Hydrology, water availability and tundra ecosystem function in a changing climate: the need for a closer integration of ideas?

Introduction

Global General Circulation Models now predict with some degree of consensus the mean temperature changes that are likely to occur within Arctic and alpine tundra regions by the middle of the next century ( Maxwell 1992; Cattle & Crossley 1995; IPCC 1996). Earlier predictions have formed the basis of detailed experimental work, both in the laboratory and field, on the responses of tundra ecosystems and their constituent organisms to changing climatic temperatures (e.g. Havström et al. 1993 ; Coulson et al. 1993 , 1996; Wookey et al. 1993 ; Chapin et al. 1995 ; Robinson et al. 1995 , 1998; Zhang & Welker 1996). Environmental temperature is a simple, easily measured parameter to which organisms usually show a direct response ( Callaghan et al. 1992 ; Callaghan & Jonasson 1995). By contrast, predictions of future patterns of precipitation are far more uncertain ( Shaver et al. 1992 ; Loaiciga et al. 1996 ; Rowntree 1996; Oechel et al. 1997 ) and realistic experimental manipulation of precipitation patterns is difficult. As a consequence, direct long-term experimental studies on the response of Arctic and alpine organisms and ecosystems to varying moisture availability are few. Similarly, it is more difficult to incorporate precipitation into climate change models for tundra ecosystems ( Starfield & Chapin 1996). The situation is complicated by the fact that water exists in various forms with distinct properties – ice, snow, free water and water vapour – each of which can have markedly different effects on the biologies of organisms or ecosystems ( Kane et al. 1992 ). Furthermore, the presence of water in its various forms has a strong mediating influence on several other important environmental characteristics such as temperature and it is difficult experimentally to isolate water as a single factor. Arctic and high alpine ecosystems are, despite an often apparent abundance of free water, areas of low precipitation. Small variations in the amount, type or timing of the yearly precipitation input, coupled with differences in evapotranspiration and drainage characteristics, can have highly significant effects on the spatial and temporal availability of water to organisms occupying contrasting parts of a complex habitat mosaic ( Kane et al. 1996 ).

The importance of water as a factor limiting the broad biogeographical distributions of organisms in parallel Antarctic ecosystems has been stressed ( Kennedy 1993, 1995). Nevertheless, the complex role of water in the functioning of tundra ecosystems remains relatively neglected ( Chapin et al. 1996 ). Many studies have examined in isolation selected effects of ‘water’ availability on particular aspects of the biologies of Arctic organisms or ecosystems (e.g. Oberbauer & Dawson 1992). Very few have attempted to demonstrate how changing patterns of precipitation will impact directly on whole tundra ecosystems with their complex internal links and interrelationships ( Chapin et al. 1992 ; Hodkinson & Wookey 1998). A key feature of most tundra ecosystems, particularly in the high Arctic, is that significant biological activity is confined to a narrow active layer of soil which supports at most a dwarf plant community. The hydrology of this surface soil layer is of paramount importance in determining the nature of the biological communities that develop and the rates of key ecological processes such as primary production, decomposition, nutrient flux and gas exchange ( Kane et al. 1992 ; Ostendorf 1996).

The purpose of this paper is to evaluate, summarize and stress the complex role and importance of water in the functioning of tundra ecosystems. It is intended to serve as a basis for a clearer understanding of the interconnected mechanisms by which future altered patterns of precipitation might change the composition of communities and thereby the functioning of such ecosystems. This is especially important because the effects of climate change are predicted to be greatest and occur most rapidly at the higher latitudes. The term ‘water’ is used as shorthand throughout to indicate H₂O in its various interchangeable forms.

Forms and availability of water

Much of the water in tundra regions exists for most of the year in biologically unusable forms, such as snow and ice. Biologically available free water, or to a lesser extent free water vapour, increases initially during snow melt at the beginning of the active season but may become locally limiting later in the year as the characteristically high saturation vapour pressure deficit begins to dry out the tundra surface ( Saunders et al. 1997 ). Tundra landscapes thus exhibit wide variation in both spatial and temporal patterns of water availability. These patterns are linked to macro-and microtopographical features and drainage patterns, interacting with rates of snow melt, precipitation and evapotranspiration ( Kane 1996).

