Previous Page Table of Contents Next Page

Ecoregion Conservation for Freshwater Systems, With a Focus on Large Rivers

Abell R. Thieme M. Lehner B.

Conservation Science Program, World Wildlife Fund, 1250 24th St. NW Washington, DC USA 19041 E-mail:


Conservation planning with the express purpose of protecting the aquatic biodiversity of large river systems is a relatively new endeavour. A conservation blueprint should be designed around the protection of sufficient habitat for the most wide-ranging and sensitive species and of the physical processes that create and maintain those habitats. WWF and several other organizations have adopted an approach to large-scale planning, referred to as ecoregion conservation (ERC). An ecoregion is a large unit of land or water containing a geographically distinct assemblage of species, natural communities and environmental conditions. The boundaries of an ecoregion encompass an area within which important ecological and evolutionary processes most strongly interact. Large river basins often fit this definition. Conservation strategies that are formulated at the ecoregion scale have the potential to address the fundamental goals of biodiversity conservation: 1) representation of all distinct natural communities within conservation landscapes and protected-area networks; 2) maintenance of ecological and evolutionary processes that create and sustain biodiversity; 3) maintenance of viable populations of species; and 4) conservation of blocks of natural habitat that are large enough to be resilient to large-scale stochastic and deterministic disturbances as well as to long-term changes. Through ERC we generate a vision for what an ecoregion should look like in 50 years if its biodiversity targets are to be maintained. These targets fall into five main categories: distinct communities, habitats and species assemblages; large expanses of intact habitats and intact native biotas; keystone habitats, species and phenomena; large-scale ecological processes; and species of special concern. The nature of freshwater systems requires that we go beyond identifying discrete aquatic areas on a map. A vision for a freshwater system must take into account the importance of lateral, longitudinal and even vertical connectivity; examine threats originating upland, upstream and even downstream; incorporate strategies for protecting hydrologic processes operating over large scales; and consider the implementation of land-based conservation strategies in the larger catchment. WWF and partners, has undertaken ERC in a number of freshwater systems, including the Amazon, Congo, Niger and lower Mekong Rivers. The many lessons we have derived from our work include the critical need to integrate the expertise of hydrologists with that of biologists, the importance of starting with catchments rather than small "hotspots" and the value of integrating freshwater strategies with parallel efforts in adjacent terrestrial and marine systems. Next steps for our work involve improving the classification of aquatic habitats so that all types can be represented in a conservation blueprint; investigating the habitat requirements and metapopulation structures of select wide-ranging focal species; forecasting future threats like climate change and incorporating that information into our strategies; and conducting research to begin to identify thresholds in land use that translate into threats to aquatic biodiversity.


Conservation planning with the express purpose of protecting the aquatic biodiversity of large river systems is a relatively new endeavour. From headwaters to mouth, these systems typically are characterized by high habitat heterogeneity with corresponding high species richness. Many also support large numbers of endemic species and may be distinguished by ecological phenomena (e.g. large-scale migrations of fish) and evolutionary phenomena (e.g. radiations of multiple species from a common ancestor).

Unfortunately, these systems’ large size also hampers the development and implementation of effective conservation strategies. There is consensus within the conservation community that strategies must be scale-appropriate, tailored to the spatial and temporal scales over which ecological processes operate (Fausch et al. 2002). Freshwater managers have long recognized the need to take a whole-basin approach to planning, as evidenced by the large number of river and lake basin planning organizations and authorities around the world. However, protecting or restoring hydrological and ecological processes over millions of square kilometres is a daunting task, especially where river systems cross international boundaries. Additionally, freshwater systems are often highly degraded, particularly in their lower reaches, having been modified extensively for irrigation, waste disposal, hydropower, flood control, navigation and other uses. Restoration of these downstream reaches can require enormous expenditures for uncertain conservation returns, but these areas are essential components of a representative suite of conservation priorities (Frissell 1997).

Understanding the trade-offs associated with different conservation strategies is critical for any large-scale planning effort. This is certainly the case for large river or lake systems, where stakeholder dynamics are nearly as complex as ecological dynamics. One well-known, basin-wide approach that focuses on trade-offs is Integrated River Basin Management (IRBM). In past IRBM projects, the maintenance of a reliable and safe water supply for human use has generally taken precedence over the protection of biodiversity, or the goals have been vaguely defined (Hooper and Margerum 2000).

