Introduction

Alpine river systems are fed by glacial icemelt, snowmelt, and groundwater (Milner and Petts, 1994; Ward, 1994; Malard et al., 1999; Ward et al., 1999; Füreder et al., 2001). Although streams originating from these water sources share common features, such as steep gradients, high flow velocities, and high dissolved oxygen concentrations, each source produces a characteristic discharge regime and a distinctive suite of physical and chemical characteristics (Milner and Petts, 1994; Ward, 1994; Tockner et al., 1997, 2002; Friberg et al., 2001). Furthermore, the distribution of snow, ice, and groundwater springs varies spatially from stream-reach to catchment scale, resulting in stream segments with characteristics reflecting the different proportions of runoff sources (Brittain and Milner, 2001). Physical variables (e.g., water temperature, channel stability, suspended sediment concentration) influence streams over time-scales ranging from diel to millennial (Smith et al., 2001), adding further complexity to abiotic mechanisms controlling the ecological processes in alpine streams.

Many aspects of the hydrology and physicochemical features of glacier-fed streams are relatively well studied, especially in Europe (Milner and Petts, 1994); in contrast, knowledge of the ecology of these systems has been relatively poor until recently (e.g., Brittain and Milner, 2001; Lods-Crozet et al., 2001, Snook and Milner, 2001; Zah et al., 2001). Recent work undertaken within a number of glaciated catchments in Europe as part of a coordinated project (Arctic and Alpine Stream Ecosystem Research [AASER]) has greatly enhanced understanding of alpine stream ecology (see Brittain and Milner, 2001; Milner et al., 2001). Importantly, this work illustrated that a range of spatiotemporal patterns in environmental conditions markedly affect the composition of benthic stream communities. Currently, the classification of alpine streams is predominantly based upon water source and temperature (e.g., Steffan, 1971; Ward, 1994), although some studies refer to habitat conditions, including turbidity and flow regimes (Ward, 1994). However, the dynamic nature of water source contributions in alpine catchments results in a greater amount of stream physical habitat variation than these traditional classifications account for. A fuller understanding of the cascade of environmental processes involved in determining benthic community assemblages is required if these unique alpine ecosystems are to be utilized as indicators of climate change (e.g., McGregor et al., 1995).

This paper identifies the range of environmental variables that affect alpine stream communities. We then review recent advances in understanding the dynamic functioning of alpine water sources and the interrelationships between environmental variables and their roles in determining benthic communities in alpine streams. We examine the traditional classifications of alpine streams into “glacial,” “snowmelt,” and “groundwater” categories based upon water temperatures and critically evaluate the suitability of this approach. The paper's central theme of habitat complexity is developed with the major linkages between environmental variables and benthic communities highlighted through a conceptual model. We propose an alternative classification scheme with the aim of providing a basis for a more complete understanding of the dynamic nature of water store contributions, along with the physical and chemical properties they impart upon stream communities. Field data collected in the French Pyrénées are used to support this new classification scheme, which we put forward as a tool for further research in alpine river catchments.

Traditional Classification and Zonation Schemes for Alpine Streams

Since stream character in alpine catchments is strongly influenced by local water source and the origin of major tributaries (Füreder, 1999), three general stream biotypes (kryal, rhithral, and krenal) have been designated based upon their water source which determines characteristic stream temperature ranges for these biotopes. These temperature ranges then play a major role in defining faunal distributions (e.g., Steffan, 1971; Ward, 1994). Glacier-melt-dominated (kryal) streams occur at high altitude close to the glacier terminus and have the lowest water temperatures. Kryal streams are usually subdivided into an upper metakryal and lower hypokryal, reflecting contrasting faunal compositions: Diamesa dominated and predominantly Diamesinae/Orthocladiinae, respectively (Steffan, 1971). Groundwater (krenal) streams are typified by relatively constant water temperature. Rhithral, or seasonal snowmelt-fed, streams are characterized by conditions intermediate between kryal and krenal (Ward, 1994).

