H. A. GLEASON'S ‘INDIVIDUALISTIC CONCEPT’ AND THEORY OF ANIMAL COMMUNITIES: A CONTINUING CONTROVERSYH. A. GLEASON’S ‘INDIVIDUALISTIC CONCEPT’ AND THEORY OF ANIMAL COMMUNITIES: A CONTINUING
CONTROVERSY
BY ROBERT P. McINTOSH
Department of Biological Sciences, University of Notre Dame, Notre Dame, I N , U S A
(Received I I February 1993 ; revised 16 May 1994; accepted 3 0 August)
CONTENTS
I. Traditions of community concept . . . . . . . . . . 11. Animal community studies from the 1950s to the 1970s and recognition of Gleason’s
concept . . . . . . . . . . . . . . . . . . . . . . . . .
( I ) Evolution and community theory . . . . . . . . . (2) Individualistic concept revisited . . . . . . . . . (3) Definition of community . . . . . . . . . . . (4) Key questions for community ecology . . . . . . . . ( 5 ) Empirical studies of communities . . . . . . . . .
IV. Community theory and questions for the 1990s . . . . . . .
VI. Acknowledgments . . . . . . . . . . . . . VII. References . . . . . . . . . . . . . .
V. Summary . . . . . . . . . . . . . .
I . TRADITIONS OF COMMUNITY CONCEPT
In the early decades of ecology in the twentieth century most ecologists took for granted that communities existed as natural, repeated, internally organized units with a considerable degree of integration which governed their structure, function, development or succession and even their evolution (McIntosh, 1985). Communities were likened to an individual organism and commonly called a superorganism (Clements, 1936) or quasi-organism (Tansley, 1935). Nicholson (1990) noted that not all ecologists accepted the extreme views of Frederic Clements, whose ideas about community dominated American ecology in its early decades, but only H. A. Gleason (1917, 1926, 1939) in three elaborations of his ‘individualistic concept ’, openly attacked them and offered an alternative concept predicated on the individualistic capabilities of species, continuous variation of the environment and diverse probabilities of arrival of propagules. Although several American ecologists had expressed similar views, none publicly supported Gleason’s heretical concept until 1947 (McIntosh, 1975, I 985). Gleason’s concept provided an alternative paradigm and the potential, at least, for revolution in the Kuhnian sense (Price, 1984b). Gleason’s (1926) version of the individualistic concept was recently recognized as one of the ‘classic papers in the foundations of ecology’ (Real & Brown, 1991).
Early marine animal ecologists such as Edward Forbes, Victor Hensen and C. G. J. Petersen had independently developed concepts of marine plankton and benthic communities largely in the traditional paradigm of discrete communities. Terrestrial
R. P. MCINTOSH and freshwater animal ecologists, notably E. A. Birge, S. A. Forbes, Victor Shelford, C. C. Adams and Charles Elton, undertook surveys of terrestrial and freshwater animal communities (McIntosh, 1985). Forbes (1880) described ‘the ideal balance of nature’ and a ‘tendency towards a just equilibrium’ in animal communities. Elton (1927) wrote, in his pioneer volume, Animal Ecology, that animals were not ‘mere assemblages’ but formed ‘closely knit communities or societies’. However, he later attacked the widely used clockwork simile for communities by noting that the animal ‘wheels’ retain the right to move to another clock and each clock has its own mainspring (Elton, 1930) ascribing to them considerable independence. Shelford eventually subscribed to Clements’s ideas and joined with him to produce a catalogue of plant and animal communities or biomes (Clements & Shelford, 1939). In 1939 a symposium was held designed to bring together plant and animal community ecologists that, however, had little influence on the already established barriers between them (Just, 1939). The commentators at this symposium raised many questions evident in recent discussions of community ecology. Both of these volumes were criticized by reviewers for the absence of the statistical and mathematical analyses which became the hallmark of later studies of animal community ecology (McIntosh, I 985).
Animal ecologists in the early decades of ecology emphasized studies of individual populations or population interactions with notable exceptions (Elton, I 927, I 930, Cole, 1946; Renkonen, 1949). Elton (Elton & Miller, 1954), for example, noted his diversion from community studies to studies of population fluctuations for 20 years after 1923 with community on the back burner. Wiens (1989) described David Lack as switching from community to population studies in the 1940s and commented that there were few studies of bird communities before the 1950s. Grant & Schluter (1984), however, dated consideration of community structure and the role of biotic interactions from Lack’s work with Darwin’s finches in the 1940s. In 1944 the British Ecological Society addressed ‘The ecology of closely allied species’ in a symposium and Elton and Lack agreed that competition was an important influence on community composition ; but others argued that the absence of equilibrium conditions precluded competition having a major influence on community composition (McIntosh, I 985). However, these inferences about competition and community were not extensively developed until the I ~ ~ O S , when Robert MacArthur introduced his ideas of community, and they only really got going in the 1960s according to Brown & Bowers (1984). Minshall (1988) identified an ‘ era of refinement and experimentation ’ in stream communities also beginning in the 1960s. Heins & Matthews (1987) dated widespread interest in stream community dynamics in North America from 1975. Esch et al. (199ob) dated quantitative approaches to helminth communities from the early 1960s along with the roots of ecological parasitology.
Although Grinnell’s early ideas of niche were seen by Wiens (1989) as paralleling Gleason’s individualistic concept, controversy about the community as a Clementsian integrated or organismic entity as against Gleason’s ( I 926) concept of community as ‘not an organism, scarcely even a vegetational unit, but merely a coincidence’ was, through the I ~ ~ O S , largely the province of plant ecologists. Some marine ecologists, however, had addressed many of the same issues and arrived at concepts similar to Gleason’s (Stephen, 1933; MacGinitie, 1939; Mills, 1969). Jones (1950) was more equivocal suggesting the fauna could be divided into communities on ‘ more-or-less
Gleason’s ‘ individualistic concept ’ 3 I 9
definite limits to the physical conditions’. He denied, however, that any assemblage of animals reacted as a unit or was structured by biological factors such as competition. Wieser (1958) recognized two approaches to marine communities - the ‘particulate’, or classificatory and the ‘ comparative ’ which, like the individualistic concept, held that a natural classification of communities was not possible.
It is commonly recognized that Gleason’s individualistic concept was a primary stimulus to the development, in the I ~ ~ O S , of the ideas of continuum and gradient applied to community in the vegetation studies of J. T. Curtis, Robert Whittaker and their students and associates (Curtis & McIntosh, 1951 ; Curtis, 1959; McIntosh, 1958, 1967, 1983, 1985, 1993; Whittaker, 1967). Their studies brought Gleason’s ideas out of limbo and added a new dimension (literally multidimensions) to the discussion of community. Gleason’s individualistic concept, and the issues it addressed, were introduced into general textbooks of ecology in the 1950s; but they were not immediately widely incorporated into animal community ecology (McIntosh, I 975, 1985). The midcentury works of Allee et al. (1949) and Dice (1952) on animal ecology were unusual in citing Gleason’s work but both accepted that communities were natural aggregations of interdependent species. Dice wrote :
The concept that most communities are composed of more or less independent individual organisms which are interrelated and to some extent coordinated so that the whole community forms a unit of organization agrees well with what is known of ecology.
Elton and Miller (1954) traced the development of ideas of the animal community and noted a change of philosophy, essentially to a community view of ecology in the recognition that the importance of biotic relations in influencing organisms required knowledge of species associations to understand the population ecology of single species. Most to the point of this discussion, they recognized Gleason’s individualistic concept and the problems it posed for the search for uniform habitats and comparison of communities. British animal ecologists, they said, found that only very small ‘centres of action’ or microhabitats, such as dung and logs, fitted their concepts of uniformity and achieved ‘broadly repeated identity’. However, the tradition in most animal ecology of the community as a highly integrated entity was recognized by Bodenheimer (1957) who, however, denied there was any evidence to support the organismic concept of community.
11. ANIMAL COMMUNITY STUDIES FROM THE 1950s TO THE 1970s AND RECOGNITION OF
Gleason’s concept, and the derivative ideas and methods of continuum and gradient in plant ecology, reviewed by McIntosh (1967,1993) and Whittaker (1967), filtered into the literature of terrestrial animal ecology. Richardson (I 980) wrote, ‘ In America the individualistic concept has been in the ascendancy for at least 30 years and there is now widespread acknowledgement that Clements pressed his organismic analogy too far ’. Perhaps the earliest specific support for it in an animal community was Whittaker’s ( I 952) study of insect communities in the Great Smoky Mountains.
Among distributional patterns charted for IOO species no two were alike; and it was not possible to fit the distributions into associations, formations or zones. Gleason’s individualistic hypothesis is asserted for animal communities.