Ecoregion conservation (ERC), a large-scale planning approach adopted by World Wildlife Fund (WWF, known also as the World Wide Fund for Nature) and several other organizations, shares IRBM’s whole-system perspective but puts biodiversity solidly first. An ecoregion is a large unit of land or water containing a geographically distinct assemblage of species, natural communities and environmental conditions (Dinerstein et al. 1995). The boundaries of an ecoregion encompass an area within which important ecological and evolutionary processes most strongly interact. For aquatic biodiversity, large river basins often fit this definition, though in some cases biogeographic barriers separate a basin into two or more ecoregions, or neighboring basins are biotically similar enough to be combined together (Abell et al. 2000; Thieme et al. unpublished data).

The first step in ERC is to develop a "biodiversity vision." The vision aims to outline those areas and processes that are essential for maintaining an ecoregion’s biodiversity features for at least the next 50 to 100 years. We then build a conservation strategy around this vision, taking into account the range of trade-offs inherent in different options. Beginning with a vision that is firmly grounded in biodiversity targets is one of the characteristics that distinguish ERC from most past IRBM endeavours.

We build a vision around a subset of biodiversity features - targets - that distinguish the ecoregion and/or serve as umbrellas for other features. These targets fall into five main categories:

- Distinct communities, habitats and species assemblages (e.g. "hotspots" of richness or ende-mism);

- Large expanses of intact habitats and intact native biotas (e.g. un-impounded rivers, assemblages without exotics);

- Keystone habitats, species and processes (i.e. features that exert a powerful influence on the composition, structure and function of ecosystems and consequently on biodiversity, such as seasonal flooding);

- Large-scale ecological phenomena (e.g. long-distance migrations of fish);

- Species of special concern (e.g. sensitive species that can serve as focal species for planning).

ERC is applicable to terrestrial, freshwater and marine systems, but the approaches for each realm are somewhat different. A terrestrially focused vision normally identifies a suite of biologically distinct areas for protection (Dinerstein et al. 2000), but the nature of freshwater systems requires that we go beyond identifying discrete aquatic areas on a map. A vision for a freshwater system must take into account the importance of lateral, longitudinal and even vertical connectivity; examine threats originating upland, upstream and also downstream; incorporate strategies for protecting hydrologic processes operating over large scales; and consider the implementation of land-based conservation strategies in the larger drainage basin (Abell et al. 2002).

Conservation strategies that are formulated at the ecoregion scale have the potential to address the fundamental goals of biodiversity conservation (modified from Noss 1992):

WWF and its partners have undertaken ERC in a number of freshwater systems around the world. Here we review a general methodology that we have developed for applying ERC to freshwater systems and we discuss variations of this methodology as applied to the Congo, Lower Mekong, Amazon and Niger River systems.


No two ERC projects have used identical methodologies, due to differences in each ecoregion’s ecology and available biodiversity data. Nonetheless, all vision-building efforts share some basic components (Table 1). A more detailed flowchart of steps is given in Abell et al. (2002).

Table 1: Basic steps for developing a freshwater biodiversity vision through ERC

Representation groundwork

  • Develop representation decision rules

  • Refine ecoregion boundaries and define biogeographic sub-ecoregions

  • Identify habitat types for representation analysis or map habitats across ecoregion

Biological importance

  • Generate overall map of important areas

  • Identify and delineate areas (e.g. river reaches, wetlands) of biological importance

  • Assign levels of importance to areas based on their relative contribution to maintaining the ecoregion’s biodiversity targets

  • Identify and delineate areas (including terrestrial) that are important for maintaining abiotic processes (e.g. hydrologically active areas)

Ecological integrity

  • Assign levels of ecological integrity to important areas

  • Map threats to aquatic biodiversity across the region of analysis (including terrestrial and/or marine areas)

  • Evaluate ecological integrity of important areas, based on habitat intactness and population/species viability


  • Prioritize among important areas based on the combination of biological importance and ecological integrity levels

Representation analysis

  • Conduct a representation analysis to ensure that all biogeographic sub-ecoregions and naturally occurring habitat types are sufficiently represented in the suite of priority areas; add to the priority areas to achieve representation, if necessary


  • Evaluate connectivity of the priority areas based on the dispersal and migratory requirements of focal species; add corridors/linkage areas to achieve connectivity, if necessary

Future threats

  • Evaluate future threats to the priority areas and across the ecoregion

Conservation interventions

  • Develop specific recommendations for conservation interventions in the priority areas

  • Develop recommendations for broad conservation interventions (e.g. aimed at all riparian zones across the ecoregion)

Biodiversity vision

  • Evaluate if the recommendations, if implemented, would achieve protection of the ecoregion’s biodiversity over the long term (50 years)

  • Modify the recommendations, if necessary, to achieve a sufficiently ambitious biodiversity vision

The foundation of a vision is a biological assessment:- a record of the distribution of species, communities and habitats in the ecoregion, of ecological processes sustaining this biodiversity and of current and future threats to its maintenance. WWF’s approach to ERC focuses on historic, rather than current, distributions of biodiversity features, with the understanding that many of these features have disappeared or are impaired. We take this approach because the vision is intended to go beyond maintaining the status quo, incorporating restoration as a tool where necessary.

