Biodiversity: towards a unifying theme for river ecology


Ward Department of Limnology, EAWAG/ETH, Überlandstrasse 133, 8600 Dübendorf, Switzerland. E-mail:


1. A broadened concept of biodiversity, encompassing spatio-temporal heterogeneity, functional processes and species diversity, could provide a unifying theme for river ecology.

2. The theoretical foundations of stream ecology often do not reflect fully the crucial roles of spatial complexity and fluvial dynamics in natural river ecosystems, which has hindered conceptual advances and the effectiveness of efforts at conservation and restoration.

3. Inclusion of surface waters (lotic and lentic), subsurface waters (hyporheic and phreatic), riparian systems (in both constrained and floodplain reaches), and the ecotones between them (e.g. springs) as interacting components contributing to total biodiversity, is crucial for developing a holistic framework of rivers as ecosystems.

4. Measures of species diversity, including alpha, beta and gamma diversity, are a result of disturbance history, resource partitioning, habitat fragmentation and successional phenomena across the riverine landscape. A hierarchical approach to diversity in natural and altered river-floodplain ecosystems will enhance understanding of ecological phenomena operating at different scales along multidimensional environmental gradients.

5. Re-establishing functional diversity (e.g. hydrologic and successional processes) across the active corridor could serve as the focus of river conservation initiatives. Once functional processes have been reconstituted, habitat heterogeneity will increase, followed by corresponding increases in species diversity of aquatic and riparian biota.

Introduction: biodiversity as an integrative concept

Whereas species diversity is a community attribute, the newer concept of biodiversity encompasses all levels of organization while integrating biotic and abiotic patterns and processes across scales (Noss, 1990; Johnson et al., 1996; Mooney et al., 1996; Naeem & Li, 1997; Peterson, Allen & Holling, 1998; Wetzel, 1999). Because of its broad scope and multidimensional nature, biodiversity has the potential to serve a unifying role (i) by linking ecology with evolution, genetics and biogeography, (ii) by elucidating how interactions between disturbance regimes and habitat heterogeneity influence niche diversification and resource partitioning, (iii) by integrating functional processes with spatio-temporal heterogeneity, (iv) by promulgating a hierarchical perspective of ecosystems, and (v) by providing a basis for effective management and restoration initiatives.

The primary goal of this paper is to develop a holistic view of biodiversity as a unifying theme for investigating river ecosystems. We believe that such an approach will lead to a better understanding of the complex behaviour of both natural and altered rivers, as well as elucidating critical interactions between the environment and the biota. However, it is not feasible here to treat all aspects of such a broad topic. Some major aspects, such as genetic diversity, are not considered. The goal is not to present a comprehensive review, but to develop some thoughts on biodiversity in river systems from a largely conceptual perspective, with selected examples drawn from our research on European rivers. In cases where empirical data are presented, the reader is referred to published works for details on sampling programmes and analytical procedures. The following focuses on habitat and landscape spatial scales and the ecological time scale, emphasizing the major role of natural disturbance in riverine flood plains, thereby serving as a logical extension of Ward, Tockner & Schiemer (1999c).

Biodiversity in a riverine context

Lotic ecology is a young discipline that developed after virtually all European river systems had been severely engineered for purposes of navigation, flood control and human use of the river corridor (Whitton, 1984; Petts, Moller & Roux, 1989). Therefore, the conceptual foundations of the discipline were somewhat constrained by a limited appreciation of the crucial roles of spatial complexity and fluvial dynamics in natural river ecosystems. This is especially apparent for floodplain reaches, in which main channels have been straightened, dredged and confined by levées. In addition, the flood plains themselves underwent major modifications to accommodate agricultural production and human habitation. Until quite recently, and with few exceptions (Antipa, 1912), river ecologists perceived river courses as stable, single-thread channels with little or no consideration of flood plains or contiguous groundwater aquifers. Such a perception produces a distorted picture of structural and functional attributes, including biodiversity (Ward & Stanford, 1995).

