Coral Reefs (1997) 16, Suppl.: S77—S91
Coral reef ecosystems: how much greater is the whole than the sum of the parts? B. G. Hatcher* Pelagic and Reef Fishes Resource Assessment Unit, CARICOM Fisheries Resource Assessment and Management Program, Tyrell Street, Kingstown, Saint Vincent and the Grenadines, West Indies E-mail:
[email protected] Accepted: 31 January 1997
Abstract. The ecosystem concept has been applied to coral reefs since the time of Charles Darwin, perhaps because of the apparent integrity of the biotic-abiotic nexus. The modern model of the ecosystem as a hierarchy with emergent properties is exemplified in reefs as massive structures formed by small colonial organisms, the selfsimilarity of these structures across large spatial scales, and the uniformity of function by diverse biological communities. Emergent properties arise through the integration of processes up the levels of organization and larger spatial and temporal scales encompassed by a whole reef. The organic response of reef morphology to hydrodynamic forcing, the constancy and conservatism of organic production across a broad range of environments, and the global persistence of reefs in the face of massive evolutionary change in species diversity are interpreted as emergent properties. Coral reefs, of course, function by the same basic laws as other ecosystems, but there is cause to view them as an end member of a continuum because of their structural complexity and high internal cycling. Well-defined boundary conditions mean that highly integrative measures of ecosystem process based on physical and biogeochemical models (e.g. community metabolism) have provided the main applications of systems ecology to questions of coral reef function. Organism-population approaches are being reconciled with form-functional models to yield new insights to ecosystem processes and interactions among reefs and adjacent systems. The form and metabolism of reef production are strongly affected by phase shifts in benthic community structure, and most reef systems are more open to trans-boundary fluxes and external forcing than the early models suggest. The attractive paradigm of the reef as a self-sufficient ecosystem is dying slowly as research focus shifts from atolls to more open fringing and bank barrier reefs, and organic inputs to system production are measured. Coral reefs contribute little in a net sense to global ecosystem processes, but on an areal basis their exports of organic * Present address: Dept. of Biology, Dalhousie University, Halifax, Nova Scotia, B3H 4J1 Canada
products are significant. Holistic models and measures of ecosystem processes incorporate the unusual whole-part relationship of reefs and are practically essential to answering the key questions facing coral reef science in the next millennium.
Introduction The aim of this study is to explore the role of the ecosystem concept (Tansley 1935; Odum 1953) in our understanding of coral reefs. Ecosystem processes maintaining biological diversity and controlling the flux of materials have been well studied because they can be examined across a broad range of spatial-temporal scales in all ecosystems. Coral reefs (like lakes, e.g., Schindler 1990) have received particular attention from ecologists working at the ecosystem level because of the apparent distinctness of their boundaries. Here I focus on the relevance, usefulness and recent application of ecosystem concepts and measurements to coral reef research. Three elementary questions emerge from the somewhat cryptic query posed in the title: 1. What are ecosystem processes in coral reefs? 2. Why (and how) do we study them? 3. Has ecosystem research advanced coral reef science? The answers cannot be stated succinctly, but the statements following summarize the relevant themes for discussion: 1. Ecosystem processes are those which link the physical environment to the interacting assemblage of organisms. Common examples include biological production, biogeochemical cycling and the evolution and maintenance of biological diversity (Table 1). As such, ecosystem processes transcend abiotic-biotic distinctions, levels of biological organization, and spatial-temporal scales of observation. Their study is as much about boundaries, scales and rate constants as about any particular physical, biological or geological entity or phenomenon. Ecosystem research focuses on
S78
universal processes, which are fundamentally no different in coral reef systems than in any other. 2. We study ecosystem processes to improve understanding, prediction and control by quantifying emergent parameters across hierarchies of organization and spatial-temporal scales. Certain measurement techniques and currencies (e.g., carbon flux) are particularly suitable for such integration in reef ecosystems, and novel models and tools (e.g., dissipation structures, feedback analysis) have been developed to deal with the complex, non-parametric data they generate. 3. Ecosystem research identifies the variables of state (e.g., net vertical accretion rate) and sets the boundary conditions (e.g., maximum sea level rise with which reefs can keep up) that can assess the status of reefs and predict their responses to environmental change. We hope that the measurement of ecosystem processes will produce scale and case-independent results, and that models based on them will usefully simplify the complexity of reefs while still explaining or predicting actual observations. Optimistically, ecosystem science aims to do for ecology what Newtonian physics did for mechanics. The morphoedaphic index (Schlesinger and Reiger 1982) or the fundamental biological diversity number (Hubbell 1997) are good examples. Coral reef ecosystems: boundary conditions ‘‘ 2 to some degree, an ecosystem is in the eye of the beholder.’’ —B. Babbitt, Secretary, ºS Department of the Interior Coral reefs are outstanding examples of marine ecosystems. Reefs embody the mixture of the minute and grand, ephemeral and permanent, simple and complex that we associate with the natural systems of this planet. ‘‘Ecosystem’’ is one of those wonderful words that communicates a fuzzy concept of near-universal consensus, but means very different things to different people when it comes to specifics. While few would agree exactly on the components, processes, scales and boundary conditions that define a reef ecosystem, many use the phrase glibly in their intercourse with scientists and others. Yet it is the specifics of ecosystem structure and function that concern scientists. The demands of society for prediction and control of anthropogenic change in the earth’s ecosystems have forced ‘‘ecosystem management’’ to the top of the environmental, resource management, and even political agendas (Christensen et al. 1996). Many have felt the effects on the review, funding and application of ecosystem research. The proceedings of the eight Coral Reef Symposia and the recent International Coral Reef Initiative (Crosby et al. 1995) nicely plot the emergence of ecosystem approaches to the research and management in coral reefs. But what is an ‘‘ecosystem approach’’ ? Surely before one undertakes ecosystem management, there must be a clear and practical definition of the ecosystem. Many investigators have tried to do this for coral reefs (e.g., Fosberg 1961; Preobrazhensky 1977; Fagerstrom 1987; Kinsey 1991). Perhaps many would agree on some general definition of a coral reef: a marine limestone structure built by
calcium-carbonate secreting organisms which, with its associated water volumes supports a diverse community of predominantly tropical affinities, at a higher density of biomass than the surrounding ocean. This definition includes the abiotic and biotic aspects, but lacks any specification of scale or integration (i.e. connectivity, selforganization, homeorhesis), which are usually associated with the ecosystem concept. For example no one defines a solitary coral polyp as a coral reef. Yet it has many of the attributes associated with coral reef ecosystems: rounded shape, distinct boundaries on its surrounding environment, symbiotic life style, high animal to plant biomass ratio, calcium carbonate precipitation, P/R ratio of about one. It is even arranged with a nutrition-gathering periphery around a digestive centre: but it’s not a reef. Similarly, a lump of coral rock on a lagoon sand sheet (i.e., a ‘‘bommie’’) is not a coral reef. Nor is the algal flat on an atoll rim: these are parts of a coral reef. A fringing reef on a Caribbean island coast is, however, considered to be a coral reef ecosystem, as is an atoll in the Maldives. The Great Barrier Reef is not a coral reef (it is a coral reef province containing thousands of reefs), but it is generally referred to as a coral reef ecosystem (which includes the vast inter-reef areas of water and sediment). It appears that our definition of coral reefs has more to do with the boundaries and scales than with the components and functions of the ecosystem. All of these examples can be studied as coral reef ecosystems (sensu stricto), even as they are not all coral reefs. It is more than semantics that the ecosystem can be defined at more than one level of organization and across a rather large range of spatial and temporal scales: it reflects the hierarchical nature of ecosystems (Webster 1979; Miller 1991). Thus, it is not so much a level or scale that we use to define a reef ecosystem as the nature of the boundary with another level or scale. Most often it is a Cartesian boundary (e.g., the transition from reef slope to open ocean). Interestingly, it can be a temporal transition (e.g., between Cretaceous and Tertiary reefs). Problems arise, however, when the components of ecosystems (e.g. fish populations) do not correspond to the Cartesian boundaries. One way around this is to use functional attributes (processes) to define the boundary conditions (e.g., the mixing line of lagoonal and oceanic water masses, or the 1% isopleth of coral larval density). That these process-derived boundaries often map onto Cartesian boundaries reflects the interaction between physical structure and biotic function that characterizes coral reef ecosystems (e.g., topographically controlled water residence time determines the relative importance of biological processes of nitrogen transformation in different reef habitats, Hatcher 1985). From this functional viewpoint, in the simplest sense, an ecosystem is defined as one in which the internal processes dominate over the trans-boundary processes. This is intuitively obvious for systems like coral reefs or lakes, which have a relatively high degree of enclosure or isolation from the larger systems in which they are embedded. The view of coral reefs as relatively closed systems, within the boundaries of which accurate budgets of biotic and abiotic materials may be derived, has profoundly influenced coral reef ecology, starting with the first application of systems ecology (e.g. Sargent and Austin 1949; Odum and Odum 1955).