In alpine regions, these zonations are generally applied to headwater streams that lie between the permanent snowline and the treeline (Brittain and Milner, 2001). In undisturbed environments, the treeline, which forms the divide between subalpine/subarctic woodland and treeless alpine/tundra areas, approximates the 10°C July isotherm (Remmert, 1980). Snowfields may be present year-round at higher altitudes. In some areas, glaciers (e.g., Brittain et al., 2001; Robinson et al., 2001) and glacier-melt-dominated streams (e.g., Burgherr and Ward, 2001; Zah et al., 2001) actually extend through the treeline, and in such situations the stream classifications are usually still applied.

Alpine drainage basins containing water from multiple sources are complex systems. Water sources such as glaciers, snow, and groundwater act as hydrological stores within which flows are attenuated; sediment can be eroded, transported and deposited; energy exchanges can affect water temperature; and physicochemical reactions and the mixing of water from various sources can result in water quality variations (Smith et al., 2001). Consequently, stream communities have to exist within, and respond to, a range of dynamic physical processes within alpine stream systems. These physical factors control both the structure and functioning of alpine stream communities. A summary of the environmental variables that contribute to the dynamic nature of alpine streams and their effect upon benthic stream communities is provided in Table 1.

The Applicability of Traditional Alpine Stream Zonation

Wide spatiotemporal variability in water store inputs contributes to the uniqueness of alpine stream systems (Smith et al., 2001). To understand how this variability may affect the physical characteristics of streams and how, in turn, these characteristics interact to determine the structure of benthic communities, it is essential to consider the range and scales of links between the physical environment and stream ecosystems. A conceptual model summarizing these interactions is presented in Figure 1. It is apparent that a range of environmental variables may dynamically shape the characteristics of alpine streams, interacting to produce a variety of habitats within which benthic communities exist.

Regional climatic influences play a major role in alpine stream dynamics by supplying and removing mass and energy at the glacier and snowpack surface, thereby determining accumulation and ablation rates of snow and ice (Hannah et al., 1999). This process drives variation in snow and ice meltwater inputs to glacial streams throughout the year. The timing and volume of bulk meltwater production (Hannah and Gurnell, 2001), along with inputs of groundwater from springs, seeps, and upwellings (Ward et al., 1999; Füreder et al., 2001; Lafont and Malard, 2001; Malard et al., 2001) result in wide spatiotemporal variation in the physicochemical properties of streams. The dynamic nature of inputs from all these sources may have a significant impact on benthic communities in alpine streams at the stream-reach scale (Snook and Milner, 2001).

Water temperature has long been recognized as a major contributor to stream community structure and functioning (Hynes, 1970; Vannote and Sweeney, 1980; Vannote et al., 1980; Sweeney, 1984; Ward, 1985). In glacial streams, water temperature plays a primary role in structuring benthic communities (Milner and Petts, 1994; Milner et al., 2001). At the level of individual organisms, water temperature controls growth rates, the timing of life cycles, and rates of primary and secondary production (Hynes, 1970; Allan, 1995). In alpine streams, water sources are the primary influence on water temperature, although climatic influences such as solar radiation input can also play an important role. Because stream water temperature may be influenced by factors other than the source, classifying streams on the basis of temperature alone is problematic. In addition, water sources impart a range of other physical and chemical properties that are discussed by current classification approaches (e.g., Ward, 1994) but rarely used in practice.

Even where water temperature is suitable, channel and bed stability may still inhibit colonization of certain taxa (Milner and Petts, 1994). For this reason, stability must be considered in conjunction with water temperature when evaluating community composition of glacial rivers (Milner and Petts, 1994; Milner et al., 2001). Channel stability is related to channel slope, substratum composition, and stream power (Resh et al., 1988). The increased sediment loads that accompany high flows, and that are characteristic of glacial streams (Gurnell, 1987, 1995), may directly affect stream communities through abrasion and reduced light penetration (Castella et al., 2001; Milner et al., 2001). Stream communities are also affected indirectly by the reworking of channel profiles (Statzner et al., 1988). Riparian vegetation can also indirectly affect stream communities by restricting local morphological and sedimentological change (Gurnell et al., 1999).