Natural alluvial rivers are characterized by shifting multiple channels and a remarkable degree of environmental heterogeneity engendered by complex interactions and transitions between surface waters, subsurface waters and riparian systems (Fig. 1), all of which are integral components of river ecosystems (Ward et al., 1998). It is only a slight exaggeration to state that the conceptual foundations of stream ecology were derived largely from studies confined to lotic surface waters, although such an approach may only be valid for certain headwater streams and canyon-constrained riverine reaches, or for highly managed rivers that have been isolated from floodplain–aquifer systems. For example, Hynes (1975), while promulgating a catchment approach, largely ignored the influence of the flood plain. The River Continuum Concept (RCC, Vannote et al., 1980), arguably the most influential paper in river ecology over the past two decades, also lacked a floodplain perspective. It required studies from tropical rivers which had escaped severe anthropogenic impacts (e.g. Welcomme, 1979; Junk, Bayley & Sparks, 1989) and historical investigations of temperate rivers (e.g. Sedell & Frogatt, 1984; Triska, 1984; Petts et al., 1989) to make lotic ecologists recognize the extent to which regulated rivers deviate from pristine conditions (Dynesius & Nilsson, 1994). As stated by Welcomme (1995), evidence ‘indicates that temperate rivers in their pristine condition acted in a similar manner to tropical systems’. In this paper, we present examples from our research on European rivers to illustrate the remarkable degree of spatio-temporal heterogeneity that may be attained where natural processes operate on a large scale.

Figure 1.

 Major environments within river corridors that contribute biodiversity to lotic ecosystems.

In this article, riverine biodiversity is discussed under the headings structural (spatial) diversity, functional (process) diversity and species diversity, although a clear separation between them is not always possible. Under Structural diversity, we consider the types and spatial array of habitat/landscape elements, including environmental gradients, as part of the landscape mosaic in river corridors. Within the subsection Landscape dynamics, we present examples to illustrate links between spatial heterogeneity and hydrologic processes at the landscape scale in river flood plains. Functional diversity from a riverine context includes myriad phenomena such as disturbance processes, nutrient cycling, energy flow and biotic interactions. In this section, the focus is on the interactions between hydrological and successional processes in flood plains. We argue that fluvial dynamics, greatly suppressed by human intervention, is a major component of functional diversity in natural river ecosystems. The section on Species diversity begins with a theoretical consideration of the determinants of species diversity, presents plots of species richness along a floodplain connectivity gradient, and ends with a hierarchical model for assessing alpha, beta and gamma diversity.

Structural diversity

The landscape mosaic

Most simply, a landscape can be considered as a spatially heterogeneous area, with structure, function and change as the primary characteristics. Spatial heterogeneity reflects the structural diversity of riverine landscapes. The landscape mosaic of natural river corridors consists of geomorphic features (e.g. channels, bars, islands) and a diversity of aquatic and riparian habitats encompassing a wide array of successional stages (Church, 1992; Amoros & Petts, 1993; Tockner et al., 1997; Ward, 1998a). Figure 2 illustrates the complex spatial array of just two landscape elements, surface waters and islands, in a river reach (bare gravel forms the landscape matrix of the active corridor). Even at this coarse level of resolution, the high structural diversity possible within a river corridor is apparent.

Figure 2.

 Distribution of vegetated islands and surface water within the active riparian corridor of the Fiume Tagliamento, Italy, river km 84–88. At low and mean water levels most vegetated islands are surrounded by alluvial gravel, not water. Large areas of alluvial gravel are maintained by the dynamic flow regime.

Lotic ecosystems are characterized by multidimensional environmental gradients (Ward, 1989). Longitudinal patterns in habitat variables along the course of rivers have served as a central theme in stream ecology (e.g. Hawkes, 1975; Vannote et al., 1980; Statzner & Higler, 1986; Tockner & Ward, 1999); upstream–downstream changes, especially for river systems traversing extensive altitudinal gradients or crossing major physiographic units, contribute to the high structural diversity of riverine landscapes at the catchment scale. Longitudinal changes also occur at finer scales of resolution, such as between reaches or as riffle-pool transitions.