S79
Ecosystem processes and models ‘‘Ecological systems become closed when transfer rates among adjacent systems approach zero or when differences in process rates between adjacent elements are so large that the dynamics of the elements are effectively decoupled from one another.’’ —J.A. ¼iens (1989) Twenty-five years ago, when systems ecology was still a young discipline, Howard Odum (1971 p. 10) published his ‘‘macroscope’’ view of the science: its message remains broadly relevant. As research moves from the descriptive to the predictive stage, we approach ‘‘How?’’ and ‘‘Why?’’ questions by making simplifying assumptions and constructing, testing and refining models based on these. The key of course is the simplifying assumption (Odum’s ‘‘detail eliminator’’). Ecological systems (especially coral reefs) are of such complexity that theoretical development always involves abstraction to another level of organization with fewer components or simpler dynamics, for which it is possible to collect data and which is easier to model (Wiegert 1988). If the abstraction is down the hierarchy of organization (e.g., modelling crown-of-thorns starfish outbreaks from laboratory measures of fecundity and larval survival) we call it ‘‘reductionism’’. If detail is eliminated
by measuring processes higher in the hierarchy (e.g., modelling COT outbreaks from sediment-climate relationships) we call it ‘‘holism’’ or ‘‘ecosystem science’’. (The two approaches are not irreconcilable, e.g., Pahl-Wostl 1993). Some of the most common ecosystem processes identified in holistic models of coral reefs are listed in Table 1. The simplifications are invariably embodied in a model, which is used to explain, to predict, possibly to control the ecosystem of interest (Reichelt et al. 1983; Jorgensen 1994). Samples from the range of reef models include a captive reef in a mesocosm (e.g., Adey and Loveland 1991), a huge convection cell (Rougerie and Waulthy 1993), a balanced biogeochemical budget (e.g., Smith 1988), a hierarchy of critical units (e.g., Bradbury 1977), or a big wheel of recycling materials with some small, but important and hard to measure, spokes radiating off (Hatcher 1996). All strive to capture the essential properties and processes of reefs. All explicitly or implicitly define system boundaries within certain spatial and temporal scales. Where they differ is in whether the reef system is viewed as essentially closed or open and the degree of complexity they incorporate (Table 2). Models of coral reef ecosystems are evolving from conceptually and methodologically simple, closed systems towards a view of reefs as open, complex systems which are more difficult to deal with but undoubtedly more
Table 1. Major ecosystem processes of coral reefs are listed with the relevant levels of biological organization. The types and spatial and temporal scales of measurements commonly made of these process are shown (ranked in decreasing scale), along with some references to examples Process
Measurements
Scales
Examples
Accretion (ecosystem, community) Biological production (community)
Vertical growth, margin expansion, infilling and erosion Net export and accumulation (P ) NEW Photosynthetic rates, community metabolism (E) Population growth, emigration and harvest rates Community respiration (R)
Barrier reef—reef matrix, 105—year Atoll—reef zone day—year Lagoon—colony week—hour Reef province—reef zone, decade—day Lagoon—colony week—hour Reef zone—matrix month—day Reef zone—matrix day—hour Ocean basin—reef matrix year—hour Atoll—reef zone year—day Lagoon—bommie hour—week Global ocean—reef zone 106—102 years
Davies and Hopley 1983 Kan et al. 1995 Smith 1988, Crossland et al. 1991 Adey and Steneck 1985 Gattuso et al. 1993 Munro 1977, Alcala and Gomez 1985 Atkinson and Grigg 1984 Hansen et al. 1987. Hatcher 1981, Riddle et al. 1990 Hopkinson et al. 1987 Sorokin 1995 Smith et al. 1981 Andrews and Muller 1983 Bilger and Atkinson 1995 Smith 1988, Crossland et al 1991 Smith and Harrison 1977 Gattuso et al. 1995 McGhee 1988, Sepkoski 1991, Jackson 1992 Pandolfi 1996.
Reef province—reef zone 103 y—month Reef province—reef bommie decade—day Reef—colony decade—colony
Connell 1978, Grigg 1983, Rogers 1992 Tanner et al. 1996 Doherty 1987, Black et al. 1991, Fisk and Harriot 1990 Buss and Jackson 1979 Wilkinson 1987
(population) Organic decomposition (community)
Biogeochemical cycling, export and accumulation (community)
Maintenance of biodiversity (biosphere, ecosystem) (community, population) (population) (population, organism)
Rates of herbivory and detritivory Microbial and anaerobic metabolism C, N and P pools and fluxes, hydrodynamics and kinematics Net export and accumulation (P ), sedimentation rates NEW Net calcification (G) Speciation, extinction, geographic radiation, adaptation and response to climate change Community succession, response and recovery from disturbance Dispersion, settlement and recruitment of reef organisms Competition, predation, parasitism and symbiosis
S80 Table 2. Models of coral reef ecosystems are classified into three basic types. Some analogs, the distinguishing characteristics, commonly used methods and examples of each type are listed Model type
Characteristics
Methods
Examples
Closed (Analogous to a coral, chemostat, oasis)
Strong boundary conditions, internalA external fluxes, selfsufficient, steady-state, retentive-conservative Trans-boundary fluxes large, strong external forcing, variable sourcesink relationships, physical transport processes dominate Thermodynamically open, energy and information dissipated rapidly, fractal, self-regulating, cybernetic
Biogeochemical budgets for hydrologic control volumes Trophodynamic biomass budgets Biogeochemical flux budgets Analytical and numerical hydrodynamic models and simulations Energy flow Pattern and network analysis Transition matrix analysis
Sargent and Austin 1949 Kinsey 1983, Smith 1988 Gattuso et al. 1995 Polovina 1984, Opitz 1995
Open (Analogous to a sponge, biofilter, river) Complex (Analogous to a dissipative structure, information network, city)
realistic (Table 2). The reality, of course, is continuum of closure in both the structural and functional attributes. In the more closed models, reefs are self-sufficient oases of high biomass in the ocean desert (Odum and Odum 1955). The reef ecosystem is analogous to an organism or a reaction chamber, with tight boundaries and dominant internal dynamics of recycling and self-seeding. Organizationally the system has high internal connectivity, is self-regulating, and has high persistence stability. Material and information are conserved, and transfers across system boundaries are small proportions of the total flux, most of which is internal. This model derives from early thinking of systems ecology, trophodynamics, island biogeography and biological accommodation, and from a focus on isolated atolls as study sites. Compared to many other marine ecosystems (e.g., estuaries, kelp forests) reefs are relatively isolated and self-contained. Closed models have useful applications (both to population and biogeochemical studies of clearly delimited reefs), but fail to capture important ecosystem processes that connect reefs to the surrounding environment (e.g., export of harvestable production). More open models portray reefs as sources and sinks embedded in a larger matrix of adjacent ecosystems (which may include other reefs). Analogous to big biofilters (Sorokin 1995), they strip plankton and nutrients from the advective stream, transforming them and accumulating or exporting the resultant materials and organisms. The system is driven by external forcing but has the capacity to attenuate oscillations. Connectivity with adjacent systems is high, and the reef can be a net importer or exporter of different materials, depending on the community structure and the physical environment in which it is embedded. Open models demand measures of boundary processes which are often methodologically difficult (e.g., larval dispersion, detrital loss) but are more amenable to the elucidation of causal relationships between forcing processes and system parameters (e.g., tidal currents and bioparticle retention in a lagoon). Thermodynamic models, long employed in the analysis of physical-chemical systems, are explicitly parameterized
Dahl et al. 1974, Smith et al. 1981, Eakin 1996 Black et al. 1991, Rougerie and Waulthy 1993, Bilger and Atkinson, 1995 Odum 1983 Reichelt et al. 1981 Johnson et al. 1993 Tanner et al. 1996
in terms of degree of closure. They have the potential to avoid the artificial dichotomy imposed by the need to balance budgets or define source or sink. In these models a coral reef is analogous to a city, with considerable structure and strongly hierarchical organization. They function within wide constraints to dissipate variable inputs of energy and information, but are subject to precipitous change of state near the limits of constraint on the faster processes (O’Neill et al. 1986; Pahl-Wostl 1993). Despite the obvious advantages of these abstractions, extensions of thermodynamic theory to complex systems, based on the thinking of Prigogine and Allen (1982) and others have been little used in reef systems to date. The Odums, of course, used energy flow as the basis for the first models of reef ecosystems (Odum and Odum 1955; Odum 1971), and the basic nomenclature of inputsoutputs, production-consumption, pools and fluxes, dissipation and control remain the dominant language of systems ecology (Odum 1983). Here ecosystem components are defined in terms of pools between which energy or materials flow and are grouped according to their mode (source or sink) and scaled according to their size and turnover rates. Energy, carbon and nutrient-based flux budgets have a long history of application to coral reefs (e.g., Johannes et al. 1972; Dahl et al. 1974), but nutrient currencies have received particular attention because nutrient supply is often assumed to limit some key process like nitrogen fixation or primary production (Smith 1984; Atkinson 1988). Other approaches use functional groupings of organisms as the system components and tropic or competitive interactions as the interconnecting processes (e.g., the topdown model of Polovina 1984). While there are clearly differences in the components and currencies used, the methods are not mutually exclusive and are often combined in ecosystem models such as the hybrid models currently being used by Pauly and Christensen (1995) to estimate the level of primary production required to support coral reef fisheries (for future reference, their number is 8.3% of reef net primary production).