Inputs of allochthonous organic matter to streams (delivered by aerial inputs, erosion, or exfiltrating groundwater: e.g., Zah et al., 2001) act as an important energy source for benthic macroinvertebrates (Cummins et al., 1973; Anderson and Sedell, 1979; Bird and Kaushik, 1981), and can influence their distribution and abundance (Vannote et al., 1980; Ward, 1992; Allan, 1995). Although inputs of vegetation to streams act primarily as a food source, they also contribute to the nature of the substratum (Minshall, 1984). The substratum consists of both inorganic (microscopic silts to boulders) and organic materials (bryophytes and land-derived vegetation) and forms the stage upon which aquatic insects move, rest, find shelter, and seek food (Minshall, 1984). In alpine streams, deposits of suspended sediment affect the nature of the substratum. The substratum is also a key control on channel and bed stability (Pfankuch, 1975). The ability of an insect to adhere, cling, burrow, or build cases is directly affected by the nature of the substratum (Mackay, 1977; Minshall, 1984). The substratum may indirectly affect the composition of benthic communities through alteration of the direction, force, and turbulence of water flows (Newbury, 1984). These flow characteristics influence the metabolism, feeding, and behavior of lotic organisms (Statzner et al., 1988).

The availability of ions from the weathering of rocks and soils along with atmospheric inputs influences the water chemistry of streams, which, in turn, may play an important role in determining benthic community composition (Hynes, 1970; Sabater and Roca, 1990). In alpine catchments, little research has been undertaken to determine specifically the influence of water chemistry on stream communities (Kownacka and Kownacki, 1972; Sabater and Roca, 1990; Robinson et al., 2001). Water analyses have confirmed that some groundwater streams have distinctive chemical characteristics that vary spatially and temporally (Sabater and Roca, 1990; Robinson et al., 2001). Such chemical variability is likely to result in habitat differences and thus to have a strong effect on benthic community composition (Sabater and Roca, 1990; Robinson et al., 2001).

Given the occurrence of high spatiotemporal variability in physical habitat in alpine streams, the traditional classification system using water temperature alone to indicate water source is misleading. For example, a stream fed only by glacial meltwater is classified as kryal in its uppermost reaches (Ward, 1994). As water flows downstream, increases in temperature occur as a result of advective heat transfer and solar heating (Füreder, 1999), leading to the classification of the stream as rhithral, even though it may have no true snowmelt influence. As such, the current classification system of kryal, rhithral, and krenal may be misleading due to the conflation of water source and water temperature. Figure 2 highlights the problem of using stream temperature to classify stream reaches. These data were collected on a glacial stream in the French Pyrénées over the 1997 melt season (see Smith et al., 2001, for methodology details). Water temperature was consistently below 4°C (kryal) at the beginning of the melt season, when snowmelt was the major contributor to stream flow. As the melt season progressed, the temperature range increased due to a greater receipt of solar radiation input, so that use of the traditional classification method would result in a rhithral classification despite the fact that snowmelt contribution was decreasing and glacial icemelt increasing (Smith, 1999).

The term “kryal” was used in a study of streams fed by a permanent snowfield but no glacier (Elgmork and Saether, 1970). In this study, five stream reaches were investigated in the Colorado Rocky Mountains between a snowfield and the first of a series of lakes. All five study sites had midsummer temperatures of approximately 1°C, and as such were deemed to be an example of kryal streams (Ward, 1994). In this situation, streams fed completely by snowmelt should be placed into a modified rhithral category, which would correctly reflect the origin and physical characteristics of these meltwaters.

A further example of the way in which traditional alpine stream categories are problematic is that outlets from cirque lakes have been classified as rhithral (e.g., Elgmork and Saether, 1970). Lakes may significantly alter the nature of headwaters by increasing channel stability downstream, modifying chemical characteristics, or increasing water temperature (Milner and Petts, 1994) to such an extent that the streams draining them are markedly different from those supplying the lake. Even though the water temperature of snowmelt streams may be similar to those of lake outlets, other physical characteristics may be markedly different enough to require a separate classification.

