Geomorphology 89 (2007) 147 – 162 www.elsevier.com/locate/geomorph
A framework for interdisciplinary understanding of rivers as ecosystems E.S.J. Dollar a , C.S. James a , K.H. Rogers a,⁎, M.C. Thoms b,1 b
a Centre for Water in the Environment, University of the Witwatersrand, Johannesburg, South Africa Water Research Centre and the CRC for Freshwater Ecology, University of Canberra, ACT 2601, Australia
Received 10 January 2005; received in revised form 27 July 2006; accepted 27 July 2006 Available online 12 October 2006
Abstract Understanding and managing the behaviour of rivers as ecosystems requires holistic, interdisciplinary approaches. However, we lack appropriate frameworks to guide interdisciplinary thinking because disciplinary paradigms lose their usefulness in the interdisciplinary arena. Conceptual frameworks are useful tools with which to order phenomena and material, thereby revealing patterns and processes. A framework for the interdisciplinary study of river ecosystems is presented in this paper. The framework presents parallel hierarchies in the geomorphology, hydrology and ecology of a river with different organizational elements and levels of organization for each discipline. It assigns spatial and temporal scales for each level of organization for the different discipline hierarchies whereby different parts can be distinguished by different frequencies of occurrence and/or rates of change. Integration of the different disciplines, within the context of a particular study, is represented by a flow-chain model that describes process interactions that can change an ecosystem from one state (a template) of biophysical heterogeneity to another (a product). The framework concept is applied by first describing in detail the relevant organizational levels that characterize the different subsystems of the river ecosystem in the context of the problem being addressed. This is followed by the identification of appropriate scales and variables within the different organizational levels. Then the interactions with the products of template/agent of change/controller interactions that may account for any feedback influences are described. A series of examples is provided to illustrate the use of the framework in various interdisciplinary settings. © 2006 Elsevier B.V. All rights reserved. Keywords: Riverine ecosystems; Conceptual frameworks; Hierarchy; Scale; Interdisciplinary research
1. Introduction Rivers are complex systems. Their form and behaviour reflect interacting geomorphical, hydrological and ecological processes. While the importance of these interactions is recognized (e.g. Phillips, 1995), solutions ⁎ Corresponding author. E-mail address:
[email protected] (K.H. Rogers). 1 Authors have equal seniority and order was determined alphabetically. 0169-555X/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.geomorph.2006.07.022
to common river problems tend to combine unconnected inputs from the several disciplines rather than taking an interdisciplinary approach. Successful interdisciplinary science requires that the separate disciplines gain a common understanding of the nature of the problem at hand, identify the scales of relevant subsystem components, the underlying processes or phenomena, and the important variables involved. Successful interdisciplinary science requires joining of many areas of understanding into a single conceptual–empirical structure (Pickett et al., 1994).
148
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162
A conceptual framework can help different disciplines work together in an integrated way by ordering phenomena and materials and, thereby, revealing patterns (Rapport, 1985). Frameworks serve as scientific maps for new areas of endeavour, where even tentative maps are useful (Pickett et al., 1999), if only because their subsequent improvement provides some measure of progress in integrative thinking. Interdisciplinary river science at present lacks a conceptual framework to bring about commonality and integration. Conceptual models in river science explain, among other things, the influences of processes on channel morphology (Leopold et al., 1964), catchments on streams (Hynes, 1975; Vannote et al., 1980), and the importance of patches in rivers (Pringle et al., 1988). But these present the perspectives of single disciplines only and cannot broadly serve the multi-dimensional decision-making environment of interdisciplinary river science. Individually, they have value but do not provide a basis for ecologists, hydrologists and geomorphologists to integrate their thinking, concepts, and data collection. This paper proposes a framework to facilitate the integration of disciplinary efforts in the understanding and management of river systems. The framework is based on hierarchy theory, which uses a set of principles to keep
track of the complex structure and behaviour of systems at multiple scales (Allen and Hoekstra, 1992). The goal of the framework is to match the description of river form (in the context of a particular problem) with appropriate fluvial processes, so that phenomena can be explained at appropriate spatial and time scales. This will facilitate understanding and prediction of the response of patterns to processes, and the influences of patterns on processes. 2. Underlying concepts The complexity of river systems challenges many traditional scientific methods. Their multi-causal, multiple-scale character limits the usefulness of the conventional reductionist falsification approach, except when applied at very small scales and within limited domains. Hierarchy theory, however, provides an approach for interpreting river complexity. A hierarchy is a graded organizational structure. A particular hierarchical level (or holon) in a system is a discrete unit of the level above it, and an agglomeration of discrete units of the level below it (Fig. 1A). A particular level in the hierarchy exerts some constraint on lower levels (O'Neill et al., 1986), especially the one immediately below; lower
Fig. 1. Nested levels of organization (A) and how they may be related to the grain and extent of scale (B). Scale is presented as being dimensionless as the final scale is dependent on the unique characteristics of individual river systems. The figure can also be used for locating the problem scale within the organizational levels. After Kotliar and Wiens (1990).