WWF has developed maps of freshwater ecoregions for North America and Africa (Abell et al. 2000; Thieme et al. unpublished data) and is in the process of finalizing a global ecoregion map (Abell et al. unpublished data). Ecoregion delineations are based on biogeography of freshwater taxa, with an emphasis on the distribution of freshwater fishes. These delineations are relatively coarse and a first step for an ERC project team is to revise the boundaries of its ecoregion based on more detailed information and regional expertise. For ecoregions that cover one or more sub-basins of a larger river or lake drainage basin, an ERC team may choose to extend the visioning effort to cover multiple ecoregions, in order to capture the entire area over which hydrological processes occur. Similarly, an ERC effort for a riverine ecoregion with diadromous species (e.g. salmon) might consider the adjoining marine environment in its region of analysis.

Most ecoregions are sufficiently large and biologically complex to justify dividing them further into biogeographic sub-ecoregions. We define sub-ecoregions for the purposes of achieving representation. If sub-ecoregions contain different species assemblages, then we should have examples from each sub-ecoregion in the vision’s ultimate set of priorities.

A second, finer representation filter relates to habitats. In each sub-ecoregion, our goal is to capture examples of all naturally occurring habitats. This approach is based on the assumption that, as with sub-ecoregions, different habitat types typically support different species assemblages. Habitat types can be mapped and classified across the entire ecoregion using satellite imagery or other data layers (van Niewenhuizen and Day 2000; Higgins 1999). Alternately, experts can assign habitat type designations to areas, based on a pre-defined list.

Representation goals - the number of habitat occurrences that should be protected in each sub-ecoregion and the size or other minimum quality that the occurrences must achieve - are specific to each ecoregion. Goals are often set based on the habitat needs of focal species, which serve as umbrellas for other taxa. A variety of criteria can be used to select focal species (Table 2). Most often these species are either wide-ranging, sensitive, or both. We attempt to define the minimum characteristics that an area would need if it were to support a viable population of the focal species.

Table 2: Attributes of focal species

Biological characteristics

1. Climatic sensitivity
2. Sensitive to pollution
3. Low reproductive rate
4. Limited dispersal ability
5. Space-demanding/wide-ranging
6. Migratory, with specialized spawning sites
7. Large-bodied/largest member of feeding guild
8. Dependent upon rare, widely dispersed habitat
9. Narrow temperature or water chemistry requirements
10. Adapted to particular flow regime, water level, flood cycle
11. Specialized dietary, habitat requirements (particularly breeding, nursery sites)
12. Population seasonally/daily concentrated and/or aggregates during part of life cycle

Population status

1. Population small or declining
2. Meta-populations with unique genetic compositions

Human-impact factors

1. New and large markets for consumptive use
2. Habitat threatened by loss, conversion, degradation, or fragmentation
3. Population threatened by direct exploitation, harassment, or ecological interactions

Once this groundwork is complete, the assessment can proceed with the identification of biologically important areas. Biologially important areas are those places, such as certain river reaches, headwater drainages, wetlands, or waterfalls, that are known (or highly suspected) to support one or more identified targets. For example, a given river reach may be distinguished by high species richness or endemism in one or more taxonomic groups; it may provide habitat for an endemic genus or even family; it may support an unusually intact species assemblage; it may represent one of the last remaining large expanses of intact habitat; it may be one of few areas where ecological processes like flooding and associated migrations occur; it may serve as an important refuge or source pool for keystone species; it may contain a rare habitat type; or it may harbor one or more species of special concern.

Typically, experts from a variety of disciplines use published and unpublished data, combined with their own observations, to identify these important areas. A strong expert group is knowledgeable about a range of taxa representing the most important biotic components of the ecoregion (Table 3). Maps of important areas for different taxa (e.g. fish, molluscs, amphibians) are often produced separately and then combined with other maps (e.g. important floodplains) to create a single depiction of areas of biological importance.

In many of the world’s river systems, however, there are scant species and assemblage data to inform this process of identifying important places. Where data are lacking, sub-ecoregions and habitat classifications can serve as proxies. In this situation, we recommend identifying areas of biological importance based on size, intactness, connectivity, or other attributes, making sure that the areas cover all habitat types occurring naturally in all sub-ecoregions.