Lotic ecologists have given much less attention to the lateral dimension. In canyon-constrained reaches, the predominant gradient is an abrupt aquatic–terrestrial transition across a narrow riparian corridor (Gregory et al., 1991). In alluvial reaches, however, the flood plain is the primary-level ecotone between the main channel and the uplands, consisting of a complex gradient of aquatic and riparian habitats that collectively contribute high structural diversity in rivers with extensive natural flood plains (Welcomme, 1979; Junk et al., 1989; Amoros & Petts, 1993).

The vertical dimension beneath surface waters, traditionally the domain of groundwater hydrogeologists, is receiving increasing attention by lotic ecologists, who are beginning to recognize contiguous aquifers as integral components of river ecosystems (Gibert, Mathieu & Fournier, 1997; Ward et al., 1998; Jones & Mulholland, 2000). Earlier regarded as being relatively homogeneous, it is now recognized that alluvial aquifers exhibit substantial structural heterogeneity (Ward & Palmer, 1994; Claret et al., 1999).

Landscape dynamics

Riverine landscapes are in fact shifting mosaics, the spatial heterogeneity of which is maintained by dynamic factors operating at different time scales (Malard, Tockner & Ward, 1999). This is well exemplified by the Fiume Tagliamento in Italy, the last morphologically intact large river in the European Alps (Ward et al., 1999a). Analyses of digitized maps and aerial photographs show that individual landscape elements (e.g. islands, gravel bars, water bodies) exhibit high turnover (≈ 30% in 3–5 years), yet the relative proportion of landscape elements remains relatively constant (Edwards et al., 1999; Kollmann et al., 1999; Tockner, Malard & Ward, 2000). The Tagliamento provides a rare glimpse of the remarkable spatiotemporal dynamics that must have characterized European rivers in the pristine state. Even highly managed European rivers may exhibit high structural diversity in remnant floodplain segments, but this reflects conditions operating prior to human intervention. In the absence of natural dynamics, rather than remaining constant, the composition of landscape elements exhibit unidirectional changes over time (Tockner et al., 2001). Nonetheless, because such river segments still retain relatively high structural diversity, they are prime locations for restoration measures based on reconstituting fluvial dynamics.

The telescoping ecosystem model of Fisher et al. (1998), originally developed to examine recovery rates following disturbance, is modified here to demonstrate the influence of flooding on environmental heterogeneity within floodplain water bodies (Fig. 3). The rising flood reduces environmental heterogeneity by inundating formerly distinct aquatic habitats (maximum connectivity). As flood waters recede, the different types of water bodies slowly recover their distinctive properties (maximum individuality), including habitat features and biotic communities. However, not all lotic systems exhibit this idealized pattern. For example, headwater flood plains, such as Val Roseg (see below), may not undergo complete inundation.

Figure 3.

 A conceptual model depicting how floodplain water bodies, that exhibit essentially uniform conditions during the period of inundation, re-establish their individuality during the dry phase (based on a modification of the telescoping ecosystem model of Fisher et al., 1998). Eupotamal refers to the main channel or side channels with both upstream and downstream connections to the main channel; parapotamal refers to side channels with only downstream connections to the main channel; plesiopotamal refers to a former braided channel that is connected only during high flow; and palaeopotamal refers to former meander loops that are reconnected with the main channel only during floods. See also Armantrout (1998).