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162
levels, conversely, can influence the structure and functioning of higher levels. In consideration of a particular level, the downward constraints and upward influences of all other levels are encapsulated by the characteristics and properties of the levels immediately above and below. The simultaneous operation of processes at different levels, within particular contextual constraints, gives rise to emergent properties. A level within the hierarchy can be characterized by a scale (O'Neill et al., 1989) (Fig. 1B). Scale defines physical dimensions in terms of grain and extent (Fig. 1B). Grain describes the smallest spatial or temporal interval in an observation set or of a study, the smallest scale of pattern to which an organism responds (O'Neill et al., 1989) or the smallest scale of influence of an ecosystem disturbance or process driver (Rogers, 2003). Extent is the total area or duration over which observations are made, the largest pattern to which an organism responds (i.e. its home range), or the largest scale at which a disturbance or process driver exerts influence on the system. Grain and extent define the upper and lower limits of resolution of the levels of a hierarchy. Scale determines the units appropriate for the variables associated with each level of a hierarchy. For example, two alluvial fans may both represent the same
149
level of organization in a geomorphological hierarchy (Fig. 2). Both can be characterized by the same variables (such as channel density, channel width and interchannel spacing). Fig. 2A shows an alluvial fan on a beach, with an extent of about 1 m, while Fig. 2B shows an alluvial fan in Death Valley, California, with an extent of about a kilometre. These scales suggest that the appropriate units for measuring channel dimensions would be millimetres in the first case and metres in the second. These two fans are, therefore, organizationally similar but very different in scale. Most geomorphological classification systems for rivers are hierarchical (e.g. Frissel et al., 1986; Van Niekerk et al., 1995; Montgomery and Buffington, 1998; Thoms et al., 2004). Schumm and Lichty (1965) use hierarchy to describe independent and dependent variables in river function. Hydrological models are formulated and developed with different orders of complexity and different descriptions of process at different scales. Large-scale models of basin yield commonly use statistically integrated representations of small-scale surface characteristics and processes that would be described explicitly in a small-scale urban flood-prediction model. Ecological organization (organism, species, community, ecosystem) is also hierarchical (Barrett et al.,
Fig. 2. Levels of organization. A is a photograph of an alluvial fan on a beach in the Eastern Cape, South Africa. B is an oblique photograph of an alluvial fan in Death Valley, California. Both photographs represent the same level of organization in a hierarchy (i.e. an alluvial fan) and can be characterized by the same variables (e.g. channel density, width etc.). However, the alluvial fan on the beach would be measured in mm–m, while the alluvial fan in Death Valley would be measured in m–km.
150
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162
1997). These three disciplinary hierarchies, however, are not compatible, and hence not easy to use in interdisciplinary applications. In the framework outlined in the following section, we propose individual disciplinary hierarchical structures that are independent of scale. We use scale as the currency for linking them. 3. Framework design To design the conceptual framework for river ecosystems, we first identify the various subsystems or hierarchies to be included in the framework, namely geomorphology, hydrology and ecology. Second, we describe the levels of organization in the three subsystems, in the context of the issue/problem being addressed. Third, we identify appropriate scales and variables within the different organizational levels. Finally, we describe the process interactions that take place between subsystem components relevant to the problem being addressed. This involves identification of the role that each component plays in river ecosystem processes, whether template, agent of change, controller or responder, and its place in an appropriate hierarchical flow-chain model. 3.1. Subsystem identification Many different disciplines study rivers. For practical purposes, the study and management of the form and
function of rivers requires a detailed understanding of the nature of and interaction between three primary disciplinary subsystems (geomorphology, hydrology and ecology). Geomorphology (in this context fluvial geomorphology) considers the landforms associated with river systems and the processes that form them. Hydrology focuses on the occurrence and movement of water through landscapes and river systems. Ecology considers the response of flora and fauna to changes in water supply, sediment movement and channel morphology, along with changes in landscape character and changes in other biotic phenomena. 3.2. Organizational hierarchies Each of the three primary subsystems in Section 3.1 can be represented as a hierarchical structure (Fig. 3) that specifies the different organizational levels. These three parallel organizational hierarchies constitute a substantive part of the framework. Most geomorphological river classification schemes assign spatial and time scales to the levels of geomorphological hierarchy. A geomorphological hierarchy could be scale-dependent, extending from drainage basin and drainage network to cross-section and size of particle (Thoms et al., 2004). A more general geomorphological hierarchy (Fig. 3) would be scale-independent, with increasing levels of detail, complexity and resolution at successively lower levels. Here, the highest level of
Fig. 3. Hierarchical descriptions of levels of organization that characterize the geomorphological., hydrological and ecological subsystems of a river.
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162
organization is considered to be the geomorphic province. Within a geomorphic province are drainage basins or catchments that share similar relief, climate, lithologic assemblages and/or fluvial evolutionary patterns, and provide water and sediment to the river. Within drainage basins are multiple macro-reaches or functional process zones (Thorp et al., 2006). Macro-reaches can be defined as stretches of river within which flow and sediment regime influences are sufficiently uniform to result in similar channel types. Within macro-reaches is a variety of channel types (such as braided or meandering). Within each channel type are geomorphic units (such as bars of different kinds) and within each geomorphic unit we can recognize agglomerations or clusters of particles. The lowest level in the hierarchy is represented by the individual sediment particle. A scaled hydrological hierarchy could link a river basin to a hillslope and on down to a surface response unit and then to the variables involved in soil physics (Kirkby, 1978). Alternatively, the hydrology of river ecosystems can be organized as a hierarchy of resolutions or levels of detail required for description (hydrological organizational hierarchy, Fig. 3). At the coarsest level of resolution is occurrence — that is, whether water occurs or not. Then comes specification of the quantity (volume) of water, which, together with topographic characteristics, enables us to predict inundation area, length, width and depth. At progressively finer resolutions, movement of water is described first by discharge (frequency and duration), then by timeaveraged velocities in 1, 2 or 3 dimensions and in 1, 2 or 3 directions, with associated shear stresses. The finest resolution describes turbulent fluctuations of these velocities, and associated Reynolds stresses. These descriptors apply over the full range of possible scales, with units depending on the scale assigned. The ecological hierarchy is also well established (Barrett et al., 1997). A population is a collection of interbreeding organisms whereas a species is a group of interbreeding natural populations that is reproductively isolated from other such groups. A community is an assemblage of individuals of different species cooccurring in space, and an ecosystem is a spatially defined area of interacting biotic and abiotic patterns and processes. The organizational hierarchies in geomorphology, hydrology and ecology are dissimilar in the nature of their holons. Geomorphological holons are physical entities; hydrological holons can be variable descriptors; and ecological holons are biological abstractions. Nevertheless, they are useful conceptual constructs and can be correlated through scale.