S81
Simplifying by moving up a hierarchy (holism) invariably assumes a degree of functional redundancy within levels of organization and groups ecosystem components at those levels into single components at a higher level. For example, the functional groupings of free-living benthic algae used in models of reef benthic community structure and function (e.g., Littler et al. 1983; Steneck 1988) is a simplifying assumption that has served us well. Similarly, DeMartini et al. (1996) compared reef fish densities on two Hawaiian reefs sampled more than a decade apart. They were unable to detect any change in species populations, which were highly variable, but when fish were classified in functional groups, the power of the tests was sufficient to detect large declines in abundance. In complex ecosystems like coral reefs, functional classification of components at higher levels of biological organization is a well-proven tool for discerning the roles of ecological processes (e.g., Done et al. 1996). Whatever the detail eliminator we choose, whatever our model of the ecosystem, the goal remains the ‘‘principle technique of scientific enquiry: by changing the scale of description, we will move from unpredictable, unrepeatable individual cases to collections of cases whose behavior is regular enough to allow generalizations to be made’’ (Levin 1992). The recognition that the variability in ocean phytoplankton communities is best predicted by spectral analysis of mixed layer dynamics (e.g., Platt and Denman 1975; Mackas et al. 1985) is perhaps the most powerful unifying concept in marine ecology currently available. In the reef context, a most elegant (if less general) example is the tuning of spur and groove topography to the ocean swell regime. Almost fifty years ago Munk and Sargent (1948) showed how a dimensionless wave number predicted the length and frequency of these complex reef formations. Forced to supply criteria for distinguishing studies of ecosystem processes (Table 1), I point first to questions posed at the higher levels of organizational hierarchies
(whether biological or physical-process). Then I follow Levin (1992) in saying that studies of ecosystem processes ‘‘ 2 first ask how much variation can be explained by variation in the physical environment, and then look to autonomous biological factors to account for the balance.’’ Physical processes have been demonstrated to define and control a wide range of reef processes (Hatcher et al. 1987; Hamner and Wolanski 1988) and provide an avenue to analysis of ecological phenomena which span broad ranges of area and time (e.g., Bradbury and Young 1981; Andrews and Muller 1983; Sebens and Johnson 1991; Black et al. 1995; see also papers in Hughes 1993). Spatial-temporal scaling ‘‘The only things that can be universal, in a sense, are scaling things.’’ —M. Feigenbaum (1979) Several things happen as we move up nested hierarchies: space-time scales increase, rate constants get slower, environmental variables which perturb systems get incorporated (Table 1; Fig. 1). Constraints on system behavior may change from within to between levels in a hierarchy, or between hierarchies (Simon 1973; Webster 1979). For example, stochastic environmental processes may control highly variable reef fish density in guilds or patch reefs, but biotic interactions may operate to hold abundances constant at the level of the community (Planes et al. 1993). Coral reef ecosystems as we conceptualize them extend to the tens of kilometers in length scales and to tens of thousands of years along the time dimension (Fig. 1). Above these we have reef provinces and cycles of reef destruction and regrowth. Biological levels of organization group within, but also extend beyond, ecosystem boundaries. A population may exist wholly within the
Fig. 1. Spatial and temporal scales of coral reef ecosystem components and processes. The nested habitats of reefs are shown as bands on the spatial axis. Components of the biological hierarchy of organization occupy overlapping domains in the timespace plot. Oceanographic, climatic and geological processes are located according to the temporal-spatial extent of their discrete or continuous (arrow) distributions. The upper boundary of reef ecosystems (incorporating both biotic and abiotic components and processes) is shown to extend to ten of kilometers and about 10 000 y
S82
limits of a reef or extend well outside of it. While they may be nested in an organizational sense, the biological levels do not nest fully in space-time, but rather overlap to varying degrees. Habitats, on the other hand, form a fully nested set along the spatial continuum. Patch reefs occur within reef zones and consist of reef matrix, which in turn defines interstitial spaces. Physical processes may be local, relatively constant phenomena occurring over a large temporal range (e.g., submarine groundwater discharge); they may be periodic or sporadic phenomena, spanning a large range of distances (e.g., waves, storms: the causes of physical disturbance); or they may be scale spanning phenomena which extend continuously through the dimensions (e.g., eddy-diffusion). In these, larger is slower, in a linear relationship (e.g., the exchange turnover of water in reef lagoons decreases with their volume). A key message from scaling exercises (Hatcher et al. 1987; Fig. 1) is that ecosystem processes can be defined and measured at many scales (not just that of the whole reef), depending on the question being addressed and the observer’s perception (Table 1). One geochemist’s tidal pool is another’s lagoon; one physicist’s turbulence is the next’s bulk flow, one mathematician’s chaos is another’s equilibrium, one ecologist’s ecosystem is another’s environment. Different patterns emerge at different scales of observation. Our ability to predict ecological phenomena depends on whether (and how) processes change within their
domains of scale (i.e., they may be scale-dependent, such as those controlling reef fish community structure: Galzin 1987; or they may be scale-invariant such as those controlling the morphology of fish prey: Aronson, 1994), and on the form of the non-linearities at the transitions between domains (Allen 1985; Holling 1992; Knowlton 1992). A more subtle message of scaling analyses is that biological hierarchies do not separate as clearly or progressively in time-space as physical-process hierarchies (O’Neill et al. 1986). One useful scheme recognizes the pervasive influence of the degree of system closure (in the population, hydrodynamic and thermodynamic senses) on ecosystem processes involving the flux of organisms and materials and its apparent relationship to reef morphology (Fig. 2) due to the dominant role of water exchange in controlling such fluxes (e.g., Smith 1984; Hamner and Wolanski 1988; Black et al. 1990). Spatial-temporal scaling is not an easy concept to grasp and apply to coral reefs. Some workers are grappling with the science of scale using dimensional and spectral analysis, geostatistics, the mathematics of fractals and other cross-disciplinary tools (e.g., see collections of papers in Giller et al. 1994; Patten and Jorgensen 1995), but there have been few thorough applications to coral reef systems. Reef scientists working at any level of organization need particularly to recognize the space-time scaling of ecological processes because of the rather large range of scales encompassed by reefs (Fig. 1).
Fig. 2. The closure continuum applied to reef ecosystems. Schematics of three characteristic reef morphologies (bank barrier, fringing and atoll) are located along a gradient of increasing degree of closure of the processes that control ecosystem processes and the exchange of organisms, energy, nutrients and information with adjacent ecosystems. The juxtaposition of net autotrophic and heterotrophic zones is depicted by distinct bands and patches. Characteristic patterns of water flow over and around reefs are shown with arrows. The relative significance of retention and recycling of organic material is reflected in the size of the dashed ellipse. As reefs become less
hydrodynamically open and more isolated from external nutrient sources, long water residence times and self-supporting communities (e.g., symbioses) favor the recycling of organics within the reef and inhibit exchanges with the surrounding ocean. Correspondingly, the relative potential of nutrient limitation changes, the excess and new productivity decreases (causing a decrease in the f-ratio), and the rate at which the system dissipates energy and information is reduced. No reef systems are completely closed, and most fall towards the more open end of the continuum
S83
Emergent properties of coral reef ecosystems ‘‘In such (scaling) analysis, natural scales and frequencies will emerge, in which rests the essential nature of the system’s dynamics.’’ —Holling (1992) One reason to study ecosystem process in reefs is that theory predicts certain emergent properties from complex, highly organized systems (e.g., Patten and Odum 1981; Prigogine and Allen 1982), and coral reefs can provide useful tests of such predictions. Few have done this explicitly (e.g., Bradbury 1977; Reichelt et al. 1981; Miller 1991), but virtually all scientists make use of emergent properties to characterize coral reef structure and function (e.g. net community metabolism to quantify the trophic status of reefs, Kinsey 1983; diversity indices to quantify succession and evolution, Sepkoski 1992). The need for measurable parameters of complex systems is by far the more common reason to search for emergent properties of coral reefs (Table 1). Let it be accepted that ecosystem processes are those that operate at the scales of ecosystems (even if logistic constraints require us to measure them at smaller scales), and transcend the boundaries among levels of organization as they map onto space-time. Thus, for example, locomotion is not an ecosystem process because its magnitude, sign and scales are species-specific and cannot be meaningfully integrated for an entire community or habitat. Migration to off-reef feeding or breeding sites is an ecosystem process because it represents a loss of material from the ecosystem, regardless of whether it is quantified for a single species or an entire community. Because they ‘‘scale up’’, ecosystem processes can be seen to emerge from the system to provide some degree of generality. I recognize three types of emergence of ecosystem properties (Table 3). Firstly, they may be the summation of the results of ongoing, small-scale process over larger and longer scales. Photosynthetic production occurs only at the level of the chloroplast-cell system, but can be expressed sensibly at all levels up to the biosphere. I call this simplest form ‘‘additive emergence’’: the process is restricted to one level of biological organization, but the form of the resulting product differs among levels, and magnitude increases additively with time and spatial scale. Speciation can act in a similar way (although it is not usually referred to as an ‘‘ecosystem process’’). The causative mechanisms (mutation and selection) operate at the
