear to be critical and their fundamental niches are essentially defined by their responses to the two variables of soil



FIQ. 10. Relative tolerances of pocket gophers to soil depth and soil texture (After Miller, 1964a).

depth and soil texture. Thus the relationships between the four species can be described as a system of included niches based on these two environmental variables. It was also shown in this study that when the ranges of two or more of these species meet, the species with the greatest range of tolerance to these factors tends to be excluded from the preferred habitat and is displaced to less favorable environments. Thus, for example, the range of Geomys bursarius interdigitates into the general ranges of T . talpoides and T . bottae along river margins where sandy soils occur. When T . bottae and T . talpoides meet along mountain slopes, T . bottae tends to occupy the deeper soils at lower elevations, while T . talpoides is displaced upslope into less favorable soils. Survival therefore depends on finding sufficient space in habitats outside the included niche of the superior competitor. The order of fundamental niche space occupied by the four species 8.6determined by soil depth and texture is: Geomys bursarius < Cratogeomys m t a m p s < Thomomys bottae < Thomomys talpoides. When competition occurs between any species pair in this group the one with the smallest niche space occupies the position of N, in the relationship illustrated for Case 1 (Fig. 2) and the other species is excluded to the difference subset N, - N,. The corresponding ranks of competitive ability are: G e m y s bursarius > Cratogeomys castamps > Thomomys bottae > Thomomys talpoides. Each case of competitive exclusion that was observed when the ranges of the two species met in Colorado, or has been reported for combinations of these species elsewhere, conformed to this general



relationship (Miller, 1964a). We conclude from this that the superior competitor is in some way more specialized and that its distribution is governed solely by its relation to the variables of soil depth and texture. These four species show morphological differences in body size and fossorial development that correspond to their competitive rankings. Gemys bursarius is the most obviously fossorial, with a relatively massive skull that is flattened doro-ventrally, reduced eyes and ears, and large forefeet and claws. Because of the pronounced fossorial character of its claws and feet it walks with difficulty, with the feet splayed outward like those of a mole. Thomomys talpoides, at the other extreme, is a more generalized, mouse-like rodent. I n body size, C. castanops is somewhat larger than G. bursarius, T . bottae is next in size, and T . talpoides is the smallest. As there is a correlation between body size and soil type in pocket gophers, with the largest individuals and subspecies occurring in deep, sandy soils, it is difficult to judge the extent to which observed differences in body size are hereditary or phenotypic. On the one hand, large body size might confer an advantage in aggressive encounters, which undoubtedly occur between these highly territorial animals, but smaller size may be advantageous when an animal is forced to live in more indurate soils. Kennedy (1954, 1959) suggests that the relatively smaller size of (2. bursarius may have adaptive value, allowing it to survive in less favorable habitats when it competes with G. personatus. The ranges of the yellow-pine chipmunk (Eutamiasamoenus) and the least chipmunk (E. minimus) overlap narrowly in the eastern foothills of the Rocky Mountains in Alberta, where E. amoenus occupies the open forests and E . minimus is mostly confined to alpine slopes. Wherever other species of Eutamias occur, E. minimus generally seems to be displaced either to sagebrush desert or to alpine habitats, although in regions of North America where there are no other western chipmunks, E. minimus also occurs in forest habitats, suggesting that its fundamental niche is in fact relatively broad. In a study of the oompetitive relationships between Eutamias amoenus and E . minimus in their region of sympatry in western Alberta, Sheppard (unpublished) observed that the most consistent difference between the habitats of the two species is in the amount of vegetative cover present. Using the classification of vegetation layers proposed by Elton and Miller (1954), Sheppard recorded the percent occurrence of each type in vegetation samples from habitats occupied by each species. As shown in Table VII, the typical habitat of E. amoenus contained more vegetation of all types, especially shrubs and trees. Both species use shrubs and trees as “ewape routes” when alarmed, but E. minimus will apparently tolerate more open habitats in this respect, moving across



greater distances in the open and making greater use of alternate forms of cover such as fell boulders and logs. Other ecological differences such as food habits were not great, and could be explained by the frequency occurrence of particular food species within the different .habitats, rather than an expressed preference. There is, however, a difference in the body sizes of the two species and corresponding differences in aggressive behavior. TABLEVII Percent Occurrence of Vegetative cover in Habitats Occupied by E. amoenus and E. minimus in WesternAlberta Percent Occurrence E. amoenus E. minimus Habitat Habitat

Vegetation type -

Field layer Shrub layer Tree layer



68.9 73.3 87.8

57 *8 14.4


Brown and Wilson (1956) suggest that character displacement reduces competition and increases isolation, and Kohn and Orians (1962) maintain that all instances of character displacement probably involve ecological displacement as well. Sheppard (unpublished) compared the body sizes of the subspecies Eutamias minimus listed by Hall and Kelson (1959). The largest subspecies in North America occur where this is the only representative of the genus, while in regions where there is potential or presumed competition with other species of Eutamias, the subspecies of E. minimus are smaller. Sheppard also compared the body lengths of E. amoenus and E. minimus from within his general study area in western Alberta where (1) the two species live in adjacent habitats and are potentially sympatric and (2) where they are sufficiently allopatric that there is little or no continued contact. This comparison is shown in Table VIII. E. amoenus is consistently larger and E. minimus smaller in areas of sympatry, although the withinspecies differences between sympatric and allopatric populations were only significant for male minimus (t = 2-57, P < 0.02). The values for female amenus (t = 1-73, P ca. 0.09) and female minimus (t = 1.82, P ca. 0.075) approach significance. While these data indicate a case of competitive exclusion, they do not show conclusively that E. minimus in this region can survive outside the habitats in which it is presently found, or that the observed distributions depend on more than different habitat preferences. Sheppard (unpublished) therefore removed resident populations of E . amoenus



from two areas of typical forest habitat and introduced approximately equal numbers of E. minimus and, likewise, introduced E. amoenus into areas previously occupied by E . minimus. Introduced populations of 10 E. amoenus in each of two areas either failed to survive or dispersed away from the alpine habitats where E . minimus had lived before. On the other hand, of 15 E. minimus transplanted into E . amoenus habitat, 3 were recovered 1 year later, 1 was still present after 2 years, and an immature E . minimus captured 2 years after the original introduction was presumably the offspring of a pair of the introduced adults. Although the survival rate for the introduced E . minimus was not particularly high, this experiment did show that they are capable of living and probably reproducing in habitats from which they appear to be excluded. TABLEVIII Body Lengths ( m n ~of) Adult Eutarniits amoenus and E. minimus from Symputric and Allopatric Habitats Malcs Mean & S.E. (Range) -.

Eutarnias amoenus Allopatric Sympatric

Eutarnias mininaus Allopatric Sympatric


Females Mean & S.E. (Range) . .