151
3.3. Assignment of scales Organizational levels between different subsystem hierarchies (Fig. 3) cannot be matched directly because they are scale-independent and noncomensurate. For example, a mayfly nymph, a salmon, and a hippopotamus all belong to the individual organism level of the ecological hierarchy, but each occupies and perceives a different grain size and extent of patchiness within the landscape. Each of these organisms responds to and influences a different range of geomorphological and flow characteristics. Similarly, an ecosystem could be a decomposing piece of wood or an entire river system, each responding to a different level in the hydrological hierarchy. An individual sand grain, a geomorphic unit, and a river macro-reach will respond differently to a particular flow. The dynamics of each can also be described in terms flow characteristics at different levels in the hydrological hierarchy. The three subsystems can only be integrated to resolve a particular problem after appropriate scaling of the respective organizational structures, thereby producing scaled hierarchies. Integration of the generic organizational hierarchies for geomorphology, hydrology and ecology (Fig. 3) for a particular purpose requires them to be scaled according to the principles presented in Fig. 1. In particular, the different levels of organization hierarchy within each subsystem are related to the grain size and extent of the proposed activity (Fig. 1B). Assignment of a scale to a level of organization implies scales for other levels, leading to a hierarchy of scales for a particular subsystem. For example, the scales of community and ecosystem would follow from nomination of an organism, such as a mayfly or a hippopotamus. This allows for the matching of scales in the geomorphological and hydrological hierarchies. Once spatial scales have been assigned, the time scales of the subsystem components can also be distinguished, hence signifying the different frequencies of occurrence and/or rates of change. Processes at higher levels have lower rates and frequencies, and therefore operate more slowly and over larger spatial arenas than those at lower levels. The geomorphological scale hierarchy for rivers (Fig. 4) is derived by assigning spatial scales to the different levels of organization, with the resulting scale levels described in the same terms as the organizational ones. Different specifications of scale may be necessary for other applications. The hydrological scale hierarchy (Fig. 4) is obtained by coupling the organizational hierarchy with the temporal characteristics of duration, timing and frequency. Hydrological processes operate at all time
152
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162
Fig. 4. Development of scale hierarchies. The hierarchies on the left side of the figure are organizational hierarchies that are developed into scaled hierarchies through acknowledgment of the grain and extent of each level of organization within the individual subsystem hierarchies.
scales and all levels of organization, and all size or spatial scales. Temporal characteristics can be structured in a scale-defined nested hierarchy: duration at one scale includes discrete events at the next smaller scale, each with its own duration, timing and frequency. For example, the flow of water in a river may be characterized at a large time scale in terms of wet/dry cycles over millennia, with each wet cycle characterized by duration, timing and frequency. Within each wet
cycle shorter cycles occur, (such as the approximately 18-year cycles in southern Africa (Tyson, 1986)), with the same temporal descriptors. Within each year of such a cycle seasonal variations occur and within each wet season individual events occur, each with the same descriptors but at different scales and measured in different units. For general application in rivers the appropriate spatial scales are suggested by links to the geomorphological scale hierarchy (Fig. 4). The
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162
153
assignment of corresponding time scales leads to the hydrological nested scale hierarchy shown in Fig. 4. The palaeoclimate scale describes sequences of wet and dry periods associated with long-term climate change over geological time. The historical climate scale describes similar changes, but over the shorter periods associated with macro-reach geomorphological processes. The hydrological regime scale includes descriptions of discharge events and associated frequencies. The channel hydraulics scale specifies time-averaged velocities and boundary shear stresses and their distributions at different resolutions. At the local fluid mechanics scale, description is in terms of local velocities and turbulence characteristics. Similar nested hydrological hierarchies for linking hydrological disturbances at various scales with appropriate ecological responses in Australian dryland river systems have been described by Walker et al. (1995), Thoms and Sheldon (2000) and Thoms and Parsons (2002). 3.4. Component interactions The first three steps of the framework define the character and relationships between components of the different subsystem hierarchies. This fourth step describes interactions between the components, and hierarchies, as an integrated system. It can be represented by a form of flow-chain model (Fig. 5). Flow-chain models have four basic components (Shachak and Jones, 1995): the abiotic or biotic agent of change, or driver; the template or substrate upon which the driver acts; controllers of the driver or agent of change; and an entity or process that responds to the driver or agent of change. Responders can be sets of processes, organisms or parts of the physical environment. Physically, interactions within a geomorphic subsystem (e.g. bedrock substratum, large woody debris, vegetation) and an agent of change (e.g. flowing water) are controlled, or modified, by one or more factors (e.g. sediment type, bed slope). The controllers may act on the agent directly, or on the susceptibility of the subsystem itself to the agent. Various organisms and/or processes respond differently to the subsystem in its different states. Floodwater, for example, acts as an agent of change by redistributing sediment. The pattern (product) of this redistribution may be controlled, in part, by large woody debris or bedrock outcrops in the channel. Riparian plants respond by colonizing the sediment. It is especially important to specify the scale of process interaction, because a responder at one scale may be a driver at another scale. For example, reeds may be agents of change when they trap sediment on a bar, or controllers when they modify flow patterns around a
Fig. 5. A flow-chain model for describing process interactions between subsystems and between different levels or scales within organizational hierarchies.