Physical effects of water

Water plays a primary role in soil formation and development and it is within the soil that much of the biological activity in Arctic and alpine ecosystems occurs. The role of ice and glacial processes in regolith formation is well known but the physical role of water in continually shaping and reshaping the tundra surface is less well understood. Similarly, the component input and output parameters of the mass water budgets of Arctic watersheds are poorly quantified and as a consequence the biologically significant features of the surface hydrology of Arctic habitats are only partially described ( Kane et al. 1990 ; Hinzman & Kane 1992; Kane 1996; Rovansek et al. 1996 ). Variations in precipitation, rates of snow melt, etc. balanced against rates of evapotranspiration determine the general availability of soil water and alteration of any of these parameters as a result of changing climate will alter soil moisture availability ( Frank & Inouye 1994; Hinzman & Kane 1992; MacDonald et al. 1994 ).

Microtopography also has an important bearing on localized water availability, with biological water availability often differing over short vertical distances, depending on site drainage patterns ( Cannon & Reid 1993). Both permafrost and free water movement characteristics play important roles in soil development ( Fitzpatrick 1997). Cryoturbation and water movement result in particular surface pattern features such as polygons, sorted circles, mud boils and solifluction stripes each with their own soil moisture characteristics ( Klimowicz & Uziak 1996). These physical processes may over time also play a significant role in maintaining substrate instability and preventing the development of more mature soils and an expanding plant cover ( Komarkova 1993).

The textural and other characteristics of the active layer of tundra soils are determined primarily by moisture status and drainage features ( Grabetskaya & Chigir 1991). Soil water and texture in turn determine soil aeration and thereby the availability of oxygen to soil animals, plant roots, etc. ( Gebauer et al. 1996 ).

Frozen water in the form of the permafrost is widely distributed within Arctic and some alpine soils where it alters and often restricts both the lateral and vertical movement of water ( Woo & Winter 1993). Biologically available water is confined to the unfrozen or active layer of soil overlying the permafrost, which continually cools the soil from below. The depth of the active layer is partly determined by soil moisture status and consequently varies amongst sites and both within and between seasons ( Woo & Xia 1996). Long-term changes in the depth of the active layer, associated with permafrost melting, has implications for several soil processes ( Waelbroeck et al. 1997 ).

In wet tundra sites the position of the fluctuating water table is an important determinant of the nature and rate of several soil processes, including decomposition, nutrient release, carbon dioxide and methane efflux, and transportation of particulate carbon. Over long time periods the permafrost, at least in wetter sites, is accretive, gradually removing water from the soil ( Hinkel et al. 1996 ). This effect will be offset against the increasing permafrost melt rates and deeper active layers associated with climate warming. At times of snow melt, on tundra with impeded drainage, surface movement of excess meltwater can act as an effective dispersal mechanism for soil organisms, such as Collembola and spiders, which are often found in aggregations at the foot of drainage slopes or along the margins of drainage channels (Hodkinson et al. unpublished).

Water has a high thermal capacity and the amounts of water present in tundra soils largely determines the soil’s thermal properties ( Rouse et al. 1992 ; Coulson et al. 1993 ; Waelbroeck 1993; Harte et al. 1995 ; Woo & Xia 1996). Dry soils warm up and cool down quickly but equivalent wet soils are thermally buffered, gaining and losing heat more slowly ( Woo & Xia 1996). Thus, within an habitat mosaic of varying moisture content, the earliest growing sites in the spring tend to be in the drier areas whereas the wetter areas are inclined to be the sites at which the commencement of growth occurs later. Often this effect is related to microtopography with the driest and earliest sites present on the raised ridges. This effect is reinforced by a positive feedback: as sites become warmer soil moisture evaporates allowing faster warming ( Harte & Shaw 1995). A similar effect has been noted in temperature manipulation experiments in which polythene tents have been used to raise soil and vegetation temperatures: an unwanted side-effect is a parallel decrease in soil moisture ( Coulson et al. 1993 ; Robinson et al. 1995 ). On a shorter diurnal cycle ground frost effects are less likely to be important in wetter soils. Bulk movement of water through the soil can also act as a potent mechanism of heat transfer.