GLEASON’S CONCEPT
3 20 R. P. MCINTOSH Some marine biologists adopted a continuum concept. Sanders ( I 960), for example, described the benthic fauna of Buzzard’s Bay as a ‘continuum’ changing with gradual change in composition of sediment. Bond (1957) and Beals (1960) adapted ordination methods, developed in plant ecology, to studies of bird communities. Beals wrote, ‘ Because different bird species seldom if ever coincide in their ecological distributions, no discrete communities can be clearly defined except where there are sharp changes of environment ’. Larimore and Smith ( I 963) asserted species individuality in associations of stream fishes.
This lack of similarity or consistency in associations seems to suggest little interdependence between species but rather dependence of certain species on certain ecological factors.
Other animal ecologists, notably invertebrate ecologists, independently pursued diverse interests in community organization (Cole, 1946; Fager, 1957, 1968; Hairston et al., 1960; Macfadyen, 1963; Root, 1973). Cole commented that cryptozoan communities under debris on the forest floor lacked the interspecific integration of a biocoenose. Fager (1957) addressed the classical question of community ecology - whether ‘recurrent’ groups of species occurred that are a nearly constant part of each others environment? and answered in the negative. Fager (1968), even in studies of invertebrates in small decaying logs, failed to find the ‘centres of action’ or the uniformity noted by Elton and Miller (1954). Using his method of ‘recurring groups’ he found much variability of invasion and re-establishment suggesting that each sample was ‘ individualistic in the sense of Gleason ( I 939) ’.
Macfadyen (1963), in an extended review of community definitions in a current textbook of animal ecology, illustrated how varied were ideas about animal community. Root (1973), in studies of the arthropod fauna of collards, found no evidence that the guild structure remained relatively constant and summarized :
The dynamic structure of the collard fauna illustrates that the concept of communities as associations with a characteristic composition, or trophic structure can be somewhat misleading. Wilbur (1972), however, had urged that ‘classical models’ of a unit organized community not be discarded, at least for amphibian communities. He wrote,
My thesis is that these natural groupings of species are organized by interspecific interactions that can be discovered and evaluated by experimentally dissecting the community into smaller components.
Some students of bird communities adopted methods similar to those of plant community ecologists and recognized the relevance of contemporary studies of plant communities influenced by Gleasonian concepts but marched to a different drummer of a widely hailed new theory of animal communities. Cody ( I 968) studied grassland bird communities and showed a pattern of individualistic species distribution (his Figs. 8 and 9) remarkably similar to that illustrated by Bond (1957) in a study of forest birds. Cody emphasized the then burgeoning theory of niche initiated by G. E. Hutchinson and Robert MacArthur and interpreted the grassland bird community on ideas of resource division, based on a community structured by competition and climatic instability controlling niche size in a saturated community. Kikkawa ( I 968) interpreted his studies of bird communities of Australian forests as ‘unique associations’ of bird species in different rain forest habitats of North Queensland contrasted with a
Gleason’s ‘ individualistic concept ’ 321
distribution ‘hierarchically arranged along the mesic-xeric gradient ’ in forests of New South Wales.
The increased interest in communities and their abstract properties, diversity and stability, gave rise to a symposium in I 969 that addressed the problems of defining and measuring what all agreed were critical attributes of community (Brookhaven Symposia in Biology 1969). It was commonly assumed at that time that increased diversity was associated with stability until later theoretical and empirical studies demolished this tradition of natural history, ecology and conservation (Goodman, I 9-75), although diversity persists as a key concept.
James (1971) used ordination methods to study bird communities, contrasting her approach to that of Cody (1968). James specifically stated that her study was conceptually related to the individualistic concept of Gleason and the derived continuum concept of J. T. Curtis, citing the prior work of Bond (1957) and Beals ( I 960). Terborgh ( I 97 I) also recognized current work of plant ecologists and offered an approach to the study of avian distributions similar to that pioneered by R. H. Whittaker (1952). Terborgh worked on an elevational gradient of 3000 m in the Andes Mountains with four physiognomically distinct vegetation types. He examined the distribution of species relative to three models based on distributional limits of species and the factors that determine their limits. Criteria for judging the predictions of the models were (a) shape of population curves, (b) shape of the curves of faunal attenuation - ‘congruity’ or similarity of adjacent samples along the gradient, (c) species distribution patterns at termini of gradients, (d) frequency distribution of ecological amplitudes of species. Faunal turnover rate was, he said, continuous with elevation, even in the context of several distinctive vegetation types, and ecotones accounted for < 20 % of distributional limits. Circumstantial evidence of competitive replacement of congeners based on narrow or abruptly terminated species amplitude curves, was interpreted as accounting for ca. 33 % of species limits. Terborgh attributed approximately 50 % of species limits to gradually changing conditions along the gradient. Distribution and shape of curves of species populations on diverse gradient are still widely used as criteria of community.
Students of aquatic organisms of various taxa in marine and fresh waters shared similar concerns about the nature of community. Mills (I 969) and Stephenson et al. ( I 972) reviewed marine community concepts with specific reference to Gleason’s ideas and subsequent studies in plant community ecology. Mills noted then current marine benthic studies that allowed ‘analysis as parts of continua of distribution along gradients’. Stephenson, and other marine biologists, joined with W. T. Williams, a statistician, to reanalyze the data of a pioneer student of marine benthic communities, C. G. J. Petersen. They expected to confirm the communities that Petersen had recognized by subjective analysis of dominant species. Their analyses revealed some ‘ Petersen-type ’ communities but these differed markedly from Peterson’s original results. Their findings revealed no groups of species with similar ecological requirements. Bourdouresque ( I 970) explicitly addressed the concepts of biocoenose and continuum and found continuity between organismic and continuum ideas in benthic taxa. Levandowsky ( I 972) used ordination techniques on phytoplankton of ponds varying in salinity and supported an individualistic hypothesis of species distribution. Sale and Dybdahl(1975) asserted that coral reef fish communities were the
322 R. P. MCINTOSH result of ‘purely chance colonization of species which do not interact with each other sufficiently to shape the community’. Briand (1976) cited the concepts of Clements and Gleason in a study of marine phytoplankton and found, ambiguously, that ‘three groups of species appear closely defined within the yearly phytoplankton community ’ but there was relative continuity among them because of linking species.
Makarewicz & Likens (1975) analysed niche relations of a zooplankton community and concluded that their findings paralleled those of Gleason and Ramensky :
The niche structure of the zooplankton community should be conceived of as an intensive or intracommunity population continuum corresponding to niche hyperspace.
Lane ( I 978) vigorously attacked them for ‘ forcing their data to fit the individualistic concept that was originally developed for terrestrial plants’. According to Lane, such reductionist approaches are of limited use for animal communities. Levandowsky and White (1977), perhaps prematurely, wrote ‘the battle between the community school and the individualistic or continuum school seems to have been fought and won by the latter in the 1950s’. In an unusually extreme interpretation, Taylor (1979), in a study of bactivorous ciliates in a pond, described independent, essentially random, species distributions. Sousa ( I 979) also recognized the Clements/Gleason division and, in studies of rocky intertidal communities, supported a non-equilibrium, disturbance- based view akin to Gleason’s ideas.
Animal community ecology in the 1960s and 1970s was, however, largely dominated by proponents of an expanded theoretical population ecology who, following Hutchinson (1957), centered their ideas on competition as the major influence in community organization (MacArthur, I 958, 1972; Lewontin, 1969 ; Schoener, 1974 ; Cody, 1974 & Diamond, 1975; Diamond, 1975, 1978; May 19764 b). Hutchinson’s (1957) ‘Concluding Remarks’, in which his theory of the multidimensional species niche appeared, stimulated a profusion of empirical and theoretical studies of communities under the umbrella of niche theory with proliferation of collateral ideas - resource partitioning, character displacement, species packing, niche shifts, ‘ etc I. Schoener (1974) reported that in the 1960s and early 1970s studies of resource partitioning ‘have grown exponentially at a rate four times that of typical scientific works’. Lewontin (1969) described this facet of ecology as transforming Clements’s idea of succession by a union of mathematics and evolution to produce ‘the bare beginning of an exact theory of the evolution of communities of organisms’. The new framework was, he said, ‘the concept of the vector field in n-dimensional space’. According to Levin (1988), the mathematical theory of animal ecologists emerged from the Clementsian approach and ‘ emphasized equilibrium, constancy, homogeneity, and predictability’. These are the key elements of the ‘New ecology’ that Colwell (1985) said dominated animal community ecology of the I 960s and I 970s, with little reference to Gleason’s concept.
The ‘bare beginnings’ of Lewontin (1969) grew into ‘new paradigms’. According to Cody and Diamond ( I 975) :
Within two decades new paradigms had transformed large areas of ecology into a structured, predictive science that combined powerful quantitative theories with the recognition of widespread patterns in nature.