clump of stems, but reeds are also products when individual seedlings take root between sand grains. Flow-chain models have been used in ecology to demonstrate, among other things, modes of change in heterogeneity (Pickett et al., 2003) but such models are not generally spatially explicit. For our framework we use a multi-level flow-chain model for integration between subsystems (Fig. 5). To illustrate the use of the multi-level flow-chain model in an interdisciplinary setting, we suggest that river form and pattern are the result of water acting as the system driver on sediment as the material, within boundary conditions defined by the physical template (Fig. 5). Here, water and sediment represent agents of change; the initial state defines the template within which processes occur to produce form and pattern as the resulting (product) state. Geomorphological or biological (e.g. vegetation) components may act as controllers, and biological entities as responders provide the links to the ecological subsystem. Water (the driver), moves sediment (the material) under constraints defined by boundary conditions (the template) to produce a product that provides the template for processes at the next lower scale level. The biological response to the product at any level provides the primary linkage with the ecological system. This structure accommodates feedback responses, allowing biotic consequences to
154
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162
contribute to product or template structure. It also allows for consideration of downward constraint by higher levels and upward integration of processes from lower levels. The relative importance of downward constraint and upward integration is different at each level of organization. The higher levels are controlled predominantly by downward influence, while features at lower levels are more manifestations of upward influence — the action of water on individual particles forms clusters, which agglomerate to form geomorphic units and ultimately channel types. At all levels the product, however, depends on the context provided from above and the integration of processes from below — the same basic sedimentary processes could produce different forms within different constraining contexts (e.g. braiding and meandering channel types). Interpreting the relationship between downward constraint and upward integration of explanation is critical in interpreting the geomorphology hierarchy. At each level of the geomorphological scale hierarchy (Fig. 4), the model will have different spatial and time scales, different hydrological and sediment descriptors, and will describe different processes. The appropriate level and characteristics for a particular problem can be specified through consideration of organizational structure, scale and process, as described earlier. As an illustration we apply these concepts to define requirements for typical river investigations in a southern African context (Table 1). In this illustration, just as the geomorphological characteristics of a river system are hierarchically organized, so too are the variables describing the associated characteristics of water as driver, and sediment as the material. The water characteristics also exhibit similar hierarchical relationships in terms of downward constraint and upward explanation. For example, at scales appropriate to channel types, the discharge in a channel is determined only by higher-level climatic, meteorological and hydrological processes, but could be synthesized by integration of velocity measurements from any lower levels. Integration of the ecological and physical (geomorphology and hydrology) subsystems is complicated by the physical heterogeneity (patchiness), biological diversity, and biological process complexity of natural ecosystems. The framework principles, however, can facilitate the ordering of investigation and pursuit of understanding. Each species perceives and reacts to template patchiness in a unique way over a range of scales, and individuals respond to patchiness differently at different life history stages. The templates of the flowchain model should, therefore, be defined to encompass the entire mosaic of patches relevant to the organism under consideration. The patches used by an organism
may extend across different parts of the physical template at the same level of system organization, or even across levels. The flow-chain model may then be required to include more than one product at a particular level, and/or account for hierarchical interaction across levels. As an example, consider a crustacean that spends its whole life cycle in a mosaic of two patch types, a pool and a riffle (distinct geomorphic units of the channel type hierarchical level). Individuals use different features at different stages of the life cycle, or for different functions of life history. A mature individual might take refuge within a particle cluster or behind an individual particle at the head of the pool and feed in a leaf pack in the slow flow of an eddy (at the particle cluster level within the pool). Juveniles will feed on the periphytic aggregate most abundant on the lee of rocks (at the particle cluster level within the riffle). Mating takes place in the sandy shallows (at the particle cluster level within the pool), and females lay eggs in coarse sand between cobbles (particle cluster within the riffle). Interbreeding individuals of a sub-population in a particular river traverse pool/riffle mosaics (channel types) throughout a particular macro-reach. Various sub-populations of the metapopulation may be spread across a whole drainage basin. Accounting for all phenomena associated with this species would, therefore, require a five-level hierarchical flow-chain model, with at least two product descriptions at the channel type level. Particular research questions, however, would focus on a smaller range and apply simpler formulations independently. By integrating scale hierarchies with process models, the framework allows scientists and managers to view the landscape as a hierarchy of nested patches generated by key formative processes, rather than as a geographic arrangement of ecosystem components. In particular, it allows explicit recognition that agents of change act across a range of scales in a manner analogous to organism response. The physical integration described above helps to identify the descriptors of physical agents of change appropriate at different scales; the biological response links in the hierarchy of flow-chain models can also facilitate similar recognition and description of biological agents of change across the spectrum of scales. The integration of the geomorphological and hydrological hierarchies considered above does not account for the role of organisms in creating or modifying patch structure, which is especially significant at the lower levels. This is a requirement for ecological integration. In an ecological time frame (as opposed to a geological time frame) the action of water as the system driver on sediment as the material, within boundary conditions defined by the physical template, defines the physical
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162 Table 1 A southern African illustration of the integration of hydrological and geomorphological physical subsystems within the framework
155
156
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162
and chemical resource base of an ecosystem. The product of these processes represents a physicochemical mosaic of patches. It is not only the physical processes, however, that engineer the river landscape; organisms do so too, when their physical actions change the form of the landscape (Naiman and Rogers, 1997) or modulate the supply of resources (Jones, 1995). These biotic engineering actions represent fundamental processes in the creation, modification and maintenance of species habitat, and in the realisation of the full heterogeneity and integrity of the overall mosaic of the river. They include the actions of beavers and hippopotami that modify the physical form of geomorphological levels of organization, the modification of sediment characteristics by feeding fish and invertebrates, and the shading and sediment-trapping effects of vegetation. Similarly, the metabolic activities of organisms (oxygen-use and production, denitrification, defecation, nutrient uptake and excretion, etc.) generate biochemical modification of the chemical patchiness of the physical subsystem. In addition to physical patchiness, therefore, the biological patchiness results from the distribution, abundance, composition and structure of organisms. This patchiness is ‘engineered’ by organism response to the physical subsystem and the population and community processes that form the basis of most ecological studies. The natural biophysical mosaic resulting from the interaction between ecological and physical subsystems is further transformed by human-induced impacts into the existing river mosaics. These biological and human agencies can be identified through the scaled hierarchies of the framework, and accounted for in the definition of controllers and agents of change in the flow-chain models (Fig. 5). 4. Examples of application of framework principles The preceding discussion has described the concept of the framework and illustrated its capabilities. The two examples of practical applications of the principles of the framework, presented below, are not direct applications of the framework procedures because they were carried out while it was being formalized. They were, however, instrumental in developing the thinking behind the framework, and exemplify the principles it embodies and their usefulness. 4.1. Example 1: A scaled understanding of plant species/physical subsystem interactions Many ecological problems arise from changes in the distribution and/or abundance of populations of single