121.3 f 0.54 (102-127) 122.2 1.40 (113-130)

125.1 0.5G (114-132) 127.4 f 1.53 (111-138)

112.4 f 0.85 (105-118) 109.5 0.76 (102-115)

115.9 & 0.82 (106-127) 113.4 & 1.14 (103-120)


Sheppard (unpublished)concluded from these and other data that the relationship between these two chipmunks is broadly similar to that described by Miller (19G4a) for pocket gophers. E. amoenus apparently occupies an included niche (N,) within the larger, fundamental niche (N,) of E . minimus, while E. amoenus is a superior competitor which in areas of sympatry displaces E. .minimus into the difference subset (N, - N,) of their two niches. To test this theory further Sheppard conducted a series of laboratory tests of the behavior of the two species during aggressive encounters. For example, for one set of experiments Sheppard constructed an apparatus consisting of a centrally located food chamber from which a screen tunnel ran for 6 feet in either direction to a pair of nest boxes. An individual E . amoenus was placed in one nest box and an E . minimus in the other and their behavior observed from be-hind a screen. In a total of 71 contests between individuals of the same sex and age group, E . amoenus won 672 aggressive encounters and lost only 23. I n 60 of C



the 71 contests the E. amoenus individual was the first to emerge from the nest box; on 43 occasions an E . amoenus individual entered the central chamber and fed, while only one E . minimus did so; and the E . amoenus spent a total of 2 185 minutes outside the nest box while the E. minimus remained outside for a total of only 983 minutes. I n these and other experiments E. amoenus individuals were consistently more aggressive and demonstrated their competitive superiority in physical encounters. This model also seems to be the best method of describing the pattern of interaction between the black rat (Rattus rattus) and the brown or Norway rat (Rattus norvegicus). The earlier arrival of the black rat in Britain, with the return of the Crusaders during the middle ages (Matheson, 1939), allowed this species to occupy most of the British Isles. The subsequent arrival of the brown rat at the begiiining of the 18th century has apparently been responsible for the disappearance of the black rat from most of the areas it occupied previously. A similar replacement is occurring in U.S., where the brown rat replaced the black rat in an area of 1 0 0 0 sq. miles in southwest Georgia between 1946 and 1954 (Ecke, 1954). The black rat is often arboreal, however, and in urban areas this allows it to coexist to a limited extent with the brown rat -the black rat occupies the top stories of buildings, travelling along power lines and rafters, while the brown rat inhabits the ground level (Southern, 1964). Relationships between the wood mouse (Apodemus sylvaticus) and the yellow-necked mouse (A.Jlavicollis)and the wood mouse and the bank vole (Clethrionomys gkreolus) invite a similar explanation. The larger yellow-necked mouse occurs entirely within the range of the more widespread wood mouse, except in northern Europe, but is absent, from large areas (Southern, 1964). I n Britain A . Jlavicollis occurs in pockets throughout populations of A. sylvaticus, but is more strictly confined to woodland habitats while A. sylvaticus is also common in fields and scrub. The habitat preferences recorded for these species in Europe also confirm this relationship (Grodzinski, 1959). The bank vole has a much more restricted distribution than the wood mouge in British woodlands, especially in winter when the ground layer of vegetation disappears (Evans, 1942; Miller, 1955; Kikkawa, 1964). This seems to be related to the fact that the bank vole is more active during the day and would be exposed to greater risk of predation if it did not restrict its movements to the immediate vicinity of protective cover. There is thus a relatively large amount of intersection in the spatial elements of the fundamental niches of these two species in summer when ground cover is abundant and the bank vole can move more freely, but interference is alleviated by the fact that the two species have different



activity rhythms and distribute their use of the environment over somewhat different times of the 24-hour cycle (Miller, 1955). I n winter, however, when movements of the bank vole are most severely restricted, food resources are less abundant, population numbers are high, and interspecies competition becomes more critical (Kikkawa, 1964). I n order for C. glareolus to survive as it does in a niche included within the larger fundamental niche of A . sylvaticus it must have some competitive advantage. Kikkawa’s observations on aggressive behavior and dominant-subordinate relations show that voles are especially aggressive and tend to form dominance hierarchies and probably do interfere with access by mice to areas occupied by the voles.



In spite of a vast literature on birds it is difficult to find well-documented information on their habitat requirements and interspecies interactions. Nevertheless, certain observations suggest that included niches occur commonly between many species. Snow (1954) concluded that most members of the Poecile group of titmice in Europe are separated by habitat and that the local absence of one member of a closely related pair has not been found to affect the habitat preference or size of the other, except in the case of Parus palustris and P. atricapillus, which are more similar than are any other sympatric Palearctic species of this genus. There is evidence that P. palustris has replaced P. atricapillus in four out of five instances, while in three cases where P. palustris is absent and its habitat available P. atricapillus has not extended into it. This is in striking contrast to the situation in North America where, with the exception of P . hidsonicus which overlaps widely with P. atricapillus, all of the Poecile group replace one another geographically, with some rather narrow zones of overlap (Snow, 1954). The system of relationships among the Paridae of North America indicates a possible system of included niches. P. atricapillus has the widest range, across the whole of the northern part of the continent, but is replaced in the southeast by P . carolinensis, in the Rocky Mountains by P. gambeli, on the western seaboard by P. rufescens and in the mountains of Mexico by P. schteri. Detailed studies of competitive relationships among these species have not been made, but where the ranges of these forms are contiguous or overlap they are separated by habitat differences, and Snow (1954) concludes that “The fact that no two species of the same genus overlap widely, except for P. hudsonicus and P. atricapillus, suggests a more recent, less complex evolutionary history for the genus in North America, a suggestion which receives support from the fact that on the whole differences between species are not nearly so great



as they are in the Palearctic.” These observations seem to suggest, in other words, that competitive exclusion may still be an active ingredient in the more recent evolutionary events affecting this genus in North America, and that P . atricapillus, with its broader geographic range and presumably greater tolerance to habitat variation, is only able to survive outside the included niches of its competitors. Thus, Snow seems to feel that the habitat differences that are observed among members of this genus in Europe, such as between the coal tit (Parus ater) which searches for insects on the trunks and larger branches of trees and the blue tit (P. caeruleus) which concentrates on twigs and leaves (Snow, 1949), are due to ecological differentiation and no longer involve competitive displacement. A more precise study of the relationships between two species of this genus in North America has been made by Dixon (1954). In the past two decades Parus rufescens has become established as a breeding bird in the district immediately east of San Francisco, but its spread is impeded by competition with its larger cogener, Parus inornatus. I n all cases where the two species come into contact, their breeding territories are mutually exclusive and, in the face of antagonism by the larger form, the adjustments permitting co-occupancyof the area appear at this stage to be made entirely by the smaller, less specialized species. These include the use of vacated or suboptimal nesting territories, modification of territorial behavior, and a more varied choice of food items. Thus, the recent entry of P. rufescens into the range of its larger cogener has been made possible by the availability of a niche space outside the more specialized, included niche of P. inornatus. The yellow-headed blackbird (Xanthocephlus xanthocephalus) is restricted in its nesting sites to emergent vegetation, e.g. bulrush (Scirpus), cattail ( T y p h ) or Phragmites in fairly deep water (Nero, 1964). The red-winged blackbird (Agelaius phoeniceus) occupies the same marshes as yellow-headed blackbirds but nests in a greater variety of situations. Most nests are located in cattails but redwings will also nest in low trees, shrubs or any weeds that will support a nest (Orians, 1961). Male redwings arrive on the breeding grounds and begin to establish territories somewhat earlier than yellow-headed blackbirds. I n the vicinity of Saskatoon, Saskatchewan the first recorded arrival of male redwings in spring was 7 to 20 days (average = 13 days) earlier than the first male yellow-headed blackbird during 6 years from 1960 to 1965 (J. B. Gollop, personal communication). I n 1964 and 1965 the author observed interspecific relationships between these two species on a small marsh near Saskatoon. Male redwings were engaged in territorial display throughout most of the marsh by the time the first yellow-headed blackbirds arrived, but within a few days the yellow-headed blackbirds had



occupied the emergent vegetation in deep water near the center of the marsh, and had displaced the redwings t o more peripheral nest sites in spite of persistent harassment by the redwings. The territories established by the yellow-headed blackbirds in their first few days on the marsh remained intact throughout the remainder of the breeding season.