The depth, duration, distribution and timing of precipitation, particularly in the form of winter snow has a major impact on temperature regimes within the soil and on rates of evaporation from the soil surface ( Coulson et al. 1995 ; Harding et al. 1995 ). Snow provides a thermal blanket that insulates the soil from the prevailing air temperatures. In the absence of snow, winter soil temperatures become isothermal, reflecting air temperature. Where snow cover is significant soil temperatures can remain several degrees above air temperature throughout the winter, permitting biological activity within an unfrozen soil ( Leinaas 1981; Rycheva et al. 1992 ; Coulson et al. 1995 ; Brooks et al. 1997 ). Delayed or reduced winter precipitation thus has a significant impact on the winter temperatures experienced by both the vegetation and the soil organisms. By contrast, summer rainfall in the high Arctic is usually low and intermittent, with organisms normally relying throughout most of the growing season on the available water released at snow melt.

Deep snow, takes longer and more heat to melt than shallow snow accumulations of similar compaction and air content. Thus exposed (usually raised) sites with thin snow cover clear quickly, providing a longer and at least initially warmer growing season than contrasting sites (usually depressions) with slower melting deep snow patches ( Sonesson & Callaghan 1991).

Water also has a high latent heat of fusion such that as soil water freezes it releases heat into the soil but when it melts it absorbs heat. These heat transfers effects can be significant in environments where low mean temperatures are linked with high soil moisture levels ( Woo & Xia 1996).

Within- and between-season variation in water inputs

The temporal and spatial distribution patterns of precipitation and the resulting water availability is a significant factor in the functioning of Arctic and alpine ecosystems. Natural variation occurs both in precipitation between sites and the amounts and type of precipitation at any one site. Winter precipitation accumulates and is stored as a snow pack for later release: summer rainfall is immediately available. However, snow meltwater may persist in temporary pools for part of the growing season ( Rovansek et al. 1996 ). These factors may result in parallel seasonal cycles of storage and release of nutrients, particularly nitrogen, where amounts of atmospheric deposition are significant ( Bowman 1992; Robinson & Wookey 1997; Woodin 1997).

There is often considerable year-to-year variation in precipitation at a given site ( Coulson et al. 1993 ; Taylor & Seastedt 1994; Walker et al. 1994 , 1995) and it is predicted that this variability will increase as a result of climate warming ( Bonan et al. 1995 ; Lliang et al. 1995 ). Precipitation is often highly seasonal and over large areas of the Arctic it falls mainly as winter snow. Thus, soil moisture levels often reach a maximum at snow melt and then decline as a result of evapotranspiration throughout the growing season, with important consequences for the biota ( Taylor & Seastedt 1994; Gold & Bliss 1995; Saunders et al. 1997 ). Sporadic drought events can have long-term impacts on both plant and invertebrate communities ( Williams 1990; Coulson et al. 1996 ; Hodkinson et al. 1996b ).