species. These may be invasive species (such as coypu, carp, zebra mussels, or water hyacinth), an endangered species, a commercially important species (such as salmon), a keystone species, or an indicator species. Most commonly these problems require detailed understanding of the structure (e.g. age structure, genetic composition) and dynamics of the populations, and an understanding of the characteristics of an organism's life history (e.g. the ‘what’, ‘where’, ‘when’ and ‘how’ of feeding, growing and breeding). These issues are the domain of ecologists, but they cannot be fully interpreted without an understanding of the controls that the physical subsystem has on the underlying processes, and of the scales at which the physical and ecological subsystems interact. The issues are well illustrated by studies of the Matumi tree (Breonadia salicena) in the Sabie River, South Africa (cf. de Fontaine and Rogers, 1999). The Sabie River, within Kruger National Park, falls within the Lowveld Geomorphic Province and consists of four to five different types of channels nested within a number of macro-reaches. Steep sections of the river provide little space for sediment storage and the river spreads out across broad bedrock outcrops to form anastomosing channels of exposed, and sediment draped, bedrock. Matumi trees are found on most substrata within this channel type and the population presents a complex bimodal demographic structure. The mixed bedrock/alluvial nature of the Sabie River gives rise to the high habitat and species diversity that is the focus of conservation efforts in the Park (Rogers and O'Keeffe, 2003). Modified flow regimes and changed land use practices threaten this heterogeneity. Understanding the response of riparian vegetation to these changes has become a central issue in subsequent river management and monitoring programmes (Rogers and Biggs, 1999; Rogers and O'Keeffe, 2003). Although Matumi seeds germinate on any damp substratum, the seedlings only establish when roots gain purchase in the cracks of rocks (Mackenzie et al., 1999). A very high proportion of small plants and a negative J demographic profile occurs on rock-rapid cataracts and a relative absence of seedlings from braid bars. These different profiles at the geomorphic unit level aggregate to present a bimodal distribution at the channel type level. In the context of the framework, the first task was to decide on the appropriate level/s of organization at which to study B. salicena. Because the species is widespread over subtropical east Africa, the Sabie River population and sub-population levels of organization were appropriate. Applying the appropriate scale to the study was complicated because different densities and
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162
heights of trees appeared to be associated with different geomorphological scale levels. The structure of the Sabie River population was, therefore, examined over a range of scales in an attempt to fully understand its relationship to the physical subsystem and the processes of sediment–water interaction that, in part, generate the template. The template was described hierarchically (Heritage et al., 1997), and the macro-reach, channel type and geomorphic-unit-scaled levels of organization were chosen as the initial foci of study. The frequency distributions, presented in Fig. 6 (A–H), show the different demographic profiles of B. salicena at different geomorphological scale levels. The trees were distributed across a wide range of geomorphological scales, but were more abundant (especially in the smaller size classes) at those with a high bedrock influence, such as bedrock-anastomosing channels (Fig. 6D) and nested geomorphic units (core bars (Fig. 6F) and pool-rapid cataracts (Fig. 6H)).
157
The form of the frequency distribution of size classes of trees describes population structure, as it has been shaped by the many processes that determine establishment, growth, and death in a population. The markedly different demographic profiles illustrate that tree– water–sediment interactions have played out differently at different geomorphological scales. The different patterns manifest at different scales result from the upward integration of patterns from lower levels. For example, the population has a bimodal structure at the level of the Lowveld Geomorphic Province because most of the population inhabits bedrock-anastomosing channels. Anastomosing channels, in turn, exhibit a bimodal population structure that is an aggregation of a negative J structure on rock-rapid cataracts and normal unimodal or bimodal structure of the sub-populations on sedimentary geomorphic units. Most ecological studies are conducted at a single, arbitrarily selected and human-centred scale. If that had
Fig. 6. Size frequency distributions of Matumi tree population at different scales and levels of geomorphic organization on the Sabie River in the Kruger National Park, South Africa.
158
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162
been the case here, an arbitrary selection of one of the demographic profiles in Fig. 6 would have occurred, and it would have been used to make inferences about processes that shape the population over time. That could lead to quite erroneous associations of vegetation patterns and processes with physical ones. The study described (de Fontaine and Rogers, 1999), which has been explicit about patchiness of the geomorphological levels of organization and about scale, however, has a much higher likelihood of generating real understanding. Having established the need to examine the population at different geomorphological scale levels, an appropriate flow-chain model (Fig. 7) can be used to integrate the ecological and physical subsystems. In terms of the scaled patch hierarchy (Fig. 1), these results established that the extent of plant response to template structure was measurable at a macro-reach scale, while the grain was defined by cracks in rocks at the individual particle scale (0.05–0.5 m). Indeed Matumi was restricted to granitic and rhyolitic rocks that did crack, and was generally absent from basaltic
rocks that are more prone to chemical than physical weathering. The presence of large, old trees but with little recruitment on sedimentary features associated with braided channels and various bars within the anastomosing channels led to the deduction that sediment storage occurred subsequent to the establishment of Matumi seedlings. Under natural or unchanged flow conditions, the increasing density of Matumi trees and foliage provides increased flow resistance and enhanced sedimentation to form lateral and core bars. This drives a negative feedback response for the Matumi population, as it reduces the availability of establishment sites. The consequence would be a normal frequency distribution (Fig. 6G) as recruitment to the population is cut off. Sequential patterns of erosion, transport and deposition of sediment would generate more complex demographic profiles (Fig. 6E, F) as sites for recruitment become periodically available. Interpretation of the demographic profiles in association with appropriately scaled physical template description, thus, engenders understanding of the
Fig. 7. A flow chain model of responses of the Matumi tree population to template patchiness and the sediment/water interactions that form the template. Template = black text in grey box; Products = white in grey box; Response = grey box with no border; Agent of change = black text in white box; Controller = elliptical border.
