2 Drivers of large-scale riverine research

2.1 Policy drivers

Management and conservation of aquatic habitats and their biota are often subject to new or adapted legal frameworks that rely on large-scale analyses. These legal frameworks may result from multiple issues such as the need for flood protection, aquatic ecosystem/water protection, nature/endangered species conservation, land use planning or conversion, rural development, or agricultural management. Meeting the needs of environmental legislation (Table 1) has pushed managers and policy-makers to demand landscape-scale products. Such products include estimates of habitat capacity for endangered species across vast areas of the Pacific Northwest (PNW) USA (e.g., Carmichael and Taylor, 2010; Burnett et al., 2007*), modeled estimates of river condition across the European Union (EU) (e.g., Pont et al., 2006*), or rigorous assessments of water body status and trends across the western USA (e.g., Pont et al., 2009; Stoddard et al., 2005*) and the conterminous USA (e.g., USEPA, 2009*; Paulsen et al., 2008*). Large-scale, multi-million dollar rehabilitation projects such as those being implemented in the U.S. for Chesapeake Bay, San Francisco Bay, south Florida, or entire states (e.g., Ohio and Oregon) require landscape-scale monitoring and analyses (Independent Multidisciplinary Science Team, 2007). Conforming to these directives and monitoring the effectiveness of large-scale restoration has created a “window of opportunity” for increased research at the landscape scale and for landscape perspectives to affect on-the-ground management (Hughes et al., 2008*).

The European Water Framework Directive (WFD), implemented in 2000, is an example of a relatively new policy directive that promotes landscape-scale analyses and management. It demands that all water bodies across Europe achieve good ecological status or good ecological potential by the year 2015 or, at the latest extension of the deadline, by 2027 (Water Framework Directive, 2000). These assessments are required for every stream and river in Europe and can be based on empirical or modeled data. The WFD specifically addresses impacts of catchment-scale land use on, for example, diffuse pollution in riverine systems. “Land use patterns, including identification of the main urban, industrial and agricultural areas and, where relevant, fisheries and forests” must be estimated (WFD Article 5 and Annex II) during the identification of human impacts in each catchment. Additionally, River Basin Management Plans (RBMP), required for all catchments, should include an “estimation of diffuse source pollution, including a summary of land use” (WFD, Article 13 and Annex VII). Because of these specific requirements, the WFD institutionalizes a large-scale approach to river research and management that requires coordination of activities such as monitoring, rehabilitation prioritization, planning, and public consultation across vast extents (Steyaert and Ollivier, 2007; Moss, 2004).

The WFD is not the only policy that promotes landscape-scale research and management across Europe. The Habitats Directive (1992) and the Birds Directive (1979) mandate a landscape approach to conservation planning that goes well beyond the traditional research scales of individual habitats and river reaches, even beyond political boundaries and frontiers. Such large-scale thinking is also reflected in the International Commission for the Protection of the Danube River (ICPDR), which focuses on the management of land including the built environment. The Natura 2000 network of conservation sites is yet another European example of a very large-scale conservation program and new policies continue to expand its scope.

There are policy drivers for landscape-scale analyses in other regions. One of the primary drivers of landscape-scale riverine research in the USA has been the Endangered Species Act (ESA; Public Law (PL) 92-205, 16 United States Code (USC) 1531 et seq. 1973). The ESA provides for the conservation of ecosystems upon which threatened and endangered species of fish, wildlife, and plants depend. Under this Act, twenty-seven evolutionarily significant units (ESUs) of anadromous Pacific salmonids, including Chinook, chum, coho, sockeye, and steelhead trout, have been listed as threatened or endangered since 1991. The cumulative area (spanning portions of the states of Washington, Oregon, Idaho, and California) of ESUs with one or more listed species is approximately 372,000 km² and encompasses thousands of kilometers of stream. To complicate matters, these fish use a wide variety of stream habitat types often spawning in low gradient habitats, rearing in smaller streams, and migrating through urbanized mainstem rivers. As a result of the ESA and the listing of Pacific salmonids over such vast extents, a critical need has arisen for estimation and prediction of species distributions and environmental conditions over large areas for which adequate field data do not exist. The Species at Risk Act adopted in Canada in 2002 serves a purpose similar to the Endangered Species Act and has promoted large-scale thinking about aquatic systems across Canada (http://laws.justice.gc.ca/eng/S-15.3/).

