A. INTRODUCTION

The McMurdo Dry Valleys (MCM) represent the largest of the ice-free oases on the Antarctic continent (ca. 4800 km²). They are among the coldest and driest terrestrial environments on earth, with an average annual temperature of ca. -20°C and total annual precipitation of 6 cm, received as snow during winter. The dry valley landscape is a mosaic of perennially ice-covered lakes, ephemeral streams, soils and glaciers. Despite the extreme conditions, biological communities exist in these lakes, streams and soils.

Much of what is known about such extreme polar deserts has been discovered by work conducted in Antarctica because few comparable systems exist elsewhere. Much of our knowledge about the structure and function of these polar desert ecosystems has been discovered only recently, resulting, in large part, from the first phase of the McMurdo Long-Term Ecological Research program (MCM-I 1993-99; see Section 1. Results of Prior Support). In MCM-I, our research explored the physical constraints controlling the structure and function of this polar desert. We discovered that subtle changes in temperature, precipitation, and albedo have profound effects on the hydrologic cycle, biogeochemistry, productivity and biodiversity within the valleys. Moreover, local effects are modified by landscape position and topography.

The MCM ecosystem is sensitive to very small variations in climate because the change between solid and liquid phase of water is delicately poised in this environment. Thus, small changes in temperature and radiant energy regimes are amplified by large, non-linear changes in hydrologic budgets that can ramify throughout the system. The presence of liquid water remains the primary limiting condition for life in Antarctica (Kennedy 1993), so the relationship of energy balance to liquid water availability, ecological function and biological diversity will continue to be a major emphasis of the proposed McMurdo Dry Valley LTER renewal (MCM-II).

A second focus of MCM-I was generated by observations that biological activities in the dry valleys were affected by the transport of water, nutrients and organic carbon between landscape units (glaciers, streams, lakes and soils). Such linkages have been shown to considerably influence ecosystem structure and function (Turner and Gardner 1991), but are most evident in extreme environments, such as deserts (Schlesinger et al. 1990) and tundra (Shaver et al. 1990; Reynolds and Tenhunen 1996). Perhaps the best example of a critical linkage among landscape units is represented by the glacier-stream-lake continuum, in which water and solutes move from glaciers to lakes by streamflow (McKnight et al. 1998; Lyons et al. 1998). This movement of water defines the streams in which biological activity is restricted to periods of liquid water availability. Water in streams stimulates weathering of minerals in streambeds and transformation of nutrients by stream communities (Lyons et al. 1998; Moorhead et al. 1998). Thus , water reaching the lakes has a very different chemical signature than glacier melt water.

Within the context of an extreme sensitivity to variations in climate and significant linkages among landscape units, it has become clear that "legacies" of past events influence the structure and function of the modern ecosystem. Geomorphic evidence indicates that the MCM environment has been remarkably stable over the past few million years, with extremely slow rates of landscape modification in comparison to temperate regions (Brown et al. 1991; Wilch et al. 1993). In addition, nutrient cycling has been slow because biological activity is limited by the harsh polar environment. As a consequence, legacies of the past, such as pools of soil organic matter deposited by paleo-lakes, and nutrients in the hypoliminia of modern lakes, directly impact present ecosystem function.

The term "legacy" refers to the carry-over or "memory" of the ecosystem of past events (Vogt et al. 1997; Little et al. 1997). A climatic legacy strongly imprints current ecosystem functions at MCM, including inundation of the entire Taylor Valley by Lake Washburn during the late Pleistocene (24,000-6,000 yr BP; Denton et al. 1989), and a subsequent cold, dry period ending about 1000 yrs BP (Lyons et al. in review a, (Table 2.1). Legacy effects at other LTER sites usually are viewed as much shorter-term consequences of historic land-use patterns, such as forest clearing (HBR, HFR, CWT, AND), grazing and resultant desertification (JRN, SEV, SGS), and conversion to agriculture (KBS), or resulting from severe weather (e.g. hurricanes, LUQ) and fires (KNZ, AND).

B. WORKING HYPOTHESES

The interrelated systems of climate, biota and physical landscape change naturally over time. The polar regions have long been considered areas where changes in climate are "amplified" (Chen and Drake 1986). In MCM-I, we hypothesized that "the structure and the function of the Taylor Valley ecosystems are primarily controlled by physical constraints." These constraints dealt primarily with the modern environment of the MCM. The small changes in climatic conditions (T and precipitation) over the past 24,000 yr led to dramatic changes in ecologically significant features of the landscape (Doran et al. 1994). These inherited landscape features, in turn, continue to affect the biological activity and diversity and the important biogeochemical processes occurring in the MCM ecosystem today. This "amplification" of small changes in climatic variables and the resultant landscape changes are the ecological legacy of MCM. Figure 2.1 presents our new conceptual model of MCM.