W. F. Blair (personal communication) has followed events for several years in a temporary pond containing populations of the bullfrog (Rana catesbiana) and leopard frog (Rana pipiens). R. catesbiana requires a habitat of standing water but R. pipiens can also live in more terrestrial environments. When drought reduced the pond to almost nil, R. catesbiana disappeared and only a few, scattered R. pipiens remained. When the pond refilled after a wet spring, the 12. pipiens population “exploded” to a very high density, far greater than the pond could normally sustain. Gradually, the numbers of R . catesbiana also increased and the numbers of R. pipiens declined. When R. pipiens was the only species present it filled the entire habitat, but when the R. catesbiana reappeared this species occupied the water environment and displaced R. pipiens t o drier sites near the edge of the pond. The salamanders Plethodon dunni and P. vehiculum are abundant in the rocky outcrop-talus slope ecosystem of the coast range of Oregon, where they occupy approximately the same habitats. P. dunni tolerates slightly lower temperatures and wetter substrates than P. vehiculunz while P. vehiculum is tolerant of lower relative humidities and higher temperatures than P. dunni, although their preferences for cool, humid conditions are not significantly different (Dumas, 1956). Although temperature does not significantly alter the humidity preference of the two species within the normal range of humidity encountered in their natural environments, there are specific differences in their substrate distributions which are apparently due t o the presence of P. dunmi ill the preferred sites and displacement of P. vehiculum t o lion-preferred but tolerated habitats. This interaction becomes especially critical in spring and fall, when the weather is variable and there is great risk of exposure to lethal environmental conditions, and when P. vehiculum is found far more often than P. dunni in suboptimal “transient habitats”. Dumas (1956) examined 36 filled stomachs of P. dunni and 37 of P. vehiculum to determine the extent of ecological differentiation in their food habits, because of the stress that is often placed on this factor in competitive interactions. He found that collembolans are the greatest food source for both species, but that P. dunni consumes a greater variety of foods than P. vehiculwm. There was a total of 39 types of food with a mean of 8.36 items per stomach in the P. dunni sample,



and a total of 28 foods with a mean of 20.84 per stomach in the P. vehiculum sample. The greater number of individual items in the P. vehiculum stomachs is explained by the fact that the somewhat smaller salamander ( P . vehiculum, snout-vent length 46 to 57 mm) consumed far more small-sized forms such as mites than did the larger species ( P . dunni, snout-vent length 48 to 65 mm). When the foods were separated by frequency index into “primary” and “secondary” foods, it was found that of 16 primary foods 10 were shared by both species but P. vehiculum had fewer secondary sources (14) than P. dunni (27). The similarity of primary food sources might suggest rather strong competition for these items, but P. dunni has more secondary sources and would be better able to survive periods of food scarcity than P. vehiculum. Dumas (1956) concluded that foods acceptable to both species are sufficiently abundant that competition for this factor will rarely be critical, but that there is rather high mortality from adverse physical conditions and there is definite competition for sites with high humidities, with P. dunni excluding P. vehiculum. D. C R U S T A C E A Connell’s (1961) study of competition between the barnacles Chthamalus stellatus and Balanus balanoides defines both the fundamental and realized niches of these species. The center of distribution for C. stellatus is in the Mediterranean and it reaches its northern limit in the Shetland Islands. Balanus balanoides is a boreal-arctic species which reaches its southern limit in northern Spain. At Millport, Scotland where this study was done, the larvae of C. stellatus settle in the marine intertidal zone between the levels of mean high water of spring tides and the mean tide level, but few survive below the mean high water of neap tide (see Fig. 11). The larvae of B. balanoides settle throughout the range of all intertidal levels from mean low water to mean high water of spring tide, but poor survival between the mean high water of spring and neap tides restricts the adult distribution in this region. Connell showed that the upper limit of distribution of Balanus is determined by mortality from desiccation, competition for space and predation by the snail Thais lapillus, but C. stellatus can survive at all levels and increased time of submergence was not a factor in the elimination of this species at low shore levels. Thus, in view of its greater ability to withstand alternate submergence and desiccation at the higher shore levels, C. stellatus has a larger niche than B. balanoides. Intraspecies competition for space is rarely observed in Chtharnalus but is common among individuals of Balanus. Comparison of the survival of Chthamalus in the presence and absence of Balanus showed that Balanus could cause considerable mortality among the Chtha.m,alus;



the Balanus settled in greater population densities, grew faster than Chthamalus, and direct observation showed that Balanus undercut or crushed individuals of Chthamalus and eliminated them from the area between mean tide level and the level of mean high water of neap tide (Fig. 11). Chthamalus which survived after a year of crowding by Balanus were much smaller than uncrowded individuals but even adult Chthamalus failed to survive when transplanted to low levels. BALANUS BALANOIDES








0 I 2


MtAN 1 l D t L l V t l

I I c







FIG.11. Distributions of adults and larvae of Balanus balanoides and Chthamalus stellatics in the marine intertidal zone (After Connell, 1961).

Similar evidence for competition in an included niche is found in competition between the crayfish Orconectes immunis and 0. virilis (R. V. Bovbjerg, personal communication). Laboratory tests have shown that both species can live in mud but prefer rock and gravel substrates when given a choice. 0. immunis is the better burrower, however, and also tolerates lower oxygen tensions than 0. virilis. Thus, ponds which are subject to summer drying, periods of low oxygen tension, and have soft mud substrates are least tolerated by the stream form, 0. virilis. Bovbjerg found large numbers of both species in oxbows after spring floods, but when the oxbows became stagnant and finally dry, the pond species 0. immunis was the only one that survived. Conversely, in spite of its demonstrated preference for rock and gravel substrates, the pond form is seldom found in streams. In one stream



where both species do coexist, 0. virilis occupies rocks and 0. irnmunis is found on mud. More concrete evidence for interspecies competition was provided by laboratory experiments in which substrate preferences were tested with both species in the same tank. A series of contests took place in which the less aggressive individuals were evicted from the more desirable crevices between stones. The stream species, 0. virilis, eventually occupied the rocks, leaving the mud substrate to the pond species, 0. immunis. Thus, 0. virilis has a more restricted niche, by virtue of its inferior ability to burrow in mud and to tolerate low oxygen tensions, but it is the more aggressive species and survives by excluding 0. immunis.