A second key policy driver across the USA is the Federal Water Pollution Control Act, commonly referred to as the Clean Water Act (CWA, PL 92-500, 33 USC 2101 et seq. 1948). The purpose of the CWA is to restore and maintain the chemical, physical, and biological integrity of the nation’s waters. As a result, the US Environmental Protection Agency (USEPA) is involved in a wide range of monitoring and analysis programs that consider impacts of land use over large spatial extents on the health and condition of surface waters (lakes, streams, rivers, estuaries, coastal waters, wetlands). These programs are designed to answer questions such as: What is the current extent of the waters that support healthy ecosystems? Is water quality improving? What are the key stressors that account for waters with poor biological condition? The evolution of CWA programs over the last decade has included a shift toward watershed-based TMDL strategies (Total Maximum Daily Loads; http://www.epa.gov/watertrain/cwa/) and rotating probabilistic surveys of all USA surface waters on an annual basis (e.g., USEPA, 2009*; Paulsen et al., 2008*; Shapiro et al., 2008*).

Japan is experiencing a similar shift toward riverine policy directives that promote a landscape-scale perspective. The River Law in Japan was originally designed solely for flood protection and water use but, in 1990, Japan launched “Nature-Oriented River Work” or “Ta Shizen Gata Kawa Zukuri,” which is an initiative aimed at conserving beautiful landscapes and riverine biodiversity. The trend toward increasing spatial extents in river conservation continued when the River Law was amended in 1997 to add “Conservation and Improvement of River Environments” as a third purpose of the law. These policies are expected to serve as a comprehensive and systematic plan to protect biodiversity and rehabilitate fluvial environments and ecosystems. One practical outcome of this strategy is the National Census on River Environment, which provides periodic surveys of the status of rivers and dams from an environmental perspective (e.g., fish survey) over large areas.

Beyond Europe, North America, and Japan, the trend toward large-scale conservation is also strong. The Ramsar Convention on Wetlands, signed in Iran in 1971 promotes a landscape perspective. In one of the latest resolutions (Resolution IX.I, Annex C) of this convention, it is stated that “management and development of wetlands must be undertaken within the context of their larger surrounding ‘waterscape’ (the river basin or catchment, including the hydrological processes and functions within the basin or catchment) as well [as] their larger surrounding landscape” (http://www.ramsar.org/pdf/lib/lib_handbooks2006_e07.pdf). Similar conditions led to the initiation of the River Health Program (RHP) in South Africa in 1994. The RHP objectives are to monitor and report on aquatic ecosystem status, trends, and emerging problems of South African rivers (Kleynhans, 1999*; Murray, 1999). In Australia, over-allocation of water coupled with a long-term shift to a drought regime resulting from climate change have stimulated a Sustainable Rivers Audit Program to monitor conditions across the nation’s major river basins (Davies et al., 2006*). In Brazil, the Manuelzão Project is directed toward public education, wastewater treatment, land and water use, and ecological monitoring across the entire Rio das Velhas Basin (Lisboa et al., 2008). Brazil has also recently passed progressive environmental protection laws designed to protect the nation’s rich natural resources, especially its water quality and biodiversity, and has established national parks to protect biodiversity and the recovery of large tracts of riparian corridors in many areas (Drummond and Barros-Platiau, 2006). Ecuador recently adopted a new constitution that formally recognizes the rights of nature “to exist, persist, maintain and regenerate its vital cycles, structure, functions and its processes in evolution” (http://celdf.org/article.php?id=712).

2.2 Ecological drivers

The nature of current conservation concerns has clearly necessitated a landscape-scale approach to a wide range of riverine issues (Table 1). A summary of all current conservation concerns is beyond the scope of this review; we simply remind readers of the large spatial extent of current issues. Managing watersheds, stream habitats, and fisheries in the face of climate change will, for example, require spatially extensive approaches to monitoring, data collection and analysis. Non-indigenous species continue to pose major threats to aquatic ecosystems (Leprieur et al., 2008b*; Lomnicky et al., 2007; Miller Reed and Czech, 2005*). Aquatic non-indigenous species range over very large areas and efforts to limit their spread or estimate their impacts must necessarily incorporate data collection and analysis techniques at the landscape scale (e.g., Sanderson et al., 2009). Increasing awareness of and concern about these threats provides a second “window of opportunity” for comprehensive, large-scale river research (Hughes et al., 2008*).