Table 3: Taxonomic groups to consider for a freshwater ERC assessment (modified from Abell et al. 2002)


1. Aquatic plants
2. Freshwater fish
3. Aquatic mammals
4. Trichoptera (caddisflies)
5. Ephemeroptera (mayflies)
6. Aquatic and semi-aquatic reptiles
7. Amphibians with aquatic life stages
8. Odonata (dragonflies and damselflies)
9. Diptera (mosquitoes, black flies, midges)
10. Neuroptera (hellgrammites, dobsonflies, alderflies)
11. Crustaceans (crabs, lobsters, copepods, ostracods)
12. Aquatic and/or wetland molluscs (snails and mussels)
13. Coleoptera (diving beetles, riffle beetles, whirligig beetles)
14. Hemiptera (backswimmers, diving bugs, water striders, water scorpions)

Areas of high biological importance cannot be protected if hydrologic and other abiotic processes fail to function within their natural ranges of variation. Maintaining these processes requires looking upstream and upland from the biologically important areas. Methodologies for identifying such "abiotic" areas are crude but evolving and almost by necessity must rely on models for large river systems. For a given ecoregion, consultation with hydrologists, biogeochemists and other physical scientists is essential, both to identify the critical processes and interpret any model outputs.

Important areas for biological targets and abiotic processes are often collectively referred to as "candidate priority areas." These areas are usually each assigned a level of importance (highest, high, moderate) based on their relative contribution to maintaining the ecoregion’s biodiversity features. This classification helps to differentiate among the areas during the later prioritisation process.

The other major input to prioritisation is an evaluation of the areas’ ecological integrity. Habitat in the areas can range from virtually intact to critically degraded. Even intact areas, however, may be unable to support viable populations of species over the long term because of insufficiencies of size, connectivity, or other characteristics. An evaluation of ecological integrity, in the context of ERC, incorporates both habitat intactness and the likelihood that the species and communities in that area can endure over the long term, barring additional disturbances. We call this latter attribute "population/species persistence." Levels of ecological integrity (e.g. intact, altered/degraded/highly degraded) are assigned to each important area based on assessments of habitat intactness and population/species persistence.

An evaluation of habitat intactness normally combines an analysis of geospatial data with expert assessment. A wide variety of geospatial information can be used (Table 4), though not all possible measures (Table 5) will be relevant to or available for a given ecoregion. Most, though not all, geospatial information and measures will relate to activities on the terrestrial landscape resulting in altered flow regimes and water quality. The analysis of habitat intactness, then, is typically conducted across each of the important areas’ watersheds.

Table 4: Possible geospatial data layers to inform an evaluation of habitat intactness



1. Indigenous areas

2. Vegetation/land cover

3. Areas of deforestation

4. Aquaculture operations

5. Cattle/livestock densities

6. Human population density

7. Species distributions (e.g. IBAs)

8. Ranges of exotic species or areas of known introductions

1. Roads

2. Canals

3. Railroads

4. Refineries

5. Toxic sites

6. Major ports

7. Industrial sites

8. Protected areas

9. Fishing centres

10. Towns and cities

11. Areas of conflict

12. Drainage projects

13. Water temperature

14. Runoff (by grid cell)

15. Pesticide application

16. Extent of floodplains

17. Power generation plants

18. Discharge (by river segment)

19. Water abstractions/Water use

20. Channelized or dyked streams

21. Erosion potential (by grid cell)

22. Pipelines (present and planned)

23. Sediment transfer (by grid cell)

24. Land uses (current and historic)

25. Mining activity and concessions

26. Logging activity and concessions

27. Irrigated and non-irrigated croplands

28. Fish passage devices (working and failing)

29. Inter-basin water transfers (present and planned)

30. River network (e.g. derived from Digital Elevation Model)

31. Impoundments and reservoirs (present and planned), plus additional barriers to passage

Table 5: Possible geospatial analyses for an assessment of habitat intactness. Additional examples are given in Abell et al. (2002)