Central Hypothesis: Past climates in polar desert environments strongly overprint present ecosystem structure and function.
Results of MCM-I research suggest that "memory" exists in all landscape units in the dry valleys. A hydrological memory exists in lake and glacier volumes, a chemical memory resides in lakes, a geomorphic memory controls stream processes and an organic carbon memory is evident in soils; the approximate ages of these memories are shown in (Fig. 2.2). In addition, biological memory is manifest in the distribution of present-day organisms across the landscape (spatial pattern of biodiversity). Our working hypotheses are derived from three dominant legacies that are now apparent in the valley ecosystem (lake, stream, and glacier characteristics; pools of organic and inorganic compounds in lakes and soils; distribution and diversity of biota). The impacts of these legacies on modern ecosystem processes are determined partly by the linkages between the soils, streams, lakes, and glaciers, cast within an overall context of the modern climate.

FOCUS 1. HYDROLOGY>
Climate controls the hydrology of this polar desert, providing water through the melt of ice and snow. Liquid water is necessary to support biological and geochemical processes, and defines important linkages among landscape units. Past climate regimes have determined many attributes of the current landscape by influencing lake levels. Periglacial and fluvial processes during past regimes have determined the geomorphic features of current streams. Understanding the structure and function of the dry valley ecosystem requires understanding hydrological responses to climate, both now and in the past. For these reasons, we propose to continue our studies of climate controls on the hydrology. We also propose to expand our studies to include an evaluation of past climatic effects on lake level fluctuations, which resulted in several important legacies.

Hypothesis 1a (Glacier Melt). The variability in timing, duration and amount of water derived from glacial melt is determined by the interaction of climate (temperature, radiant energy, snow cover) and landscape position of glaciers (slope, aspect, elevation).

Hypothesis 1b (Streamflow). The variability in timing, duration and amount of stream discharge is controlled by glacier melt and by long-term periglacial and geomorphic processes.

Hypothesis 1c (Legacy). Subtle changes in temperature and precipitation have determined the hydrologic balance of lake levels over the last 24,000 yrs through direct control over mass balance of glaciers providing melt water.

FOCUS 2. BIOLOGICAL ACTIVITY AND DIVERSITY
The distribution and activity of biota in the dry valleys are determined largely by the availability of water. Results of MCM-I indicated that other factors also constrain biological activities (e.g. temperature, nutrients, energy, etc.), but that linkages among landscape units can ameliorate some of these limitations. Legacies of nutrients and organic carbon that have accumulated through past biological and geochemical processes within ancient hydrological regimes may also drive modern communities. Thus, current patterns of biological activity and diversity partly reflect past distributions of water, nutrients, organic carbon and biota.

Biological activity in streams primarily is limited to microbial mats that are found on stable substrates (Algers et al. 1997; McKnight et al. 1998), which are active when liquid water is present and nutrient availability is adequate. Primary production in lakes is limited by available light and nutrients, but sources of nutrients differ with respect to location in the water column. Surface waters receive nutrients from streams flowing into lakes, whereas phytoplankton in deeper water rely on the upward diffusion of ancient nutrient pools from deep water and sediments. Also, the upward diffusion of ancient DOC from deep water and sediments apparently results in the current, negative values of net primary production observed in phytoplankton communities. The activities of soil communities are limited by a number of factors, including temperature and the availabilities of water and organic carbon. Sources of organic carbon include in situ production, modern transport from remote sources, and ancient carbon residing in soil profiles.

Hypothesis 2a (Stream Biota). The distribution, abundance, community structure and activity of stream biota are dictated by the timing, duration and amount of water derived from glacial melt, in conjunction with the stability of streambed habitat created by periglacial processes.

Hypothesis 2b (Lake Biota). The distribution, abundance, community structure and activity of lake biota are controlled by the water balance and the availabilities of radiant energy, ancient and modern nutrient and dissolved organic carbon.

Hypothesis 2c (Soil Biota). The distribution, abundance, community structure and activity of soil biota are controlled by allochthonous inputs of organic carbon from modern and ancient sources, within the context of favorable soil moisture, temperature and salinity regimes.

FOCUS 3. BIOGEOCHEMICAL PROCESSES
Biogeochemical processes responsible for the immobilization and mineralization of nutrients, and geochemical weathering, require a medium of liquid water. Hence, the timing, duration and location of biogeochemical processes in the past and the present are controlled by water availability. Also, hydrologic linkages among landscape units provide both a medium for these processes and a conduit for transport of chemical moieties. It is for these reasons that legacies of carbon, nutrients and other chemicals are present throughout the modern dry valley landscape.