The influence of fluvial dynamics on landscape diversity (structural diversity at the landscape scale), however, varies for different geomorphic styles. This differential response is illustrated in Fig. 4, based on the response of four European river segments to changing water levels. The Val Roseg, a glacial flood plain in the Swiss Alps, exhibited a positive relationship between flood levels and landscape diversity over virtually the entire range of discharge. This relationship was derived from detailed surveys of channel patterns at different flow levels, coupled with aerial photographs (colour and infrared images) and a suite of physicochemical measurements (Tockner et al., 1997; Malard et al., 1999; Ward et al., 1999b; Tockner et al., 2000). In Val Roseg, landscape diversity is structured by the shifting dominance of subcatchment water sources (snow melt, englacial water, subglacial water, ground water) and flow paths (surface, alluvial aquifer, hillslope aquifer; cf. Fig. 7), and reflects the fact that the system accommodates major spates without inundating the entire flood plain. The Danube within the Alluvial Zone National Park near Vienna is a large lowland river (Schiemer, Baumgartner & Tockner, 1999). Landscape diversity is the greatest at relatively low inundation levels, reflecting a floodplain surface that has been structured by flow regulation since about 1850. Functional processes within the river–floodplain complex represent a shifting dominance between autogenic and allogenic factors (Tockner et al., 1999a). The Fiume Tagliamento (already referred to above), a ‘flashy’ Mediterranean river characterized by extensive island-braided reaches, attains the highest landscape diversity at intermediate levels of inundation. The Morava, a meandering river tributary to the Danube, appears to exhibit progressively declining landscape diversity with increasing inundation, although few data are available (Tockner et al., 2000). The relationship between discharge and landscape diversity is a functional characteristic of any river–floodplain system that is likely to exert major control on biodiversity patterns. These examples demonstrate that changes in landscape diversity during expansion and contraction events are controlled by both regional processes operating at the catchment scale (e.g. Val Roseg) and by local processes occurring at the floodplain scale (e.g. Danube). For example, in Val Roseg environmental conditions within individual water bodies are determined largely by the relative importance of different hydrologic reservoirs and flow paths operating within the catchment (Malard et al., 1999). In contrast, the nature of Danubian floodplain water bodies reflects the frequent shifts between short-term periods of hydrological connectivity and the dominance of internal processes during isolation (Tockner et al., 1999a).

Figure 4.

 Idealized conceptual relationships between landscape diversity (sensuMalard, Tockner & Ward, 2000) and water level for a glacial flood plain (Val Roseg), an island-braided reach (Tagliamento, middle section), a lowland river-floodplain complex (Danube) and a meandering reach (Morava). Landscape diversity is determined from categorical maps and indices such as patch richness and Shannon diversity for physicochemical parameters of the water column (for more details: see Malard et al., 2000; Tockner et al., 2001).

Figure 7.

 Expansion and contraction cycle of the channel network of the Val Roseg flood plain (four seasons); and the relative contribution of different water sources to total floodplain discharge. HA: hillslope aquifer; MW: glacial melt water (englacial and subglacial water); GW: alluvial ground water; SM: snow melt water. Based on a mixing model, the relative contribution of hillslope ground water to total floodplain discharge ranges from 1% in summer to > 70% in winter (K. Tockner & F. Malard, unpublished data).

The influence of water level dynamics on landscape diversity varies for the environmental variable considered. In a Danubian flood plain, differential responses for specific conductance, sestonic chlorophyll, nutrients (e.g. Nitrate-N), or shoreline length were observed (Fig. 5). We used the coefficient of variation (CV) to quantify landscape heterogeneity, as CV adjusts the sample variance by the mean and thus is a better comparative measure of heterogeneity than sample variance alone (Crowl et al., 1997; Palmer, Hakenkamp & Nelson-Baker, 1997). Because no single indicator will possess all the desirable properties, a set of complementary indicators is required to understand biodiversity (cf.Noss, 1990). Developing an overall index of landscape diversity in relation to water level will require a greater understanding of the dynamics of the key variables. It is probable that landscape-level responses will vary among catchments as a function of geomorphic structure and climatic regime.

Figure 5.