1. Percentage of area grazed, by sub-basin

2. Average population density, by sub-basin

3. Length or area of floodplain habitat cut off from river

4. Urban expansion or population growth, by sub-basin

5. Sediment contribution or erosion potential, by sub-basin

6. Number of impoundments per stream length, by sub-basin

7. Road density or number of road-stream crossings, by sub-basin

8. Number of pipeline-stream crossings, or length of pipeline, by sub-basin

9. Average discharge, flow accumulation, or runoff of grid cells, by sub-basin

10. Length or percentage of streams with riparian vegetation cover, by sub-basin

11. Degree of protected area coverage (all areas, or only aquatic habitats), by sub-basin

12. Number or coverage of mining, logging, or other resource extraction operations, by sub-basin

13. Percentage of land-use classes within fixed-width buffer of streams or other water bodies, by sub-basin

14. Percentage of land-use classes, by sub-basin (e.g. 20 percent forest, 40 percent agriculture, 10 percent urban)

15. Percentage of headwaters (defined by elevation, gradient, stream order) with original land cover, by sub-basin

16. Number or length of free-flowing streams, divided by number or length of impounded streams, by sub-basin

17. Length of stream habitat lost as a result of channelization (requires historic and current stream morphology maps)

18. Length of stream flooded by impoundments, or length of stream above impoundments made inaccessible to migrating species, by sub-basin

Evaluating the population/species persistence of a given area is more of a challenge than evaluating its habitat intactness, because we generally have little or no information about species’ life cycles, habitat requirements and metapopulation structures. Layering that information for all or a subset of species historically occurring in the area would allow evaluation of the overall population/species persistence of the area. For obvious reasons, detailed assessments like these are many years off. Where there is literally no information available to evaluate population/species persistence, the assessment of habitat intactness can be used alone to signify ecological integrity.

Biological importance and ecological integrity levels are typically the two main inputs used to prioritise among important areas. A matrix with levels of importance on one axis and levels of ecological integrity on the other provides a simple tool for assigning priority levels. ERC teams have often chosen to take a "triage" approach, assigning lower priority to those areas considered to be highly degraded and probably beyond repair (Table 6). An alternative approach might assign highest priority both to those degraded areas most urgently in need of protection to stem further habitat loss and to those intact areas representing rare opportunities for preservation. Once the matrix is designed, priorities are assigned to important areas to highlight those that should be given attention first.

Table 6: Example of an integration matrix for assigning priority levels to important areas

Ecological Integrity

Biological Importance












Highly degraded




This ERC effort focused only the subset of Amazonian rivers with associated flooded forests or grasslands.

Following the prioritisation, a representation analysis is undertaken to ensure that all sub-ecoregions and habitat types have been captured in the suite of priorities. Elevating the priority level of certain areas or adding new areas to the set fills gaps in representation.

Similarly, new areas may be added to address issues of connectivity. Important areas that are functionally isolated could theoretically be reconnected through the restoration of intervening areas or the removal of a structural barrier.

This process yields a set of areas - a combination of linear and polygonal features - that represent those parts of the ecoregion that are most important from a biodiversity conservation perspective. This collection of places does not necessarily constitute a vision. There remain the issues of impending threats, conservation interventions for the important areas and conservation strategies needed more broadly within the ecoregion and perhaps even outside of it.

Forecasting future threats - their form, direction, location and magnitude - is an inexact science, but developing a conservation plan without an eye to the future is surely shortsighted. Some future threats, such as structural developments and land concessions, are gazetted and thus have a degree of predictability. Others, such as population growth, may be forecasted based on current trends. Climate change models can yield predictions about future changes in water availability and water temperature over large scales, providing an idea of possible impacts and their extent; different models, however, often show inconsistencies in their results, which demand careful interpretation. We recommend combining quantitative information with expert assessment to identify those areas in need of urgent attention if impending threats are to be forestalled. A future threats assessment can also suggest actions to be implemented across the ecoregion.

With information on current and future threats, it is possible to recommend conservation interventions for each important area, for the ecoregion as a whole and for areas of intermediate size. These recommendations will likely relate to the type of protection required (e.g. creating buffer zones along a river, reducing water withdrawals), rather than to the exact approach (e.g. land purchases, regulations) for achieving it.

The biodiversity vision is the sum total of these outputs. An ERC team looks at the final set of recommendations and evaluates if these actions would, in its best judgement, result in protection of the ecoregion’s biodiversity features over the long term. If not, then the vision is probably not ambitious enough and requires modification. Once the vision is complete, the next step is development of an actual implementation strategy, based on a host of biological and socio-economic considerations. A vision, though, should never be so final that new scientific and socio-economic information cannot be incorporated as it becomes available.


The approach described above is based on a theoretical ecoregion. In the real world, especially for large river systems, data and expertise limitations prevent undertaking many of the steps. Here we briefly describe the variations that resulted from applying ERC to the Congo, Lower Mekong, Amazon and Niger River systems.