E. I N S E C T S The ant's Messor barbarus and M . aegyptiacus are harvesters which make long processions from their nests to harvest grass seeds. Pickles (1944) studied the relationships between these two species in Algeria and found that the foraging territory of a nest of M . aegyptiacus was 2 348 yd2 (1 963 m2) with a foraging distance of 82 f t (25 m). This nest was within a larger space occupied by M . barbarus, which had a foraging territory of 7 857 yd2 (6 569 m2)and a foraging distance of 45 f t (14 m). Processions of M . [email protected] frequently led over the nest of the M . barbarus and beyond it, showing that the presence of M . barbarus did not deter M . aegyptiacus from foraging in this direction. On only a few occasions, however, did M . barbarus forage in the direction of the M . aegyptiacus nest, and then only a few single individuals were involved. Eventually a series of battles took place during which there was considerable damage to members of the M . barbarus population and they abandoned the old nest and constructed a new one farther away. These two species apparently eat the same foods and have very similar habit'at requirements. M . aegyptiacus can occupy areas within the foraging territories of M . barbarus because of its greater competitive ability, which in this case involves direct aggression. M . barbarus can, at the same time, survive outside the immediate foraging territory of M . aegyptiacus because M . barbarus has a greater foraging distance. This system of relationships therefore allows coexistence of both species within the same broad habitat. While it is unlikely that the niche of one species will be entirely contained within the niche of another, it is evident that the critical factors which affect the outcome of competition between two species can often be reduced to a few simple variables which do have this relationship. This would seem to be especially true for homiotherms which, because of their adaptations to climatic factors, are subject to fewer controls and are perhaps more likely to have evolved a strong



interference element in their competitive interactions. The plains pocket gopher, Geomys bursarius, can subsist on certain grasses which will not sustain individuals of the mountain pocket gopher, Thomomys talpoides (Myers and Vaughan, 1963), so that what has been referred to as an “included niche” for G. bursarius contains elements outside the niche of T . talpoides; but both species can survive on such a wide variety of foods that it is unlikely that food is a critical factor in their habitat or geographic distribution or in competition between them (Miller, 1964a), and their aggressive, territorial behavior indicates that competition for space overrides this source of ecological differentiation. Thus the critical factors of soil depth and texture determine the suitability of such space for the different species, and therefore the outcome of competition between them. Similary, while the salamander Plethodon dunni can tolerate somewhat lower temperatures than P. vehiculum, the niche of P . dunni is included within the niche of P. vehiculum in the critical range of temperature and humidity between the preferred cool and humid conditions and the suboptimal conditions of high temperature and low humidity. Thus, the included niche is an adequate description of interspecies relationships as long as it accounts for the critical range of factors that govern competition and survival, and in this sense we should be prepared to accept reasonable approximations. It should also be emphasized, however, that we cannot assume that intersection between fundamental niches will inevitably lead t o competitive exclusion -this is an outcome which should first be demonstrated or, in the absence of direct evidence, only inferred when it is the most reasonable explanation of observed events. Given that the examples in this section are of competitive exclusion under the conditions of a n included niche in which N, is the proper subset of N, and the realized niche of 8, is the difference subset N, - N,, we can examine the properties of these interactions with respect to: (1)the characteristics of the competition process, (2) the factors which give the more specialized species S, its competitive advantage and (3) the factors which allow the less specialized species S, to survive competition from S,. I n each of the preceding examples the process of competition involved a form of interference which led to competitive exclusion within the niche space of the intersection subset, and the competition was for some kind of space or a structural feature within a given space. The space requirements of different taxonomic groups are obviously different, depending on their specific needs, and we may infer that the factors affecting the form of interference also were different or had different values, but whatever these factors and mechanisms may be they apparently lead t o comparable end results through a similar set of general conditions. O*



The following table categorizes forms of interference and the factors which seem to be related to their effectiveness: Forms of Interference Related Factors I. Indirect 1. Chemical communication 11. Direct A. Physical Contact 1. Population growth B. Threat 2. Aggression C. Epidiectic Display 3. Territory 4. Body size 5. Coloration 6. Voice A requirement of indirect ipterference is that a signal be produced which will be effective in the owner's absence, and this will almost invariably involve some form of chemical communication. 'Wilson and Bossert (1964), as mentioned earlier, have surveyed this topic, mostly with respect to intraspecies relationships, but there is a need for research on how chemical signals are transmitted between species and how effective they are. Regardless of whether a chemical is produced to act specifically as a signal, or whether such chemicals as metabolic wastes inadvertently condition the environment, the effect in either case may be indirect interference with access to parts of the environment or the resources it contains. The simplest form of direct interference in the preceding examples was the physical destruction of Chthumalus stellatus by populations of Balanus balanoides. This was due to the greater rate of growth of the Balanus populations, greater tolerance of crowding and, eventually, the reduced size and efficiency of the crowded Chthumalus. Competitive superiority of the ant Messor aegyptiacus was also due to direct physical aggression and a relatively primitive form of territoriality. The chipmunks Eutamias amoenus and E . minimus showed a kind of interference intermediate between physical combat and threat. Chases resulted in the establishment of dominance of E . amoenus, but with no apparent physical harm to either animal. The larger titmouse Parus inornatus is the more successful aggressor in competition with the chickadee, P. rufescens (Dixon, 1954), and the red-headed woodpecker (Melanerpes erythrocephalus) excludes the smaller downy woodpecker (Dendrocopus pubescens) from desired nest holes (Schwab and Monnie, 1959). I n competition between the hummingbirds Calypte anna and Selasphorus sasin (Legg and Pitelka, 1956), the former exercises dominance by earlier breeding and territory occupation, more effective flight displays and larger body size (see also Hartley, 1950; Marler, 1956). If the process of interference consists of direct or overt aggression, body size would seem logically to be advantageofis, and this hypothesis is supported by



numerous examples among competing species of vertebrates. Additional evidence is found in the fact that character displacement in the direction of greater difference in body size is pronounced in zones of overlap between closely related and presumably competing species (Brown and Wilson, 1956; Hutchinson, 1959). We may conclude that larger animals tend generally to be more aggressive and more successful in competition, although N. Tinbergen has pointed out in personal conversation that the universal tendency of animals to avoid aggressive encounters may be as important, or more so, in maintaining dispersion of individuals and species (see also Ripley, 1961). Klopfer (1962) also emphasizes that smaller size may have adaptive value in allowing the subordinate animd to escape the effects of competition and survive, as suggested by Kennerly (1959) to account for size difference between Geomys bursarius and G . personatus. The smaller titmouse, Parus caeruleus, can feed from the extremities of twigs that cannot be reached by its larger competitor, P . major (Snow, 1949), and (Klopfer, 1962) the shorter-billed downy woodpecker (D. pubescens) can deal with smaller branches than the more heavily-weaponed hairy woodpecker (D.villosus). I n fact, in the examples given for species pairs in included niches there was generally an inverse relation between body size and niche size, suggesting that ( 1 ) larger body size may confer competitive advantage but (2) animals with smaller body size are adapted to a wider range of ecological conditions and are better able to survive outside preferred habitats, often on smaller food particles. Miller (1964a) postulated a correlation between body size, Competitive ability or aggression, and territory size, assuming that a larger body. size would enable an animal to sustain its aggressive drive or to display over a greater area, thereby restricting the movements of its smallei competitors to smaller and perhaps less favorable areas. Unfortunately, territory is highly variable and difficult to measure. Territory size depends on topography and local habitat conditions (Beer et al., 1966), and accurate data relating territory to competitive ability are seldom available. What little information does exist shows that in some cases the animal with the larger territory is the superior competitor, while in others the opposite relationship exists. The crimson-crowned bishop (Euplectes hordeacea) has a relatively large territory which is not much affected by the abundance of breeding males, whereas its cogener the Zanzibar bishop (E. nigroventris) has a smaller territory which is highly compressible, according to population density (Moreau and Moreau, 1938). E. nigroventris occupies less favorable habitats and its smallest territories are often those with the most obvious digadvantages of lack of food and orowding by other species. Colonies of the tri-colored blackbird (Agehius tricolor) do not have territories (Orians, 1961) and are also