level of the organism (or lower, —the gene), but are manifest at population, community and global levels. Aronson (1994) provides some marine examples of additive emergence from evolutionary biology. Emergent properties can also result from the upscale integration of faster, smaller processes into slower, larger processes which have some different properties and cannot be fully predicted from the sum of the component processes (this is the hierarchical model, Webster 1979; O’Neill et al. 1986). For example, reef accretion occurs within the coral-microalgal symbiosis as the CaCO skele3 ton is laid down, within the matrix as corals and other calcifiers produce, trap and bind carbonate framework and sediments, within the lagoon as it fills in with these products and within the entire reef ecosystem as it grows and infills in excess of loss processes such as dissolution and off-reef sedimentation. Reef growth is the result of the complex interaction of many different, smaller scale processes. We can think of this as ‘‘integration emergence’’: many small-scale processes interact (sometimes non-linearly) to produce the greater whole, which operates at larger, longer scales (Holling 1992). Thirdly, ecological emergence occurs as the manifestation of exchanges among levels of organization or across boundaries of ecosystems. In contrast to primary production, for example, the consumption of organic matter is a richly variable process (whether the mechanism is detritivory, herbivory or predation) that occurs at, but differs profoundly among, several levels of organization in reef ecosystems. For example, grazing fish exhibit a remarkable range of adaptations to reef plant characteristics and distribution, but they all convert plant tissue into gas, detritus and fish flesh. Subsequent transfers of organic material up to top predators (including man) or into microbial communities of the sediment are constrained by the net result of the grazers’ various modes of plant consumption. We could call this ‘‘difference emergence’’: the whole is defined by its balance of exchange with other levels in the system’s hierarchy. (This is the trophic dynamic property: Lindeman 1942; Polovina 1984). All three types of emergent properties have many examples in coral reef ecosystems, but the themes of similarity of pattern across scales (scale-independence, Aronson 1994: fractal geometry, Bradbury et al. 1984) and simplification up levels of organization persist. Ecosystem science attempts to capitalize on these effects by focusing measurement and analyses at the higher levels of organization. In doing so, it is both constrained and empowered
Table 3. Emergent properties of coral reef ecosystems are classified into three modes by which they arise from ecosystem hierarchies. The essential characteristics are listed along with examples of processes exhibiting that mode Mode of emergence
Characteristics
Example process
References
Additive
Processes sum linearly up levels of hierarchy Processes incorporate up hierarchy Processes differ across levels and interact nonlinearly up hierarchy
Photosynthetic production Reef growth (vertical accretion) Phase-shifts in benthic community dominance
Kinsey 1983, Atkinson and Grigg 1984 Davies and Hopley 1983 Miller 1991 Done 1992, Knowlton 1992 Aronson 1994
Integrative Differential
S84
to include the physical and geochemical processes which often dominate system dynamics at the larger and longer scales (Fig. 1). Looking for emergent properties in ecosystem processes requires empirical observations. As higher levels of organization generally encompass larger and longer scales, ecosystem science is faced with some difficult measurement challenges. One either measures a process at manageable scales and then sums up-scale, hoping the process is scale-invariant (additive emergence, Table 3), or one measures a variate which hopefully integrates the relevant processes at the scale of interest (integrative or difference emergence). The small-scale complexity (i.e. ‘‘fine grain’’, Wiens 1989) of coral reefs hinders the first approach, while their physiography and hydrodynamics greatly aid integrative measurements of ecosystem processes. Consider the example of primary productivity in reef ecosystems (Hatcher 1990, 1996). Photosynthesis and respiration can be measured at the level of the individual plant, the patch reef or the sediment community by enclosing them and sampling the controlled water volume to detect change in oxygen or carbon dioxide. The diel integral of the processes at the level of the plant is net photosynthetic productivity, while at the level of the enclosed community it is net community production (animal as well as plant respiration is included). At higher levels (and larger scales) of habitat integration the latter measure becomes the ecosystem parameter ‘‘E’’ (excess production: Kinsey 1983). It can be measured by following water flowing across a reef flat, or sampling the retained water in an entire lagoon. It can also be approximated by measuring the net flux of nutrients essential to community metabolism in or out of control volumes, as long as the ratio of nutrient to carbon or energy in the materials produced are known. In the old and elegant methods (Sargent and Austin 1949) the water mass integrates the process across time and space, allowing single, integrated measures of the metabolic performance of entire communities up to the scales of a lagoon-reef ecosystem (Smith 1988). The underlying processes of photosynthesis and respiration are the same from polyp to reef, but the magnitude of the net production term differs among them because more and different components contribute to the integrated signal at larger scales. It should be possible to calculate E for a lagoon by area-weighted averages of net community production for all the systems nested within it. The fact that such estimates never match those of the integrated process reflects not just the compounding of measurement error but the fact that the measurement time scales appropriate for a coral colony are far shorter than those for a lagoon (i.e., there is a measurement scale mismatch: Smith and Kinsey 1988). Ecological research is rife with such mismatches, it is certainly not a problem unique to reefs (Wiens 1989; Levin 1992). Simplistically, ecosystem processes are the emergent properties of ecosystems. While not unique in nature, the holistic integrity that arises from the complexity of physical and organic interactions makes coral reef ecosystems particularly significant to the scientists who study their processes.
Population versus systems approaches ‘‘One of the greatest barriers to the development of interfaces between population biology and ecosystems science is the perceived mismatch, especially as regards evolutionary processes 2 . There is considerable overlap between the spatial and temporal scales of interest to the population biologist and the ecosystem scientist.’’ —S.A. ¸evin (1992) Understandable and appropriate differences of opinion arise on the issues of which models, methods and spatialtemporal scales of observation sets are best suited to answering a given question about coral reef dynamics. This is not a trivial issue, since it has to do with the goals of science, processes of abstraction, model development, and the determination of boundary conditions. Some of the questions commonly addressed by reef scientists clearly define these issues. For example, the question asked by Kan et al. (1995) whether the growth of a fringing reef at high latitude could keep up with sea level rise during the Holocene clearly establishes the model as net vertical accretion rate versus local sea level rise and specifies the space-time scales and methods of measurement. One would not think of answering this question by measuring coral colony expansion, bioerosion and sedimentation rates on the present reef slope. The net accretion variate, as measured from dated strata in vertical cores, is the time space integral of all these processes which is scaled to the sea level curves. Most of the questions we ask about reefs cannot be so unambiguously assigned to ecosystem levels and scales. For example, the question of how storm waves alter the structure of coral reef communities asked by Massel and Done (1993) has been approached at levels of organization from the individual coral colony to entire reef tracts, at scales of meters to miles, and minutes to millennia. Not surprisingly, no two studies provide the same answer: undetectable to massive effects have been measured at all levels and scales. Physical disturbance is a strongly scaledependent process, such that the pattern of ecological effect measured is a function of the scale of observation (Wiens 1989; Aronson 1994; Christensen et al. 1996). There is yet to emerge a fundamental parameter for quantifying and predicting storm effects on reef communities (see papers in Hughes 1993). In this case, the scale of the observation set defines the effect of the process, and it is an ecosystem process only when expressed at the ecosystem (entire reef) scales. Though we may make different simplifying assumptions, apply different models, or measure with different tools when studying coral reef structure or function, we can all agree that the starting point is to identify the time and spatial scales of the system of interest and the processes being measured (Hatcher et al. 1987). Only then, can the environment, degree of enclosure (Fig. 2) and type of constraints (i.e. the boundary conditions) of a problem be set (e.g., Platt et al. 1984). Reef ecologists increasingly design their sampling or experiments to span a range of scales (Andrews and Muller 1983; Doherty 1987; Rogers 1992). For example Galzin (1987) measured reef fish distribution at spatial scales of meters to hundreds of
S85
kilometers to demonstrate that variability decreased with increasing scale. I have so far skirted the issue of differences between population-based and energy or mass-based approaches to the study of reef ecosystem processes. Mixing examples of models for coral species and chemical species purposely counters the conventional view that ecosystem processes are defined and measured only in terms of materials fluxes. It is time to discard this misconception, as community ecology has made major contributions to our understanding of ecosystem processes (e.g., O’Neill et al. 1986; PahlWostl 1993; Christensen et al. 1996). The two schools of ecology have produced different views of ecosystem processes. One holds the ecosystem concept as a fuzzy idea: useful perhaps for purposes of description but not as a model for prediction and control in systems of such taxonomic diversity and functional complexity as reefs. To meet those objectives, we must further develop quantitative models of the biotic components of ecosystems and their interactions: the organism, its species populations, and the communities they form. It is on these components that the bulk of ecological research has been conducted and to which the best-tested theorems apply (e.g., Connell 1978; Sale 1991). Deterministic or stochastic models of these systems, developed at the scales of the underlying mechanisms may be sufficient to predict the trajectory of a coral reef ecosystem through time-space. What is required now are better methods to combine and extend these measurements and models up to the levels of organization and scales at which we define coral reefs. One recent application uses simplified rules of interactions between components in transition matrix models to predict the structure of benthic communities at whole-habitat, decade scales which characterize many forms of disturbance in reef ecosystems (Tanner et al. 1995). A population-based approach is not inconsistent with the ‘‘macroscope’’. Indeed, ‘‘macroecology’’ continues to develop as a legitimate branch of the field (e.g., Brown 1995), which empirically relates statistical patterns of diversity, distribution and abundance to abiotic factors. The other view sees an ecosystem as a machine that processes material. The emergent properties which arise from interactions between the components demand quantification of processes in different modes and scales than those defined by the biotic components alone. To predict the behavior of a complex system like a coral reef may require the measurement and modelling of a small number of high-order variables which integrate component behavior and capture the essential functions of the entire system (Simon 1973; Field et al. 1989). Energy, chemical or information currencies are usually used in these models: in part because good physical-chemical theory and engineering models are available; in part because they integrate and are measurable at the larger, longer scales of ecosystems. Recent applications use network analysis of carbon fluxes to predict significant change in the trophic structure and function of reef communities in response to shifts from micro to macro-algal dominance (e.g., Johnson et al. 1995), or use whole-reef scale biogeochemical budgets to quantify the small contribution of reef metabolism to global climate change (e.g., Smith and Buddemeier 1992).