 Relationships between water level (cm) and variability for selected environmental factors in a Danubian flood plain (study area: 4.5 km2). The coefficient of variation (CV) was calculated as a compound index of landscape heterogeneity (see Crowl et al., 1997; Palmer et al., 1997). CV was calculated for specific conductance, nitrate-N and sestonic chlorophyll (number of replicates per sampling date: 35–49; curves are fitted by eye; after Tockner et al., 2000). Shoreline length (km) was calculated from topographic inundation maps (K. Tockner & C. Baumgartner, unpublished data). mw: mean water level.

Functional diversity

Ecological succession

The balance between rejuvenation (erosive flooding) and terrestrialization processes sustains a diversity of successional stages in riverine flood plains (Fig. 6). The wide variety of water bodies and riparian vegetation stands characterizing natural river corridors is attributable, in part, to hydrarch and riparian succession operating on time scales of decades and centuries. Because different successional stages contain distinct biotic communities (e.g. Castella et al., 1984; Copp, 1989), species richness is increased by successional diversity. In contrast, river regulation tends to reduce erosive flooding events below dams, thereby upsetting the balance between rejuvenation and terrestrialization, typically resulting in an unidirectional trajectory towards low seral diversity.

Figure 6.

 The diversity of successional stages in natural and regulated riverine landscapes as a function of the relationship between rejuvenation and terrestrialization processes (see text).

Fluvial dynamics as functional diversity

We believe that lotic ecology would benefit by recognizing natural disturbance from fluvial dynamics as an important type of functional diversity. A dramatic example of functional diversity in this context is the ‘flood pulse’ that, upon inundating the flood plain, sets off a series of interactive processes that reverberate throughout the entire ecosystem (Junk et al., 1989).

Ecosystem expansion and contraction without overbank flooding, a common phenomenon in lotic systems, has been given scant attention by river ecologists (Stanley, Fisher & Grimm, 1997). Such ‘flow pulses’ (in contrast to flood pulses) also induce major structural and functional shifts (Tockner et al., 2000). The channel network in the flood plain of Val Roseg, which expands and contracts in response to annual glacial melt cycles, illustrates this clearly (Fig. 7). Hydrochemical indicators and tritium dating of ground water were employed to link the expansion/contraction cycle with shifts in dominant hydrological processes (Tockner et al., 1997; Malard et al., 1999; Ward et al., 1999b). During the period of maximum contraction in winter, subsurface hydrological processes predominate, habitat fragmentation is the highest, and ground water from deep alluvial and hillslope aquifers constitutes the predominant water source. In spring, expansion of the channel network largely reflects an extended period of snowmelt runoff, which commences on the lower slopes, moves to higher slopes and terminates with the melting of snow deposited on the glaciers. Snowmelt on the catchment slopes infiltrates morainic and fluvioglacial deposits and enters the flood plain via subsurface flow paths, whereas snowmelt from the glaciers enters the flood plain as surface runoff. During maximum expansion in summer, surface hydrological processes predominate and glacial meltwater is the main water source. Surface and shallow subsurface flowpaths dominate. Although highly turbid water containing glacial flow permeates the system, some clear water refugia occur along the floodplain margin. As temperature declines in autumn the channel network contracts, especially in the upper part of the flood plain (Fig. 7). The upper floodplain is fed largely by subglacial water that enters as surface flow in the main channel and seeps into the alluvium. In the lower part of the floodplain, surface flow is sustained in autumn by upwelling of ground water from deep alluvial and hillslope aquifers. Ongoing investigations are exploring the adaptations of aquatic animals that exploit the shifting dominance of hydrological processes.