Much of this theory was first justified, and later criticized, based on studies of bird
Gleason’s ‘ individualistic concept ’ 323 communities. Stearns (1982), in an essay on 25 years of progress in community biology, asserted ‘it was birds that inspired the idea that communities are static systems in competitive equilibrium in the work of MacArthur (1958)’. This comment seems justified, in that the major volume on animal community ecology published in the mid- I ~ ~ O S , celebrating two decades of the new theory associated with the work of Robert MacArthur (Cody & Diamond, 1975), had birds cited on 216 pages, mammals on 35, lizards on 28, insects on 6 and fish not at all. According to Grant (1986) this volume ‘marked the final flourish of an era of unbridled enthusiasm for the notion that interspecific competition is important in conferring structure upon communities of organisms’. The proportions of citations were more equal a decade later in the volume in which Grant’s article appeared. Birds were cited on 89 pages, fish on 63, lizards on 62, insects on 47 and mammals on 40 pages (Diamond & Case, 1986). Wiens (1986) described animal community ecology from 1960 to 1975 as follows:
it seemed clear that many ecologists believed that natural systems were at or close to equilibria determined by resource limitation, that selection on resource-utilization traits was more or less incessant, and that competition between species was the primary component of this selection and therefore played the major role in structuring communities.
The similarity between early plant ecology and the theoretical animal community ecology of the 1960s and 70s was noted by Law & Watkinson (I 988) :
From this blend of field study and theory there emerged a picture of communities neatly ordered by competition into groups of compatible species (Diamond, 1978), not altogether dissimilar to that suggested by plant ecologists of earlier generations.
These ideas initially enjoyed widespread and enthusiastic acceptance but soon provoked discussion that was later described in a commentary in Science (Lewin, 1983) as, ‘a debate as acerbic and acrimonious as any that has stirred the combative instincts of academia’. The crux of the more heated discussions revolved around the idea that species distributions, particularly of birds, were determined by interspecific com- petition. This had culminated in the assertion that bird communities were organized according to ‘assembly rules’ delineated by Diamond (1975). The concept of rule- governed community organization was vigorously and, if one followed the statistics, rigorously disputed by Connor & Simberloff (1979). Caswell (1976) had introduced a null model of community structure that examined distribution of species abundances assuming no interspecific interactions among them. The core of the later debate was the necessity of distinguishing apparent pattern in community organization from random events. Connor & Simberloff argued that the concept of a community organized by assembly rules as a consequence of competition was false and Diamond’s assembly rules were ‘tautological, trivial or a pattern expected were species distributed at random ’. They urged the need of a null hypothesis of random species assemblies to be tested by appropriate statistics before a pattern of organized communities based on competition could be sustained. Schoener (1982) noted an additional pole of the dispute, that favoured predation as the structuring force as suggested by the work of Paine (I 966) and Connell (1975) in marine littoral communities.
May (1976) suggested, like others before and since, abandoning consideration of species to focus on more general aspects of community organization such as energy flow, trophic groups, body size, species number or diversity which would, he said,
324 R. P. MCINTOSH reduce the ‘chaotic and vagarious’ level of individual species to ‘constant and predictable ’ properties of community organization. Transferring the long familiar object of community research from the species and number of individuals to such generalized categories recalls Robert Frost’s saying that writing poetry in blank verse was like playing tennis without a net. Certainly the evolutionary connection is lost.
The diverse interpretations of the results of animal community studies during the I 960s and 1970s recalls Macan’s ( I 974) question about community concept :
How far is a community an entity any alterations to a part of which must affect the whole, and how far is it a collection of independent species ? Results so far indicate that it is not an entity to the extent that some have suggested.
The relatively spasmodic attention to animal communities evident in the infrequent earlier symposia and texts gave way to a rush of symposia and attendant volumes in the 1980s and 1990s (Barnes & Minshall, 1983a; Strong et al., 1984b; Price et al., 1984; Kikkawa & Anderson, 1986; Diamond & Case, 1986a, Gee & Giller, 1987; Gray et al., 1987; Matthews & Heins, 1987; Grubb & Whittaker, 1988; Hastings, 1988; Morris, et al., 1989; Polis, 1991 ; Ricklefs & Schluter, 1993). The thread of the traditional dispute concerning the organized community and Gleason’s individualistic concept is woven through this literature. I hope to follow this thread and examine, if not untangle, some of the knots, Gordian or otherwise. Some of the dispute was complicated by frequent misstatements about Gleason’s concept and belaboring straw men set up in place of Gleason’s ideas. Although Gleason emphasized chance arrival of immigrants, he did not assert that a community was a random assemblage of species selected solely by their response to the physical environment of a site. Gleason recognized patterns, each species responding individualistically to the complex of biotic and abiotic forces. He asserted that early arrival and control of a site by a species could inhibit the success of later arrivals as competition did restrict a new species once the ground was fully occupied (McIntosh, 1975).
111. ANIMAL COMMUNITIES IN THE 1980s
The controversy of the I 970s about animal communities precipitated by theoretical population ecology and its attendant ideas of niche, resource partitioning, competitive exclusion, coexistence, character displacement, saturation and equilibrium continued in the 1980s amid pleas for tolerance and pluralism in ecology (Schoener, 1982, 1987; McIntosh, 1987). Animal community ecology in the 1980’s was variously described as experiencing, ‘disappointment’ (Brown, 1981), a ‘ dream world’ (Price, 1984b), ‘growing disquietude’ (Wiens, 1986), ‘fierce counterattack’ (Schoener, 1986d), ‘morbid self-evaluation’ (Cody, 1987), and being in ‘a state of flux’ (Kareiva & Anderson, 1988) or ‘ in disarray’ (Morin, 1989). Nevertheless, extended empirical studies of communities and symposia on communities proliferated, theoretical models were constructed and deconstructed and speculation about the nature of community continued. Richardson ( I 980) reviewed the familiar dichotomy between Frederic Clements’s organismic community and H. A. Gleason’s individualistic concept of community. According to Richardson, the re-emergence in the I 970s of organismic ideas of community, that were then waning among plant ecologists, was attributable to animal ecologists who were emphasizing the importance of biotic controls on the
Gleason’s ‘ individualistic concept’ 325 structure of the community. Richardson said that animal communities were more conducive to organismic notions than plants and both concepts were acceptable since communities in stable environments are more organismic because, ‘ species evolve toward accommodation with others in the enduring species complex, ’ whereas successional communities, in contrast, are ephemeral and individualistic. Collins et al. ( I 982) expressed doubt about Richardson’s statement that animals are more likely than plants to occur in tightly co-evolved organismic communities. Any lagging confidence in the much criticized superorganism was renewed by Moore ( I 983) who described the ‘revival of the organismal heresy’. Wilson & Sober (1988) gave added evidence of the resilience of the superorganism writing :
Imposing consistency clearly shows that groups and communities can be organisms in the same sense that individuals are. Furthermore, superorganisms are more than just a theoretical possibility and actually exist in nature.
A landmark in changing ideas of animal ecology was a symposium held in I 98 I . The organizers of the symposium (Strong, et al . , 19843) expected to find:
fairly commonly.. . the community with so few strong interactions that organization arises primarily from mutually independent autecological processes rather than from synecological ones.
Since some of the participants were among the major promulgators of the idea of community as organized by interspecific interactions, especially competition, significant differences of opinion concerning the evidence, appropriate experiments, theory and methods of data analysis were to be expected (Diamond & Gilpin, 1982). Various participants came down on either side of the controversy. Gilpin & Diamond (1984), reiterated Diamond’s ‘ assembly rules ’ for bird communities and stoutly defended these against the requirement that a null hypothesis of random aggregations be tested before a pattern in a community could be asserted. Their position was rebutted by Connor & Simberloff (1984) followed by reciprocal ‘rejoinders’ (Gilpin et al., 1984). Gilpin and Diamond’s arguments were criticized as ‘obfuscatory goo ” by Connor & Simberloff. Some may have judged the exchange of goo a draw but Gilpin subsequently, if unilaterally, declared victory, asserting that he and Diamond had shown that Connor and Simberloff s ‘statistical observations were invalid and that a proper statistical analysis confirmed the reality of permitted and forbidden communities according to Diamond’s assembly rules (Gilpin et al., 1986). T. Underwood (1986), however, reviewed this controversy and found the arguments of Simberloff and co-workers cogent and that the rebuttals of Diamond & Gilpin ‘make no concrete progress’ on the underlying issue of competition.