There is much to be learned about coral reefs from developing and exploring the contrasts between population-community and mass-balance approaches to the study of ecosystems. Certainly there is no shortage of relevant literature (e.g., Jones and Lawton 1994; Patten and Jorgensen 1995) but few good examples from reef research. In assessing ecosystem processes it is a mistake to focus only on the physical and biogeochemical approaches of systems ecology, because the abundance and behavior of species populations can have profound effects on material fluxes and budgets (e.g., Smith et al. 1981 showed how the presence of a single species of algae strongly affected a reef ecosystem’s response to change in sewage input). Reef ecologists, like those who seek emergent properties in other clearly-bounded ecosystems (e.g., lakes: Edmondson and Lehman 1981; Schlesinger and Reiger 1982), are not yet in a position to identify a single approach as ‘‘right’’: again, it depends on the question posed and the scale of the phenomenon under investigation. The alternative approaches are effectively reconciled in the ‘‘dual hierarchies’’ of O’Neill et al. (1986 p. 198), which recognize their differences but also their similarities. The traditional hierarchy of organization is split into a set of biotic levels and a set of functional levels, but the essentials of hierarchy theory (Webster 1979) are thought to apply to both (Pohl-Wostl 1993). It is necessary to consider both approaches, even when focusing on one. Some are working to understand how the details of lower levels inform processes at higher levels, others are working to predict system behavior with as few details as possible. The models, methods, time and space scales of our observation sets are appropriately diverse. It is a tribute to the maturity of coral reef science that this dichotomy is becoming a non-issue: I think we all recognize hierarchical organization and the cruciality of scaling in reef ecosystems. Regardless of the hierarchy and scales in which we choose to work, the behavior of that system may not necessarily predict the behavior of others nested within or without. Reef ecosystems (no matter how defined) are more than the sum of their nested parts (sub-systems). We are all, in some sense, striving to discover how much more: in this sense at least, we are all studying ecosystem processes. Current contributions of ecosystem research to coral reef science The existence of such a large, multidisciplinary scientific society as the ISCRS, focusing for a week on a single, rare type of seabed reflects the pervasive and compelling concept of the coral reef as an ecosystem, and demands the recognition of processes operating at the levels and scales of entire reefs. Management based on single species population models has been demonstrated to be impractical in the socioecological context of tropical reef ecosystems, while (largely untried) management based on higher levels of organization (i.e., habitat, ecosystem) is increasingly perceived as the only hope of conserving coral reefs (e.g., Bohnsack and Ault 1996; Christensen et al. 1996). There is a set of questions facing coral reef science which either explicitly or implicitly forces us to consider
S86
processes at the higher levels and larger-longer scales of ecosystems. They are the questions that now, and probably in the near future at least, will continue to set priorities for research: 1. In face of anthropogenic environmental change, will coral reefs continue to exist in form and function as they did during the pre-industrial Holocene? 2. What are the forms and magnitudes of organic and inorganic products and services man can remove from coral reefs on an ecologically sustainable basis? 3. In what ways and to what extent do coral reefs interact with adjacent, non-reef ecosystems? Scarcely a piece of reef research cannot be embedded in at least one of these questions, no matter the hierarchy, level or scale of the process it seeks to investigate. Ecosystem models and measurements of ecosystem processes are essential to all three. The first question is essentially the topic of the SCORE Working Group 104 — Coral reef responses to global change (Buddemeier 1995). Key ecosystem approaches here are the application of geological models of reef performance during the Pleistocene and Holocene to predict change under various greenhouse scenarios. Smith and Buddemeier (1992) convincingly use ecosystem process measurements and analytical tools to scale reef metabolism to carbon flux over a broad range of cases, demonstrating that normal reef metabolism usually results in small net efflux of CO to the atmosphere. There is, of 2 course, significant variation in the rate, but whether the net flux is positive or negative, the role of reefs in exacerbating or ameliorating atmospheric CO concen2 tration will always be minuscule on human time scales (Buddemeier, 1996), although possibly significant at geological scales (Opdyke and Walker 1992). Another aspect of the system response to environmental change question concerns shifts between alternate stable states for reef benthos: micro or macroalgal dominated (Hatcher 1984; Done 1992; Knowlton 1992). Models incorporating biotic forcing and non-linear interactions within and among system levels, which can be extended to whole reef and decadal scales, are needed here (e.g., May 1977; Field et al. 1989; Holling 1992). This is a tall order that should foster new applications of ecosystem analysis to reefs. The second and third questions of harvestable exports and transfers to and from adjacent ecosystems are more tractable, given current understanding of reef ecosystem processes and the models, tools and measurements at our disposal. The advantage is that reasonable answers to these questions can be made based primarily on established physical and chemical principles, and rather crude assumptions about the biology. The essential model is a biogeochemical budget in which the total fluxes of inorganic and organic matter are huge, but the net fluxes are so small that measurement becomes problematic, even with good, system level techniques (Smith 1988; Smith and Kinsey 1988). For example, excess production (E) by most reef ecosystems studied to date is only about 3% of their gross photosynthetic production, and the absolute magnitude of E approximates the same low value per unit area as the oligotrophic ocean (Crossland et al. 1991). Yet this
is the amount of organic material available to support both sustained reef growth and export to man and adjacent ecosystems. Whether the excess production of reefs in mesotrophic oceans is correspondingly higher remains an outstanding question because so few system-level measurements have been made of such reefs. Averaged estimates of excess production from the handful of wholereef measures available suggest that only about 15% of E accumulates in reef structure and that 10% at most is harvested by man (Crossland et al. 1991). The remaining 75% of E is assumed to be exported from the reef system. Recent measurements by Erez (1990) based on near-reef concentration gradients suggest that this number is of the right order for Red Sea fringing reefs. The magnitude of the proportion of E harvested and available for harvest by man from reefs is a key issue that clearly requires ecosystem-level models and the measurement of ecosystem processes. Expressed on an areal basis, the value derived from measures of community metabolism (Crossland et al. 1991) corresponds to about one twentieth of the primary production Pauly and Christensen (1995) estimated was required to sustain existing fish catches from reefs, based on the ECOPATH (top-down) model. The discrepancy between these estimates does not necessarily result only from error compounding in trophodynamic models (e.g., Grigg et al. 1984 obtained rather close agreement on estimates of net system productivity using ECOPATH and community metabolism measures). It suggests that the average reef (or at least the types that man fishes most intensely, such as fringing reefs around densely-populated coasts) has a higher E than most reefs measured to date, or that more of the E remains in these systems in a form suitable to support harvestable secondary production. The question of variability amongst reef systems used to calibrate and verify these models is a crucial one. The bulk of the data come from about sixty measurements of reef flat metabolism at sea level (Kinsey 1983, 1991). They give the tight modal results that have led to the sweeping generalization that entire reef systems (and, by areal scaling, the global reef signal) have P/R ratios only slightly greater than one, corresponding to values of E only slightly greater than zero (Smith 1988; Crossland et al. 1991; Gattuso et al. 1995). The outliers are interesting in that they are either reefs which have not yet reached sea level (the high E reefs, e.g., Adey and Steneck 1985) or which receive substantial inputs of organic material (the negative E reefs, Kinsey 1983). Such reefs are abundant throughout the world as bank barrier and fringing types, which are generally closer to human development than emergent barriers and atolls. It follows that a more representative model of coral reefs’ capacities to sustainably export organics (and perhaps a reconciliation of the discrepancies between reductionistic and holistic measurements of ecosystem processes) will now be better served by studies of submerged, near shore, and nutrient enriched reefs than by more sampling of reef flats at sea level in the oligotrophic ocean. Measuring community metabolism in such systems (which are often very open hydrodynamically and biogeochemically) poses serious methodological problems, but they are not insurmountable (e.g., Smith and Harrison 1977; Rogers 1979; Adey and Steneck 1985).
S87
New production in reef ecosystems ‘‘The concept of new production provides a conceptual basis for the relationship between nutrient supply and ecosystem function, and how these relate to the earth’s biogeochemical cycles.’’ —¹. Platt, P. Jauhari and S. Sathyendranath (1992) One of the most informative ecosystem measures results from the partitioning of organic production into a maintenance term and a net term. Derived originally from energy budgets for individual organisms (another example of the scale-independence of some ecosystem models), the parameterization recognizes that the component of production not consumed within an ecosystem is available for accumulation or export. Obviously this bears directly on the questions of nutrient supplies to, and organic exports from, reefs. The key concept here is that of ‘‘new production’’ (P ), as originally defined in nitrogen currency for the NEW photic zone of the pelagic ecosystem by Dugdale and Goering (1967). It is the component of primary production which is derived from allochthonous nutrient inputs. They used the net nitrogen flux in a dynamic equilibrium box model of the mixed layer to partition the primary
Fig. 3. The concept of new production (P ) applied to the coral N%8 reef ecosystem. A schematic diagram splits the primary production of organic material (P ), resulting from the processes of photosynT05!thesis (P) and nutrient assimilation, into a component (P ) that N%8 draws on allochthonous nutrients and a complementary one (P ) R%'%/ that uses nutrients remineralized from organic material already in the system. The analogous terms formalizing community metabolism (excess production and respiration by the entire community) are shown in brackets. For simplicity the components and processes of secondary production based on P are not shown, and the orT05!ganic pool includes both living and detrital organic material. Net inputs of nutrients to the reef (most of the total input is simply advected through) are depicted from the atmosphere (as dinitrogen), and from the ocean as both dissolved inorganic nitrogen and phosphorus and particulate organic material containing these nutrients, which are delivered to the reef hydrodynamically. The main pathways of nutrients in simplified marine systems for which P has N%8
