Allocation of River Flows for Restoration of Floodplain Forest Ecosystems: A Review of Approaches and Their Applicability in Europe
- First Online:
- Cite this article as:
- Hughes, F. & Rood, S. Environmental Management (2003) 32: 12. doi:10.1007/s00267-003-2834-8
- 652 Views
Floodplain forests are flood-dependent ecosystems. They rely on well-timed, periodic floods for the provision of regeneration sites and on tapered flood recession curves for the successful establishment of seedlings. These overbank flood events are described as “regeneration flows.” Once floodplain forest trees are established, in order to grow they also require adequate, although variable, river stage levels or “maintenance flows” throughout the year. Regeneration flows are often synonymous with flood flows and only occur periodically. There is a disparity between this need for varied interannual flows over the decadal time frame and the usual annual cycle of flow management currently used by most river management agencies. Maintenance flows are often closer to established minimum flows and much easier to provide by current operational practices.A number of environmental flow methodologies, developed in North America, Australia, and South Africa are described in this review. They include the needs of the floodplain environment in the management and allocation of river flows. In North America, these methodologies have been put into practice in a number of river basins specifically to restore floodplain forest ecosystems. In Australia and South Africa, a series of related “holistic approaches” have been developed that include the needs of floodplain ecosystems as well as in-channel ecosystems. In most European countries, restoration of floodplain forests takes place at a few localized restoration sites, more often as part of a flood-defense scheme and usually not coordinated with flow allocation decisions throughout the river basin. The potential to apply existing environmental flow methodologies to the management of European floodplain forests is discussed.
KeywordsFlow allocationFloodplain forestRegeneration flowsMaintenance flowsRiver managementRiver restoration
Floodplain ecosystems have important roles to play in maintaining the integrity of river corridors that are under increasing pressure from population growth and intensive development. Concerns are focused on the economically important natural functions that floodplains carry out and with issues of biodiversity and habitat conservation (e.g., Gopal and others 2001). It is recognized that river flows are the key to the revitalization of floodplain habitats, and an interest in managing flows for this purpose has emerged in recent years, growing out of a longer-established interest in managing flows for instream habitats. Much of this interest has developed in North America, Australia, and South Africa, where methodologies for flow allocations have been developed as a response to heavy demands on the total volume of the water resource, particularly in semiarid and arid areas. As yet there are few signs of flow allocation methodologies being developed and applied to revitalize floodplain ecosystems in Europe. Instead, floodplain restoration takes place at discrete sites along river corridors (Hughes and others 2001), unconnected to the overall planning of water resources use and management within the river basin.
Methods for deciding how to allocate river flows to benefit instream ecological targets such as fish species or aquatic invertebrates have been developed since the early 1980s. Most widely used and written about is a set of techniques collectively known as the Instream Flow Incremental Methodology (IFIM) (developed initially by Bovee 1982 and subsequently by many authors such as Stalnaker and others 1995), of which the best known and most utilized component is PHABSIM. This is a physical habitat simulation model aimed at matching the habitat needs of target species at different parts of their life-cycles with the river flows that provide those needs. This model has been applied to many single-species ecological targets, although there have also been multispecies applications (e.g., Gore and others 1992,2001Bullock and Gustard 1992, Maddock and others 2001). PHABSIM has many limitations; it is difficult to interpret (Gan and McMahon 1990), it can only be used in small rivers since it requires intensive and expensive data inputs (King and Tharme 1994, King and Louw 1998), and it concentrates on minimum and optimal flows for target species at the expense of overall ecosystem richness and ecological processes (Pusey and others 1998, Gore and others 2001).
In the 1990s, an urgent need to allocate river flows for dependent terrestrial ecosystems as well as instream targets was recognized. Instream methodologies have been extended to floodplain ecosystems and influenced more holistic approaches to water allocation, which include ecological targets alongside traditional water uses.
In many river systems forests form the natural vegetation of floodplain zones, their exact nature being determined by their location in a river basin (Tabacchi and others 1996). These forests are linear, flood-dependent ecosystems that are limited in their lateral extent by the cross-sectional shape of the river valley, the depth of the water table away from the active channel, and the width of the area subjected to flooding. Thus in low-order streams (sensu Horton 1945) or partially incised rivers they may be narrow strips of trees, while on wide, geomorphologically complex, meandering or braided rivers they tend to mimic this complexity in a mosaic of different forest communities. In this review we refer to all of these as floodplain forests. In many European countries floodplain forests have disappeared or become greatly reduced as a result of river management activities, and they are regarded as a threatened habitat (UNEP-World Conservation Monitoring Centre 2000, Hughes 2002).
While most aquatic species and nonforested floodplain ecosystems require particular intraannual flow patterns to satisfy their life cycle needs, floodplain forest ecosystems additionally require interannual variations in flow patterns to satisfy theirs. This paper specifically reviews approaches to prescribing flows for forested floodplain ecosystems and describes some of the methodologies currently in use to implement them. It begins by describing the essential linkages between floodplain forests and their adjacent rivers. It then describes flow allocation methodologies that have specific geomorphological or ecological targets before describing more holistic methodologies aimed at providing “ecological” or “environmental” flows within the wider context of river basin water resource planning. A series of case studies describes the implementation flow methodologies for floodplain forests. The final section discusses the potential for the application of these methodologies in Europe.
Linkages Between the River and Floodplain Forest Ecosystems
Early efforts to understand the relationships between the abiotic inputs to floodplain ecosystems (mostly in the form of river discharges) and ecosystem responses on (often forested) floodplains have led to the development of simulation models. Some of these earlier models have contributed to subsequent flow allocation methodologies. For example, Franz and Bazzaz (1977) use a probabilistic model based on niche differentiation in a floodplain forest to stimulate the effects of a modified environmental gradient (due to hydrological change) on species distribution in floodplain forests, ahead of a proposed reservoir. In early transition models, vegetation alters in response to incremental changes in environmental variables. The new vegetation state then becomes the starting point for the next incremental change (Mitsch and Gosselink 1993). A number of transition models are based on growth rates of individual species and their response to the environment. A good example is SWAMP (Phipps 1979), a tree-growth simulation model that charts tree growth rates for different river discharges in a bottomland forest in Arkansas, USA. FORFLO, based on SWAMP, but developed later by Pearlstein and others (1985), has been applied to evaluate the impact of an altered hydrologic regime on succession in a bottomland forest in South Carolina, USA.