Hypothesis 3a (Stream Chemistry). The chemistry of stream water is dictated by the timing, duration, amount and chemistry of water derived from glacial melt, in conjunction with stream location, geomorphology, mineralogy of streambed materials, and biota.

Hypothesis 3b (Lake Chemistry). The chemistry of lake water is dictated by the (1) timing, duration, amount and chemistry of stream water, (2) biogeochemical processes occurring within the lake, and (3) retention of past inputs and products of biogeochemical processes by the lack of hydrologic flow through stable stratification and restricted interaction with the atmosphere.

C. BACKGROUND

Most of Antarctica is covered by ice that may exceed 3,000 m in depth, but the McMurdo Dry Valleys (76°30' to 78°30'S, 160° to 164°E) form the largest ice-free expanse on the continent (ca. 4800 km²). The MCM are among the most extreme deserts in the world, far colder and drier than any other LTER site (Fig. 2.3). They contain perennially ice-covered lakes, ephemeral streams, glaciers descending from the encompassing mountains, extensive areas of soils and expanses of exposed bedrock.

Taylor Valley is the main research area for the MCM-LTER and is composed of four main basins (Fig. 2.4). The Lake Fryxell and the Lake Bonney basins are enclosed basins and are the two largest watersheds in the valley. They are divided by the Nussbaum Riegel, a hill of ca. 700 m elevation. The Lake Hoare basin also is an enclosed basin and is formed by the Canada Glacier, which blocks flow to Lake Fryxell. The Commonwealth watershed is adjacent to the coast and all streams in this basin flow to the ocean (McMurdo Sound).

The biological diversity of MCM is relatively simple, represented almost entirely by microbiota (no higher plants or animals) that comprise relatively short food chains. Cyanobacteria, bacteria and yeasts are abundant throughout the MCM. Many species of eukaryotic algae occur in the lakes and streams, microzooplankton are present in the lakes, and nematodes are found in the soils. Despite this taxonomic simplicity, the basic ecological processes of primary and secondary production, decomposition, nutrient cycling, and energy flux within the MCM demonstrate complex interactions among biological entities, as well as between biota and abiotic factors of the environment. The extreme environment, coupled with a paucity of higher trophic groups, implies strong "bottom up" controls on ecosystem structure with minimal influence of "top-down" controls exerted by predation common to other environments (e.g. Carpenter 1988). Thus, MCM offers a natural laboratory for study of ecosystems and basic ecological processes wit hout the complications introduced by a plethora of higher organisms.

C.1. Streams
MCM streams flow for 6-12 weeks per year, between late November and early February. In addition to current active channels, relict channels associated with past advances of glaciers or with fluvial sediment deposition can be observed throughout MCM. Annual variations in flow and duration are related to landscape position (Lyons et al. in review b), with streamflow in Bonney basin (Fig. 2.5) being more stable than in Fryxell basin, due to the microclimatic variations between the basins (Fountain et al. 1998). Increases in solute concentration from the glacier source to the lake outlet can be several orders of magnitude, and stream chemistry varies in relation to stream length (Fig. 2.5a), landscape location and the abundance of algal mats (Lyons et al. 1998). The longer streams in Fryxell Basin have higher concentrations of silicate weathering products, such as H4SiO4 and K+, and streams entering Lake Bonney have higher marine aerosol salts components, such as Cl- (Fig. 2.6). Moreover, streams with abundant al gal mats are depleted in nutrients (McKnight et al. in review a; Moorhead et al. 1998).

In many streams, dense mats of filamentous cyanobacteria grow during summer streamflow and are freeze-dried during the rest of the year. Abundance ranges from extensive, blanket-like coverings on rocks and streambed to sparse patches restricted to the margins of active channels or on the undersides of rocks (Vincent 1988; McKnight et al. 1998). The cyanobacterial mats are composed of an upper layer, enriched in accessory pigments, and an under layer with greater chlorophyll-a content (Vincent 1988). Mats that have remained desiccated through the winter can begin photosynthesizing within 20 minutes of being wetted (Vincent and Howard-Williams 1986) and mats that have been desiccated for years to decades can achieve high biomass and rapid growth within 2 weeks (McKnight et al. in review b). Mosses occur on the streambanks, where they are wetted for short periods or in flat areas that become damp (Alger et al. 1997).

Algal mats are most abundant at sites which have moderate gradients and streambeds composed of large cobbles arranged in a flat stone pavement (McKnight et al 1998; Alger et al. 1997). These stone pavements have been created by periglacial, freeze-thaw processes probably acting over centuries. Abundance is lower at high gradients and on deltaic deposits. Black, orange, and green algal mats occur at most sites (Fig. 2.7). Black mats are found near the channel margins and are composed of Nostoc sp., and green mats are found on the underside of rocks and are composed chiefly of Prasiola calophylla or P. crispa. Orange and red mats occur in flowing water, either in the main channel or in parafluvial seeps draining at the stream margins, are composed of species of Oscillatoria and Phormidium and are rather diverse in composition.