( I ) Evolution and community theory The idea of evolution of species producing an organized community was implicit in
ecological thought long before coevolution became a familiar term. It became a matter of considerable debate in the 1970s and 1980s. Oddly, the originators of the organismic- individualist community dispute did not advance Darwinian evolutionary consider- ations. Clements was essentially neo-Lamarkian, and Wilson ( I 983) commented that Gleason never used the evolutionary argument which, he said, is an obvious part of
3 26 R. P. MCINTOSH community ecology today. Commentators on community theory and evolution, however, commonly alluded to its Gleasonian interpretations and coevolution was held by some to be a bulwark of the theory of the organized community. Advocates of competition based theory argued that if competition could not be demonstrated in the present it must have occurred in the evolutionary past (Rosenzwieg, 1979), which Connell ( I 980) termed ‘the ghost of competition past ’. The ghost haunted some animal community ecology and was joined by ghosts of predation past and theory of community, sometimes touted as ahistorical, was reduced to appealing to spooky historical events to explain its presumed patterns. In spite of these difficulties, Rosenzweig held out hope for theoretical ecologists writing that Einstein ‘might have congratulated us for organizing a bewildering chaos into the rich rococo tapestry that ecological science is becoming ’.
Connell (1980) contrasted the two schools of community thought. One, he wrote, denied the likelihood of coevolution and competition as a major force in structuring communities and was led by Gleason, and his contemporary, the Russian botanist, Ramensky, who had advanced the individualistic hypothesis independently (Mclntosh, 1983). In this corner Connell included the animal ecologists H. G. Andrewartha, L. C. Birch 8z J. Wiens. In the opposite corner of supporters of the organized community he located the classical plant ecologists F. E. Clements, a similarly classical animal ecologist, A. J. Nicholson, and the more recent participants, R. H. MacArthur, M. L. Cody 8z J. Diamond, who supported a new theory of an organized community, predicated on competition. Connell asserted that until adequate evidence was produced, ‘ I will no longer be persuaded by such invoking of the Ghost of Competition Past’. Gilpin 8z Diamond ( I 984) later sounded a more moderate tone asserting that the school of ecology they represented had not claimed that competition was ‘unusually dominant’, which may have surprised some of their critics.
Stearns ( I 982) asserted that theoretical predictions in evolutionary ecology stood up to testing although theoretical models propounded for community ecology were shown to be false. He noted the increasing recognition among ecologists that ‘the world is an unpredictable and contingent place’ and said that the hoped-for order must be on a larger scale and a product of more complex processes than previously thought. In Stearn’s vision, field experimentation was the key to community ecology. Gilpin et al. ( I 986) argued, conversely, that field studies of community suffered ‘ crippling disadvantages ’ and that it was necessary to understand laboratory communities before field communities could be comprehended.
Walter et al. (1984) wrote that ‘The concept of an “organized” community, particularly if it carries evolutionary connotations, is unrealistic. ’ Schoener ( I 984b) argued, to the contrary, that competition was a significant component of evolution and community organization. Colwell ( I 985), considering ‘The Evolution of Ecology’ itself, introduced conceptual and philosophical issues of history and mechanism asserting that Jacques Monod’s ‘ Chance and Necessity ’ ( I 97 I ) ‘ have taken an increasingly central role in both ecological and evolutionary thinking ’. According to Colwell a shift had occurred in ecology ‘toward a contemporary version of Gleason’s (1939) idea that natural communities are composed of ‘individualistic species’. He concurred with the press by Connor and Simberloff for null models in ecology by asserting that the individualist concept ‘is currently regarded by many ecologists as a
Gleason’s ‘ individualistic concept’ 3 27 kind of ‘ null model ’ for community organization - the case to be disproved with sound evidence’ - specifically indicating that lack of such evidence had been at the heart of heated debates in the 1970s and 1980s.
Futuyma ( I 986) considered evolution and coevolution in communities and maintained that in situ evolution in a local community accounts for little because species have not been associated for very long. Community structure, he said, ‘is largely a consequence of ecological sorting among species that evolved their properties in a variety of other arenas’. Diamond (1986) returned to the ‘ghost hypothesis’ and provided a test based on what he has termed a ‘natural experiment’ in Montane birds of New Guinea. He discredited Connell’s ( I 980) anti-ghost argument that he said was a consequence of a logical fallacy. According to Diamond, the ghost hypothesis, that competing sympatric species evolved to reduce competition between them, was true for most montane birds in New Guinea. Wilson & Sober (1989), contrary to other evolutionists such as John Maynard Smith, asserted that the properties of organism, including evolution, can be extended to include superorganisms and predict where they do or do not exist. Although evolutionary ecology flourished in the 1980s, the significance of evolution to community organization remains problematic.
( 2 ) Individualistic concept revisited Citations of Gleason’s articles propounding the individualistic concept increased
five-fold from 1965 to 1990 and an increasing proportion of citations was by animal ecologists who strongly supported it (Price, I 984a; Schoener, 1 9 8 6 ~ ) . Strong (1983) posed alternative forces in community structure to counter the then prevalent theory of community with ‘a singular emphasis on competition ’ as explained by Roughgarden ( I 983). Strong (1986) cast the individualistic concept of Gleason as an alternative to classical superorganism theory and ‘ to another extreme, to excessive competitionism ’. He proposed a concept of ‘ density-vague ’ population dynamics with varying intensity of interspecific competition as ‘ akin to the individualistic view of species in communities ’. Brown ( I 984) examined the relationship between abundance and distribution of species and argued that the prevalence of relatively rare species ‘ appears to support Gleason’s ( I 926) classical “ individualistic ” concept of species distribution and community organization ’. Price ( I 984a) was most explicit :
The most important factor in assembly of specialists appears to be the individualistic response of species to the display of resources in a habitat. This view mirrors that held by phytosociologists from Ramensky (1924) and Gleason (1926) to Whittaker (1967).
Price was correct in asserting that Gleason’s concept referred to more than response of species to abiotic conditions and included the ‘display of resources’. It is incorrect to pose Gleason’s concept as purely based on response of species to abiotic forces as against an alternative theory based on response to biotic forces, as some have done (Pound, 1988; Simberloff & Dyan, 1991). Gleason is also sometimes interpreted as asserting that an association or community is a purely random collection of the available species, perhaps stemming from later association with Connor and Simberloff s null model. Communities, Gleason said, occurred as a consequence of environmental variation that was continuous and fluctuating and the variable and fortuitous immigration of plants. He wrote that the community ‘is merely one minute part of a
3 28 R. P. MCINTOSH vast and ever-changing kaleidoscope of vegetation, a part which is restricted in its size, limited in its duration, never duplicated except in its present immediate vicinity and there only as a coincidence, and rarely if ever repeated’ (Gleason, 1939). His successors in the continuum concept similarly argued that not all things were possible, only some, and though chance was important, communities were not random aggregations of species (Curtis & McIntosh, 195 I). Schoener (19866), cited the long-standing debate between proponents of Clementsian ideas and Gleason’s concept, and, correctly, said that null models propounded by Connor & Simberloff, ‘The Florida School’, had an individualistic view more extreme than Gleason’s.
(3) Definition of community One of the difficulties of the ‘new’ community ecology’, as of the ‘old’, was
succinctly stated by Giller & Gee ( I 987) :
Community ecology may be unique amongst the branches of science in lacking a consensus definition of the entity with which it is principally concerned.
Many ecologists reverted to Elton’s ( I 927) usage and preferred to speak of ‘ assemblage ’. Simberloff (1990) asserted that the switch to assemblage was due to the notion that community had come to denote integration and order almost as great as that depicted in the holistic, superorganismic community of Clements and his associates ’. The new animal community theory of the 1960s and 1970s had revived the notion of a highly organized community. A common alternative was that an assemblage was a group of co- occurring organisms without interspecific interactions, hence lacking organization or structure (Connell, I 975). Assemblage was particularly favoured by students of fishes and insects, although many animal ecologists used it, not always consistently. Cody & Diamond (1975) had ‘ species assembled non-randomly into communities ’ and the ‘fine structure of such assemblages ’ determined by physical and biological environment. Dayton ( I 984) referred to both ‘ very loosely organized assemblages ’ and ‘ well organized, self-regulating assemblages ’. Grant & Schluter (1984) similarly refer to ‘ structured ’ assemblages of finch species. Gee & Giller ( I 987) edited a volume entitled Organization of Communities with a subhead on Spatial and Temporal Organization in Contemporary Communities under which were categories of Terrestrial, Microbial, Decomposer and Aquatic ‘ assemblages ’ most of which included articles about ‘communities’. O’Connor (1987) had alternatives of ‘communities are random assemblages ’, or ‘ communities are highly self-organized entities’. Assemblage by some definitions, avoids any intimation of interaction among a group of organisms. It begs the question, however, at what level of interaction have the ‘rules of organization’, if any, transformed a ‘mere assemblage’ into a ‘closely knit’ or ‘structured ’ community ? The structured assemblage ’ and ‘random community’ became oxymorons by some usages.