A more recent example of modeling floodplain forest ecosystems is given by Johnson (1992), who uses a simple mathematical model based on measured rates of forest succession and river meandering for pre- and postdam periods to predict future changes in forest composition in the Upper Missouri River in the United States. The model he uses is analagous to many ecosystem models. In it, five compartments represent the major forest cover types, including pioneer communities, and the fluxes of material substances (such as energy and nutrients) are described by linear, differential equations. Stromberg and Patten (1990) relate streamflow requirements to the growth rates of two riparian tree species, Populus trichocarpa and Pinus jeffreys along Rush Creek in California, USA. They discover that the growth of Populus trichocarpa is opportunistically related to high river flows through the year while the vernal growth habit of Pinus jefferys depends on prior-year flooding. They then use this information to model future growth rates under different flow scenarios. A later study by Stromberg and others (1996) demonstrates that as well as particular river discharges, maintenance of local water tables within floodplain zones is also critical to the growth and survival of floodplain species.
Importance of Variable Flows
The models described above are primarily designed to elucidate relationships between streamflows and tree growth patterns. They concentrate on variable flows contained within the channel wetted perimeter and which, by lateral seepage, replenish water tables in adjacent floodplains and promote growth. These flows could be called “maintenance flows.” A parallel and large literature developed through the 1980s but most prolifically through the 1990s on the flow needs for the regeneration of riparian trees. Field studies and greenhouse experiments investigated the relationships between flood events, shape of the hydrograph, and regeneration patterns through time and showed that flood events are an important driving force in the development and progression of floodplain forest ecosystems. There are a number of reviews of this literature including Junk and Welcomme (1990), Brinson (1990), Kangas (1990), Malanson (1993), Décamps (1993, 1996), Gurnell (1995), Naiman and Décamps (1997), Hughes (1997), and Hughes and Rood (2001).
The literature shows that flood timing, flood stage, and the shape of the hydrograph through the first growing season are critical in determining the level of successful regeneration in any year. In many cases, the flows that were found to promote a regeneration event were medium-to-large magnitude flow events, capable of causing channel movement and of creating sedimentation sites that tree seedlings could colonize. The pattern and spatial scale of regeneration across a floodplain and the volume of flows that promote that regeneration vary with channel pattern. In meandering rivers, parallel scroll bars commonly support lines of even-aged trees with the youngest trees nearest to the river. In braided and anastomosed river reaches, spatial patterns of tree ages are more random as islands are eroded and deposited across the floodplain. Flow events that promote widespread regeneration across a floodplain could be termed “regeneration flows” and are, in general, larger and therefore occur less frequently than maintenance flows. The distinction between these different types of flow needs is important if we are to allocate flows within “whole of river” management frameworks. Maintenance flows can be accommodated within instream flow allocation methodologies and can often be equated with minimum flows, but regeneration flows are less predictable and require acceptance of the notion of planned or controlled overbank floods. Temporal variability is a key characteristic of both types of flow and needs to be reproduced for effective flow management.
The importance of different sized floods to carry out different geomorphological and ecological tasks is emphasized by Hill and others (1991). They distinguish between flows that are needed for fisheries, channel maintenance, riparian habitats, and valley maintenance in the Salmon River, in Idaho, USA (in increasing order of magnitude and decreasing order of frequency). Their paper has been very important in the development of thinking in North America on the provision of flows for different physical and biological purposes because it considers simultaneously the linkages between instream and out-of-stream flows and between the geomorphological and ecological “functions” that different flows might perform. Elsewhere, other methodologies have used similar conceptual approaches e.g., the holistic approach and its derivatives devised in Australia by Arthington and others (1992).
A number of methodologies have been and are being developed to bring these conceptual ideas into the practical and applicable domain. These can broadly be classified into four principal categories (Table 1). Methodologies in two of these categories only deal with instream targets, but the two other categories include flow allocation for floodplain ecosystems as well as instream habitats. It is primarily with these latter two categories that this review is concerned. The greatest effort has been put into these methodologies in countries that experience semiarid or arid conditions and in which water quantity is frequently a limiting factor to economic development, for example, in Australia, South Africa, and the semiarid southwestern United States.
Classification of flow allocation methodologies
Instream and floodplain
Flow allocation approach
Flow allocation approach
Geomorphological or ecological targets only
• IFIM (instream flow incremental methodologies), e.g., PHABSIM (physical habitat simulation models). Usually used for single-species targets.
Very data intensive, limited to small rivers, can be difficult to interpret. Applied over the intraannual time frame to recommend-flows for habitat provision.
• Flushing Flows (e.g., Kondolf 1998). Designed to achieve instream or floodplain. geomorphological objectives.
• Floodplain maintenance flows (Whiting 1998). Flows to maintain physical properties of a floodplain based on dominant floodplain building processes.
Usually used in conjunction with minimum environmental flow criteria, these can be used to mimic the effects of floods, usually downstream of regulatory structures. These flows cannot deal with restoring sediment supply downstream of dams. The flows can be varied on an inter-annual basis.
• 2-stage refinement of IFIM recommended by Gore and others (2001) for long-lived species like fresh water mussels.
Recommends linkage of historical flow patterns to recruitment years to allow provision of necessary inter-annual variability in flow patterns. This allows provision of all life-cycle needs for a greater range of aquatic species.
• Recruitment box Model (Mahoney and Rood 1998). Designed specifically for initiating recruitment of floodplain tree species.
This allows fine-tuning of planned releases downstream of dams to allow both well timed flood peaks and well-tapered flood recession curves. Flows can be varied on an inter-annual basis depending on water supply
Broad environmental targets
• Minimum environmental flow methods, usually using a statistical threshold e.g., CAMS (catchment abstraction management strategies) United Kingdom strategy for minimum flow levels, using daily Q95 or minimum flows determined by SDAGE (Schémas Directeurs d’Aménagement et de Gestion des Eaux) plans in France.
Typically, these flows are the current operational flows in many countries. They are determined on a monthly basis and therefore vary intraannually but have no scope for deliberate inter-annual variations. Such methods consider instream hydrology but tend to ignore geomorphological processes and floodplain ecosystems governed by inter-annual discharge even
• A multiple flow methodology by Hill and others (1991). A conceptual methodology for measuring different types of flow regimes.
This describes a series of methodologies for identifying and analysing flow requirements of instream in and out-of-stream ecosystems in order to recommend multiple flow regimes. The multiple flow analysis routine described uses methods currently in common use to ensure wide applicability.
• Bottom-up approaches e.g., EPAM (Expert Panel Assessment Methodology), Australia (Swales and Harris 1995); BBM (Building Block Methodology), South Africa (King and Louw 1998). Drift (Downstream Response to Imposed Flow Transformation), South Africa (King and others 2003).
These construct a modified flow regime by adding flow components to a zero flows baseline. they can build in inter-annual flow variability but sometimes rely heavily on ‘expert’ opinion rather than scientific data for determining the flow components.