2.3 Technological drivers

The incorporation of landscape thinking into riverine research has been catapulted by the availability of new technologies (Johnson and Host, 2010*; Johnson and Gage, 1997*). Habitat assessments in the past were necessarily transect-based and field intensive. Research and analyses had to proceed by sampling discrete reaches of very large, diverse, and dynamic systems. Therefore, despite conceptual models that placed streams in their landscape context (e.g., Schlosser, 1991*; Vannote et al., 1980*; Hynes, 1975*), aquatic research methods have historically been focused on individual stream reaches.

Aerial photography, in use since the 1940s, was the first technology to provide a landscape-scale perspective but a suite of recent new technologies has dramatically improved our ability to conduct research over large areas. High spatial resolution (e.g., 1-meter pixel resolution) digital imagery provides powerful new tools for riverine landscape mapping and feature identification. Remote sensing has previously enabled synoptic views of entire rivers and their catchments and can now provide data at resolutions that capture reach-scale riparian and instream habitat structuring (Hall et al., 2009*; Silva et al., 2008*; Legleiter, 2003*; Sawaya et al., 2003; Leuven et al., 2002; Marcus, 2002*). Geographic Information Systems (GIS) have made landscape pattern analysis possible. As well, digital terrain modeling has provided tools for defining catchment boundaries and stream flow in flexible and easily repeatable ways.

Spatial data can describe relatively immutable natural conditions such as elevation, geology, soil type, air temperature, and precipitation. Spatial data are also available to describe anthropogenic stressors such as road density, land use, urbanization, migration barriers, water withdrawals, dams, and mine claim density. Mertes (2002*) provided a comprehensive review of remote sensing advances and their applications to riverine environments, and discussed the newly arising capabilities to examine “spatial and temporal relationships among biota, hydrology, and geomorphology across scales from microhabitats to channel units to valleys to catchments”. Mertes (2002*) surveyed the increasing variety of sensing technologies available for mapping both landscape and water properties including optical sensors, light detection and ranging (LiDAR), forward looking infrared (FLIR), and radio detection and ranging (RADAR), deployed from multiple platforms from low altitude tethered balloons, to helicopter and fixed wing aircraft, to satellites. These new techniques allow us to capture information with a high degree of both lateral and longitudinal resolution. It is now also possible to directly sense water properties including elevation of the water surface, width of flooded area, distribution of wood, surface temperature, turbidity, and sediment movement (Smikrud et al., 2008; Smikrud and Prakash, 2006; Mertes, 2002*). And, we can generate stream network maps at much finer scales using LiDAR-based digital elevation models (DEM) than with satellite generated data (Mouton, 2005) as well as bathymetric delineations (Jones et al., 2008; McKean et al., 2008*).

In addition to advances in spatial resolution, new sensors have increased spectral resolution. “Hyperspectral” sensors image in hundreds of finely defined wavelength bands, in contrast to the 3–6 broadly defined wavelength bands of color aerial photography and multispectral land imaging satellites (e.g., Landsat, SPOT). Legleiter et al. (2002) and Marcus (2002*) compared capabilities of multi- and hyperspectral sensors for mapping instream features and both found that higher spectral and spatial resolutions increased mapping accuracies over lower resolution data. One outcome of these studies was the interesting prospect that mapping from high-resolution hyperspectral imagery might be more accurate than field based mapping (Marcus, 2002). Datasets generated from hyperspectral imagery have more stringent requirements for accurate preprocessing and require increasingly sophisticated classification techniques (Aspinall et al., 2002; Wulder et al., 2004); however, the spatial and spectral resolution provided by high-resolution hyperspectral systems is opening new avenues for ecosystem mapping, including mapping aquatic habitat features in both large and small streams (Figure 2*). Traditional reach-based models, for example, have been limited in their ability to represent lateral habitat features but, using these new techniques, we can now map floodplains and large rivers accurately across two and even three dimensions (Kinzel, 2009).

Figure 2: Aerial photo of a tributary of the Sarufutsu River, Japan, by DMC (digital mapping camera). Field-validated instream features such as woody debris (red lines), log jam (thick orange line), and deep pools (green circles) are superimposed. The field mapping of these features closely matches the photo imagery.