interspersed in apparently less favorable nesting sites than those occupied by the territorial red-winged blackbird ( A .phoeniceus). Moore (1964) measured the steady densities of the territorial males of 15 species of dragonfly in comparable habitats and found that territory size, as reflected in density, was correlated with the size of the species. On the other hand, Pickles (1944) found an opposite relationship between territory and foraging distance and competitive ability in the ants Messor barbarus and M . aegyptiacus. Gibb (1956) states, “Specific differences in size of territory have never yet been satisfactorily explained in terms of any particular biological requirement”, and we must conclude at this stage that if territory size is significant in competitive interactions between species it does not have a value which is consistent for different taxonomic groups. I n the absence of interspecies competition the fundamental niche of each species will eventually be fully occupied as a result of the pressures of intraspecies population growth, with the more desirable parts of the niche space occupied by the stronger individuals of the population. In the case of an included niche, for example, intraspecies and interspecies population pressures will tend to oppose each other in the manner shown in Fig. 12. The most preferred space of the two fundamental

Case 1

FIQ.12. Forces of population pressure within and between species in their realized niches N, and N,.



niches is in the region of the values x, and yl,and this space will be occupied by the strongest individuals of species S,, with pressure on other individuals of this species forcing them toward the less desirable regions of N,. Although S , may be forced by competitive exclusion to live in the niche space N, - N,, there will be pressure from within its own population to expand into the preferred space of N, as well as toward the region of lowest survival at the periphery of N, (Xg,y6). Presumably, according to this hypothesis, the weakest individuals of 8,will be most directly in competition with the strongest individuals of S,. Depending on the strength and effectiveness of interference from S,, the region of immediate exclusion may exist as a “tension zone” where the niche distributions are not clearly delineated. This may explain distributions of Planaria naontenegrina and P . gonocephala when they occur together in the same stream, as compared with the temperature tolerances they express when they live in separate streams. Rather than a clear separation of niches at the extreme limit of tolerance of P . motenegrina (16 to 17”C), population pressure from P . gonocephula shifts the point of demarkation in their realized niches to approximately 13 to 14°C. This problem, while vaguely defined at present, may have considerable bearing on the evolution of ecological differentiation and of ecological and ethological isolating mechanisms.

VII. SPECIES DIVERSITY Klopfer (1962) suggests the following ways in which it is hypothetically possible to increase the number of species in a fixed area: (1) by increasing the amount of time during which speciation can occur, (2) by increasing the “space” within which niches can be provided for different species, and (3) by reducing the size of niche required by each species. Niche size in this sense is the measurable range of conditions which determine the presence or absence of a species. Klopfer and MacArthur (1960) conclude that the increased faunal diversity of the tropics is a result of the time available for speciation and smaller niche sizes. One might also add that the other factor (2) of space is also important, in that the structural diversity of tropical habitats provides a greater amount of niche space in a given area. Klopfer (1962) carries this line of reasoning further to state that a reduction in the volume of the niche of a species also implies that the behavior of the animal has become stereotyped, as reducing the niche size also reduces the range of objects in the environment (or environmental variables) to which the animal responds by feeding, nesting, or taking shelter. Hence, if niche size is relatively large there is a wider range of behavior and a given area will support fewer of such species. Support for this argument is found in the relative abundance of passerine birds in temperate



regions. Klopfer maintains that passerines have a wider behavior range than non-passerines and are therefore capable of occupying a greater range and less stable set of conditions, and that this is the reason that members of this group are relatively more common in temperate habitats. Simpson (1964) analysed the species density of mammals in terms of the number of species in areas of North America and found two major trends: the most dominant trend is an increase in number of species from north to south, with a lesser trend toward increase in species with altitude. He decided that niche size is not an important contributing factor, although no particular reason was given for this conclusion. Van Allen (1965) analysed morphological variation in six species of birds in relation to niche size. The niches of these species are known to be broader on some islands than on the mainland and in each case there is also greater variation in bill measurements in the island birds, with the exception of one species in which the niche on the Canary Islands is known to be smaller than on the mainland. Van Allen concludes that niches on the zoogeographic mainland are relatively tightly packed together by the action of stabilizing selection imposed, presumably through competition, by ecologically adjacent species. On the islands the available environment is partitioned into wider niches with weaker stabilizing selection. Thus, the wider niches on the mainland would permit greater phenotypic variation if phenotype is controlled to a significant extent by the adaptive diversity of the niche (Van Allen, 1964). I n examples of included niches, in which the niche size of the included species is smaller by definition, there is a corresponding trend toward greater specialization. This is especially evident among pocket gophers in which Geomys bursarius, the species with the smallest niche, is the most highly evolved in fossorial behavior and morphology (Miller, 1964a). It has also been noted that in these and other species in which interference probably takes the form of overt aggression or threat, there is a trend toward greater body size. It would be interesting to compare examples of this sort with the exceptions that are known to exist (e.g. burrowing rodents) to Bergmann’s Rule. As the realized niche of an animal is the phenotypic expression of its responses to both its biotic and physical environments, specialization for competition may to some extent explain exceptions t o rules based on physical factors. It is also interesting to note that, while the fossorial development of Geomys bursarius may make it more efficient in sandy soils, this specialization apparently makes it less well adapted than less specialized forms (e.g. Thomomys talpoides) to a wide range of soil types. Caution must be used, however, in forming generalizations about the



causes and especially the mechanisms of species interactions. Attempts to derive general theories of population control have been seriously hampered by the indiscriminate application of results from studies of one taxonomic group t o observations of relationships in another, and the same can be said of competition theory. There is no reason to assume that all animals respond in the same way to the same factors and are, therefore, subject to the same laws of population growth or competition. Gammarus duebeni has a large amplitude with respect to abiotic factors but low productivity, while its competitors G. salinus and G. zaddachi have high productivities, faster development rates, higher growth rates and higher reproductive potentials within a narrower range of conditions (Kinne, 1954). On the basis of research with these and other marine organisms Kinne (1956) suggests that animals with a narrow range of tolerance to abiotic factors tend to have a high biotic potential. However, it would obviously be a mistake to assume that a high replacement rate is a universal criterion of biological success. I n species with ill-defined space requirements, competing mainly through exploitation rather than interference, replacement rate may have high value; but in species with strong interference elements in their competitive relationships and rather rigid dispersion mechanisms, it may be more advantageous to have a relatively low replacement rate and greater population stability. I n other words, animals living in restricted niches may be successful in maintaining their populations either through high biotic potential and exploitation or by competitive interference. This article suggests a t least two major sources of species diversity. When competition is primarily through exploitation and the system is under strong environmental control, it is likely that fluctuations in factors affecting reproduction and survival will continually alter the outcome of the competitive interaction, allowing coexistence of mixedspecies populations. If there is no interference, the more similar the species the more likely that their interaction values will be equal and correspondingly less environmental change will be required in order to allow coexistence. If there is a strong element of interference resulting in competitive exclusion, species diversity may still be increased if one of the species becomes specialized and, in so doing, reduces its niche size in such a way that both species are able to exist in the same biotope, e.g. when the more specialized competitor occupies an included niche which is small enough to allow the other species to survive in the difference subset of their niches. -4s noted earlier, coexistence is a, relative term depending on the size of area one chooses to measure. If there is a strong element of interference and the critical habitat features are uniformly distributed, competitive exclusion may operate over