• Top-down approaches e.g.,Benchmarking Methodology, Australia (Arthington 1998, Brizga 2000).
These link change in a flow statistic from its pre-regulation value to ecosystem impact. They determine the degree of modification that can occur before ecosystem degradation takes place. Their main use is in risk assessment for flow scenario building which can include inter-annual variability.
• Combined bottom-up and top-down approaches e.g. Flow Restoration Methodology, Australia (Arthington and others 2000).
These allow more precise definition of flow regimes than either top-down or bottom-up approaches and allows inter-annual flow variability.
Water Allocation Methodologies
Flow Allocation for Geomorphological Goals
In this section, we emphasize the importance of different flow types in creating floodplain landforms rather than instream channel forms and aquatic habitats. A number of authors have given flow definitions (or produced typologies) to describe the geomorphological capabilities of different flow types, although many concentrate on in-channel processes. For example, flushing flows are those flows needed to maintain substrate properties of the channel (Reiser and others 1989) and channel maintenance flows are those that define the ability of a channel to maintain sediment transport processes (Rosgen and others 1986, Andrews and Nankervis 1995). It is important to define appropriate flows for a specified set of physical objectives. In a discussion of the development of flushing flows for channel restoration on Rush Creek, California, Kondolf (1998) points out that the choice of flows involves a compromise between the replication of predam flows and channel processes, and the maximization of water storage and regulation of flows for downstream users. They also depend on the age of the dam and the degree to which adjustments have already been made in the channel downstream. Kondolf and Wilcock (1996) group methods for estimating flushing flows into three categories (Table 2). It is also important to monitor the impacts of designated flushing flows. Kondolf (1998) points out that unless a program is in place to test the effects of flushing flows in achieving their objectives, there is a severe limitation on the degree to which any particular experience can inform the next restoration initiative.
Methods for estimating flushing flowsa
Sediment entrainment methods—estimate flows needed to mobilize gravel bed, assuming that bed mobilization is a suitable surrogate for flushing
Direct calibration methods—require observation of bed mobilization, bedload transport or changes in bed material size
Self-adjusted channel methods—use a statistic drawn from the pre-dam (or upstream, unregulated) flow regime as the flushing flow, assuming some kind of equilibration between channel and pre-dam flow
The development of maintenance flow approaches to describe flows needed to maintain both physical and biological functions of floodplains are reviewed by Whiting (1998). He proposes that floodplain maintenance concepts can be applied to “the quantification of water rights, the preservation of natural floodplains and their ecosystems, and the restoration of altered streams” (p. 160). He argues that while there is a suggestion that vertical accretion is a key to maintaining floodplain habitats (Kondolf and Wilcock 1996), there has been little consideration of the processes maintaining the physical properties of the floodplain. The six main mechanisms involved in floodplain formation are lateral point bar and vertical accretion, braided channel accretion, oblique and counterpoint accretion, and accretion in abandoned channels (Whiting 1998, summarized from Nanson and Croke 1992). It is clear, from the extensive literature on the geomorphological processes that shape floodplains, that the dominant mechanisms vary among rivers in different bioclimatic regions and with different caliber sediment loads and channel pattern types. The exact morphological and sedimentological characteristics of any particular floodplain will be the direct result of these dominant mechanisms and will strongly influence the vegetation communities found there (Hughes 1997). It seems likely that floodplains that depend most on vertical accretion processes are those experiencing most frequent overbank flood events. In many instances, these are the flow events that decrease in frequency with river management practices such as dam construction and intense water abstraction.
Whiting (1998) develops a conceptual model to facilitate quantification of the flows required to maintain a floodplain. In all cases flows must allow for the movement and deposition of the sediment making up the floodplain. Thus, on floodplains built mainly from lateral accretion or braid bar accretion, estimation of floodplain maintenance flows is based on a calculation of the threshold for movement of sediment as bedload, i.e., the fluid drag matches the gravitational force. Where a floodplain is mainly built through vertical accretion, maintenance flows are defined as flows that promote movement of sediment in suspension, i.e., those where shear velocity is greater than the particle settling velocity. The dominant mechanisms operating over a floodplain through time can be identified through interpretation of sediments and organic layers and the debris contained within them. In addition to flow magnitudes, flow frequency and duration are important parameters in the volume of sediment deposited over time and hence floodplain maintenance. These flow characteristics are best calculated using hydraulic models coupled to sediment transport equations. Whiting (1998) uses these concepts to calculate the flows maintaining the Chagrin River floodplain in Ohio, USA, as those with a recurrence interval of four or more years. (cf. the commonly quoted 1.5 years determined for a number of North American rivers and reviewed by Leopold 1994 ).
On large river floodplains, occasional high flows capable of causing considerable channel movement are needed in addition to channel-forming and floodplain maintenance flows. The extensive floodplain forest ecosystems found on large floodplains are characterized by a complex mosaic of different aged vegetation communities whose initiation and establishment are linked to the deposition of lateral or point-bars during a flood event. Viewed over time, these floodplain forests are a constantly shifting vegetation mosaic in which the proportions of pioneer and mature vegetation types reflect the turnover rates of floodplain sediments and the speed of channel movement (Hughes 1997).
Hill and others (1991) comment on the lack of methodologies for determining the flow volume or duration needed to maintain riparian habitats and their floodplains. They propose a conceptual model that has been discussed above but also demonstrate how it can be used to develop multiple flow recommendations on the Salmon River, Idaho, USA (Figure 1). Their methodology uses a combination of PHABSIM flow analysis for fishery flows, evaluation of bankfull flows for channel maintenance flows, HEC-2 analysis to identify out-of-channel riparian flows, and finally, peak discharges that usually exceed Q25 as valley maintenance flows. The last are the most difficult flows to establish on the basis of relatively short hydrological records because, over time, most valleys have been formed by a multiplicity of factors, other than fluvial processes (Hill and others 1991).
Flow Allocation for Ecological Targets on a Floodplain
A number of authors have carried out research to determine flows necessary to support the biotic components of floodplain ecosystems (Hughes and Rood 2001). Stromberg and Patten (1991) related annual radial growth rates for several poplar species to various flow parameters along Bishop Creek, California, USA. They determined that to maintain healthy growth rates, the flow volume needed is about three to four times greater than diversion flows and about 40%–60% of estimated natural flow volumes.