Our first major attempts to apply ERC to large rivers were for the Congo and Lower Mekong systems. These efforts were undertaken consecutively, in March 2000. In each of these situations, we relied solely on expert assessment workshops to develop visions, with few geospatial data sets available to inform the process. Each of the workshops lasted three days and the expert groups were comprised of about a dozen individuals each. Experts identified biologically important areas on maps, evaluated the areas’ ecological integrity and developed recommendations for those areas and the ecoregion as a whole. Data insufficiencies required the experts to take coarse approaches and their frustration with the lack of information led them to focus their recommendations largely on how to fill data gaps.

Within the Amazon River system, with its vast size and almost complete lack of species data for all but a few locations, we took a different approach based almost entirely on geospatial data. Instead of identifying important areas for biodiversity, a team of scientists first divided the basin into sub-ecoregions, for the purposes of representation. They then divided the sub-ecoregions into major sub-basins, which were the units of analysis for the remainder of the assessment. Landsat TM imagery was used to map habitat types within the floodplains and associated rivers. Biological importance for each sub-basin was based on a combination of calculated habitat diversity (applying a Shannon-Weaver Index), the presence and extent of special habitats (lakes, secondary rivers, islands, cataracts) and the results of prior biodiversity assessments (i.e. PROBIO and ProVárzea/PPG7).

Ecological integrity of the Amazonian sub-basins was also assessed primarily as a function of geospatial indicators. These indicators were percent natural vegetation coverage within floodplain habitats; percent natural vegetation coverage within each sub-basin; size of population centres within each sub-basin; presence and assessed degree of impact of urban, petroleum, mining and farming within each sub-basin; and number and location of dams within each sub-basin. Population/species persistence was not explicitly addressed. The biological importance and ecological integrity values were used to identify a highest priority sub-basin in each sub-ecoregion, in order to achieve representation at a very coarse scale. Additional areas were added to achieve connectivity within the Amazon River main stem.

To identify major threats and opportunities for each priority sub-basin, experts again relied heavily on geospatial data and developed general recommendations for conservation interventions. The next step will be conducting more detailed assessments of the priority sub-basins to identify smaller priority areas. In effect, the ERC process will be repeated for these sub-basins, most of which are as large as entire ecoregions found elsewhere.

The Niger River Basin effort differed from the Congo, Mekong and Amazon efforts in two ways: it used a more balanced combination of expert assessment and geospatial data and it explicitly included hydrological considerations. The experts were provided with hardcopy maps displaying land cover (USGS 2001), protected areas (WWF data), roads (ESRI 1993), Important Bird Areas (BirdLife International data), dams (FAO 2001), agricultural suitability (FAO 2000) and population density (ORNL 2001). Data on runoff generation from a global hydrological model, provided by the University of Kassel, Germany (Döll, Kaspar and Lehner 2003), were used as a starting point for discussions about which sub-basins were most important for maintaining the flow regime.

Three taxonomic expert groups (for fish, birds and other vertebrates) and one for hydrological processes worked to delineate areas of importance across the Niger Basin. The experts selected 19 priority areas for conservation action. They assigned a level of threat to each area and developed a list of conservation actions that should be undertaken in the priority areas and at the level of the basin.

The visions for these four river systems are in various stages of completion (Baltzer, Nguyen Thi Dao and Shore 2001; WWF 2001; WWF 2002; Wetlands International unpublished data). Maps illustrating the results are available upon request from WWF.


Our experiences applying ERC to the four river systems of the Congo, Lower Mekong, Amazon and Niger have taught us a range of lessons. We have incorporated these lessons into each successive effort, but the nature of ERC is that all ecoregions present unique circumstances that require flexibility and innovation.

The most important and perhaps obvious, lesson is that terrestrial approaches to ERC translate imperfectly to freshwater systems. For instance, evaluating the ecological integrity of an aquatic area requires looking beyond it, since upland, upstream and downstream activities affect it and barriers to dispersal and migration can be impassable. Terrestrial areas are affected by activities outside their boundaries, but most often to a lesser extent.

A second lesson relates to the importance of expanding our analysis beyond pure biological information. Ever since we first applied ERC to freshwater systems, we have known that a robust vision must incorporate information on hydrologic processes. But, it was only with the Niger River Basin effort that quantitative hydrologic information was used and to relatively good effect. However, the global hydrologic dataset used for the modelling was coarse and the project would have benefited from the development of finer-scale data. The Niger experience has also taught us the value of explicitly linking the hydrological and biological parts of an assessment and in future efforts we intend to identify those places with the greatest hydrologic impact on biologically important areas.