Water and plant communities

Availability of soil water in Arctic and alpine environments plays a significant role in determining the species distribution, spatial pattern, community composition and productivity of the vegetation ( Bliss & Wiegolaski 1973). For many tundra plants their growth response with respect to soil moisture is parabolic, reaching an optimum at a preferred soil moisture level and declining as conditions deviate to either the wetter or drier side of this optimum. Plant growth form, related to resource acquisition and use, determines the growth response of individual species to soil moisture ( Bowman et al. 1995 ). In Dryas otcopetala, for example, individual leaf weight and leaf biomass are reduced by supplementary water and the efficiency of water use, as measured by carbon isotope discrimination, appears strongly coupled to water availability ( Welker et al. 1993 ). Patterns of plant community distribution can often be clearly associated with soil moisture profiles, either within a habitat mosaic or along a moisture gradient ( Kincheloe & Stehn 1991; Squeo et al. 1993 ; Timoney et al. 1993 ; Cooper & Andrus 1994). Changing patterns of precipitation are likely to produce a re-assortment of these communities along these environmental moisture gradients ( Harte & Shaw 1995; Harte et al. 1995 ). As these communities differ in their productivity and resource requirements this has implications for total ecosystem productivity ( Bliss et al. 1981 ; Chapin et al. 1992 ). Individual species may often grow under a range of soil moisture conditions but their optimum productivity is limited by water availability and the frequency of drought conditions ( Nosko & Courtin 1995). Changes in plant cover are also likely to have a direct feedback effect on soil moisture by altering rates of evapotranspiration ( Holdsworth & Mark 1990; Kane et al. 1990 ; Plochl & Cramer 1995). By contrast, in wetter tundra communities, water movement through the soil can enhance productivity ( Chapin et al. 1988 ).

The composition and productivity of plant communities is also strongly influenced by the year to year presence/absence and persistence of winter snowpatches ( Talbot et al. 1992 ; Stanton et al. 1994 ; Ferrari & Rossi 1995; Scott & Rouse 1995). For some species snow cover provides protection from extreme winter temperatures but for others early release from snow cover and a longer growing season are more vital ( Sonesson & Callaghan 1991; Ferrari & Rossi 1995; Galen & Stanton 1995). Experimental increase in winter snow cover by snow fences can produce significant alteration in soil moisture and temperature conditions over time, leading to changes in plant community composition ( Scott & Rouse 1995).

Water and invertebrate communities

Many host specific insect herbivores depend upon particular host plants and their response to water availability is mediated through their host. Foliar water content is known partly to determine the palatability of leaves to insects ( Scriber 1977) and there is growing evidence that plant stress, induced by soil waterlogging or drought, can produce alterations in leaf chemistry, particularly the mobilization of soluble nitrogen. This in turn may increase leaf palatability, leading to improved insect growth performance and occasionally to population outbreaks ( Mattson & Haack 1987). In addition, micrometeorological conditions per se can have a effect on insect activity, although humidity usually plays a subsidiary role to temperature ( Bergman et al. 1996 ; Hodkinson et al. 1996a )

Soil animals and their corresponding functional roles are, by contrast, influenced more directly by soil moisture ( MacLean 1981; Petersen & Luxton 1982; Hodkinson et al. 1994 ; Leakso et al. 1995 ). Some groups such as nematodes are dependent for movement on water films in the soil ( Kuzmin 1992), others such as Enchytraeidae ( Sømme & Birkmoe 1997) and many Diptera larvae, particularly chironomid midges, are highly desiccation susceptible and ‘prefer’ wet soils ( Bengtson et al. 1974 ). Their abundance and vertical distribution within the soil alter in direct response to experimental moisture manipulation ( Briones et al. 1997 ). At the other extreme some Arctic cryptostigmatic mites, while tolerating moist conditions, are strongly desiccation resistant ( Hodkinson et al. 1996b ). Many animals such as Collembola actively move to track the changing moisture status of their habitat ( Hertzberg et al. 1994 ). Arctic and alpine soil invertebrate communities therefore are highly variable within the habitat mosaic and often reflect the moisture status of their immediate microhabitat, as well as the more general soil water characteristics ( O’Lear & Seastedt 1994; Klironomos & Kendrick 1995; Grossi & Brun 1997). Changes in soil water status, associated with experimental climate warming, have been shown to alter selectively the populations, biomass or diversity of the mesofauna ( Harte et al. 1996 ; Coulson et al. 1996 ) thus producing a shift in the functional composition of the invertebrate animal community.