In the semiarid regions of western North America riparian woodlands have been severely degraded by human impacts, including direct clearing and the downstream impacts of dam construction (Rood and Mahoney 1990). The conservation and restoration of these woodlands has developed as a major environmental priority and the current and imminent relicensing of many of these dams is dependent on reconsideration of operations practices and environmental impacts. Efforts are underway to restore more natural instream flow patterns that will enable recovery of the natural processes of ecosystem regeneration (Rood and others 1998). Dam removal is also seen as a potential river restoration tool (Bednarek 2001), but so far the few case studies have concentrated on restoring aquatic habitats and fisheries.
The reestablishment of erosion and deposition to create barren nursery sites for recruitment of native riparian tree species requires some flooding. However, the timing of the high flows is important. Peak flows need to precede dispersal of tree seeds so that germinants established prior to the flood peak are not scoured away by the rising stream. The timing of flood peaks is somewhat variable, but within a given region, native plants tend to be adapted to the range of naturally occurring hydrographs in unmodified rivers and synchronized to the flood peak, even though seed dispersal is often directly cued by photoperiod (daylength) and modified by temperature (Mahoney and Rood 1998). Dams can substantially alter the timing of peak flows, creating an unnatural scheduling that may promote the encroachment of invasive exotics that compete with and may exclude native plants (e.g., Stevens and Waring 1985, Sher and others 2000). Ecologically informed dam operation would involve consideration for the seasonality of hydrograph patterns that could serve as a management tool to favor the native plants and prevent further proliferation of undesirable exotic species (Shafroth and others 1995).
The recession limb of the flood hydrograph is also extremely important for the successful recruitment of floodplain vegetation (Mahoney and Rood 1992, Segelquist and others 1993). Seedlings that are dependent on the moisture associated with the capillary fringe above the riparian water table are especially vulnerable to drought stress (Mahoney and Rood 1998). Gradual recession ramping rates are required in which stream stages and corresponding riparian water table levels decline at rates that can be matched by seedling root growth. After initial recruitment of riparian vegetation, sufficient minimum instream flows are required during the subsequent hot and dry periods to provide adequate water to maintain the riparian phreatophytic vegetation (Rood and Mahoney 1990).
The floodplain “recruitment box model” (Mahoney and Rood 1998) has been used to describe quantitatively the combined requirement for appropriately timed high flows to create and saturate suitable floodplain sites and subsequent gradual flow recession to permit seedling survival. In the model, ideal stage recession rates following peak flows should be about 2.5 cm/day or less, although this decline rate is influenced by floodplain substrate texture, plant species, and the ambient weather conditions related to water demand, particularly temperature, rainfall events, wind and sunshine (Mahoney and Rood 1992, Hughes 1997 , Mahoney and Rood 1998, Barsoum and Hughes 1998).
How far can flow prescriptions for riparian restoration along different rivers and in different regions be generalized?
How applicable are common flow prescriptions to different riparian trees and shrubs?
Typical present-day flow prescription strategies are actually based on river and floodplain stage (elevation) rather than river discharge (flow). This is appropriate for riparian processes since it is the floodplain stage elevation rather than the amount of water passing through that primarily determines the physical and ecological conditions for recruitment (Mahoney and Rood 1998). The expression of all data in terms of stage produces quantitative consistency across a broad range of stream sizes. For example, seedlings of riparian cottonwoods are established in bands at specific floodplain elevations that parallel the stream channel. These bands occur at relatively consistent elevations ranging from about 60 to 200 cm above the base stream stage, the low river elevation that usually occurs towards the end of the plants’ growing season. This elevational range is relatively consistent through about 20 degrees of latitude across western North America and for streams ranging from relatively small creeks to large rivers. However, the generalization of flow prescriptions becomes more difficult where there is an element of bedrock control along a river channel, as in the Sabie River of South Africa (Moon and others 1996 , Van Coller and others 1997).
The second question about common instream flow prescriptions concerns their applicability to different riparian tree and shrub species. Studies along the Truckee River, Nevada, USA , the Oldman, St. Mary, and Bow rivers in Alberta, Canada (Rood and others 1998, 1995), and the Bill Williams River, Arizona, USA (Shafroth and others 1998) consistently indicate that native willows (Salix spp.) and cottonwoods (Populus spp.) and exotics like salt cedar (Tamarix ramosissima), have generally similar stream stage requirements. Differences in phenology and timing of seed release between native and exotic species could provide the basis for deliberate management of flows to favor desirable native trees and shrubs and discourage the encroachment of undesirable exotic species (Sher and others 2000), although the seeds of exotic species can remain viable for some years. Rivers whose vegetation types and associated flow needs change downstream present a more complicated picture (Nilsson and others 1991, Tabacchi and others 1996, Cordes and others 1997). In these cases, since the flow needs for regeneration of individual riparian species are only satisfied in certain years under unmanaged conditions, flow allocations for regeneration are also only needed in selected years. A programe of different flow prescriptions can therefore be identified in different years to satisfy overall demands of the particular riparian ecosystem. There is evidence from dryland rivers in Australia that clusters of flood years are commonly associated with the El Niño Southern Oscillation (ENSO) and have cumulative effects on floodplain habitats associated with the persistence of discharge patterns (Puckridge and others 2000). Such phenomena should be taken into account while designing dam operating rules and suggest that decadal and inter- and intraannual flow patterns are important.
Flow Allocations for a Broad Range of Environmental Goals
Water resources are logically managed within the context of river basins. This principle has recently been adopted in the European Water Framework Directive of 22 December (EU, 2000), which will require all member states to manage the “environmental quality” of water within river basins. It is less well established that ecosystem needs for water should have at least the same status as other water uses such as agriculture, industry, and domestic drinking water supply (Naiman and others 2002). However, recognition of the economic as well as less easily definable noneconomic functions of instream and dependent terrestrial ecosystems has brought ecosystem needs onto the agenda in many countries, especially those with the most severe water shortages. Notable in this context are a series of flow allocation methodologies now used in South Africa and Australia. These emerged from a joint initiative to develop an holistic approach, which was a conceptual approach to water allocation in rivers in both countries developed during the late 1980s and early 1990s (Arthington and others 1992). The holistic approach stresses the need to include all components of the riverine ecosystem and is underpinned by the “natural flows paradigm” (Poff and others 1997) and basic principles for the restoration of river systems (Ward and others 2001). The objective is to determine the water needs of instream and dependent terrestrial ecosystems and then to negotiate with water resource planners as to what can be achieved in practical terms given costs, design constraints, and competing water demands (Arthington and others 1992). The water allocation methodologies developed from the holistic approach are described in more detail below. One of their main characteristics is adaptability to the circumstances found in a particular river (Arthington and Pusey 1993).