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162
underlying processes. On the basis of the above interpretations, a flow-chain process model could be developed (Fig. 7). This explains processes over two nested geomorphological levels, and allows description of vegetation response and feedback. Kruger National Park managers seek to maintain the dynamism between the physical and ecological subsystems (Fig. 7) and the associated biodiversity. The model structure of Fig. 7 will help them manage the river in the face of decreasing flow and increasing sediment supply that can lead to reduced bedrock exposure, homogenization of the physical subsystem, and consequent loss of the biodiversity. 4.2. Example 2: Determination of environmental water allocations Allocating river water to sustain natural ecosystems, to rehabilitate rivers degraded by over-abstraction or flow regulation, and to protect biodiversity has become a key management objective worldwide. Management of environmental flows is frequently concerned with the question: ‘how much water is required to protect and conserve river function?’. The answers to this question depend on the academic discipline of the person answering the question. From a geomorphological perspective, water allocation is required to maintain the structure and function of natural physical features of the river channel (Thoms and Sheldon, 2002). From a biological perspective, water allocation is required to maintain individuals, populations, communities and ecosystem processes (e.g. Tennant, 1976). An interdisciplinary framework, however, sees environmental water allocations in a spatial and temporal context that considers the key links between geomorphology, hydrology and ecology, and also recognizes constraints from higher levels and influences from lower levels in the organizational hierarchies that may come from the various environmental and institutional processes. Hence, the framework changes the focus of water allocations from one that is orientated to single species within a single disciplinary context to one that considers complex multi-scale interactions among biota, physical structure, and hydrological processes. Many different approaches and techniques are available to assist in the determination of environmental water allocations. A recent review of 207 individual methods by Tharme (2003) noted the application generally occurred at two levels; a reconnaissancelevel approach that relied on hydrological methods such as the Tennant Method (Tennant, 1976); and, a set of more comprehensive methods of assessment that either
159
relied on hydraulic simulation such as the Instream Flow Incremental Methodology (IFIM) or ‘holistic’ methods. The methods reviewed by Tharme (2003) are inherently reductionist in approach and use data collected at a site, or a river reach, to build up an environmental flow regime. Reliance on methods, like IFIM (Bovee and Milhous, 1978) and the Building Block Methodology (King and Louw, 1998), while extremely useful in the initial period of environmental water allocations, fails to recognize the limitation of bottom–up influences in hierarchical organizations such as river ecosystems. The influence of lower levels of organization within complex systems gets progressively weaker at successively higher levels, hence limiting their domain of influence. Advances in understanding environmental water allocations and management, therefore, require bottom–up and top–down approaches that capture a continuum of hierarchical influences. To invoke the concept of hierarchy within the context of environmental water allocations, we must recognize top–down constraints. Employing a conservation ecology analogy, a top–down approach would recognize the character of the hydrological landscape (Thoms et al., 2004) at different scales, and the management objective would be to maintain the diversity or heterogeneity of the landscape. Managing landscape diversity or heterogeneity is an essential component in conserving system resilience (Pickett et al., 2003) and in the context of environmental flows, the resilience of riverine ecosystems. Many strategies for the management of environmental flows view rivers as uniform and fail to fully consider the spatial and temporal hydrological complexity. A recent study by Thoms and Parsons (2003) demonstrated a complex spatial pattern of hydrological character in a large Australian dryland river system. The study along the Condamine Balonne River identified six river zones of different hydrological character using multivariate analysis of the hydrological character of 44 gauging stations (Fig. 8). These hydrological zones represent ‘patches’ within the hydrological landscape mosaic of the Condamine Balonne River system. The Thoms and Parsons (2003) study also notes that the dominant time scale of hydrological influence differed between the river zones. The hydrological character of the headwater zones in the Condamine Balonne was characterized by short-term variables that corresponded to individual floods whereas longer time scale variables, characteristic of event sequencing, better represented those zones lower in catchment (Fig. 8). The spatial and temporal complexity identified at this larger scale requires environmental water allocations to be managed
160
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162
Fig. 8. Hydrological zones identified in the Condamine Balonne River, Australia by Thoms and Parsons (2003). Hydrological zones correspond to groups of nodes with similar hydrological character. Groups were derived from classification analysis and a Gower association is given at each major split of the dendrogram, where lower values indicate greater similarity.