production into new and recycled components (Fig. 3). The convenient nitrogen dynamics of the photic zone mean that new production is well-estimated by the flux of nitrate into the system (mainly through vertical mixing), while the remineralization of organic matter within the system is the main source of ammonium. Thus, ammonium assimilation estimates the recycled production: that component of primary production derived from autochthonous nutrients. The two types of production are obviously complementary and additive to yield the total net primary production in the system. The assumed proportionality of nitrogen flux to new organic production means that the magnitude and type of system production can be inferred simply by measuring and budgeting the two inorganic nitrogen species in a controlled volume of seawater (be it a bottle or an upwelling zone). Subsequent work has developed the concept of new production into a powerful generality by using the dimensionless ratio ‘‘f ’’ (i.e., the ratio of P to total primary NEW production) to effectively classify the degree of closure and trophic status of pelagic production systems in different parts of the ocean (despite significant discrepancies between the theoretical and operation definitions of the terms, e.g., Platt et al. 1992). The concept is directly relevant to the questions of limitation, accumulation and
been estimated are shown with bold arrows: the uptake by plants of new DIN delivered by vertical mixing, the uptake of ‘‘old’’ DIN remineralized from organic production within the system, and the loss of organic materials from the system due to the sinking of organic particles. In reef systems atmospheric and advective inputs of nutrients (including gaseous and organic forms respectively) take on a greater significance, resulting in other nutrient pathways (dashed arrows). The uptake by plants of nutrients derived from the capture, metabolism and subsequent remineralization of ‘‘new’’ particulate organic material (N ) is a potentially significant contribuPOM tion to P of reefs, which is not reflected in estimates of net system N%8 production based on community metabolism (i.e., E is less than P by an amount equivalent to the respiration of the N ). N%8 POM Export of system-produced organics greatly exceeds the loss to sedimentation in reef systems, and in dynamic equilibrium advective losses and harvests will approximate P but may be underN%8 estimated by E
S88
export that characterize the study of ecosystem processes in coral reefs. The measurement and calculation of P (and the f-ratio) for coral reefs, however, is much NEW more difficult than for the pelagic ecosystem because of the variety and magnitude of nutrient supplies to support primary production (Fig. 3). The reef is fixed relative to the water mass, so advective processes such as topographic upwelling, boundary-layer mixing and wave pumping deliver large amounts of nutrient in both dissolved and particulate form (Atkinson 1988; Hamner and Wolanski 1988). Dinitrogen fixation (atmospheric input) is also a significant contributor of allochthonous nitrogen to support new production in reef systems, and losses to the atmosphere (denitrification) are also likely to be substantial in some reef habitats (D’Elia and Wiebe 1990). The complexity of the nitrogen cycle in reefs means that simple nitrate and ammonium fluxes cannot be used to partition new and recycled production, but estimates of P based on budgets of phosphorus should be possible NEW (Smith 1984, 1988). The source of inorganic nutrients to support P in NEW atolls has been postulated to be the deep ocean water upwelled through the reef matrix by thermal convection (Rougerie et al. 1992). An impressive body of biogeochemical research demonstrates the existence of geoendothermal upwelling, but the significance of its contribution to P is difficult to assess because of the operational NEW problems in distinguishing the new nutrient supply from the rapid regeneration of nutrients from organic matter in the coral reef matrix (Tribble et al. 1994). The fact that much of the advective flux of nutrients to reefs is in the form of particulates further complicates the estimation of P . Some of the material is captured by NEW the reef as plankton (i.e., by the ‘‘wall of mouths’’, Hamner and Wolanski 1988) and subsequently metabolized. Inputs of nutrients in organic form must first be remineralized before they can support new production (Fig. 3), but they represent a potentially large (and until recently, overlooked) supply of allochthonous nutrients to reefs (Kinsey 1991). Of course, great quantities of organic material are also advected out of reef systems as detritus, so the net result of particulate exports minus imports provides another means of estimating P (assuming the NEW system’s organic pool is in steady state). Estimating the magnitude of particulate fluxes at whole-reef scales is difficult, but measures by Erez (1990) and Ayukai (1995) demonstrate that the nutrient inputs in the form of particulates captured by the reef can be of the same order as the net system production! In defining P for reefs it is tempting to equate it with NEW excess production (Kinsey’s ‘‘E’’: 1983). Although the indices are conceptually similar, the operational definitions and methodological constraints differ. Excess production is derived from community metabolism measurements that balance total system photosynthetic production with the respiration of all reef organisms. The remainder (E) underestimates P for reef ecosystems because partiNEW culate organic inputs must first be respired before the nutrients they contain can be used to support new production. Reef systems in which organic nutrient inputs elevate P will have a correspondingly higher community respiNEW ration, thereby depressing E. Thus may explain in part
why E has been found to be so constant across a rather large range of reef ecosystems (Kinsey 1991), despite large discrepancies in measures of yield (export production). The bottom line is that P of reefs differs from, and is NEW potentially greater than, E by some variable, but possibly large amount. The implication is that the proportion of production available for harvest or export to adjacent ecosystems based on measurements of community metabolism may be underestimated for some reef systems. The discrepancy will be most pronounced in more open reef systems (Fig. 2) receiving significant particulate inputs and supporting high macroalgal biomass (i.e., those most likely to be subject to anthropogenic influence). These are the types of reefs for which we have the fewest measures of ecosystem processes and for which the demand for management science is greatest. Indices of ecosystem function like E and P capture NEW emergent properties of entire communities (e.g., their trophic status or degree of internal and external coupling) and can be used to assess their responses to variation in the environment (e.g., as the f-ratio predicts plankton community performance in oligotrophic and eutrophic areas of the ocean, Platt et al. 1992). Large-scale (geographic) differences in ecological processes in reef ecosystems also appear to be closely related to variations in the nutrient supply regime (Hallock and Schlager 1986; Birkeland 1996). Models that predict the response of reef production processes to local and global increases in nutrient supply are the most urgent requirement of ecosystem science. The question of nutrient limitation of reef productivity is of particular interest from both a theoretical and management perspective (Atkinson 1988). In contrast to much of the ocean, the assumption of nutrient limitation of net primary productivity is questionable in reef systems. Experimental work at scales of small portions of reef habitat (e.g., Kinsey and Davies 1979; Bilger and Atkinson 1995), up to entire reef lagoons (e.g., Grigg et al. 1984; Smith 1988) suggest that the nutrient supply rate exceeds that required to support the observed net productivity of reef ecosystems (even though restrictions on uptake may limit gross production rates). If nutrients do not regulate primary production on reefs, however, it challenges the current formalization of new production (Platt et al. 1992). Finally, because the major pool and pathway of export from reef systems is detrital organic material (Alongi 1988; Sorokin 1995), measurements of magnitude, form and fate of detritus in coral reef ecosystems provide indices of the trophic status and processes controlling export production. For example, reefs that export large quantities of macroalgal detritus (e.g., Killar and Norris 1988) exhibit an imbalance of primary production and grazer consumption compared to other reefs (e.g., Crossland and Barnes 1983) that may indicate progress towards a benthic community phase shift long before it is measurable as an increase in macroalgal biomass. Although decomposition processes are more diverse and difficult to measure than production processes, their study is becoming a growth industry (e.g., Ducklow 1990; Riddle et al. 1990; Sorokin 1995) with major benefits for our understanding of the function of coral reef ecosystems.
S89
Synthesis and closure ‘‘The likelihood that measurements made at a particular scale will reveal something about ecological mechanisms is a function of the openness of the system’’ —J.A. ¼iens (1989) In trying to synthesize this discussion of reef ecosystem processes, I return to the concept of the closure continuum (Fig. 2). For many of the questions discussed here, the appropriate models and methods are largely determined by the position of the ecosystem on this continuum. In reef ecosystems the degree of hydrodynamic closure is the primary determinant of that position, thereby affecting a broad range of ecosystem processes from nutrient uptake to competitive dominance of certain corals. Reef ecosystems range from very open linear bank barrier reefs far from shore, which have not yet reached sea level, through fringing reefs with nearshore lagoons, to emergent atolls with enclosed lagoons and multiple patch reef systems. With increasing hydrodynamic, thermodynamic and organizational closure, the physical and biological processes that enhance larval retention, nutrient recycling and homeorhesis increase, and the connections with adjacent systems and production available for growth and export decrease. To answer the key questions facing coral reef science in the next millennium, I believe that we should attempt both population-community and materials-flux measurements across a range of spatial-temporal scales in the more open types of reef systems. Comparisons of ecosystem processes among reefs far apart on the closure continuum will improve our models and contribute to the quest for some unified body of theory. Acknowledgements. Thanks to Richard Aronson, Annamarie Hatcher and Donald Kinsey for stimulating discussions, and to Nancy Knowlton, Steve Smith and Bill Wiebe for helpful reviews which significantly improved the manuscript. I acknowledge the many reef scientists whose work educated and inspired me but was unable to cite herein. A. Hatcher’s help drafting the figures is much appreciated. Support was provided through the Canadian International Development Agency and the organizers of the Eighth International Coral Reef Symposium, for which I am grateful.