In South Africa, the history of interest in managing river flows for environmental purposes is described by King and Louw (1998). This subject was first addressed in 1987, when two workshops were held on the subject and coincided with a policy shift within the Department of Water Affairs (DWAF) from one of provision of water in response to demand, to one of holistic management of the nation’s water resources. There was recognition that “the riverine environment is not a user of water in competition with other water users but is the base of the resource itself, and needs to be actively cared for if development is to be sustainable” (King and Louw 1998 p 110). There was also recognition that in a country of about 45 million people, 12 million did not have access to adequate supplies of drinking water. Scientific understanding of river functions was also increasing through the 1990s, and the applicability of the instream flow incremental methodology was assessed in two of the largest rivers flowing into Kruger National Park (King and Tharme 1994). This assessment showed that for many South African rivers, lack of data and time made application of this methodology too difficult.
This experience and the policy shifts occurring in the South African government showed the need for a rapid and practical methodology for assessing instream flow requirements in association with proposed plans for water resource developments, such as new dams. From this need grew the building block methodology [BBM; unrelated to the riparian restoration scheme of Petersen and others (1992) also called the building block methodology]. The BBM relies to a large extent on “best available knowledge and expert opinion” (King and Louw 1998).
The biota associated with the river can cope with low flow conditions that naturally occur in it often and may be reliant on higher flow conditions that occur in it at certain times.
Identification of what are felt to be the most important components of the flow regime and their incorporation into the modified regime will facilitate maintenance of the natural biota.
Certain kinds of flow influence channel geomorphology more than others. Incorporation of these flows into the flow regime will aid maintenance of the natural channel structure.
The BBM is restricted to stating the case for riverine (including riparian) uses and does not directly take into account offstream uses. However, it is recognized that water resource development will involve use of water and that constructing an instream flow requirement (IFR) that is the same as the natural regime is likely to be inappropriate.
The BBM produces two recommended flow regimes, one for river maintenance and one for drought conditions. The results of the BBM workshop process, the IFR, is incorporated into project planning at the prefeasibility and feasibility stages. A development of the BBM called DRIFT (downstream flow response to imposed flow transformation) has, in addition, the flexibility to negotiate flow regimes and address social impacts (King and others 2003).
In parallel, the holistic approach led to the development of a series of related Australian methodologies which are reviewed by Arthington (1998), Arthington and others (1998), Arthington and Zalucki (1998), and Arthington and Pusey (2003) . Arthington and her coworkers were instrumental in developing these approaches and divide the methodologies in use into bottom-up and top-down approaches. Most of the bottom-up approaches are similar in basic concept to the South African BBM, where the environmental flow regime is built up by flows requested for specific purposes from a starting point of zero flows. They include the holistic approach, expert panel assessment method (EPAM), and habitat analysis method. Some methodologies, like EPAM, have relied too heavily on general statements by panels of experts that have served to set broad limits on water abstraction from a river system but lack a rigorous scientific basis for defining important flow patterns. They all depend on the natural science knowledge of participants in the process and the availability of reliable data about the river system. Bottom-up approaches start with consideration of the flows required to maintain geomorphological processes and channel structure and go on to consider the habitat and life-cycle requirements of aquatic plants and invertebrates, fish, and floodplain vegetation. The needs of water-dependent wildlife are increasingly being included in the equation, as are the flows that influence water quality. Flows are also being recommended to maintain various ecological processes such as nutrient and energy exchanges between the river channel and its floodplain, and the ecological linkages between rivers, estuaries, and coastal waters (Arthington and others 1998, Bunn and Lonergan 1998, Loneragan and Bunn 1999). A recent field study by Pettit and others (2001) shows strong linkages between various attributes of the natural flow regimes and life history characteristics of riparian vegetation in two contrasting river types in Western Australia. Data of this kind allow determination of the ecologically significant river flows for riparian ecosystems and therefore strengthen the applicability of these bottom-up methodologies.
An alternative to defining flows from the bottom-up is a top-down approach in which the environmental flow regime is developed by determining the maximum acceptable departure from natural flow conditions (Brizga 1998). The benchmarking methodology is the only top-down method currently used in Australia while the flow restoration methodology combines a bottom-up and top-down approach (Arthington and Pusey 2003). The benchmarking methodology was developed for rapid assessment of future water resource development options at a whole catchment scale and determines how a river would respond as the flow regime progressively departs from its natural state. It takes interannual variability of flows into consideration as one of a range of flow statistics that describe the flow needed to maintain geomorphological processes, and instream and floodplain species and communities. It is essentially a risk assessment approach rather than a definitive prescription of a modified flow regime (Arthington 1998).
The flow restoration methodology (Arthington and others 2000) compares hydrological characteristics of a river system in its regulated and unregulated states to assess the effect on river-dependent species of reinstating or not reinstating various flow characteristics. It is able to define more precisely than other methodologies how to alter the flow regime of a regulated river so that it most closely achieves preregulation ecological characteristics (Arthington and Pusey 2003).
The choice of methodology used in different river systems in Australia is often as much a political as natural science based decision, and there seems to be little transfer of lessons learned between river basins with resultant proliferation of closely related methodologies (Arthington and Pusey 2003 ). Development of flow allocation methodologies in Australia has given rise to a recommended best-practice framework, which is described by Arthington and others (1998) (Figure 3). It recommends an iterative process involving assessment of catchment and river characteristics, including the impacts of river regulation, preliminary qualitative flow recommendations by a team of experts, further detailed studies, quantitative flow recommendations, modeling and evaluation of selected scenarios, evaluation of socioeconomic implications of selected scenarios and final recommendations, monitoring with feedback, and ongoing research.
Implementation of Flow Allocation Methodologies: Case Studies
Instream flow prescriptions have been implemented for various streams in North America through the 1990s. Sustained interventions in dam operation practices have been made to restore floodplain processes along the Truckee River in Nevada, USA (Rood and others 2003 ), the Oldman and St. Mary rivers in Alberta, Canada (Rood and others 1998b, Rood and Mahoney, 2000) and the Bill Williams River in Arizona, USA (Shafroth and others 1998). Rood and Mahoney (2000) show that major floods in 1964 and 1975 in the postdam period in the St. Mary River in Alberta did not trigger regeneration of Populus deltoides, while a similar sized flood in 1995, which was deliberately managed to try to promote regeneration did so successfully (Figure 4). In the Truckee River there has now been a 20-year period of revised flows for ecological targets. High levels of cottonwood (Populus fremontii) recruitment took place in 1987 following dam releases designed to restore populations of the endemic fish, the cui-ui sucker (Chasimistes cujus) (Rood and others 2003 ). From 1995 to 1999 dam releases were successfully implemented specifically to promote cottonwood recruitment. This case study shows that recovery of riparian ecosystems can be rapid and that releases for specific ecological targets need only be implemented in years when water supply and demand allows such releases (Rood and others 2003, Cordes and others 1997).