There is little merit in providing separate terms for a new dichotomy in ecology for groups of interactive and non-interactive species. The distinction has already gone the route of other dichotomies in ecology, such as r and K species, in being described as a continuum from interactive species to non-interactive species (Cornell & Lawton, 1992). Community has covered the gamut for generations of ecologists. If one insists that interactions be thoroughly understood before recognizing a community there will
Gleason’s ‘ individualistic concept ’ 329 be very few communities. Ecologists would be hard pressed to identify a site demonstrably lacking in interspecific interactions to qualify as an assemblage, sensu stricto.
Community has commonly served as a catchall for any group of organisms with terms such as association, guild or biome applied to sometimes murky ideas about aggregations of organisms. The influx of animal ecologists into community ecology in the last several decades added new nuances to the old meanings of community but often they worked familiar ground. One thing the new community ecology shares with the old is the lack of a ‘consensus definition’, as Giller & Gee noted. This lack of consensus disappointed Pimm (1984) who had hoped for better things following MacFadyen’s (1963) extended review of definitions. All that came of it was, he said, the addition of a qualifier e.g. ‘bird community ’ limiting community to ‘this newer definition ’ of organisms on the same trophic level. Such communities were simply ‘horizontal ’, whereas Pimm preferred to consider the ‘vertical’ property of community that extended across two or more trophic levels appropriate to analysis of food webs. Colwell (1984) described the traditional definition of community as ‘ no more than a set of interacting or potentially interacting populations that coexist in a habitat’. Strong et al. (1984b) also offered a somewhat equivocal definition : ‘ Ecological communities are groups of species living closely enough together for the potential of interaction’. The awkward words are ‘closely’ and ‘potential ’. Wiens (1984b) added some complicating factors in asserting : ‘Aggregates of populations and underlying resource systems occurring together in an area over some period of time make up communities.’
Diamond & Case (19863) provided a ‘flexible’, some would say ‘vague’ definition of community: ‘the populations of some or all species coexisting at a site or in a region’. Later in the same volume, Roughgarden & Diamond (1986) considered communities and recognized as a ‘unifying theme ’ ‘limited membership ’ thus reverting back to earlier ideas of restricted composition. Schoener (19863) described the ‘most general ’ definition of community as ‘a set of species populations that occur in one place’ and said it is commonly nearly equivalent to ‘guild’, a term with its own ambiguities (Simberloff & Dyan, 1991).
The nature of community as including horizontal and vertical interactions, within and between trophic levels, was reiterated by Kitching (1986) who reasserted the view that a community is a ‘coevolved whole’ continuing the organismic tradition. Yodzis (1986) ambitiously defined community ‘to mean the set of all living things at some given location’. He noted the tradition of studying smaller components of the total community and suggested that too little has been done to show that such studies make sense; although such studies have ‘certainly shaped the myths and metaphors in terms of which we currently think about communities’. Yodzis’s resolution is to lump the biospecies, with which community ecology is usually concerned, into ‘ trophospecies ’ comprising several species with similar food habits. Yodzis, however, reported that even his simple models when perturbed ‘ are in significant measure indeterminate when viewed on a relatively long temporal scale ’. Terborgh & Robinson ( I 986) joined Yodzis and May ( I 976 b) in anticipating that transtaxonomic recognition of guilds will become ‘the standard currency of ecologists in their efforts to understand community relationships ’. Holmes & Price (1986) defined community as ‘a group of organisms in a particular place, without any preconceptions on whether they interact or not’. They
3 30 R. P. MCINTOSH also called on Root’s (1973) categorization of communities as ‘component communities’ of specific microenvironments and ‘ compound communities ’ as a ‘ complex mixture of component communities that interact to varying degrees ’.
Schoener ( 1 9 8 6 ~ ) expressed reservations ‘about adding a new term to an already jargon-laden biological vernacular ’, but saw no alternative but to add ‘similia- community’ (from similiu = similar things). Similia-communities are sets of species in different places that ‘ are similar with respect to crucial organismic and environmental traits’. Similarity and its measurement are a long-term concern of ecology, the crucial question generally being how similar must things be to be similar or similia- communities ? Rosenzweig ( I 987) coined ‘ anomic ’ community for a group of organisms in which the presumed ‘rules of organization’ have not organized them into a proper community. Like Frederic Clements in the early era of ecology, Rosenzweig preferred Greek sources and it is probable that ‘anomic’ community will go the way of most of Clements’s classic terminology. Giller & Gee ( I 987) provided a hierarchy of levels of organization in community ecology (community, subcommunity, guild, taxon guild) and a ‘plea for consistent terminology’ echoing the sentiments of all too many earlier ecologists with, perhaps, no more hope of success. Gee and Giller optimistically noted, ‘ a pleasing parallel between mathematical models and empirical analyses of the effect of spatial scale on the perception of the equilibria1 status of the community’ - a parallel not equally apparent or pleasing to other ecologists.
(4) Key questions for community ecology
Ecology has a long history of questions. Vas ist ein Pflanzengemeinschaft ? or Why are there so many kinds of animals? (Hutchinson, 1959), which Schoener (1974) described as a ‘celebrated riddle’: there have been many responses if not answers. Animal community ecologists in the I 980s produced numerous additional questions some new, some old, some clear, some abstruse. A. J. Underwood (1986), Roughgarden & Diamond (1986) and Southwood (1987) explicitly addressed the old and difficult question - What is a community ? Underwood argued that study of community properly deals with species and individuals and wrote ‘it is probably more profitable for ecologists to pay more attention to how often, and how consistently, various combinations of species occur together ’. Study of interactions and interdependence, he wrote, must be done at several scales ‘without reference to organized communities or superorganisms ’. Roughgarden & Diamond agreed with Underwood that community ecology is concerned with the variety and abundance of organism, in their term ‘ limited membership ’. The problem has always been what constitutes limited membership - association, character species, constancy or fidelity of species in a nearly forgotten lexicon. Limited membership, according to Roughgarden & Diamond, has three causes all long familiar to ecologists : physical environment, dispersal limits and interactions among species. Roughgarden & Diamond posed a subsidiary question, What is community structure? Structure takes a bewildering range of meanings and is loosely synonymized with organization or pattern, each of which has its own nuances (Connell, 1975). Included in structure, beyond the classical attributes of spatial and temporal abundance of species (composition), are resource allocation, niche relations, species- area, trophic levels or food webs, body size relations, foraging techniques, age distribution, vertical and horizontal distribution (literally in space or abstractly in
Gleason’s ‘ individualistic concept ’ 331 trophic levels) temporal distribution and morphology. A perennial, and perhaps the oldest, concern in assessment of community structure, or indeed any attribute of community, was rediscovered by May (1984)’ ‘any attempt to elucidate patterns of community structure must deal with the question of how to delimit the community’. For ecologists who wished to work with real communities rather than mere assemblages, Hairston (1984) stated ‘The greatest need is for a legitimate means of identification of interacting groups of species ’.
The major problem of quantification of a community is entitation, which is often overlooked. Southwood’s ( I 987) answer to ‘What is a community ? ’ first required that a community be recognized, then went on to the nature, properties and function of communities including horizontal links on the same trophic level and vertical links between trophic levels. He joined Underwood and Roughgarden & Diamond in asserting that the most fundamental description of a community is the number of different species and the abundance of individuals of each. Southwood concluded, noting two of Schoener’s (1987) seven axes of controversy about community, ( I ) tightly linked structured groups vs. individualistic groups, (2) organism-driven (biotic) vs. environment driven (abiotic). His ‘forecast’ is that the poles of neither axis will be the answer ‘for ecology deals with a mixture of pattern and probabilism. ’
Strong et al. (19846) provided a series of ‘contemporary questions in community ecology’ all turning on interactions of species but they commented that ‘the most profound issue of contemporary ecology.. . stochasticity makes deductive answers to these questions doubly difficult to find’. Nevertheless they raised ‘one of the ultimate community questions ’ - How do communities really behave ?
Many questions were generated under the umbrella of the ruling school of animal community theory of the 1960s and 1970s and continued into the 1980s. Lewontin ( I 969) had asked, ‘ Can there be more than one stable community composition in a given habitat?’ Chesson 8z Case (1986) asked ‘To what extent are the attributes of natural communities predictable ? ’ noting the derivation from community theory predicated on equilibrium and reviewing new theoretical directions beginning with a new question ‘What is a non-equilibrium explanation ? ’ Their answer is in substantial part damage control for the traditional theories that constructed models based on competition and stable equilibria. One response to the evident difficulties posed by the proponents of Gleason’s concept was to urge pluralism which is progress following an era of monistic community ecology (McIntosh, 1987). T. Underwood (1986) raised collateral questions about the worth of community theory.