Soil moisture also plays an important role in the overwintering strategies of many soil arthropods that actively reduce their susceptibility to freezing by desiccation ( Worland 1996; Bale et al. 1997 ). Avoidance of contact with free water prevents the external inoculation of ice crystals which can trigger the freezing of the animals body fluids. At the other extreme in wetter soils, encasement of soil animals in pockets within the ice can lead to conditions of anoxia ( Sømme & Conradi-Larsen 1977). Body water plays a critical role in the cold hardiness of terrestrial arthropods ( Block 1996). Partial desiccation can significantly effect the cold resistance of both freezing tolerant and freezing susceptible forms. An increase in the amount of unfreezable water in arthropods exhibiting both survival strategies may occur by the production of low molecular weight sugar alcohols (polyols) and by desiccation which promotes glycerol and trehalose synthesis in some species. Also partial dehydration in winter may reduce or mask ice nucleator activity. Thus cold and drought may not be competing mortality factors but may act as complementary adaptations which aid survival ( Block 1996).

Water and microbial communities

Direct studies on the population dynamics and community composition of the Arctic and alpine microflora and their response to water are difficult to conduct as numerical estimates of abundance are unreliable and it is often difficult to distinguish between active and inactive cells. Consequently, most experimental work focuses on microbial activity and function rather than on numerical abundance and biomass ( Mancinelli 1984; Panikov 1994; Robinson & Wookey 1997). The relationship between soil moisture and bacterial and fungal populations and/or biomass can, however, be inferred from correlations across sites of varying moisture status, although water emerges as just one of a group of intercorrelated variables. Tundra bacteria, for example, appear most abundantly at wet and warm sites of high nutrient status ( Holding et al. 1974 ). Similarly, total fungal mycelium length is positively correlated with a combination of soil moisture, temperature and pH. At very wet sites, however, fungal biomass is suppressed ( Dowding & Widden 1974; Miller & Laursen 1974). Within a given site the normal seasonal increase in mycelial growth is suppressed in dry years and precipitation can induce a rapid increase in hyphal growth within periods as short as 4 days ( Miller & Laursen 1974).

Much of the fungal mycelium in tundra soils belongs to sterile forms: fruiting fungi, including mycorrhizal forms, are relatively scarce. Water, nevertheless, is important in the life history of these fruiting forms: surface water movement actively disperses spores and high humidities prevent spore desiccation ( Dowding & Widden 1974; Gardes & Dahlberg 1996).

Water and ecosystem processes

Soil moisture is one of the key variables controlling the rates organic matter decomposition in tundra soils ( Nadelhoffer et al. 1992 ; Smith et al. 1993 ; Rastetter et al. 1996 ) through its impact on decomposer performance, as measured by parameters such as CO₂ efflux or microbial cellulase activity ( Smith et al. 1993 ; Nizovtseva et al. 1995 ). Rates of aerobic organic matter decomposition are again, like plant growth, an approximately parabolic function of soil moisture, becoming maximal at intermediate soil moisture levels ( Moorhead & Reynolds 1993). Variations in soil moisture or snow cover over the tundra landscape produce parallel patterns of variation in organic matter decomposition rates resulting from differences in the abundance and activity of decomposer organisms ( O’Lear & Seastedt 1994). Similarly, seasonal differences occur in decomposition rates linked to soil moisture as well as to temperature. Mineralization rates are enhanced by alternate wetting and drying of the soil that stimulates microbial activity ( Chapin et al. 1988 ). Interestingly, in wet tussock tundra, rates of litter (and N) loss are higher over the winter compared with the summer period, possibly resulting from significant losses associated with spring runoff over permafrost ( Hobbie & Chapin 1996). Freeze-thaw events play a significant role in the physical breakdown of litter in tundra environments where typical comminutors such as earthworms are absent ( Hobbie & Chapin 1996).