Single flows identified for a particular target (whether instream or in the riparian zone) can impose a relatively coarse and static managed flow regime. Whitaker and Shelby (2000) suggest that there is a need for alternative flow scenarios that protect the diversity of natural flows as well as volumes at specified times. They compare three managed flow regimes in the Dolores River, Colorado, USA, including fixed time requests, trigger requests, and percentage-based requests (Figure 5). A fixed-time request flow regime has also been devised for the Kafue River below the Itezhitezhi Dam in Zambia. The hydrograph is severely stepped and suffers from not always providing an adequate water regime for either the ecosystems or traditional users of the Kafue Flats (Figure 6) (Scudder and Acreman 1996).
A particular case study in Australia, now much publicized, is that of the River Murray, within the Murray-Darling Basin, which lies mostly within the State of Victoria. A draft proposal on how to divide up Victoria’s River Murray water resources was arrived at in 1997 by the Murray Water Entitlement Committee, made up of many stakeholder groups including water users, irrigation industry groups, water authorities, environmental groups, catchment management authorities and the Department of Natural Resources and Environment (Murray Water Entitlement Committee 1997). One of the key principles has been to introduce capping of water removed from the river in the light of a growing abstraction rate, which had reached 80% of natural flows at the Murray mouth in South Australia. Various mechanisms have been put in place to effect this cap, including reducing the amount of water which is sold, increasing the levy on water sales, introducing particular measures for times of drought in such a way as to equitably “share the pain,” restricting water rights in some years, reducing urban bulk water entitlements back to levels based on maximum usage up to 1993–1994, and significantly improving environmental provisions, particularly to ensure adequate watering of important forests and wetlands along the river (Murray Water Entitlement Committee 1997). This last clause is particularly interesting in relation to this review.
Crucial to provision of seasonally useful water for the environment is exactly how the cap on water use is managed and spread between users, seasons, and years. Having a cap prevents more water being taken from the river and in particular helps retain what is left of the spring floods. However, spring floods have already been reduced to levels that require boosting for effective watering of important wetlands such as the Barmah-Millewa floodplain forests. The water requirements of these nationally important floodplain forests and other wetland areas naturally inundated by medium-sized floods were determined by examining historical records and the needs of key animal and plant species. A model was then developed to evaluate a number of scenarios using different hydrological inputs (Bren 1991). It was considered that the target floodplain areas should be watered for 4 months and that the longest period without floods should be 5 years. At the time of the proposal, the forests only received 100 Gl/yr. By modeling the impact of this low flow, it was demonstrated to be inadequate to water the forests, even when it was stored and then discharged as a single release, in the spring, from the upstream Hulme Dam. It was then proposed to accumulate the annual 100-Gl allocation over several years when there was available space in the upstream reservoirs.
Attempts were made to model the release of water in more effective ways, e.g., high releases of 550 Gl/month over two weeks and lower flows the rest of the time, but then gaps of up to 13 years between floods occurred. With additional “low-security” water being made available from other sources, some successful forest watering scenarios were developed by the modeling process (Table 3). These proposals are now being implemented while decisions on who bears costs and risks are still being debated (Murray Water Entitlement Committee 1997, Tarlock 2000, Arthington and Pusey 2003). In addition, a strong plea has been made to develop and implement methods for monitoring the success of these environmental water allocations (Reid and Brooks 2000). Given the longevity of floodplain woodland communities, indicators of forest change are most likely to be found in forest understorey layers and in regeneration patterns of woody seedlings and young saplings (e.g., Vasicek 1991).
Modeling results for the Barmah floodplain forest in the Murray-Darling Basin, Australiaa
Floods (N/100 years)
Periods 5 years without floods (yr)
3 longest periods without floods (years)
5, 3, 3
15, 14, 12
Full take-up 100 gL
13, 9, 6
Extra 50 gL low security
9, 6, 6
This case study from the Murray River highlights the importance of maintaining variation in interannual flows and the provision of periodic high flows when restoring floodplain forest ecosystems. Since the recruitment needs of these ecosystems need only be satisfied in some years, there can be considerable flexibility in the management of flows between years and in the setting of stakeholder priorities from year to year. In this regard, stakeholder representation is vital and highlighted by Scudder (1991), who reviews a number of protocols for releases downstream of dams in Africa. A typical consensual flood operating rule is given by Bruwer and others (1996) in a discussion of the community-based management of the Phongolo floodplain in South Africa (Table 4). Stakeholder involvement is also integral to the adaptive flow management advocated by Irwin and Freeman (2002).
Flood operating rules for Phongolopoort Dam on the Phongolo River in South Africaa
Activities at each stage in the flood operating rule
Combined water committees for the river send a request to plan a flood release to the Department of Water Affairs.
Combined water committees arrange site meeting following collection of information relevant to interested and affected parties. Issues debated at this meeting include, the season, dam capacity, future possible release requirements. It results in clear identification of compatibilities, conflicts and constraints regarding a proposed release.
Combined water committees advise tribal authorities to call a community meeting in individual areas of jurisdiction for community-level debate.
Feedback and further debate is carried out iteratively until a consensus is reached among combined water committees. This consensus involves negotiation, conflict resolution, site inspections, option generation and crucially, anticipation of issues important to other interested and affected parties. At all times the combined water committees must have a community-based mandate for their decisions.
At a final stage, a liaison committee meeting is held at which all interested and affected parties are represented. At this meeting, flood size, duration, timing and all other issues related to flooding are decided and agreed upon.
Notice of the flood is given via all parties concerned plus radio and media.
The flood release is made from the impoundment with intensive monitoring executed by the Department of Water Affairs and other necessary institutions. On average, flood levels rise by 4 m above normal river flows and take 10 days to pass through the floodplain.
Postflood feedback meetings are held with the combined water committees
Currently, most water management institutions plan abstraction levels on a monthly basis and try to ensure minimum flow targets through the year. This time frame for water quantity management reflects the emphasis on providing guaranteed flows for fish populations and the fact that political and planning time frames are often only a few years long. As a result, a disparity has developed between the time- frames used by institutions planning and managing water resources and the more recently identified time-frames that are important to the functioning of floodplain forest ecosystems. This is problematic for their effective restoration.
Application of Water Allocation Methods to Restoration of Floodplain Forests in Europe: What Potential?
Floodplain forests are a rare and endangered habitat in Europe. Data from 45 countries suggest that 88% of alluvial forests have disappeared from their potential range (UNEP-World 2000 Conservation Monitoring Centre 2000). The main causes of loss are conversion to agricultural use and changes to flow patterns downstream of river control structures. Remaining floodplain forests are particularly affected by the latter because, as we have shown above, the life histories of trees found in floodplain forests are dependent on particular intra- and interannual river flow patterns for their regeneration and growth.