How well will the currently available studies stand up when analyzed by less personally involved historians of ecology at some point in the future when today’s hypothesis and theory might be as interesting and relevant as Ptolemy’s? If in the search for new theory today’s is found wanting, will future ecologists view today’s experiments excitedly, as an abandoned gold-mine to be reopened, or as the ideological equivalent of a used-car lot, littered with the rusty remains of vehicles of self-advancement, now abandoned by careless speedsters down the vanished highways of former ecological theory ?
Sadly, Underwood’s ‘message for future historians of ecology is ‘do not look - spare yourselves ’. Historians, however, find as much interest in misguided or failed science as in successful science.
Kitching ( I 986) posed a methodological problem for community ecologists claiming
332 R. P. MCINTOSH that to answer community-level questions one must carry out community level studies rather than make inferences from lower level population studies. Carpenter (1989) applied this rule to the study of lakes. Rosenzweig (1987), to the contrary, asked ‘ Is it possible to learn anything useful about large systems of interactions by focusing on one ? ’ He suggested a mathematical theorem that ‘ proves that one can make accurate inferences about the dynamics of entire systems just by studying (carefully) the dynamics of any single species that belongs to it’. However, he leaves the original question unanswered leaving the reader to wonder why he brought it up at all.
Schoener ( 1 9 8 6 ~ ) addressed the question that is at the hub of the traditional debate between Clements and Gleason and their many successors. Is each community effectively unique, or is there a modest number of ‘types ’ of communities ? Diamond & Case (1986b) asked the same question and, foregoing the hope of devising a model that would apply to all communities, expressed a more modest hope that it might be possible to ‘devise a model for each type’. Bush & Aho (1990), in an attempt to summarize a symposium on parasite communities, generalized that there had been no central conclusions offered but emerged with a unifying central question - What happens to empty space? - the classical question of succession.
Some questions raised by animal community ecologists are more pithy if not more clear than those posed above. Rathke ( I 984), for example, asked ‘ Quo Vadis ? ’ calling on the abilities of her readers in Latin as well as in speculation about the future. Rathke’s response to her rhetorical question was that random models had led to rejection of a competition hypothesis about flowering phenology. Colwell (1984) was similarly cryptic, but at least in plain English, asking ‘What’s New ? ’ What was new was a move toward a community ecology that rejected the basic assumptions on which the mathematical theory of ecology of the 1960s and 1970s was erected, leading to the ‘Death of the “old ecology”?’. The question mark possibly implied that it wasn’t surely dead and, like Count Dracula, might be around to haunt its opponents, as it has.
Schoener (1987) identified seven axes of controversy in community ecology that provide a summary of the history of community ecology and of the several recent articles asking the grand question -What is a community?, or inquiring about the nature of community. Schoener’s axes orient the empirical studies on axes different from, or complementary to, the classical organismic-individualistic community axis. Several of Schoener’s seven axes may be seen as implicit in the classical axis but segregated out of it in the context of more recent disputes and disputants in community ecology. Others are not neatly seen as part of the classical dichotomy. The poles of several of Schoener’s axes, e.g. unpatterned, vs. patterned, random vs. nonrandom, physical vs. biological, interactive vs. noninteractive, do not match the classical axis. Both organismic and individualistic communities are patterned but in quite different ways, organismic communities are non-random but so are Gleason’s individualistic communities, contrary to a common misrepresentation. Schoener suggested two ‘emollients’ for the scars of controversies that have developed in community ecology over the several axes he outlines. First, a pluralistic approach which is tolerant of variation among communities and ‘many faceted theory’. The second is a mechanistic approach for a theory of community ecology based on concepts from lower levels, which is what ecologists usually mean by reductionism.
Hengeveld ( I 988) considered deterministic vs individualistic hypotheses of
Gleason’s ‘ individualistic concept ’ 333 biological invasions in the context of community theories. Hengeveld argued that species invasions can be viewed as individualistic species responses, a position clearly kin to Gleason’s. Russell (1992) tried to bridge the gap asking, ‘How does an assembled community fit into the concept of spatial and/or temporal continua? ’ The questions about trophic controls posed in the famous HSS (Hairston et al., 1960) article continued to exercise ecologists and a new set of terminology was introduced to ecology in its wake. Keystone species gave rise to trophic cascades and discussions of top-down versus bottom-up control of community, usually in aquatic communities. Strong (1902) asked, ‘Are trophic cascades all wet ? ’ but described finding forces extending through as many as four trophic levels as ‘one of the most important in all of ecology of the last decade ’.
The ‘enduring debate in American ecology about the nature of the biological community’ (Richardson, I 980), ‘temporal schizophrenia’ (A. J. Underwood, I 986), ‘current uncertainties and controversy in community ecology ’ (Southwood, 1987) and seven ‘axes of controversy’ (Schoener, 1987) persisted through the 1980s and on into the 1990s. Many studies of diverse taxa or habitats addressed the grand questions of community just described. Some commentators on ecology deplored the tendency of ecologists to ask such large questions. Price ( 1 9 8 6 ~ ) offered a ‘critical review of questions and approaches and attributed the controversy to ‘ miscommunication fueled by failure to focus on well-defined questions’. Slobodkin (1986) suggested the possibility that ‘the “ big questions’’ of ecology are simply too big to be answered’. He advocated ‘ minimalism ’ or choosing the smallest questions recognizable in a professional field. Peters (1991), in a universally negative critique of ecology, stated that ecologists must shun unanswerable or intractable grand questions. Among his bad examples was ‘What is a community? ’ Community ecologists, nevertheless, addressed questions large and small and, as Slobodkin noted, this may add to the ‘volume and vituperative quality ’ of current ecological publications. Slobodkin ( I 992) wrote, ‘Ecology may be the most intractable, legitimate science that has ever developed’, an unpromising prospect for community ecologists who engage in the most complex aspects of ecology. Menge (1992) deplored the tendency of ecologists to pose questions like ‘ Do top-down (e.g. trophic interactions) or bottom-up (e.g. nutrients) effects control communities ? ’ He allowed that both could affect community structure and more modest questions should ask how, or when, they interact and what mechanisms operate in each.
( 5 ) Empirical studies of communities
Numerous empirical studies of animal communities of diverse taxa and habitats provided a spectrum of evidence and interpretations concerning community in the 1980s. These ranged from explicit support of either pole of the traditionai axis to various positions that ranged from intermediate to indeterminate. The number of studies in each of the three groups was roughly the same, no position enjoying a plurality. Proponents of either extreme, or the intermediate positions, appeared in studies of all major taxa and many different habitats. Examples of these are considered below in sequence : ( I ) those at or near the individualistic, Gleasonian, or continuum pole; (2) those at or near the organismic, Clementsian, deterministic pole; and (3) those intermediate or variable.
334 R. P. MCINTOSH Wiens & Rotenberry (1981) concluded that few significant correlations were evident
among breeding birds of shrubsteppe, ‘suggesting that bird populations in this system vary largely independently of one another ’. James et al. (I 984), noted that the niche of the wood thrush was more aligned with the individualistic approach of Gleason. Bock (I 987) wrote of Arizona land birds :
It is apparent that these assemblages usually are dominated numerically by widespread species whose past histories and present dynamics cannot have anything to do with communities as we have delimited them. This leads me to advocate an individualistic approach to avian ecology.
Wiens (1989) assembled two volumes on ecology of bird communities. He noted the dominance of the competition - based ‘ MacArthurian paradigm’ in the 1960s followed by increasing studies of species characteristics. According to Wens :
This pattern may be a manifestation of the resurgence of an individualistic Gleasonian view of communities among ecologists at that time, a movement that apparently did not influence those publishing in ornithological journals.
Marine biologists had long participated in discussion of the nature of community (McIntosh, 1967, 1985). Dean & Hurd (1980) continued the tradition applying the continuum concept to the marine fouling community. Williams et al. (1981) were more extreme finding distributions of marine planktonic diatoms were ‘ entirely probabilistic’. They looked enviously at terrestrial ecology where they said ‘the meaning of the term “community” can largely be rid of ambiguity’, a situation not apparent to terrestrial ecologists. Gray (198 I) asserted that the continuum viewpoint agrees with most marine benthic studies. Jones (1984) asserted that absence of classificatory groups in coral-reef sediments showed that community composition varies continuously along a sedimentary gradient. According to Sale (I 984) presumed orderly patterns of animal communities in coral reef fish were not true. There was, he wrote, little evidence of biotic interactions influencing distributions and ‘ assembly rules, if they exist, must be very subtle’. Dean & Connell(1987) stated that patterns of a marine invertebrate community did not meet predictions based on expectations of a highly organized community. The community was unorganized, with only predation possibly important but it was weak. Ebeling & Lauer (I 986) offered an individualistic explanation of resource partitioning in surf perches.