Vertical and lateral flux rates for the greenhouse gases CO₂ and CH₄ in tundra soils are closely linked to soil moisture characteristics ( Slobodkin et al. 1992 ; Howard & Howard 1993; Tenhunen et al. 1995 ; Weller et al. 1995 ; Moosavi et al. 1996 ). The relationship, however, is not simple. Waterlogging of the soil active layer and associated changes in the position of the water table relative to the position of the permafrost, controls the potential for aerobic vs. anaerobic decomposition, the rates of CO₂ or CH₄ production and the rates of accumulation or loss of organic matter ( Rouse et al. 1995 ; Gilmanov & Oechel 1995; Tenhunen et al. 1995 ; Johnson et al. 1996 ; Tenhunen 1996). Aerobic respiration, usually measured as CO₂ efflux, is normally positively related to temperature but this response can usually be modified by soil moisture ( Oberbauer 1991; Howard & Howard 1993). Respiratory Quotient (RQ) values may be higher at low soil moisture levels but the effects of moisture on metabolism may be less pronounced at lower temperatures ( Fischer & Bienkowski 1987; Fischer 1995). Beyond a particular moisture content, rates of CO₂ efflux are negatively correlated with soil moisture. A seasonal slowing of the rate of CO₂ efflux has similarly been noted in more xeric habitats and is associated with declining soil moisture levels ( Oberbauer et al. 1996 ).

Rates of methane efflux from tundra soil surfaces reflects the balance between production of CH₄ and its oxidation by methanotrophic microoranisms and is temperature dependent ( Torn & Chapin 1993; Vourlitis et al. 1993 ). Methane is a product of anaerobic decomposition in waterlogged soils or in saturated horizons within an active layer of a partially saturated soil profile. The sites and rates of methane generation therefore are related to the position of the water table within the active layer ( Vourlitis et al. 1993 ; Johnson et al. 1996 ). This layer itself may also vary in depth, producing differences in methane generation both amongst sites and seasonally within sites ( Torn & Chapin 1993; Vourlitis et al. 1993 ). Methane oxidization occurs throughout a nonsaturated soil profile but is often highest in the surface layer of the water ( Vecherskaya et al. 1993 ).

As organic matter is mineralised nutrients are released into the soil. Both the rate of mineralization ( Marion & Miller 1982; Robinson et al. 1995 ; Robinson & Wookey 1997) and the subsequent movement of nutrients in the soil are dependent on water. At certain levels of soil water, addition of extra water to tundra soils often stimulates the mineralization of N, although it may also stimulate microbial immobilization of N ( Van Gestel et al. 1993 ; Moorhead & Reynolds 1993; Binkley et al. 1994 ). Oxidation and reduction processes, coupled to varying soil moisture, govern the availability of some plant trace nutrients such as iron and manganese ( Marion 1996). Movement of water through soils can produce spatial and temporal gradients in nutrient concentrations ( Pecher 1994). Rates of soil nitrification, but not denitrification, are often linked to soil moisture, being higher in drier sites ( Stark & Firestone 1995; Chapin et al. 1996 ). By contrast, rates of nitrification by surface cyanobacteria appear to be positively correlated with soil moisture ( Chapin et al. 1991 ). Tundra soils are often regarded as nutrient ‘leaky’, with soluble nutrients being rapidly lost through runoff into streams ( Marion 1996).

Water and pollutants in tundra ecosystems

Despite often being remote from industrial sources, Arctic and alpine ecosystems are subject to continual inputs of a number of important pollutants, including heavy metals and acidifying oxides of nitrogen and sulphur ( Wadleigh 1996; Steinnes 1997; Woodin 1997). The presence of a pollutant haze within the Arctic atmosphere is well documented, with pollutant levels reaching a maximum in late winter and early spring ( Shaw 1995). Precipitation, falling as either snow or rainfall (or even as condensing mist), scavenges these compounds from the atmosphere and deposits them on the tundra surface. Deposition rates are thus determined by the rate and form of precipitation. Falling snow is a less effective scavenger than rain or mist and this produces seasonal differences in deposition rates, usually with the lowest deposition rates occurring in winter ( Woodin 1997). Pollutants once deposited can be stored in snow accumulations for later pulsed release ( Cheam et al. 1996 ), whereas those carried in rainfall will be dispersed immediately in soluble or particulate form, both by surface and subsurface processes. Changes in the patterns of precipitation are therefore likely to produce changes in the rates of pollutant input and dispersal and influence the effects of pollutants on the ecosystem.