References Adey WH, Loveland K (1991) Dynamic aquaria: building living ecosystems. Academic Press, San Diego, pp 513—540 Adey WA, Steneck RS (1985) Highly productive eastern Caribbean reefs: synergistic effects of biological, chemical, physical, and geological factors, In: Reaka ML (ed) The ecology of coral reefs, NOAA Undersea Res Program, Rockville, MD 3 : 163-187 Alcala AC, Gomez ED (1985) Fish yields of coral reefs in the central Philippines. Proc 5th Int Coral Reef Symp 1 : 75—81 Allen PM (1985) Ecology, thermodynamics, and self-organization: towards a new understanding of complexity. In: Ulanowitcz RE, Platt T (eds) Ecosystem theory for biological oceanography. Can Bull Fish Aquat Sci 213 : 3—26 Alongi DM (1988) Detritus in coral reef ecosystems: fates and fluxes. Proc 6th Int Coral Reef Symp 1 : 29—36 Andrews JC, Muller H (1983) Space-time variability of nutrients in a lagoonal patch reef. Limnol Oceanogr 28 : 215—227 Aronson RB (1994) Scale-independent biological processes in the marine environment. Ann Rev Oceanogr Mar Biol 32 : 435—460
Atkinson MJ (1988) Are coral reefs nutrient limited? Proc 6th Int Coral Reef Symp 1 : 157—166 Atkinson MJ, Grigg RW (1984) Model of a coral reef ecosystem II. Gross and net benthic primary production at French Frigate Shoals, Hawaii. Coral Reefs 3 : 13—22 Ayukai T (1995) Retention of phytoplankton and planktonic microbes on coral reefs within the Great Barrier Reef, Australia, Coral Reefs 14 : 141—147 Bilger RW, Atkinson MJ (1995) Effects of nutrient loading on mass transfer rates to a coral community. Limnol Oceanogr 40 : 279—289 Birkeland CE (1996) Geographic differences in ecological processes. In: Birkeland CE (ed) Life and death of coral reefs. Chapman and Hall, New York, pp 273—287 Black KP, Gay SL, Andrews JC (1990) Residence times of neutrallybuoyant matter such as larvae, sewage or nutrients on coral reefs. Coral Reefs 9 : 105—114 Black KP, Moran P, Burrage D, De’ath G (1995) Association of low-frequency currents and crown-of-thorns starfish outbreaks. Mar Ecol Prog Ser 125 : 185—194 Black KP, Moran PJ, Hammond LS (1991) Numerical models show coral reefs can be self-seeding. Mar Ecol Prog Ser 74 : 1—11 Bradbury RH (1977) Independent lies and holistic truths: towards a theory of coral reef communities as complex systems. Proc 3rd Int Coral Reef Symp 1 : 1—8 Bradbury RH, Young PC (1981) The effects of a major forcing function, wave energy, on a coral reef ecosystem. Mar Ecol Prog Ser 5 : 229—241 Bradbury RH, Reichelt RE, Green DG (1984) Fractals in ecology: methods and interpretation. Mar Ecol Prog Ser 14:295—296 Brown JH (1995) Macroecology. University of Chicago Press, Chicago, Ill 286 pp Bohnsack JA, Ault JS (1996) Management strategies to conserve marine biodiversity. Oceanogr 9 : 73—82 Buddemeier RW (1995) Coral reef responses to global change: the role of adaptation. Reef Encount 17 : 7—8 Buddemeier RW (1996) Coral reefs and carbon dioxide. Science 271 : 1298—1299 Buss LW, Jackson JBC (1979) Competitive networks: nontransitive competitive relationships in cryptic coral reef environments. Am Nat 113 : 223—234 Christensen NL, Bartuska AM, Brown JH, Carpenter S, D’Antonio C, Francis R, Franklin JF, MacMahon JA, Noss RF, Parsons DJ, Peterson CH, Turner MG, Woodmansee RG (1996) The report of the Ecological Society of America committee on the scientific basis for ecosystem management. Ecol Appl 6 : 665—691 Connell JH (1978) Diversity in tropical rain forests and coral reefs. Science 199 : 1302—1310 Crosby MP, Drake SF, Eakin CM, Fanning NB, Paterson A, Taylor PR, Wilson J (1995) The United States Coral Reef Initiative: an overview of the first steps. Coral Reefs 14 : 1—3 Crossland CJ, Barnes DJ (1983) Dissolved nutrients and organic particulates in water flowing over coral reefs at Lizard Island. Aust J Mar Freshwater Res 34 : 835—844 Crossland CJ, Hatcher BG, Smith SV (1991) The role of coral reefs in global carbon production. Coral Reefs 10 : 55—64 Dahl AL, Patten BC, Smith SV, Zieman JC (1974) A preliminary coral reef ecosystem model. Atoll Res Bull 172 : 7—36 Davies PJ, Hopley D (1983) Growth fabrics and growth rates of Holocene reefs in the Great Barrier Reef. BMR J Aust Geol Geophys 8 : 237—251 D’Elia CF, Wiebe WJ (1990) Biogeochemical nutrient cycles in coral reef ecosystems. In: Dubinsky Z (ed) Coral reefs. Elsevier, New York (Ecosystems of the World, vol 25) pp 49—74 DeMartini EE, Parrish FA, Parrish JD (1996) Interdecadal change in reef fish populations at French Frigate Shoals and Midway Atoll, northwestern Hawaiian islands: statistical power in retrospect. Bull Mar Sci 58 : 804—825 Doherty PJ (1987) The replenishment of populations of coral reef fishes, recruitment surveys and the problems of variability manifest on multiple scales. Bull Mar Sci 41 : 411—422
S90 Done TJ (1992) Phase shifts in coral reef communities and their ecological significance. Hydrobiologia 247 : 121—132 Done TJ, Ogden JC, Wiebe WJ, Rosen BR and 26 co-workers in the BIOCORE Working Group (1996) Biodiversity and ecosystem function of coral reefs. In: Mooney HA, Cushman JH, Medina E, Sala OE, Schulze E-D (eds) Functional roles of biodiversity. A global perspective. J. Wiley & Sons, New York, pp 393—429 Ducklow HW (1990) The biomass, production and fate of bacteria in coral reefs. In: Dubinsky Z (ed) Coral reefs. Elsevier, New York (Ecosystems of the World, vol 25) pp 265—290 Dugdale RC, Goering JJ (1967) Uptake of new and regenerated forms of nitrogen in primary productivity. Limnol Oceanogr 23 : 196—206 Eakin CM (1996) Where have all the carbonates gone? A model comparison of calcium carbonate budgets before and after the 1982—1983 El Nino at Uva Island in the eastern Pacific. Coral Reefs 15 : 109—119 Edmondson WT, Lehman JT (1981) The effect of changes in the nutrient income on the condition of Lake Washington. Limnol Oceanogr 26 : 1—29 Erez J (1990) On the importance of food source in coral-reef ecosystems. In: Dubinsky Z (ed) Coral reefs. Elsevier, New York (Ecosystems of the World, vol 25) pp 411—418 Fagerstrom JA (1987) The evolution of reef communities. WileyInterscience, New York, pp 3—15 Feigenbaum M (1979) The universal metric properties of nonlinear transformations. J Stat Phys 21 : 669—706 Field JG, Wulff F, Mann KH (1989) The need to analyze ecological networks. In: Wulff F, Field JG, Mann KH (eds) Network analysis in marine ecology. Coast Est Stud 32, Springer, Berlin, pp 3—12 Fisk DA, Harriot VJ (1990) Spatial and temporal variation in coral recruitment on the Great Barrier Reef: implications for dispersal hypotheses. Mar Biol 107 : 485—490 Fosberg FR (1961) Qualitative description of the coral atoll ecosystem. Atoll Res Bull 81 : 1—11 Galzin R (1987) Structure of fish communities of French Polynesian coral reefs: 1. Spatial scales. Mar Ecol Prog Ser 41 : 129—136 Gattuso JP, Pichon M, Frankignoulle M (1995) Biological control of air-sea CO fluxes: effect of photosynthetic and calcifying organisms and2ecosystems. Mar Ecol Prog Ser 129 : 307—312 Giller PS, Hildrew AG, Raffaelli DG (eds) (1994) Aquatic ecology: scale, pattern and process. Blackwell Scientific, Oxford, 649 pp Grigg RW (1983) Community structure, succession and development of coral reefs in Hawaii. Mar Ecol Prog Ser 11 : 1—14 Grigg RW, Polovina JJ, Atkinson MJ (1984) Model of a coral reef ecosystem III. Resource limitation, community limitation, fisheries yield and resource management. Coral Reefs 3 : 23—27 Hallock P, Schlager W (1986) Nutrient excess and the demise of coral reefs and carbonate platforms. Palaios 1 : 55—64 Hamner WM, Wolanski E (1988) Hydrodynamic forcing functions and biological processes on coral reefs: a status review. Proc Sixth Int Coral Reef Symp 1 : 103—113 Hansen JA, Alongi DM, Moriarty DJW, Pollard PC (1987) The dynamics of benthic microbial communities at Davies Reef, central Great Barrier Reef. Coral Reefs 6 : 63—70 Hatcher AI (1985) The relationship between coral reef structure and nitrogen dynamics. Proc 5th Int Coral Reef Symp 2 : 166—172 Hatcher BG (1981) The interaction between grazing organisms and the epilithic algal community of a coral reef: a quantitative assessment. Proc 4th Int Coral Reef Symp 2 : 515—524 Hatcher BG (1984) A maritime accident provides evidence for alternate stable states in benthic communities on coral reefs. Coral Reefs 3 : 199—204 Hatcher BG (1990) The primary productivity of coral reefs: a hierarchy of pattern and process. Trends Ecol Evol 5 : 149—155 Hatcher BG (1996) Primary productivity and decomposition. In: Birkeland CE (ed) Life and death of coral reefs. Chapman and Hall, New York pp 140—174 Hatcher BG, Imberger J, Smith SV (1987) Scaling analysis of coral reef systems: an approach to problems of scale. Coral Reefs 5 : 171—181