Freshwater biologists commonly described various communities as in accord with the individualistic hypothesis. Reice (1980) wrote that benthic fauna supported ‘the individualistic hypothesis of community structure, common in plant ecology, dating back to Gleason (1926)’. Matthews & Hill (1980) wrote that stream fish ‘species associations were transitory ’ with ‘ a few stable interspecific interactions ’. Grossman et al. (I 982) deplored the ‘ overburdened ecological vernacular ’ and attempted to clarify it in finding that stream fish assemblages were regulated by stochastic factors. Yant et al. (I 984) argued that Grossman et al. had biased their results by using an atypical site. Rahel et al. (1984) and Herbold (1984) asserted that Grossman et al. had used a faulty definition of the fish assemblage and studied it at the wrong scale, an increasingly frequent accusation in the 1980s. Grossman et al. (1985) rebutted each criticism saying their conclusions had only been strengthened by the exchange. Matthews (I 982) found no more structure of fish communities than random aggregation with little interspecific interaction. Heins & Matthews (I 987) provided an historical perspective on studies of
Gleason’s ‘ individualistic concept ’ 335 ‘North American stream fish communities’ noting that by the late 1970s ecologists were being cautioned against an uncritical acceptance of competition as structuring fish communities. Norton (1991) said that patterns of distribution and abundance of cottid fishes are more consistent with Gleason’s individualistic concept than with a ‘ synecological mechanism ’.
Mammal communities were frequently exploited in examining community questions in the 1980s (Morris et al., 1989). Dueser & Brown (1980) thought that competition was insufficient to structure a community of rodents by assembly rules. Dueser & Porter ( I 986) found competition in a small mammal community, although ubiquitous, was relatively weak. Brown & Kurzius (1987) noted that recent discussions of animal communities continued the traditional Clements vs. Gleason debate. They put animal ecologists R. MacArthur, J. Diamond and P. Grant in the Clements camp and a largely botanical group, R. Whittaker, L. Cole, B. Huntly and H. J. B. Birks, in the Gleason camp. By 1987 numerous animal ecologists might have been added to the Gleason camp. Brown & Kurzius, themselves, noted that desert rodents were distributed individualistically making co-evolution unlikely and efforts to study ‘assemblages of many species in terms of pair-wise interactions ’ ineffectual or even ‘ misleading’. Owen (1990) described mammals in Texas as fortuitous assemblages and wrote :
This supports the concept of an open community of mammals analogous to that espoused by phytosociologists. Spatial structure of mammalian distributions is thus concordant with the continuum concept of community.
Studies of amphibian and reptile communities proliferated in the 1970s and early 1980s (Toft, 1985). Scheibe (1987) studied temperate lizard communities comparing field data with randomly generated null communities and found ‘no evidence to support the limiting similarity hypothesis ’ associated with communities structured by competition. Haefner (1988) compared a range of random and competition models with data of Anolis lizards. Competition models were no better than some random models Gascon (1991) assessed distribution of rain forest tadpoles and produced an interesting concatenation of Gleasonian ideas.
Insects of diverse taxa and habitats were studied to address community questions. Vepsalainen & Pisarski (1982) found that ant communities did not fit into ideas of assembly processes and that early arrival was an important consideration. Boecklen & Price (1989) studied sawfly ‘assemblages’ on willow clones and found that every clone had a unique arrangement of sawflies with independent species responses. Gilbert & Owen (1990) denied a Clementsian ‘holistic vision’ of community and provided a remarkable parallel to Gleason’s ideas :
We believe that any ‘structure’ is a biological epiphenomenon, a statistical abstraction, a descriptive convention without true emergent properties but only collective ones, wholly referable in its properties to those of the constituent species, populations, and individuals. Thus we believe that animal ecology is learning what plant ecologists learned many years ago.. . , and lean to the view that at least some communities of syrphids are merely coincidences of species in space and time.
Perhaps the major contribution of paleontology to ecology stems from consideration of micro- and macro-fossil communities. Birks (1981), Davis (1986), and others, found that putatively stable or climax communities of Clements segregated into species that
336 R. P. MCINTOSH migrated separately to different refugia in the Pleistocene. Springer & Bambach (1985) began a study of marine invertebrate fossils with a review of Gleasonian and post- Gleasonian concepts. They found ideas of discrete communities less informative than gradient analysis and no evidence that species interdependence was important in community structure. Janssens et al. (1986) found fossil mosses, that had once coexisted, occurred in very different phytogeographic contexts. They wrote :
Such fossil records support Gleason’s ( I 926) individualistic concept of vegetation, the uniqueness of past communities (Birks 1981) and the short term nature of vegetation units.
Graham (1986) found individualistic responses of mammalian species during the late Quaternary, the species having been ‘massively and repeatedly reshuffled ’ . Coope ( I 987) interpreted insects in the late Quaternary as forming individualistic species associations :
It is clear that the insect community did not react en bloc as an integrated whole, but each of its component species responded by moving at its own rate as the climate underwent rapid and intense changes.
In contrast to the foregoing, many ecologists supported concepts of organized and integrated communities often in the same taxa and habitats used to demonstrate individualistic communities. Haefner ( I 98 I ) found well defined community structure in birds, the species being clustered according to rules. Noon (1981) inferred that competition was important in a community of thrushes. Moulton & Pimm (1986) found evidence of extensive competition in avian communities and recalled the ‘ intellectual blood spilled’ in earlier debates about the need for a null hypothesis.
Marine biologists also discerned kinds of community concepts. Mook (1981) interpreted marine fouling organisms converging to the same community. Perks (1982) preferred organismic ideas to continuum concepts in marine communities. Dayton et al. ( I 984) expressed hope for a Clementsian deterministic stability of kelp communities that offered some hope, perhaps, to animal ecologists. Pearson & Rosenberg (1987) emphasized the importance of food in marine benthic communities and claimed that succession of these communities was predictable and produced characteristic communities.
Freshwater communities of diverse types were sometimes interpreted as organized, integrated, distinct entities. Naiman et al. (1988) found:
Compelling reasons for examining the stream-river profile as a series of discrete patches or communities with reasonably distinct boundaries rather than a gradual gradient or continuum.
Tonn & Magnuson (1982) found fish ‘assemblages’ in northern Wisconsin lakes resulted from deterministic mechanisms. Power 8z Matthews (1983) gave evidence of vertical structure or ‘trophic cascade’ in a sequence of effects of piscivorous bass on herbivorous minnows affecting attached algae. Moyle & Vondracek (1985) found a fish ‘assemblage’ with the characteristics of a ‘highly structured community’. Matthews et al. (1988)’ contrary to some of Matthews earlier studies, found persistent and stable faunas and many individual locations had relatively stable fish ‘ assemblages ’. Wikramanayake ( I 990) argued a co-evolutionary adjustment in a tropical stream ‘assemblage’ to reduce interspecific competition. According to Meffe & Sheldon (1990) fish ‘ assemblages ’ recovered following defaunation. They wrote :
Gleason’s ‘ individualistic concept’ 337 These assemblages were not randomly structured units but were largely deterministic systems highly predictable from local habitat structure.
Kodric-Brown & Brown (1993) examined relative importance of deterministic and stochastic factors in assembly of fish communities of desert springs in Australia. They concluded that ‘ ecological relationships that determine community structure may be highly deterministic with the proviso, ‘When the influence of historical and environmental factors can be assessed ’. Gilliam et al. ( I 993) reported that interspecific interactions structured a stream fish community in tropical Trinidad.
Amphibian and reptile communities also illustrated the extremes of interpretations. Woodward (1983) found anurans of desert ponds were segregated into species characteristic of temporary and permanent ponds which he attributed to predation ‘operating in the past ’. Wilbur ( I 984) suggested that contemporary predation determined frog survival in long-lived ponds, whereas competition was more important in short-lived ponds. Roughgarden (1986) asserted:
Now there is no doubt that interspecific competition occurs between anoles and that its strength depends on the similarity in body size of the species.
Southerland ( I 986) noted the controversy about equilibrium (competition based) and non-equilibrium (non-interactive) communities and claimed that salamanders of streamside habitats occurred as discrete ‘ assemblages ’ representing the former.
Insect communities were seen as organized less frequently than those of other taxa. Joern & Lawlor (1980) asserted that resource use patterns of grasshoppers were not ‘ merely the result of fortuitous occurrences among the individual species but reflect biotic interactions among these species’. Lawton (1984) suggested that the insect community on bracken fern was predictable and controlled by density dependent events operating independently of other species, at least those on the same trophic level.
Fossils were also interpreted as organized communities in some instances. Van Devender (1986) called on the ghost of competition past to explain niche separation in Pleistocene packrat middens. Webb ( I 987) applied rules of assembly and disassembly to Cenozoic mammals and reported faunal equilibrium that ‘ implies the probability of higher-structured (interactive) tetrapod communities.