geographic distances, as it does with pocket gophers which can only be said to coexist within relatively large areas or at the point of disjunction between soil types. If, however, the competitors occupy small territories and the habitat is suitably diverse, coexistence may be possible within relatively small areas of habitat. Thus the area of minimum coexistence is proportional to the area of interference and varies inversely with habitat diversity -in the last analysis no two individuals or species can occupy precisely the same point and a t this extreme cannot coexist. The possible parallels between mechanisms of population control and of competitive interactions have received very little attention. Just as it is impossible to account for natural control of all populations with one general theory based on stress, food or climate, we cannot expect) that the mechanisms of interference or exploitation will be the same in every animal population. It would be interesting to know whether species whose numbers tend to be controlled by climate also, as was suggested earlier, tend to have weak interference elements and compete mainly through exploitation; or, conversely, whether a strong interference component also suggests intraspecies control based on similar mechanisms. Another parallel may also exist between different kinds of isolating mechanisms and forms of interference. We know, for example, that birds have evolved strong ethological isolating mechanisms based on auditory and visual stimuli (Sibley, 1961), which also seem to be important factors in territorial behavior and competitor interactions. It would appear that the mechanisms which function in reproductive isolation may be used also as interference mechanisms, and that in each case there is a common evolutionary history. I n certain fishes, on the other hand, ethological mechanisms are poorly developed and species isolation depends more on physical factors in the habitat. The breakdown of isolation and subsequent hybridization occurs rather frequently in these species - does this also mean that interspecies interference is uncommon in such groups? Many insects (e.g. Drosophila) have evolved pre-mating, ethological mechanisms of isolation but also seem to depend on successive post-mating mechanisms for complete reproductive isolation (Mayr, 1963). Species recognition in these animals depends entirely on the adults, which suggests that this is the only stage during which ethological interference mechanisms would be effective, but if species recognition is relatively weak, interspecies interference might also be weak even among adults. There is, in other words, a possible parallel between mechanisms of population control, interspecies competition and species isolation, all of which influence the species diversity of natural communities.



REFERENCES Andrewartha, H. G. and Birch, L. C. (1953). Aust. J . 2001.1, 174-177. The Lotka-Volterra theory of interspecific competition. Andrewartha, H. G. and Birch, L. C. (1954). “The Distribution and Abundance of Animals,” 782 pp. Chicago: University of Chicago Press. Bagenal, T. B. (1951). J . Anim. Ecol. 20, 242-245. A note on the papers of Elton and Williams on the generic relations of species in small ecological communities. Bakker, K. (1961). “An Analysis of Factors which Determine Success in Competition for Food among Larvae of Drosophila melanogaster,” 281 pp. Groningen: J. B. Wolters. Beauchamp, R. S. A. and Ullyott, P. (1932). J . Ecol. 20, 200-208. Competitive relationships between certain species of fresh-water triclads. Beer, J. R., Frenzel, L. D. and Hansen, N. (1956). WiLon Bull. 68, 200-209. Minimum space requirements of some nesting passerine birds. Birch, L. C. (1945). Austral. J . Exp. Biol.a i d Med. Sci. 23, 29-35. The influence of temperature on the development of the different stages of Calandra oryzae L. and Rhizopertha dominim Fab. (Coleoptera). Birch, L. C. (1948). J . Anim. Ecol. 17, 15-26. The intrinsic rate of natural increase of an insect population. Birch, L. C. (1953). Ecology 34, 698-711. Experimental background to the study of the distribution and abundance of insects. I. The influence of temperature, moisture and food on the innate capacity of increase of three grain beetles. Birch, L. C. (1957). Amer. Nut. 91, 5-18. The meanings of competition. Birch, L. C. and Battaglia, B. (1957). Ecology 38, 165-166. The abundance of Drosophilu willistoni in relation to food in natural populations. Bray, J. R. (1958). Ecology 39, 770-776. Notes toward an ecologic theory. Brian, M. V. (1056). J . Anim. Ecol. 25, 339-347. Exploitation and interference in interspecies competition. Brown, W. L., Jr. and Wilson, E. 0. (1956). Syst. 2001.5, 49-64. Character displacement. Burkholder, P. R. (1952). Amer. Sci. 40, 601-631. Cooperation and conflict among primitive organisms. Carson, H. L. and Stalker, H. D. (1951). Ecology 32, 317-330. Natural breeding sites for some wild species of Drosophila in the eastern United States. Chiang, H. C. and Hodson, A. C. (1950). Ecol. Monogr. 20, 175-206. An analytical study of population growth in Drosophila melunogaster. Clements, F. E. and Shelford, V. E. (1939). “Bio-ecology,” 425 pp. New York: John Wiley and Sons. Cole, L. C. (1960). Science 132, 348-349. Competitive exclusion. Connell, J. H. (1961). Ecology 42, 710-723. The influence of interspecific competition and other factors on the distribution of the barnacle Chtlmmalus atellatus. Crombie, A. C. (1947). J . Anim. Ecol. 16, 44-73. Interspecific competition. Darwin, C. (1859). “On the Origin of Speciss by means of Natural Selection, or the Preservation of Favoured Races in the Struggle for Life.” London: John Murray. Dice, L. R. (1952). “Natural Communities”, 547 pp. Ann Arbor: University of Michigan Press.



Dixon, K. L. (1954). Condor 56, 113-124. Some ecological relations of Chickadees and Titmice in Central California. Dumas, P. C. (1956). Ecology 37, 484-495. The ecological relations of sympatry in Plethodon dun& and Plethodon vehiculum. Ecke, D. H. (1954). J . Mammal. 35, 521-525. An invasion of Norway rats in southwest Georgia. Elton, C. (1927). “Animal Ecology”, 209 pp. London: Sidgwick and Jackson. Elton, C. (1935). J . Anim. Ecol. 4, 127-136. A reconnaissance of woodland bird communities in Englmd and Wales. Elton, C. (1946). J . Anim. Ecol. 15, 54-68. Competition and the structure of ecological communities. Elton, C. (1958). “The Ecology of Invasions by Plants and Animals”, 181 pp. London: Methuen and Co. Elton, C. and Miller, R. S. (1954). J . Ecol. 42, 46CL496. The ecological survey of animal communities: with a practical system of classifying habitats by structural characters. Evans, F. C. (1942). J . Anim. Ecol. 11, 182-197. Studies of a small mammal population in Bagley Wood, Berkshire. Frank, P. W. (1952). Physiol. Zool.25,178-204. A laboratory study of intraspecies and interspecies competition in Daphnia pulicaria (Forbes) and Simocephalw vetulus 0. F. Muller. Gause, G. F. (1935). “La thdorie mathdmatique de la liitte pour la vie”, 61 pp. Paris: Hermann et Cie. Gause, G. F. and Witt, A. A. (1935). Amer. Nut. 69, 59&609. Behavior of mixed populations and the problem of natural selection. Gibb, J. (1956). Ibis 98, 420-429. Territory in the genus P a m . Gilbert, O., Reynoldson, T. B. and Hobart, J. (1952). J . Anim. Ecol. 21, 310-312. Gause’s hypothesis; an examination. Gordon, C. (1942). Nature 149, 610. Natural breeding sites of Drosophila obscura. Greenway, J. C. Jr. (1958). Amer. Comm. for International Wildlife Protection, Spec. Publ. N o . 13. Extinct and Vanishing Birds of the World. Grinnell, J. (1917a). Amer. Nut. 51, 115-128. Field tests of theories concerning distributional control. Grinnell, J . (1917b). Auk 34, 427-433. The niche-relationships of the California thrasher. Grinnell, J. (1924). Ecology 5, 225-229. Geography and evolution. Grinnell, J. (1928). Univ. Calif. Chron. Oct. 1928. pp. 429-450. Presence and absence of animals. Grodzinski, W. (1959). Ekologia Polska 7, 83-143. The succession of small mammals communities on an overgrown clearing and landslip mountain in the Beskid Sredni (western Carpathians). Hall, E. R. and Kelson, K. R. (1959). “The Mammals of North America”, Vol. I. 546 pp. New York: Ronald Press. Hardin, G. (1960). Science 131, 1292-1297. The competitive exclusion principle. Harrison, J. 0. (1964). Ecology 45, 508-519. Factors affecting the abundance of Lepidoptera in banana plantations. Hartley, P. H. T. (1949). Ibis 9, 393-413. The biology of the mourning chat in winter quarters. Hartley, P. H. T. (1950). Symp. SOC. E z p . Biol. No. 4, pp. 313-336. An experimental analysis of interspecific recognition.