In general, the inclusion of one or more hydrologists and perhaps biochemists as well, may be the most important way that an ERC team can improve upon a standard visioning process. Aquatic biologists and physical scientists rarely have opportunities to interact and many hydrologists have never been challenged to put their expertise to use in biodiversity conservation. Conserving aquatic biodiversity is as much about maintaining physical processes as it is about focusing on species.

All four ERC efforts have underscored the importance of entering an assessment effort with as much geospatial data on hand as possible. The most successful efforts have allowed experts to use and react to map-based information and to the results of geospatial analyses. In an ideal situation, a geographic information system (GIS) would be employed to divide an ecoregion into component sub-basins; the optimal scale of these sub-basins would vary by ecoregion according to the precision of digital elevation models and the scale of available geospatial data (e.g. there is no need to delineate very small sub-basins if land cover data are at a coarser scale). A variety of calculations related to the distribution of habitat types and threats could be undertaken for each sub-basin prior to an assessment, in order to inform it.

Geospatial data have weaknesses as well as strengths. They provide relatively standardized information about parts of ecoregions with which experts are unfamiliar and they allow for quantitative measures. On the other hand, geospatial data will always only be a proxy for direct measures of habitat intactness. They can also be outdated, inaccurate, or misleading (e.g. when a protected area is in fact no more than a paper park).

Expert assessment can complement and validate the results of a geospatial analysis, because experts make their assessments based on observations of actual aquatic habitats. But, because aquatic biologists may be unfamiliar with land-based threats occurring at a distance from their study sites, it is important to involve additional individuals with a detailed knowledge of activities occurring on the landscape. A usual challenge is finding experts who understand how specific land uses affect aquatic species and habitats (e.g. how plantation forest differs from native forest in terms of hydrologic and nutrient flows).

We have also learned the value of a pre-assessment habitat classification in enabling an automatic representation analysis once priority areas are delineated. A sophisticated classification would go beyond simply identifying different habitat types on a map and would assign classes based on similarities of geomorphology, ecological processes and environmental gradients (Higgins et al. 1999). Of course, the accuracy of these classifications must be ultimately checked on the ground, a task that can be daunting for remote, isolated areas of large river systems.

Habitat classifications and geospatial analyses cannot replace expert assessments entirely, but for ecoregions where biodiversity data are virtually nonexistent and experts are unfamiliar with large areas, the classifications and analyses can provide a preferred alternative to guesswork. Ecoregions that are the most data-poor in terms of biological information can be least suited to expert assessment, because the experts have a frustratingly small amount of information on which to base their decisions. In these cases, data surrogates and/or predictive models can provide a first cut at an assessment and experts can then review the results. On the other end of the spectrum, an ecoregion that is data-rich, with comprehensive information on species distributions, is a good candidate for the use of a systematic algorithm to assist priority-setting (Margules and Pressey 2000). In the middle are those ecoregions where some species and habitat distribution data exist, but data are not available for the entire region of analysis or they are largely unpublished. In this case, an expert workshop may be the most appropriate approach for conducting an assessment.

Our experiences in the Congo, Lower Mekong and Amazon have reinforced the value of integrating freshwater strategies with parallel efforts in adjacent terrestrial and marine systems. These realms are intimately connected, yet we tend to pursue independent planning efforts due to resource limitations, a lack of crossover expertise and a need to highlight the conservation of normally neglected aquatic biodiversity. Integrating separately derived results into a single vision is possible, but a true integration of planning for freshwater, marine and terrestrial biodiversity requires more than a simple overlay of priority areas. Incorporating hydrologic concerns into terrestrial and even marine planning may provide a good platform for a true integration.

Our methodology for conducting biological assessments and developing biodiversity visions in freshwater systems is still quite young and is evolving rapidly. We expect to improve our approach to classifying aquatic habitats by working with partners like The Nature Conservancy, which has developed a methodology for habitat classification in data-rich systems. We hope to catalyze investigations of the habitat requirements and metapopulation structures of select wide-ranging focal species, like the giant Mekong catfish (Pangasianodon gigas (Chevey), with the intention of exporting methodologies for species investigations in other similar systems. We are developing tools for forecasting future threats and incorporating that information into our strategies; climate change in particular is expected to have drastic consequences for aquatic biota in certain regions, yet we have failed to incorporate this information into the design of conservation plans. Finally, we hope to conduct research aimed at identifying thresholds in land use that translate into threats to aquatic biodiversity, so that we can begin to answer the critical question of ERC, "How much is enough?"