It is recognized by water planners in Europe that to make sound decisions about the allocation of runoff to potentially competing uses, a set of management objectives for river systems has to be established (Marsh and others 2000). Acreman and Adams (1998) consider that a framework for development and implementation of objectives needs to be established at two levels. At the first level, the overall management objectives must be set. At the second, the river flows required to meet the overall objectives need to be defined. In general, management objectives in the European context have involved water allocation to satisfy the needs of agriculture, industry, drinking water supply, and some instream biota, usually target fish species. Management objectives have not involved satisfying the water needs of floodplain ecosystems. An interest has developed in the restoration of floodplain areas, although the drivers behind this interest are complex and not always related to each other. There is an imperative to carry out ecosystem conservation and restoration, driven by the EU Habitats Directive (1992 and subsequent amendments) and the EU Birds Directive (1979 and subsequent amendments). In addition, flood defense strategies are receiving a major overhaul following destructive flooding in 1998, 2000, 2001, and 2002 in many European countries. Attitudes to flood defense are changing from those driven by hard engineering solutions to those that consider setting back flood defenses and reusing floodplains as areas of water storage in the catchment, a process now renamed as flood management.
Conservation-driven floodplain restoration has spawned a series of site-specific restoration initiatives, some of which involve plans to plant floodplain forests and relatively few of which can include the possibility of improving hydrological connections to the floodplain in order to stimulate natural regeneration. A number of these projects are reviewed by Zockler (2000). New flood defense approaches liberate floodplain land and might encourage growth of trees in the flood retention zones created. In some cases blocks of floodplain trees might be used as a “friction factor” to slow down flows. Studies are being carried out on the effects of tree planting on the hydraulic properties of floodplains e.g., by the RIZA Institute for Inland Water management and Waste Water Treatment in the Netherlands. Neither the conservation-led approach to floodplain restoration nor the flood defense-led approach to reusing floodplain land is integrated with decisions on water allocation for ecological purposes. It is still not widely understood that to achieve functional floodplain forests, it is not sufficient to plant trees on floodplains. Specific flow regimes that favor channel movement and natural regeneration of vegetation also have to be provided.
In Europe, high population densities and intensive development on floodplain land make it difficult to consider restoration outside the context of site-specific initiatives (Hughes and others 2001). Bayley (1995) and Gregory (1998) discuss the importance of integrating biophysical knowledge with socioeconomic knowledge for choosing candidate sites for floodplain restoration in the United States. They emphasize the importance of understanding the economic mechanisms that influence the values attributed to floodplain land. For example, subsidized flood insurance encourages floodplain development, which in turn raises the call for armoring of river banks to prevent meander formation; on the other hand, tax relief, production subsidies or conservation easements (covenants) can encourage participation and long-term allocation of land to restoration goals. In Europe, opportunities for restoration of floodplain forests at specific sites are highly variable and determined by combinations of biophysical, sociopolitical, and institutional factors. It is difficult to be strategic and easier to be opportunistic in site selection. Long histories of land tenure and highly complex stakeholder relationships make even site-based restoration initiatives difficult to implement. Nevertheless, there is great merit in considering the possibilities of integrating these site-based schemes with water allocation policies in river basins in order to improve flow arrival patterns at restoration sites. The possibilities for this integration are very variable across Europe, as shown by examples described below.
In the United Kingdom, Marsh and others (2000) suggest that overall management objectives for key habitats may be set by conservation targets of the UK Biodiversity Action Plan (the UK’s response to the EU 1992 Habitats Directive and EU 1979 Birds Directive) or statutory conservation designations [Special Areas of Conservation SACs), Special Protection Areas (SPAs), or Sites of Special Scientific Interest SSSIs)]. Within the UK Biodiversity Action Plan, floodplain forests are included in the Wet Woodland Habitat Action Plan (Latham and Kirby 1998) but have largely disappeared from the landscape. The few residual alluvial forests have been proposed as SACs and the creation of four large (> 50 ha) floodplain forests with appropriate hydrological inputs is promoted within the Wet Woodland Habitat Action Plan. Setting restoration objectives for individual sites may involve a huge number of participating agencies and stakeholders, which can lead to inertia. Just among government agencies, there are often strongly held, opposing views. e.g., the UK Environment Agency, which is also the operational agency for flood defense has generally been opposed to tree-planting on floodplains while the Forestry Commission has promoted growing trees on floodplains for a number of reasons including the production of broad-leaved timber (Kerr and Nisbet 1995).
Flow allocation is also within the UK Environment Agency’s remit. Allocations are based on pollution dilution requirements and on the needs of a few instream biota (usually fish), and flow objectives are determined on a monthly basis. The new UK Catchment Abstraction Management Strategies (CAMS) concentrate on instream targets with the protection of Q95 flows (Q95 is, on average, the flow exceeded 95% of the time) to ensure aquatic environmental quality. The possibility of allocating flows for floodplain ecosystems is not yet included in the CAMS, although consideration of the water table needs of stream-dependent wetlands can take place through water level management plans determined for individual sites. It seems likely that floodplain trees will reappear in the United Kingdom with changed views on flood management and with increasing availability of floodplain land. However, protocols for developing new catchment flood management plans currently underway by the UK Department for Environment, Food and Rural Affairs (DEFRA) are still being determined. It is already a considerable shift in thinking that these plans will consider flood management in an integrated way through a catchment rather than piecemeal at individual sites.
In the Netherlands, where floodplain forests have also largely disappeared (Wolf and others 2001), sites for their restoration have already become available as a result of changes in approaches to flood defense. Here, flood defense is already planned at a catchment scale, and in many locations decisions have been made to set back defenses rather than build them up as a response to rising sea levels and increasing flood frequencies. In this way, the rivers are allowed more room for channel movement and floodplain development with accompanying opportunities for restoration of floodplain forests within a mosaic of different floodplain habitats. A good example is the Millinger Waard Nature Reserve on the Waal channel of the Rhine River, where natural regeneration of black poplar (Populus nigra) is already taking place. Ideas on river restoration in the Netherlands have shifted from a sectoral to an integral approach where multifunctional projects involving flood defense and floodplain restoration have become the norm (Cals and others 1998).