The use of the term “rhithral” has also been questioned with reference to its application to precipitation as well as snowmelt-fed streams (e.g., Illies, 1961; Illies and Botosaneanu, 1963; Ward, 1994). More recently, the terms “ombral” and “chial” have been proposed to describe rainfall- and snowmelt-fed streams, respectively (Schütz, 1999; Lencioni, 2000). However, even this division is problematic because streams are rarely influenced by a single water source. Snowmelt may flow overland into glacial- or groundwater-fed streams; precipitation may also fall directly (typically <1% contribution) or flow overland or through hillslopes into streams. In these situations, the stream would still be classified by its temperature. In a similar manner, spring-fed tributaries may markedly increase the temperature of kryal streams below confluences (e.g., Ward, 1994). Therefore, a stream reach downstream of a groundwater input to a glacial stream could be classified as rhithral if thermal conditions were substantially altered, regardless of the absence of meltwater from snowpacks.

In streams close to glacial margins, the subdivision of the kryal into an upper metakryal and lower hypokryal is also problematic. Snowpacks cover the ablation zone of glaciers, melting upglacier over the summer (Hannah et al., 2000a), and their meltwaters may flow into the proglacial zone, resulting in a mixed water source. However, as temperature is the principal physical variable used as a classification criterion, little account is taken of the influence of the snowmelt contributions.

The quantity and quality of water source inputs to alpine streams may also vary over a number of temporal scales from diel to intraannual (Fig. 3). This variety can result in changes in stream classification over an annual cycle. For example, springtime flows may be predominantly snowmelt driven. In summer, glacier melt often dominates, while toward the end of the melt season and through winter, groundwater inputs may be the principal water source for streams (Malard et al., 1999; Smith et al., 2001). Similarly, changes in water routing can occur diurnally. Groundwater sources may dominate stream channels at low flows through the late evening and night, when glacial icemelt or snowmelt is minimal. As solar radiation increases toward an afternoon peak, the proportion of snow- and icemelt increases, and streams shift to be either snowmelt or icemelt dominated. The current system does not take these temporal shifts into account unless they are accompanied by marked changes in water temperature.

An Alternative Approach to Alpine Stream Classification

Given the level of inconsistency in classifying alpine streams using traditional approaches, we suggest that future studies should concentrate on determining the actual relative contributions of different water sources rather than assigning streams to broad categories based upon their principal water source and associated temperature regime. Hence, we propose nine new categories for classification based upon the relative proportions of glacial meltwater (kryal), snowmelt (nival), and groundwater (krenal) present (Table 2). A greater number of categories, giving qualitative explanations of the stream type as determined by source waters alone, may improve comparisons of stream reaches between studies and allow changes in source waters within alpine catchments to be tracked over temporal scales ranging from diel to seasonal. A simple triangular diagram (Fig. 4) to illustrate the proportion of source waters to a stream reach allows ease of classification.

The classification principally aims to categorize streams based on their relative proportions of source waters. Variations in environmental characteristics can be predicted within the nine categories, providing better resolution (Table 3). Characteristic temperature ranges for each category are not included due to the complicating influences of climate and distance from the source. The important role of water temperature in alpine streams is acknowledged, but its use as a primary classification tool is severely limited, particularly when water temperature is equated with water sources. However, to aid comparison between studies, water temperature ranges should be stated alongside the new classification categories to facilitate proper interpretation of habitat characteristics within a stream reach and so enable comparison between studies.

The new classification system accounts only for stream conditions between precipitation events, which is a limitation. However, given that precipitation events often provide volumes of water that exceed those from other sources (Matthaei et al., 1997), their contribution should be evident from the examination of stream hydrographs (Hannah et al., 1999, 2000b). Nevertheless, since precipitation events are usually episodic, they are likely to result in relatively short-term deviations from the conditions described by a classification. However, it is important to acknowledge these short-term deviations because peak flows caused by precipitation events are likely to have an effect on stream communities (e.g., Matthaei et al., 1997).