at scales that capture appropriate patterns of hydrological character in the system in question. The spatial and temporal complexity of hydrological character within the Condamine Balonne River is an example of a heterogeneous hydrological landscape. Recognition of hydrological mosaics has several implications for environmental flow strategies. The time scale of flow variables associated with the spatial arrangement of the different hydrological patches needs to be recognized so that management intervention can be placed at the appropriate spatial and time scale. Different targets of flow restoration need to be set for individual hydrological zones whereby the attributes of flow must be manipulated, restored or conserved in accordance with the different time scales of hydrological influence. Maintaining the hydrological integrity of individual zones would allow maintenance of the diversity of the broader mosaic of the hydrological landscapes within a catchment. Environmental water allocations are effected through manipulation of the hydrological regime. At what scale should these hydrological manipulations be made to predict physical and biological responses? At a particle scale, flow hydraulics influences the character of the riverbed substratum (Lancaster and Belyea, 1997) and if macro-invertebrates are the diagnostic fauna, the
corresponding level of biological organization may be that of an individual organism. At a larger scale, the frequency of a flow partly determines the morphology of river zones (e.g. macro-reaches) and the corresponding level of biological organization is that of a macroinvertebrate community. In Australia and elsewhere, macro-invertebrate communities, collected at the local scale, are commonly used as primary biological indicators in environmental flow assessments. These community-level attributes, however, may be inappropriate because of the inherent spatial and temporal complexity in hydrologic and geomorphological character. For example, given the dominance of short-term pulse scale hydrological variables in the headwater zones of the Condamine Balonne River, it would be more appropriate to monitor populations of individual organisms at small patches within a reach. In those zones located farther downstream that are characterized hydrologically by events over longer time scales, community-level attributes could be monitored. In other words, the biological indicators used to monitor environmental flows should match the appropriate scales of physical and hydrological processes that occur in the river system. Incorporation of this multidimensional spatial and temporal
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162
approach into existing environmental flow strategies will advance the application of the natural flow paradigm and by association, may improve ecosystem responses to managed flows (Thoms and Parsons, 2003). The framework places the collective use of top– down and bottom–up methods in a multi-dimensional context and facilitates an adaptive management approach for the setting and management of environmental water allocations. The management and conservation of diversity in the hydrological landscape is a higher-level objective within which methods like IFIM and the Building Block Methodology could be used to assess flow needs in specific patches of the mosaic. Determining important biophysical flows then has a context; important biophysical flows, whether they be for a single species at a site or a range of organisms within multiple river patches become the focus for hypothesisbased monitoring. Monitoring through testing hypotheses about the functions of certain flows and the implications of changing these flows in river ecosystems overcomes the much-critiqued approach of monitoring for the sake of monitoring. 5. Conclusion The framework developed here for the interdisciplinary understanding of rivers as ecosystems provides a map that enables links between different disciplines to be made at appropriate scales. It recognizes subsystems that contribute to the study of rivers and the relevant levels of organization within each, assigns appropriate scales and variables within the different subsystem organizational hierarchies, and describes interactions between the agents of change, templates, and controllers via a multi-level flow-chain model. Most river science, at present, tends to be locked into or work from a descriptive base with a strong emphasis on classification, description of pattern and process, or modelling of selected attributes of rivers. Much river research is conducted without explicit consideration of scale; it is therefore scale-less. A tendency also exists to focus on a limited scale range — indeed much work is done at finer scales, with the result that it is often inappropriate to the question or problem being investigated (Pickett et al., 1999). Interdisciplinary river science needs to address multi-causal relationships that occur at many different scales so it can better understand these complex systems and interface with policy-management. Effective collaboration between river scientists, and their communication of knowledge to the water industry, can only improve with the development of a common framework and common set of concepts.
161
Acknowledgements Many people have contributed to this manuscript both wittingly and unwittingly. The staff and postgraduate students of the Centre for Water in the Environment (South Africa) and the River and Floodplain Laboratory (Australia) are all thanked. The Andrew Mellon Foundation, Water Research Commission of South Africa, Land and Water Australia and the Australian Research Council funded much of the work on which this manuscript is based. MCT thanks the Centre for Water in the Environment for support during time in South Africa. References Allen, T.F.H., Hoekstra, T.W., 1992. Towards a Unified Ecology. Columbia University Press, New York. Barrett, G.W., Peles, J.D., Odum, E.P., 1997. Transcending processes and the levels of organisation concept. Bioscience 47, 531–535. Bovee, K.D., Milhous, R.T., 1978. Hydraulic simulation in instream flow studies: theory and techniques. US Fish and Wildlife Service Biological Services Program, Cooperative Instream Flow Service Group, Instream Flow Information Paper, vol. 5. FWS/OBS-78-33. 