Species diversity

Determinants of species diversity

Aspects of structural and functional diversity interact in complex ways to determine the species diversity within communities. Similar unimodal responses to three inter-related variables are hypothesized in Fig. 8. The Intermediate Disturbance Hypothesis (IDH; Connell, 1978) predicts low species diversity in habitats exposed to high levels of disturbance, where only highly tolerant/rapid colonizing species occur (Fig. 8a). Low species diversity also occurs under very low levels of disturbance, where superior competitors monopolize resources. Species diversity is maximum where environmental perturbations are at an intermediate level of intensity/frequency because a variety of taxa coexist. Support for the IDH is available from studies of marine, terrestrial and lentic environments (e.g. Taylor, 1973; Menge, 1979; Flöder & Sommer, 1999). Ward & Stanford (1983) suggested that the IDH may also provide an explanation for species diversity patterns in both natural and altered lotic ecosystems; empirical support from running waters is provided by the investigations of McAuliffe (1984), Minckley & Meffe (1987), McCormick & Stevenson (1989) and Townsend, Scarsbrook & Doledec (1997), among others. The disturbance gradients that occur at different scales on riverine flood plains provide an excellent setting for rigorous testing of the IDH.

Figure 8.

 Idealized curves of the relationships between species diversity and disturbance, ecotone density, patch size and connectivity (see text).

It has been proposed (Naiman et al., 1988) that species diversity will also maximize at some intermediate level of ecotone frequency (Fig. 8b) because large homogeneous patches provide little habitat for edge species, whereas small patch size excludes interior species. Maximum diversity should occur where there is an optimal mix of patch and edge habitat. Habitat patches within river corridors offer an appropriate setting to investigate this relationship.

Connectivity refers to the transfer of energy and matter (including biota) across ecotone boundaries, with major implications for biodiversity (Lachavanne & Juge, 1997; Ward, 1998b). We postulate that species diversity will be maximum at some intermediate level of hydrologic connectivity (sensuAmoros & Roux, 1988) between patches within river corridors (Fig. 8c). At low connectivity, species diversity should be reduced by the absence of the fluvial dynamics that sustains a diversity of successional stages within the river corridor, whereas excessive connectivity will keep all communities in pioneer stages. Bornette, Amoros & Lamouroux (1998), who studied aquatic plant assemblages in a flood plain of the French Rhone, found the influence of connectivity more complex than anticipated. They concluded that effects of connectivity on species richness are determined by the type and intensity of connectivity and by geomorphic attributes of the cut-off channels. For example, connectivity may reduce species diversity if excessive nutrient loading in the river leads to eutrophication of connected floodplain water bodies (e.g. Van den Brink et al., 1996).

Because species diversity also varies as a function of productivity, Huston (1979, 1994) combined the roles of disturbance and productivity in the dynamic equilibrium model (Fig. 9). The model predicts that communities in unproductive environments (scarce resources) with high disturbance will be controlled by abiotic factors and will contain non-interactive assemblages of some fugitive species. In contrast, species diversity in environments rich in resources but with minimal disturbance will be constrained by competitive exclusion. According to the model, species diversity attains the highest levels in environments with an optimal balance of disturbance and resource levels (along the oblique band in Fig. 9). River flood plains provide ideal environments for testing the interactive roles of disturbance and productivity in structuring species diversity at different spatial scales. At the floodplain scale, for example, comparisons could be made between dynamic flood plains that differ in productivity or between flood plains with different degrees of hydrologic connectivity. Investigations could also compare water bodies within flood plains that vary in productivity (successional stage) or connectivity (fluvial dynamics).

Figure 9.

 Postulated roles of disturbance and resource levels in structuring species diversity [greatly modified from Huston’s (1979; 1994) dynamic equilibrium model]. The model predicts that the level of disturbance required to attain maximum species diversity varies as a function of resource level.

Species richness along a connectivity gradient

Plots of species richness for different faunal and floral components at six sites along a connectivity gradient (Fig. 10) provide an explanation for the high species diversity in river corridors. Sites are situated along a transect from the Danube channel (Site 1) to isolated water bodies at the edge of the flood plain (Site 6). Each of the biotic components shown attains peak values at different sites, resulting in a series of overlapping diversity maxima arrayed along the connectivity gradient. When all groups are combined species diversity peaked at an intermediate level of connectivity. Tockner et al. (1999b), however, have demonstrated that beta diversity displayed an almost inverse relationship with alpha diversity, with the highest average values in isolated and fragmented floodplain channels.