Many ecologists got mixed or equivocal signals from their data, and some warned against undue expectations of either extreme of individualistic or organized community. Schemske & Brokaw (1981) compared bird distributions in treefall gaps and closed forest and found some species in either but ‘the majority of species overlap broadly along the gap-mature continuum’. Grant (1986) wrote ‘we are far from being able to generalize confidently about the frequency and intensity of competition for food among these birds, or indeed among any animals’. Homes et al. (1986) struck a plural view of community structure between competition models and non-equilibrium models, although they found little evidence of a tightly organized community of birds at any one spatial scale.
In marine communities less explicit positions were also encountered. Venrick (1982) attributed individualistic distributions to the ghost of competition past but identified two distinct recurring associations of species. Grossman ( I 982) expressed surprise that fish in a rocky intertidal habitat fitted a deterministic model but algal and invertebrate ‘ assemblages ’ were stochastically regulated. Fenchel ( I 987) described a community of
338 R. P. MCINTOSH microorganisms as species populations that interact or are confined in space or time but indicated that scale of size and interaction time restricts the likelihood of strong interaction. Wilson ( I 99 I ) argued that soft-sediment communities required a distinct paradigm but that at present a unifying theory of such communities is not attainable.
Freshwater communities produced similarly equivocal or variable results. Hockin ( I 982) interpreted the copepod community as switching from a non-interactive community in spring to an interactive community in a time of less disturbance. Beckett & Miller (1982) described two endpoints of an invertebrate community with composites of individualistic species responses in between. Minshall et al. (1985) described the stone dwelling fauna of streams as changing from equilibrium status in summer to non- equilibrium in autumn. Hildrew & Townsend (1987) said that neither deterministic or stochastic factors in structure of freshwater benthic communities was primary. Wevers and Warren ( I 986) explored the unlikely possibility of ‘ integrating organismic and individualistic views ’ of stream communities. They considered the stream community as organized hierarchically as a group of subsystems on five levels which, they said, are indistinct.
Freshwater fish communities produced similarly variable interpretations. Peckarsky & Dodson (1980) suggested that stream community structure differed in ‘harsh ’ (physically controlled) streams from ‘benign ’ (biotically controlled) streams. In the latter, competition is mitigated by predation that reduces populations of prey. Schlosser (1985) allowed that both deterministic and stochastic factors regulate stream fish ‘ assemblages ’. Tonn ( I 985) conjectured that factors determining fish community structure differ from ‘assemblage’ to ‘assemblage’, season to season and year to year. Grossman ( I 990) reviewed the thorny question of ‘ assemblage ’ stability in stream fishes noting a lack of concensus which remains as a spur to further investigation.
Ross ( I 986) reviewed resource partitioning in fish ‘ assemblages ’ from seven global habitats, including freshwater streams, lakes and the oceans, based on the literature from 194-1983. The pace of events is evident in Ross’s bibliography, which includes 5 articles from 1940-50, 7 from 1951-60, 23 from 1961-70, I I I from 1971-80 and 63 from 1981-83. Ross stated that differential resource use is widely documented but underlying mechanisms are ‘obfuscated ’ which showed ‘the difficulty of formulating broad generalizations of community control ’.
Students of parasite communities introduced the terms ‘ interactive ’ and ‘ isolationist ’ to the community lexicon (Price, I 984b). Isolationist communities are dominated by individualistic responses with weak interspecific interactions. Interactive communities are dominated by interspecific interactions and are closed (saturated). Bush & Holmes (1986) said that helminth communities of lesser scale formed a continuum between interactive and isolationist extremes. Kennedy et al. (1986) found helminth com- munities in fish were isolationist, whereas those of birds were interactive. Holmes ( I 990) considered the poles of isolationist and interactive communities and asked : ‘Where along this continuum do communities of gastrointestinal parasites lie ? ’ Not surprisingly, the answer was, ‘The evidence is contradictory’. Aho ( I 990) reported that helminth communities of amphibians are characteristically non-interactive or iso- lationist, but the infracommunity helminth community in birds could be determined by either biotic or stochastic processes.
Efforts to clarify the controversy concerning community organization were
Gleason’s ‘ individualistic concept ’ 339 complicated in some sense by Schoener’s ( I 987) multiplying the axes from one to seven. Hubbs’s (1987) summary of the symposium in which Schoener’s paper appeared responds to a question posed by one of Schoener’s axes.
Are stream fishes regulated by stochastic or deterministic factors? The proper answer is yes. Depending on circumstances stochastic or deterministic factors are the most important. The interesting questions now revolve about when and where do the controls differ.
Strong et al. ( I 984a) recognized a spectrum of theoretically ideal communities from essentially random assemblages with little density dependence to highly deterministic systems, ‘structured by strong interspecific competition’. They found only 38-39 yo of herbivorous insect populations were not density dependent but populations fluctuated markedly between generations and ‘ communities of phytophagous insects are only moderately constant in their structure’. Wilbur (1987) joined other ecologists in writing: ‘A general answer to the classic question in community ecology of what regulates the distribution and abundance of species remains elusive ’. Wilbur’s results, he said, suggested ‘the futility of arguments between proponents of predation and competition as the single unifying force structuring communities ; and he asserted the importance of chance events in structuring communities.
The above review of animal community studies, although incomplete, makes evident the lack of concensus lamented by Giller & Gee (1987). Interpretations of community pattern, structure or organization, and the reasons therefore, ranged from firmly deterministic, integrated communities based on biotic interactions, or the ghosts thereof, to explicitly individualistic communities predicated on interactions of species with their abiotic and biotic environment coupled with stochastic events. Rarely, a group of species was seen as approaching the null model of a random aggregation; more frequently it was somewhere between the integrated-individualistic poles.
IV. COMMUNITY THEORY AND QUESTIONS FOR T H E 1990s In the mid 1980s an anonymous editor posed the rhetorical question. ‘Community
ecology: back on its feet again? If volume of research activity and publication addressing the myriad questions posed about community were the basis for answering the editor’s question, the answer would have been a resounding -Yes! If achieving consensus on defining community and its terminology, agreement about the existence of community structure and about processes or ‘ rules ’ that structured communities were the criteria, the answer would have to be - No! Questions concerning communities persisted. Williamson ( I 987) asked, almost plaintively, ‘Are communities ever stable ? ’ Crawley ( I 987) asked ‘ What makes a community invasible ? ’ J. H. Lawton ( I 987) asked ‘Are there assembly rules for successional communities ? ’
Perhaps the most cogent, if least tractable, question was the oldest - What is the community ? The many definitions reviewed ranged from simple groups of closely related taxa to grandiose representations about all species in a specified site. Many of the characteristics ascribed to community (richness, diversity, stability, etc.) depended in the first instance on the entity chosen for study. Not a few criticisms of studies reporting community attributes were based on claims that the original study was dependent on a biased or non-representative choice of study area and the now frequent
340 R. P. MCINTOSH problem of scale. A corollary of the above question is whether the community studied at a particular site (the concrete community of an older if not wiser literature) is a replicate of a larger community repeated at different sites (the abstract community of the earlier lexicon, now sometimes designated as meta-community). Depending on the response to these questions is whether the functions of any species as a member of the community are to be described specifically as in Charles Elton’s famous analogy, ‘There goes the vicar’. That is, do species have functional attributes that mandate certain combinations of species and numbers of individuals in repeated instances of a community.
The questions that are engaging the attention of animal community ecologists in the 1990s were posed in the late 1980s and commonly involve a search for ‘rules’. ‘Are there assembly rules for successional communities ? ’ (Lawton, 1987). In a broad sense this is in the tradition emanating from Frederic Clements who posited ‘laws’ of community formation in the early decades of the twentieth century and perpetuated by animal ecologists, notably Eugene Odum ( I 969)’ who anticipated an orderly, predictable process of succession, community development or community assembly. Lawton’s response to his question is that there are some broad rules but the details are ‘fuzzy’.
Distinguishing between the effects of chance and determinism on community assembly undoubtedly constitutes one of the most important and difficult problems in contemporary ecology.
Although ‘assembly rules’ as described by Diamond (1975), and the theory and empirical studies of community on which they rested, had been severely criticized throughout the I 980s, the search for assembly rules and many of the assumptions about community on which they rest persisted unabated. Brown (1987) suggested that the functional organization of communities could be characterized by two classes of rules. ‘ Capacity rules ’ include all extrinsic processes, both physical and biotic, that affect the capacity of the environment to support the community in question. ‘Allocation rules’ are biotic interactions, such as competition, intrinsic to the community. Brown stated that extrinsic rules ‘ultimately determine the outcome of the interactions within the assembl