Heatwole, H. and Davis, D. M. (1965).Ecology 46, 140-150. Ecology of three sympatric species of parasitic insects of the genus Megurhyssa (Hymenoptera: Ichneumonidae). Hinde,R.A. (1956).Ibis98,340-369.Thebiologicalsignificanceofterritoriesofbirds. Hutchinson, G.E. (1944).Ecology 25, 3-26. Limnological studies in Connecticut. VII. A critical examination of the supposed relationship between phytoplankton periodicity and chemical changes in lake waters. Hutchinson, G. E. (1948).Ann. N . Y . Acad. Sci. 50, 221-246. Circular causal systems in ecology. Hutchinson, 0. E. (1957). Cold Spr. Harb. Symp. quant. Biol. 22, 415-427. Concluding remarks. Hutchinson, G. E. (1959).Amer. Nut. 93, 145-159. Homage to Santa Rosalia, or why are there so many kinds of animals? Hutchinson, G. E. and Deevey, E. S. (1949). “Survey of Biological Progress. Vol. I”, 325 pp. New York: Academic Press. Kennerly, T. J. Jr. (1954).T e z w J . Sci. 6 , 297-329. Local differentiation in the pocket gopher (Geomys personatus) in soiithern Texas. Kennerly, T. J. Jr. (1959).Evolution 13, 247-263. Contact between the ranges of two allopatric species of pocket gophera. Kikkawa, J. (1964).J. Anim. Ecol. 33, 259-299. Movement, activity and distribution of the small rodents Clethrionomys glareolus and Apodemus sylvaticus in woodland. Kinne, 0. (1954).Interspezifische Sterilpaarung als konkurrenzoligischer Faktor bei Gammariden (Crustacea, Peracarida). Sonderbook aus die Naturwissenechaften 41, 434. Kinne, 0.(1956).tfber den Wert kombinierter Untersuchungen (im Biotop und im Zuchtersuch) fur die okologisehe Analyse. Sonderbook aus die Naturwissenschaften 43, 8-9. Klopfer, P. M. (1962).“Behavioral Aspects of Ecology”, 161 pp. Englewood Cliffs, New Jersey: Prentice-Hall. Klopfer, P. M. and MacArthur, R. H. (1960).Amer. Nut. 94, 293-300. Niche size and faunal diversity. Kohn, A. J. (1959).Ecol. Mongr. 29, 47--90.The ecology of Conus in Hawaii. Kohn, A. J. and Orians, G. H.(1962).System 2001.11, 119-126. Ecological data in the classification of closely related species. SOC. 2001. Bot. Fen. Vannma 13, 1-91. Kontkanen, P. (1950).A n n . (2001.) Quantitative and seasonal studies on the leafhopper fauna of the field stratum in open areas in North Karelia. Kostitzin, V. A. (1939).“Symhiose, parasitisme et evolution (Etude mathematique) Actualit& Scientifiques”. Paris: Hermann. Lack, D. (1944).Ibis 86, 260-286. Ecological aspects of species-formation in passerine birds. Lack, D. (1954).“The Natural Regulation of Animal Numbers”, 343 pp. Oxford: Clarendon Press. Lanyon, W. E. (1956).Ibis 98, 485489. Territory in the meadowlarks, genus Sturnella. Legg, K. and Pitelka, F. A. (1956).Condor 58, 393-405. Ecologic overlap of Allen and Anna Hummingbirds nesting a t Santa Cruz, California. Lotka, A. J. (1932).J . Wash. Acad. Sci. 22, 461-469, The growth of mixed populations, two species competing for a common food supply. MacArthur, R.H. (1957).Proc. iNat. Acad. Sci., U.S. 43, 293-295. On the relative abundance of bird species.



MacArthur, R. H. (1958). Ecology 39, 599-619. Population ecology of some warblers of northeastern coniferous forests. MacArthur, R. H. and MacArthur, J. W. (1961). Ecology 42, 594-598. On bird species diversity. Marler, P. (1956). Ibis 98, 496-501. Territory and individual distance in the Chaffinch Fringilla coelebs. Marshall, J. T. Jr. (1948). Condor 50, 233-256. Ecologic races of song sparrows in the San Francisco Bay region. Part 11. Geographic variation. Matheson, C. (1939). J . Anim. Ecol. 8, 76-93. A survey of the status of Rattus rattus and its subspecies in the seaports of Great Britain and Ireland. M a p , E. (1963). “Animal Species and Evolution”, 797 pp. Cambridge: Belknap Press of Harvard University Press. Merrell, D. J. (1951). Amer. Nut. 85, 159-169. Interspecific competition between Drosophila funebris and Drosophila melanogaster. Miller, R. S. (1955). Proc. 2001. Soc., Lond. 125,505-519. Activity rhythms of the wood mouse, Apodemus sylvaticus, and the bank vole, Clethrionomys glareolus. Miller, R. S . (1964a). Ecology 45, 256-272. Ecology and distribution of pocket gophers (Geomyidae) in Colorado. Miller, R. S. (1964b). Ecology 45, 132-148. Larval competition in Drosophila melanogaster and D . simulans. Miller, R. S. (1964~).Amer. Nut. 98, 221-238. Interspecies competition in laboratory populations of Drosophila melanogaster and Drosophila simulans. Milne, A. (1961). Symp. Ezp. Biol. No. 15. pp. 40-61. Definition of competition among animals. Mishima, J. (1964). Sci. Rep., Tokyo Kyoiku Daigaku 11, 277-301. The influence of environmental temperature upon the interspecific competition in Drosophila larvae. Mitchell, R. (1964). Ecology 45, 546-558. A study of sympatry in the water mite genus Arrenurius. Moore, J. A. (1952). Evolution 6, 407-420. Competition between Drosophila melanogaster and Drosophila simulans. I. Population cage experiments. Moore, N. W. (1964). J . Anim. Ecol. 33,49-71. Intra- and interspecific competition among dragonflies (Odonata). Moreau, R. E. (1948). J . Anim. Ecol. 17, 113-126. Ecological isolation in a rich tropical avifauna. Moreau, R. E. and Moreau, W. M. (1938). J . Anim. Ecol. 7,314-327. The comparative breeding ecology of two species of Euplectea (Bishop Birds) in Usambara. Myers, G. T. and Vaughan, T. A. (1963). Colorado Cooperdive Pocket Gopher Project Annual Progress Report, January, 1963. Food habits of the plains pocket gopher in Eastern Colorado. Nero, R. W. (1964). Wilson Bull 75,376-413. Comparative behavior of the yellowheaded blackbird, red-winged blackbird and other icterids. Neyman, J., Park, T. and Scott, E. L. (1958). Gen. Systems 3, 152-179. Struggle for existence; the Tribolium model: biological and statistical aspects. Nicholson, A. J. (1933). J . Anim. Ecol. 2, 132-178. The balance of animal populations. Nicholson, A. J. (1937). J . Council Sci. Ind. Res. (Australia)10, 101-106. The role of competition in determining animal populations. Nicholson, A. J. (1955). Austral. Jour. 2001.2, 9-65. An outline of the dynamics of animal populations. Odum, E . P. (1959). “Fundamentals of Ecology”, 546 pp. Philadelphia and London: W. B. Saunders Co.