Water, disturbance and tundra ecosystems

Water, through the mechanisms described above, plays a highly important part in creating and maintaining natural physical disturbances of the tundra landscape ( Walker & Walker 1991; Lynch & Kirkpatrick 1995 ). Glacierization, cryoturbation, solifluction, erosion, etc. maintain, over varying time scales, a natural level of disturbance that is unique to polar and montane ecosystems, preventing over large areas the development of mature soils and their associated plant, invertebrate and microbial communities. Changes in precipitation patterns, resulting from climate change, are therefore, likely to produce changes in natural disturbance patterns and to have important impacts on both the composition and functional aspects tundra communities.

Water, through its presence as permafrost, also plays a highly significant role in the way in which tundra landscapes respond to direct physical disturbance by man ( Truett & Kertell 1992). The interaction between permafrost and soil surface temperatures are particularly important. Slight temperature alterations produced by poorly insulated buildings or by vehicle or pedestrian tracks, which produce slight depressions of the tundra surface resulting in soil compression, lead to subsequent melting of the permafrost, subsidence and in extreme cases the development of thermokarst features ( Racine & Ahlstrand 1991; Kevan et al. 1995 ). Similar changes may occur alongside the route of laid roads ( Auerbach et al. 1997 ). Often these changes are associated with alterations in soil nutrient characteristics, vegetation composition and soil animal community structure ( Forbes 1992; Emers et al. 1995 ). Wet sedge meadows are particularly vulnerable and a single vehicle passage can produce channelling of surface water and initiate gully erosion ( Kevan et al. 1995 ).

Discussion

Climate change will almost certainly result in altered patterns of precipitation in Arctic and alpine regions but the precise changes are uncertain ( Callaghan & Jonasson 1995). Some areas will become drier, others wetter. There will be changes in the amount, seasonal distribution and form of precipitation. Changes in moisture availability will have both direct and indirect effects on ecosystem function and will have implications for the thermal regimes of tundra systems. Reduced winter snowfall will result in markedly lower soil temperatures during winter and reduced moisture input into the soil at snowmelt, coupled with perhaps an extended growing season as the absence of snow cover leads to more rapid soil warming in the spring ( Coulson et al. 1995 ). Drier soils will also warm more quickly during summer and this may affect the long-term integrity of the permafrost. Increased soil moisture may lead to greater evapotranspiration, increased cloud cover, reduced radiation input and an increased buffering against temperature extremes. Conversely, reduced cloud cover and drier conditions may increase the frequency of both high and low temperature extremes.

The Arctic and alpine water environment is thus exceedingly complex and interactive, making the precise effect of any alteration of water availability on ecosystem structure and function difficult to predict ( Hodkinson & Wookey 1998). The complex water relationships of the biological organisms and the rates of many ecosystem processes are ultimately determined by surface hydrology. Thus, there is a compelling need to integrate biological models of water use with hydrological models that describe the form, distribution and movements of water within the landscape with respect to both space and time ( Hinzman & Kane 1992). Tools to provide these necessary key data are now becoming available, e.g. satellite imagery used to measure the moisture content of the soil active layer ( Kane et al. 1996 ), etc. At present the resolution is crude (25 × 25 m) and the correlation with ground validation measurements is not particularly strong. Detailed ground mapping of soil moisture on a microtopographical scales is still required to resolve the spatial variation hidden within the broader landscape measurements. Ultimately, integrated biological/hydrological models, which use precipitation as a driving variable, will be needed to explore the complex ramifications of changing precipitation patterns on tundra communities. Until this approach is adopted water, despite its overwhelming importance, will remain a neglected and only partially understood factor in climate change studies of tundra ecosystems.