A Test Application of the Alternative Alpine Stream Classification

The new classification has been tested using data collected from a small alpine glacierized catchment (Taillon-Gabiétous) within the Cirque de Gavarnie region, French Pyrénées. The catchment contains two cirque glaciers (Taillon and Gabiétous), large valley snowpacks, and a karst groundwater system that contributes a number of springs to the lower catchment. The maximum altitude is 3144 m, comprising a catchment area of 6.4 km² at the farthest downstream point sampled. Permanent snow and ice cover within the catchment is 5.4% (Smith et al., 2001).

Temporal changes in the percentage of source water contributions to stream flow for the Taillon-Gabiétous catchment, 1997 ablation season (Smith, 1999), are shown in Figure 5. Close to the Taillon Glacier margin, the stream is predominantly classified as Nivo-kryal, reflecting the large input of snow and glacial meltwaters to the proglacial stream. Toward the end of the ablation season, the proportion of snowmelt input to the stream decreases to less than 20%, icemelt waters become dominant, and the classification becomes Kryo-nival. These subtle changes, which may be of importance to benthic communities, have not been documented previously since water temperature was always <2°C and the stream reach was classified simply as kryal throughout the melt season (Snook and Milner, 2001). Other physical variables, notably diel discharge fluctuations and suspended sediment concentration also varied as water sources changed (Smith et al., 2001). Classification approaches based principally upon temperature do not adequately describe this variability.

The classification may also be used to describe changes in streams with distance from glacier margins. Figure 6 shows proportions of source water contributions to the Taillon Glacier stream, averaged over the 1997 ablation season (for site descriptions, refer to Snook and Milner, 2001). At the glacier margin (Site 1) and at Site 2, the stream is classified as Kryo-nival, highlighting the mix of waters from the snow-covered glacier. The classification changes to Nivo-kryal as distance from the glacier increases, reflecting the increased proportion of snowmelt input as the catchment area upstream of the stream reach increases and the percentage covered by glacier decreases. Between Sites 3 and 5 (Site 4 was located on a side tributary), approximately 1 km downstream of the Taillon Glacier, a number of groundwater springs enter the Taillon stream, thus accounting for the change in classification to Kreno-nival. These two examples illustrate the spatiotemporal variability in water source contributions that may be identified using the alternative classification. In comparison, traditional classification approaches would not identify this dynamism. Further evaluation is required to determine the effectiveness of this alternative classification system, in describing source water dynamics over longer time periods (interannually). Testing is also required to establish whether the approach adequately describes variations in source water inputs and the physicochemical characteristics of other alpine glacierized catchment streams.

Summary

Stream-reach characteristics in alpine catchments are strongly influenced by water source. Classifications based predominantly upon water temperature are problematic since the conflation of water source and water temperature is often inaccurate. Furthermore, alpine streams exhibit wide spatial and temporal variation in a range of physical and chemical characteristics, both within and between streams, which is largely unaccounted for by current classification approaches.

Even within the relatively small Taillon-Gabiétous catchment, temporal and spatial changes in water sources are evident throughout the melt season, and therefore, alternative stream classifications occur. In a much larger catchment (Val Roseg, Switzerland), groundwater contributions have been shown to vary from <10% to >70% over the course of an ablation season (Tockner et al., 2002), potentially leading to much greater variability in stream classification than is shown in this paper for the headwater Taillon-Gabiétous catchment.

Further examination of temporal changes in stream classification may yield important information on associated changes in stream physical and chemical characteristics, which, in turn, may be important determinants in the timing of life cycles of aquatic insects and benthic macroinvertebrate community structure. Water temperature and channel stability are recognized as fundamental stream characteristics influencing benthic communities of alpine streams at the reach scale (Milner and Petts, 1994; Milner et al., 2001). However, recent work has highlighted the roles of other environmental variables, such as hydrochemistry, suspended sediments, and particulate organic matter, in determining alpine stream community structure (e.g., Castella et al., 2001; Lods-Crozet et al., 2001; Milner et al., 2001). Although it may be argued that these variables are less important than water temperature and channel stability, further work is necessary to quantify their influence, especially at smaller spatial scales (i.e., patch) where their relative importance may be higher. This classification provides a framework for future research that should provide a more refined resolution for understanding dynamic changes in alpine stream physical characteristics and benthic communities.