Presentation of this paper was made possible by the support of Jamie Pittock of WWF’s Living Waters Programme and Marc Goichot of WWF’s Mekong Programme. We would like to thank those individuals within WWF and its partner organizations who have led ecoregion conservation efforts in the Amazon, Lower Mekong, Congo and Niger River systems.


Abell R.A., Olson D.M., Dinerstein E., Hurley P.T., Diggs J.T., Eichbaum W., Walters S., Wettengel W., Allnutt T., Loucks J.J. & Hedao P. 2000. Freshwater Ecoregions of North America: A Conservation Assessment. Washington, DC, Island Press.

Abell R., Thieme M. Dinerstein E., & Olson D. 2002. A Sourcebook for Conducting Biological Assessments and Developing Biodiversity Visions for Ecoregion Conservation. Volume II: Freshwater Ecoregions. Washington, DC, World Wildlife Fund. Available online at:

Baltzer M.C. Nguyen Thi Dao & Shore R.G. eds. 2001. Towards A Vision for Biodiversity Conservation in the Forests of the Lower Mekong Ecoregion Complex. WWF Indochina/WWF US, Hanoi and Washington.

Dinerstein E.G., Powell D., Olson E., Wikramanayake R., Abell C., Loucks E., Underwood, Allnutt T., Wettengel W., Ricketts T., Strand H., O’Connor S. & Burgess N. 2000. A workbook for conducting biological assessments and developing biodiversity visions for ecoregionbased conservation. Volume I: Terrestrial Ecoregions.Washington, DC, World Wildlife Fund.

Dinerstein E., Olson D.M., Graham D.J., Webster A.L., Primm S.A., Bookbinder M.P. & Ledec G. 1995. A Conservation Assessment of the Terrestrial Ecoregions of Latin America and the Caribbean.Washington, DC, The World Bank.

Döll P., Kaspar F., & Lehner B. 2003. A global hydrological model for deriving water availability indicators: Model tuning and validation. Journal of Hydrology, 270: 105-134.

Environmental Systems Research Institute (ESRI) 1993. Digital chart of the world. Redlands, California, Environmental Systems Research Institute.

Fausch K.D., Torgensen C.E., Baxter C.V. & Li H.W. 2002. Landscapes to riverscapes: Bridging the gap between research and conservation of stream fishes. BioScience, 52: 483-498.

Food and Agricultural Organization of the United Nations (FAO). 2000. Global Agro-Ecological Zones Global - AEZ. Rome, Italy and Laxenburg, Austria: FAO and International Institute for Applied Systems Analysis IIASA. (Available at:

Food and Agricultural Organization of the United Nations (FAO). 2001. Atlas of Water Resources and Irrigation in Africa. Land and Water Digital Media Series CD 13. Rome, FAO. (Available at:

Frissell C.A. 1997. Ecological principles. In J.E. Williams, C.A. Wood & M.P. Dombeck eds. Watershed Restoration: Principles and Practices. Bethesda, Maryland, American Fisheries Society: 96-115.

Higgins J. Lammert M. & Bryer M. 1999. Update No. 6: Including aquatic targets in ecoregional portfolios: Guidance for ecoregional planning teams. In Designing a geographaphy of hope update: Reflections of current thinking and conversations. Arlington, Virginia, The Nature Conservancy.

Hooper B.P. & Margerum R.D. 2000. Integrated watershed management for river conservation: Perspectives from experiences in Australia and the United States. In P.J. Boon, B.R. Davies & G.E. Petts eds. Global Perspectives on River Conservation: Science, Policy and Practice. New York, John Wiley & Sons. pp. 509-517.

Margules C.R. & Pressey R.L. 2000. Systematic conservation planning. Nature, 405:243-253.

Noss R.F. 1992. The Wildlands Project: Land conservation strategy. Wild Earth Special Issue - The Wildlands Project 10-25.

Oak Ridge National Laboratory ORNL 2001. Landscan 2001 Global Population Database. (Available at:

United States Geological Survey USGS, IGBP 2001. Global Land Cover Characterization. (Available at:

Van Nieuwenhuizen G.D.P. & Day J.A. eds. 2000. Cape Action plan for the environment: The conservation of freshwater ecosystems in the Cape Floral Kingdom. Cape Town, SA, Freshwater Research Unit, University of Cape Town.

WWF 2001. Biological Priorities for Conservation in the Guinean-Congolian Forest and Freshwater Region.Washington, DC, WWF-US/CARPO.

WWF 2002. Biodiversity vision for the Amazon River and Floodplain Ecoregion. Technical report.Washington, DC, WWF-US. 86 pp.

Previous Page Top of Page Next Page