130 pp. de Fontaine, M.L., Rogers, K.H., 1999. Relationships between demographic structure of Breonadia Salicina (Delile) sub-populations and geomorphic features of the Sabie River, Kruger National Park. Centre for Water in the Environment Technical Report, vol. 4/95. Frissel, C.A., Liss, W.J., Warren, C.E., Hurley, M.D., 1986. A hierarchical framework for stream classification viewing streams in a watershed context. Environmental Management 10, 199–214. Heritage, G.L., van Niekerk, A.W., Moon, B.P., Broadhurst, L.J., Rogers, K.H., James, C.S., 1997. The geomorphological response to changing flow regimes of the Sabie and Letaba River systems. Water Research Commission Report No 376/1/97. Water Research Commission, Pretoria. Hynes, H.B.N., 1975. Edgardo Baldi Memorial Lecture. The stream and its valley. Verhandlungen der Internationalen Vereinigung für Theoretische und Angewandte Limnologie 19, 1–15. Jones, C.G., 1995. Linking species and ecosystems. Editor, Chapman Hall, New York. King, J.M., Louw, M.D., 1998. Instream flow assessments for regulated rivers in South Africa using the Building Block Methodology. Aquatic Ecosystem Health and Management 1, 109–124. Kirkby, M.J, 1978. Hillslope Hydrology. Wiley, Chichester. 398 pp. Kotliar, N.B., Wiens, J.A., 1990. Multiple scales of patchiness and patch structure: a hierarchical framework for the study of heterogeneity. Oikos 59, 253–260. Lancaster, J., Belyea, L.R., 1997. Nested hierarchies and scaledependence of mechanisms of flow refugium use. Journal of the North American Benthological Society 16, 221–238. Leopold, L.D., Wolman, M.G., Miller, J.P., 1964. Fluvial Processes in Geomorphology. W.H. Freeman and Company, New York. Mackenzie, J.A., van Coller, A.L., Rogers, K.H., 1999. Rule-based modelling for management of riparian systems. Water Research Commission Report No 813/1/99. Water Research Commission, Pretoria, South Africa. Montgomery, D.R., Buffington, J.M., 1998. Channel processes, classification, and response. In: Naiman, R.J., Bilby, R.E. (Eds.),
162
E.S.J. Dollar et al. / Geomorphology 89 (2007) 147–162
River Ecology and Management: Lessons from the Pacific Coastal Ecoregion. Springer, Verlag, pp. 13–42. Naiman, R.J., Rogers, K.H., 1997. Large animals and the maintenance of system-level characteristics in river corridors. Bioscience 47, 521–529. O'Neill, R.V., DeAngelis, D.L., Waide, J.B., Allen, T.F.H., 1986. A Hierarchical Concept of Ecosystems. Princeton University Press, Princeton. O'Neill, R.V., Johnson, A.R., King, A.W., 1989. A hierarchical framework for analysis of scale. Landscape Ecology 3, 193–205. Phillips, J.D., 1995. Biogeomorphology and landscape evolution: the problem of scale. Geomorphology 13, 337–347. Pickett, S.T.A., Kolasa, J., Jones, C.G., 1994. Ecological Understanding: The Nature of Theory and the Theory of Nature. Academic Press, San Diego. Pickett, S.T.A., Burch, W.R., Grove, J.M., 1999. Interdisciplinary research: maintaining the constructive impulse in a culture of criticism. Ecosystems 2, 302–307. Pickett, S.T.A., Cadenasso, M.L., Benning, T.L., 2003. Biotic and abiotic variability as key determinants of savanna heterogeneity at multiple spatiotemporal scales. In: Du Toit, J.T., Rogers, K.H., Biggs, H.C. (Eds.), The Rivers of Kruger National Park, South African Water Research Commission, Pretoria, pp. 22–40. Pringle, C.M., Naiman, R.J., Bretschko, G., Karr, J.R., Oswood, M.W., Webster, J.R., Welcomme, R.L., Winterbourne, M.J., 1988. Patch dynamics in lotic systems: the stream as a mosaic. Journal of the North American Benthological Society 7, 503–524. Rapport, A., 1985. Thinking about home environments: a conceptual framework. In: Altman, I., Werner, C.M. (Eds.), Home Environments. Plenum Press, New York, pp. 255–286. Rogers, K.H., 2003. Adopting a heterogeneity paradigm: implications for management of protected savannas. In: Du Toit, J.T., Rogers, K.H., Biggs, H.C. (Eds.), The Rivers of Kruger National Park, South African Water Research Commission, Pretoria, pp. 41–58. Rogers, K.H., Biggs, H.C., 1999. Integrating indicators, endpoints and value systems in strategic management of the rivers of the Kruger National Park, South Africa. Freshwater Biology 41, 439–452. Rogers, K.H., O'Keeffe, J., 2003. River heterogeneity: ecosystem structure, function and management. In: Du Toit, J.T., Rogers, K.H., Biggs, H.C. (Eds.), The Rivers of Kruger National Park, South African Water Research Commission, Pretoria, pp. 189–218.
Schumm, S.A., Lichty, R.W., 1965. Time, space and causality in geomorphology. American Journal of Science 263, 110–119. Shachak, M., Jones, C.G., 1995. Ecological flow chains and ecological systems: concepts for linking species and ecosystem perspectives. In: Jones, C.G. (Ed.), Linking Species and Ecosystems. Chapman and Hall, New York, pp. 280–294. Tennant, D.L., 1976. Instream flow regimes for fish, wildlife, recreation and related environmental resources. Fisheries 1, 6–10. Tharme, R.E., 2003. A global perspective on environment flow assessment: emerging trends in the development and application of environmental flow methodologies for rivers. River Research and Applications 19, 397–441. Thoms, M.C., Parsons, M.E., 2002. Ecogeomorphology: an interdisciplinary approach to river science. International Association of Hydrological Sciences 276, 113–120. Thoms, M.C., Parsons, M.E., 2003. Identifying spatial and temporal patterns in the character of the Condamine Balonne River, Australia, using multivariate statistics. River Research and Applications 19, 443–458. Thoms, M.C., Sheldon, F., 2000. Lowland rivers: an Australian introduction. Regulated Rivers: Research and Management 16, 375–383. Thoms, M., Sheldon, C., 2002. An ecosystems approach for determining environmental water allocations in Australian dryland river systems: the role of geomorphology. Geomorphology 47, 153–168. Thoms, M.C., Hill, S.M., Spry, M., Mount, T.J., Chen, X.Y., Sheldon, F., 2004. The geomorphology of the Darling Basin. In: Breckwodt, R., Boden, R., Andrew, J. (Eds.), The Darling. Murray Darling Basin Commission, Canberra, pp. 68–105. Thorp, J.H., Thoms, M.C., Delong, M.D., 2006. The river ecosystem synthesis: biocomplexity in river networks across space and time. River Research and Applications 22, 123–147. Tyson, P.D., 1986. Climatic Change and Variability in Southern Africa. Oxford University Press, Oxford. Van Niekerk, A.W., Heritage, G.L., Moon, B.P., 1995. River classification for management: the geomorphology of the Sabie River in the eastern Transvaal. South African Geographical Journal 77, 68–76. Vannote, R.L., Minshall, G.W., Cummins, K.W., Sedell, J.R., Cushing, C.E., 1980. The river continuum concept. Canadian Journal of Fisheries and Aquatic Sciences 37, 130–137. Walker, K.F., Sheldon, F., Puckridge, J.T., 1995. A perspective on dryland river ecosystems. Regulated Rivers : Research and Management 11, 85–104.