Oriam, G. H. (1961). Ecol. Monogr. 31, 285-312. The ecology of blackbird ( A g e h i u s ) social systems. Park, T. (1948). Ecol. Monogr. 18, 265-308. Experimental studies of interspecies competition. I. Competition between populations of the flour beetles, Tribolium c o n j m m Duval and Tribolium cmtaneum Herbst. Park, T. (1954). Physiol. 2001.27, 177-238. Experimental studies of interspecies competition. 11. Temperature, humidity, and competition in two species of Tribolium. Pickles, W. (1944). J . Anim. Ecol. 13, 128-129. Territories and interrelations of two ants of the genus Measor in Algeria. Pratt, D. M. (1944). Biol. Bull. 85, 116-140. Analysis of population development in Daphnia at different temperatures. Ripley, S. D. (1961). A u k 78, 366-371. Aggressive neglect as a factor in interspecific competition in birds. Robertson, F. W. (1960). Genet. Re.?. Camb. 1, 288-304. The ecological genetics of growth in Drosophila. I. Body size and developmontal time on different diets. Lond. B., 132: 258-277. Robertson, F. W. and Sang, J. H. (1944). Proc. Roy. SOC. The ecological determinants of population growth in a Drosophila culture. I. Fecundity of adult flies. Sang, J. H. (1949). Physiol. 2001.22, 183-202. The ecological determinants of population growth in a Drosophila culture. 111. Larval and pupal survival. Sang, J. H. (1950). Bwl. Rev. Cambridge Phil. SOC.25, 188-219. Population growth in Drosophila cultures. Schwab, R. G. and Monnie, J. B. (1959). Wilson Bull. 71, 190. Strife over a nesting site between Downy and Redheaded Woodpeckers. Sheppard, D. H. (1964). “Ecology ofthe chipmunks Eutamias a m e n u s luteiventris (Allen) and E . minimus orocetea Merriam”. Ph.D. Thesis, University of Saskatchewan, 1964. Unpublished. Sibley, C. G. (1961). I n “Vertebrate Speciation” (W. F. Blair, ed.), pp. 69-88. Austin: University of Texas Press. Simmons,K. E. L. (1951). Ibis 92, 407-413. Interspecific territorialism. Simmons,K. E. L. (1956). Ibis 98, 390-397. Territory in the Little Ringed Plover, Charadrius dubiua. Simpson, G. G. (1964). System. Zool. 13,57-73. Species density of North American recent mammals. Slobodkin, L. B. (1954). Ecol. Monogr. 24,69-88. Population dynamics in Daphnia obtwa Kurz. Slobodkin, L. B. (1961). “Growth and Regulation of Animal Populations”, 184 pp. New York: Holt, Rinehart and Winston. Sluiters, J. E. (1954). Limosa 27, 71-86. Waarnemingen over de brie bij Amsterdam broedende pluviersoorten (Leucopolius a. alexandrinus, Charadriua dubius curonicus en Ch. h. hiaticula). Snow, D. W. (1949). Var Fagelwadd 8, 156-169. Jamforande studier over vara mesarters naringssokande. Snow, D. W. (1954). Ibis 96, 565-585. The habitats of Eurasian tits (Parus spp.). Snow, D. W. (1956). Ibis 98, 438-447. Territory in the blackbird Turdus nteruh. Sokoloff, A. (1955). Ecol. Monogr. 25, 387-409. Competition between sibling species of the Pseudoobscura group of Drosophila. Sokoloff, A. (1957). Cold Spr. Harb. Symp. qwznt. Biol. 22, 268-270. Discussion following paper by A. Milne: Theories of natural control of insect populations.



Solomon, M. E. (1949). J . Anim. Ecol. 18, 1-35. The natural control of animal populations. Southern, H. N. (ed.) (1964). “The Handbook of British Mammals”, 465 pp. Oxford: Blackwell Scientific Publications. Tinbergen, N. (1956). Ibis 98, 401-411. On the functions of territory in gulls. Townes, H. and Townes, M. (1960). U.S. Nut. Mus. Bull. 216, 1-676. Ichneumonflies of America north of Mexico. 2. Subfamilies Ephialtinae, Xoridinae Acaenitinae. Udvardy, M. F. D. (1952). Oikos 3. 98-123. The significance of interspecific competition in bird life. TJdvardy, M. F. D. (1959). Ecology 40, 725-728. Notes on the ecological concepts of habitat, biotope and niche. Ullyett, G. C. (1950). Phil. Trans. B : Biol. Sci. 234, 77-174. Competition for food and allied phenomena in sheep-blowfly populations. Van Allen, L. (1965). Amer. Nut. 99, 377-390. Morphological variation and width of ecological niche. Vaurie, C. (1951). Proc. X Int. O m . Congr. pp. 163-166. Adaptive differences between two sympatric species of nuthatches (Sittu). Volterra, V. (1926). Mem. R . Accad. Lincei ser. 6, 1-36. Vartatzioni e fluttuazioni del numero d’individui in specie animali conviventi. Von Haartman, L. (1956). Ibis 98, 460-475. Territory in the pied flycatcher Mwcicapa hypoleuca. Weatherly, A. H. (1963). Nature, Lond. 197, 1 P 1 7 . Notions of niche and competition among animals, with special reference to freshwater fish. Williams, C. B. (1947). J . Anim. Ecol. 17, 11-18. The generic relations of species in small ecological communities. Williams, C. B. (1951). J . Anim. Ecol. 20, 24G253. Intraggeneric competition aa illustrated by Moreau’s records of East African bird communities. Williams, C. B. (1964). “Patterns in the Balance of Nature”, 324 pp. London and New York: Academic Press. Wilson, E. 0. and Bossert, W. H. (1964). Recent Progress 19, 673-716. Chomicel communication among animals. Windsor, C. P. (1934). ColdSpr. Harb.Symp.q m n t . Biol. 2,181-187. Mathematical analysis of the growth of mixed populations. Wynne-Edwards, V. C. (1962). “Animal Dispersion in Relation to Social Beheviour”, 653 pp. Edinburgh and London: Oliver and Boyd.