Figure 10.

 Species richness peaks for different faunal and floral components (including native and non-native species) along a Danube-floodplain transect (idealized curves, modified from Tockner, Schiemer & Ward, 1998). The connectivity gradient extends from the main channel to the edge of the flood plain, a distance of about 1 km.

Hierarchical model

Gamma diversity, the regional species pool, is a function of the number of species in each habitat (alpha diversity), the number of habitat types (habitat diversity), and the turnover of species between habitats (beta diversity). The spatial extent of the ‘region’ is not fixed, however, and may vary according to the goals of the study or other criteria. For example, in some cases a biome may be the appropriate regional study unit, whereas in other cases the species within a catchment or within a flood plain may be considered as gamma diversity. Therefore, we have developed a hierarchical framework in which alpha, beta and gamma diversity are arrayed hierarchically (Fig. 11). At the uppermost hierarchical level all of the species (of fishes, for example) occurring in the Alps constitute gamma diversity; alpha diversity is the number of species in each catchment; and beta diversity is the turnover of species between catchments. At the second hierarchical level, gamma diversity becomes the number of species in one of the catchments, alpha diversity the number of species in each river reach, and beta diversity the species turnover between reaches. Although only four hierarchical levels are shown in this example, additional levels could be added at the top (e.g. central Europe) or bottom (e.g. habitats; microhabitats) and the levels may be configured in a different manner. The identification of hierarchical levels (or ‘domains’) may help to isolate factors constraining or generating biodiversity patterns and processes. For example, hydrological connectivity is expected to control biodiversity at the floodplain scale, whereas biogeographic factors primarily operate at the catchment scale.

Figure 11.

 Hierarchical model for biodiversity in the Alps (modified from Ward et al., 1999c).

Conclusions and perspectives

A widened concept of biodiversity could provide an integrative perspective for aquatic ecologists and a unifying theme for holistic studies of rivers as ecosystems. Biodiversity integrates ecology with evolution and biogeography. At the ecological scale, biodiversity integrates pattern (structure) with process (function) and biotic variables with abiotic variables. Biodiversity links spatial and temporal phenomena across hierarchical scales and levels of biological organization and elucidates the roles of functional processes in ecosystems. An operational understanding of biodiversity links basic research with applied research by providing a basis for effective river conservation and management initiatives.

Research needs include integrated studies of biodiversity that investigate all major riverine environments (Fig. 1). Biodiversity baseline data for near natural reference rivers and river segments are badly needed to develop viable conservation and management strategies. Application of landscape indices from a biodiversity perspective may provide the appropriate scale for initiating river restoration programmes (e.g. Frissell & Bayles, 1996). Quantification of successional diversity for both riparian and aquatic components may lead to greater understanding of the maintenance of biodiversity, with important management applications. The inclusion of independent measures of beta diversity, usually given scant attention by river ecologists, in comprehensive studies of biodiversity should yield additional insight into the interactions that structure biotic communities. The focus on habitat and landscape presented here, although perhaps the most relevant scale for management of riverine biodiversity, should be expanded to encompass population and genetic levels of organization.


We are grateful to the following for providing ideas or contributing data used in this paper: D.B. Arscott, Dr C. Baumgartner, Prof. P.J. Edwards, Prof. A.M. Gurnell, Dr J. Kollmann, Dr F. Malard, Prof. G.E. Petts, Dr C.T. Robinson, Prof. F. Schiemer, and Dr U. Uehlinger. We also thank Profs J.D. Allan and A.G. Hildrew and two anonymous reviewers for valuable suggestions to improve the presentation. Research results reported herein supported by grants from the Swiss National Science Foundation (SNF grant 21–49243.96) and the ETH Forschungskommission (0–20572–98).


  1. Based on a keynote lecture presented at the Symposium for European Freshwater Sciences, Antwerp, Belgium, 25–28 August 1999.