Holling CS (1992) Cross-scale morphology, geometry and dynamics of ecosystems. Ecol Monogr 62 : 447—502 Hopkinson CS, Sherr BR, Ducklow HW (1987) Microbial regeneration of ammonium in the water column of Davies Reef, Australia. Mar Ecol Prog Ser 41 : 147—153 Hubbard DK (1988) Controls of modern and fossil reef development: common ground for biological and geological research. Proc 6th Int Coral Reef Symp 1 : 243—251 Hubbell SP (1997) A theory of immigration, extinction, and the relative abundances of species in rain forests and coral reefs. Proc 8th Int Coral Reef Symp (in press) Hughes T (ed) (1993) Disturbance: effects on coral reef dynamics. Coral Reefs 12 : 115—233 Jackson JBC (1992) Pleistocene perspectives on coral reef community structure. Am Zool 32 : 719—731 Johannes RE and 22 others (1972) The metabolism of some coral reef communities: a team study of nutrient and energy flux at Eniwetak. BioScience 22 : 541—543 Johnson CR, Klumpp DW, Field J, Bradbury RH (1995) Carbon flux on coral reefs: effects of large shifts in community structure. Mar Ecol Prog Ser 126 : 123—143 Jones C, Lawton J (eds) (1995) Linking species and ecosystems. Chapman and Hall, New York, 396 pp Jorgensen SE (1994) Fundamentals of ecological modelling. Elsevier, Amsterdam (Developments in environmental modelling, vol 19) pp 5—15 Kan H, Hori N, Nakashima Y, Ichikawa K (1995) The evolution of narrow reef flats at high-latitude in the Ryuku Islands. Coral Reefs 14 : 123—130 Killar JA, Norris JN (1988) Composition, export, and import of drift vegetation on a tropical, plant-dominated, fringing reef platform (Caribbean, Panama). Coral Reefs 7 : 93—103 Kinsey DW (1983) Standards of performance in coral reef primary production and carbon turnover. In: Barnes, DJ (ed) Perspectives on coral reefs, Australian Institute of Marine Science Townsville, pp 209—220 Kinsey DW (1991) The coral reef: an owner-built, high-density, fully-serviced, self-sufficient housing estate in the desert — or is it? Symbiosis 10 : 1—22 Kinsey DW, Davies PJ (1979) Effects of elevated nitrogen and phosphorus on coral reef growth. Limnol Oceanogr 24 : 935—940 Knowlton N (1992) Thresholds and multiple stable states in coral reef community dynamics. Am Zool 32 : 674—682 Levin SA (1992) The problem of pattern and scale in ecology. Ecol 73 : 1943—1967 Lindeman RL (1942) The trophic dynamic aspect of ecology. Ecol 23 : 399—418 Littler MM, Littler DS, Taylor PR (1983) Evolutionary strategies in a tropical barrier reef system: functional-form groups of marine macroalgae. J Phycol 19 : 229—237 Mackas DL, Denman KL, Abbott MR (1985) Plankton patchiness: biology in the physical vernacular. Bull Mar Sci 37 : 652—674 Massel SR, Done TJ (1993) Effects of cyclone waves on massive coral assemblages on the Great Barrier Reef : meteorology, hydrodynamics and demography. Coral Reefs 12 : 153—166 May RM (1977) Thresholds and breakpoints in ecosystems with a multiplicity of stable states. Nature (London) 269 : 471—477 McGhee GR (1988) The late Devonian extinction event: evidence for abrupt ecosystem collapse. Palaeobiology 14 : 250—257 Miller IIIrd W (1991) Hierarchical concept of reef development. Neus Jahrb Geol Palaontol Abh 182 : 21—35 Munk WH, Sargent MC (1948) Adjustment of Bikini Atoll to ocean waves. Trans Am Geophys Union 29 : 855—860 Munro JL (1977) Actual and potential fish production from the coralline shelves of the Caribbean Sea. FAO Fish Rep 200 : 301—321 Odum EP (1953) Fundamentals of ecology. Saunders, Philadelphia, 574 pp Odum HT (1971) Environment, power and society. Wiley-Interscience, New York, 331 pp
S91 Odum HT (1983) Systems ecology, Wiley Interscience, New York, 644 pp Odum HT, Odum EP (1955) Trophic structure and productivity of a windward reef community on Eniwetok Atoll. Ecol Monogr 25 : 291—32 O’Neill RV, DeAngelis DL, Waide JB, Allen TFH (1986) A hierarchical concept of ecosystems. Princeton University Press, New Jersey, 253 pp Opdyke BN, Walker JCG (1992) Return of the coral reef hypothesis: basin to shelf partitioning of CaCO and its effect on atmospheric CO . Geology 20 : 733—736 3 2 Opitz S (1995) Quantitative models of trophic interactions in Caribbean coral reefs. ICLARM Tech Rep 43, Manila, 87 pp Pahl-Wostl C (1993) The hierarchical organization of the aquatic ecosystem: an outline how reductionism and holism may be reconciled. Ecol Model 66 : 81—100 Pandolfi JM (1997) Limited membership in Pleistocene coral assemblages from the Huon Peninsula, Papua New Guinea: constancy during global change. Palaeobiol 22 : 152—176 Patten BC, Jorgensen SE (eds) (1995) Complex ecology : the wholepart relation in ecosystems. Prentice Hall, New Jersey, 705 pp Patten BC, Odum EP (1981) The cybernetic nature of ecosystems. Am Nat 118 : 886—895 Pauly D, Christensen V (1995) Primary production required to sustain global fisheries. Nature 374 : 255—257 Planes S, Levefre A, Legendre P, Galzin R (1993) Spatio-temporal variability in fish recruitment to a coral reef (Moorea, French Polynesia). Coral Reefs 12 : 105—113 Platt T, Denman KL (1975) Spectral analysis in ecology. Annu Rev Ecol Syst 6 : 189—210 Platt T, Lewis M, Geider R (1984) Thermodynamics of the pelagic ecosystem: elementary closure conditions for biological production in the open ocean. In: Fasham MJR (ed) Flows of energy and materials in marine ecosystems. Plenum Press, New York, pp 49—84 Platt T, Jauhari P, Sathyendranath S (1992) The importance and measurement of new production. In: Falkowski PG, Woodhead AD (eds) Primary productivity and biogeochemical cycles in the sea. Plenum Press, New York, pp 273—284 Polovina JJ (1984) Model of a coral reef ecosystem 1. The ECOPATH model and its application to French Frigate Shoals. Coral Reefs 3 : 1—11 Preobrazhensky BV (1977) Problems of studying coral-reef ecosystems. Helg Wiss Meeresunters 30 : 357—361 Prigogine I, Allen PM (1982) The challenge of complexity. In: Schieve R, Allen PM (eds) Self organization and dissipative structures. Texas University Press, Austin, TX pp 3—21 Reichelt RE, Green DG, Bradbury RH (1983) Explanation, prediction and control in coral reef ecosystems 1. Models for explanation. Proc Great Barrier Reef Conf, James Cook University Townsville, pp 231—236 Riddle MJ, Alongi DM, Dayton PK, Hansen JA, Klumpp DW (1990) Detrital pathways in a coral reef lagoon. Mar Biol 104 : 109—118 Rogers CS (1979) The productivity of San Cristobel Reef, Puerto Rico. Limnol Oceanogr 24 : 342—349 Rogers CS (1992) A matter of scale: damage from hurricane Hugo (1989) to US Virgin Island reefs at the colony, community and whole reef level. Proc 7th Int Coral Reef Symp 1 : 127—133 Rougerie F, Waulthy B (1993) The endo-upwelling concept: from geothermal convection to reef construction. Coral Reefs 12 : 19—30
Rougerie F, Fagerstrom JA Andrie C (1992) Geothermal endoupwelling: a solution to the reef nutrient paradox. Cont Shelf Res 12 : 785—798 Sale PF (1991) Reef fish communities: open, non-equilibrial systems. In: Sale PF (ed) The ecology of fishes on coral reefs. Academic Press, San Diego, pp 564—598 Sargent MC, Austin TS (1949) Organic productivity of an atoll. Am Geophys Union Trans 30 : 245—249 Schindler DW (1990) Experimental perturbations of whole lakes as tests of hypotheses concerning ecosystem structure and function. Oikos 57 : 25—41 Schlesinger DA, Reiger HA (1982) Climatic and morphoedaphic indices of fish yields from natural lakes. Trans Am Fish Soc 111 : 141—150 Sebens KP, Johnson AS (1991) Effects of water movement on prey capture and distribution of reef corals. Hydrobiologica 226 : 91—101 Sepkoski JJ (1991) Phylogenetic and ecological patterns in the Phanerozoic history of marine biodiversity. In: Eldredge N (ed) Systematics, ecology and the biodiversity crisis. Columbia University Press, New York, pp 77—100 Simon HA (1973) The organization of complex systems. In: Pattee, HH (ed) Hierarchy theory. Brazillier, New York, pp 3—27 Smith SV (1984) Phosphorus versus nitrogen limitation in the marine environment. Limnol Oceanogr 29 : 1149—1160 Smith SV (1988) Mass balance in coral reef-dominated areas. In: Jansson, B.-O. (ed) Lecture notes on coastal and estuarine studies, Vol.22 Springer, Berlin, New York Heidelberg pp 209—226 Smith SV, Buddemeier RW (1992) Global change and coral reef ecosystems. Ann Rev Ecol Syst 23 : 89—118 Smith SV, Harrison JT (1977) Calcium carbonate production of the Mare Incognitum, the upper windward reef slope, at Eniwetak Atoll. Science 197 : 556—559 Smith SV, Kinsey DW (1988) Why don’t budgets of energy, nutrients and carbonates always balance at the level of organisms, reefs and tropical oceans? Proc 6th Int Coral Reef Symp 1 : 115—122. Smith SV, Kimmerer WJ, Laws EA, Brock RE, Walsh TW (1981) Kaneohe Bay sewage diversion experiment: perspectives on ecosystem responses to nutritional perturbation. Pac Sci 35 : 279—395 Sorokin YI (1995) Coral reef ecology. Springer, Berlin, New York Heidelberg, 465 pp Steneck RS (1988) Herbivory on coral reefs: a synthesis. Proc sixth Int Coral Reef Symp 1 : 37—49 Tanner JE, Hughes TP, Connell JH (1996) The role of history in community dynamics: a modelling approach. Ecol 77 : 108—117 Tansley AG (1935) The use and abuse of vegetational concepts and terms. Ecol 16 : 284—307 Tribble GW, Atkinson MJ, Sansone FJ, Smith SV (1994) Reef metabolism and endo-upwelling in perspective. Coral Reefs 13 : 199—201 Webster JR (1979) Hierarchical organization of ecosystems. In: Halfon E (ed) Theoretical systems ecology. Academic Press, New York, pp 119—131 Wiegert RG (1988) Holism and reductionism in ecology: hypotheses, scale and systems models. Oikos 55 : 267—269 Wiens JA (1989) Spatial scaling in ecology. Func Ecol 3 : 385—397 Wilkinson CR (1987) Interocean differences in size and nutrition of coral reef sponge populations. Science 236 : 1654—1657
.