Whereas many western European countries have little or no remaining floodplain forests (often in the order of 1% of their total forest resource), eastern and central European countries have retained many impressive tracts, notably on the River Danube system. Descriptions of the floodplain forest resource in many Eastern European countries are given in Klimo and Hager (2001). For example, 290,000 ha remain in Croatia (Vukelic and Raus 2001), 82,000 ha in Poland (Sienkiewicz and others 2001) and 31,600 in the Czech Republic (Klimo 2001). However, the status of these forests is highy variable and in many cases the natural forest has been replaced with plantations of poplar hybrids or black locust (Robinia pseudacacia). In Hungary, for example, most of the significant remaining floodplain forests are located around the Danube and Tisza rivers, but 40% of their area has been converted to plantations (Haraszthy 2001). Floodplain land also continues to be converted to agriculture. For example, in Austria, the Danube floodplain forests decreased from 33,000ha in 1930 to 8000 ha in 2001 (Hager and Schume 2001).
Conservation of variable annual flow patterns is crucial in maintaining the ecological integrity of the remaining forests. On the Danube River in northwest Hungary, floodplain forests of the Szigetkoz area have demonstrated reduced growth rates and deteriorating health since 1992 when annual floods ceased following construction of the Gabcikova Barrage system upstream in Slovakia (Somogyi and others 1999). Following a ruling by the European Court in the Hague in 1997, water should be released into the Danube system below the dam (Kern and Zinke 2000) but bilateral negotiations between the two countries on the timing and volume of releases have been, so far, without significant results. Decisions will require information on discharge volumes and equivalent stage, timing and duration of floods needed by the Danube forests. This case highlights the value of understanding and quantifying the links between hydrological inputs, geomorphological processes and forest regeneration and growth. It also demonstrates the inherent difficulties in reaching agreement on water allocation where rivers cross international boundaries, a situation that is very common in Europe.
It is clear from these descriptions that integration of flow allocation and river corridor management objectives are very important (Schaefer and Brown 1992). However, in many countries, water budgeting, including determination of abstraction rates in a catchment, flood defense, and the setting of conservation priorities are managed by separate government agencies and are hardly coordinated (De Jong and others 1994, Lorenz and others 2001). The needs of floodplain forests cannot generally be fully met by current institutional arrangements. While the water allocation methodologies described earlier in this paper could not easily be transferred to European rivers, the principles on which they are based could be (Table 5). An assessment of their applicability at reach and catchment scales by setting up demonstration projects is urgently needed.
Potential for flow allocation approaches to restore European floodplain forests
Flow allocation approach
Potential for restoration of floodplain forests in Europe
These could be used in some parts of northern Europe, e.g., northern Sweden, below headwater dams in mountain areas throughout Europe, and in some small streams elsewhere if weir structures allow flow control. In these locations they are unlikely to have significant effects on regeneration potential of floodplain forests.
Recruitment box model (Mahoney and Rood 1998)
This model was developed for use below dams and could only be used in its orginal context in limited parts of northern Europe. However, the principles of flood timing and tapered recession flows for floodplain forest regeneration could be used widely in flood storage areas on floodplains if engineering structures allow control of water releases. The principles are applicable to large and small floodplain rivers.
Multiple flow methodology (Hill and others 1991)
Identification of four main flow types for: instream/fisheries, channel maintenance, riparian ecosystems, and valley maintenance would be possible in most European rivers. However, in order to identify flow regimes relevant to floodplain forests, lags between present-day floodplain forest types and the processes that initiated their regeneration should be identified. This is important because the range of contemporary, available flows may not include regeneration flows.
Bottom-up approaches, top-down approaches, and combined approaches developed in South Africa and Australia (Swales and Harris 1995, King and Louw 1998, Arthington 1998, Brizga 2000, Arthington and others 2000, King and others 2003).
The expert-workshop approach could be adapted to specified reaches or whole river systems in Europe to produce flow-related scenarios for river managers to consider. The transboundary nature of many European rivers will complicate the process. The top-down approach could produce estimates of the percentage of mean-annual flows needed for different ecosystem processes. These estimates can lead to specification of thresholds of water use below which ecosystem health deteriorates and to caps on water use. The use of caps on water use might most benefit Mediterranean countries where water deficits are the most severe but are relevant to many other rivers. All these approaches are only useful for floodplain forests if occasional well-timed flood flows can be included in the flow scenarios.
Opportunities for more integrated approaches to the restoration of floodplain forests should arise in Europe through the implementation of the European Water Framework Directive (EU 2000), with its specific attention to “dependent terrestrial ecosystems” and its general “whole catchment” framework for all decision-making. However, there are also foreseeable constraints associated with its long implementation time, the many possible loopholes, and the fact that it has no accompanying financial instruments. First steps towards implementation of the European Water Framework Directive (EU 2000) have highlighted many problems associated with the spatial misfit between the physical and biological resources of a river basin on the one hand and the remits of the institutions responsible for their management on the other (Moss 2001).
In some countries such as Australia and South Africa, there is recognition of the need to revitalize floodplain forest ecosystems, and their flow needs are included within new flow allocation methodologies for multiple users. We argue that in order to conserve or restore these ecosystems, planning of river flows must take place over decadal time frames and that simply meeting minimum monthly flow standards in each water year is inadequate. Most of the holistic approaches to flow allocation descibed here recognize this aspect. While in some river basins, varied interannual flow needs can be provided through cooperation between river managers and natural scientists, especially where flows can be manipulated downstream of dams, in many places the situation is more complex. In most European countries high population densities and intensive development on floodplain land preclude such strategies. Here, site-specific restoration initiatives are more feasible and opportunities for floodplain forest restoration are arising through new flood defense strategies. In some locations such as at the Regelsbrunner Au restoration project on the River Danube in Austria, the flood pulse still exists and reactivation of floodplain forest regeneration has been achieved by removing artificial embankments. In other places, site-specific restoration needs to be integrated with water allocation strategies in the whole river basin, in order to promote opportunities for natural regeneration of floodplain trees.
This research has been funded by EU research contract: EVK1-CT-1999-00031FLOBAR2 [Floodplain Biodiversity and Restoration 2 (FLOBAR2): Integrated Natural Science and socio-economic approaches to flow management]. We would like to thank Bill Adams, Angela Arthington, Jacky Girel, Chris Joyce, Jackie King, Tim Moss, and Keith Richards for providing references, comments, and advice on the manuscript. Our thinking on the subject of environmental flows has been much helped by discussions with the FLOBAR2 advisory committee, including Peter Allen-Williams (DEFRA), Russell Cryer (RSPB), Mark Diamond (Environment Agency), Nigel Holmes, Gary Kerr (Forestry Commission), Keith Kirby (English Nature), George Peterken, Christoph Zockler (UNEP-WCMC), and people who attended the FLOBAR2 workshop in Cambridge in May 2002. Owen Tucker kindly redrew the figures.