In streams with abundant mats, the assimilation capacity for nitrate and phosphate is very high resulting in low nutrient concentrations (McKnight et al. in review a). However, streams with sparse mats have higher nutrient concentrations. Gross rates of photosynthesis in microbial mats of the streams approach an upper limit of 4-8 g/Cm/h which increase with temperature (Hawes and Howard-Williams 1998). In the thicker mats, gross photosynthesis essentially is equivalent to respiration, and the development of high biomass in some locations is hypothesized to require a lack of physical disturbance (Hawes and Howard-Williams 1998).

C.2. Lakes
Despite their proximity, the lakes in Taylor Valley differ in many respects. The differences in geochemistries of the bottomwaters imply distinct histories and/or different sources of solutes (Lyons et al. in review a), i.e. bottomwater of Lake Hoare is fresh, Lake Fryxell is brackish, and Lake Bonney is hypersaline (Fig. 2.8). Lyons et al. (1998) noted that surface waters from Lake Hoare and Lake Bonney would evolve to Na(+), Mg(2+), SO4(2-), Cl(-) - rich waters upon sublimation and/or evaporation, while Lake Fryxell would evolve to Na(+), HCO3(-), and CO3(2-) - rich waters.

The MCM lakes have abundant planktonic and benthic microbial populations (Vincent, 1988; Lizotte and Priscu 1998). Benthic microbial mats primarily are composed of bacteria, cyanobacteria, pennate diatoms and eubacteria (Wharton et al. 1983). Differences in biomass and net primary production occur between the lakes, presumably as a result of differences in nutrient availability, light intensity caused by lake ice thickness, and grazing losses (Priscu, 1995). Maximum rates of primary production coincide with peaks in phytoplankton biomass, often occurring where favorable light regimes overlap zones where nutrients are diffusing from deeper water (Fig. 2.9). Gross primary production rates vary markedly between the lakes, with deep photosynthetic maxima of ca. 10 gC/L/d at 8 m in Lake Fryxell and at 13 m in the west lobe of Lake Bonney, respectively. Profiles from Lake Hoare and the east lobe of Lake Bonney generally decrease with depth, from maximum values < 2 gC/L/d near the surface. Respiration, estimat ed by ETS activity, is highest in Lake Fryxell, reaching a maximum rate just below the oxic/anoxic boundary. Lake Fryxell also contains the highest DOC values of all the lakes (McKnight et al. 1991, 1993).

Prior to MCM-I, little was known about the planktonic food web in dry valley lakes. Primitive rotifers constitute the only metazoa present in the plankton, but flagellates and ciliates are abundant and their abundance differs dramatically among lakes (Laybourn-Parry et al. 1997; James et al. 1998). Large interannual variability exists, probably resulting from variations in streamflow and food availability driven by meteorological conditions. The planktonic food web consists of phytoplankton, bacterioplankton, heterotrophic nannoflagellates (HNAN), ciliates and rotifers (Fig. 2.10). Mixotrophy is an important mode of metabolism and provides a mechanism for over-winter survival (and overall biodiversity) of certain phytoplankton species (Roberts unpublished; McKnight et al 1998). Grazing by ciliates, rotifers and nannoflagellates and settling of filamentous cyanobacteria reduces planktonic biomass during the six months of winter darkness (Lizotte et al 1996; McKnight et al 1998).

Because these stream-lake systems are unusual in their biogeochemistry, species distribution and evolutionary history, they serve as interesting comparisons to more temperate systems. Because of the minimal amount of terrestrially produced organic matter entering these systems (Aiken et al. 1996), unlike streams at other LTER sites (e.g. HBR, CWT, KNZ, NWT, AND), they serve as important end-members in the understanding of carbon dynamics in aquatic systems.

C.3. Soils
Many ecosystem processes are mediated by soil organisms (Freckman et al. 1997; Brussaard et al. 1997). For example, they control rates of organic matter decomposition and, consequently, regulate the mass of stored carbon. Despite their importance, the ecological and edaphic factors controlling the distribution and populations of soil fauna are poorly known (Hendrix et al. 1986; Freckman and Caswell 1985; Freckman 1994). Research on soil communities with low diversity, such as in Antarctica, provides an opportunity to elucidate relationships between species diversity, physical and chemical factors, and ecosystem functioning that otherwise are masked by the overwhelming complexity of soil biodiversity found in most terrestrial ecosystems.

Antarctic soils are the oldest, coldest, and driest on Earth (Pastor and Bockheim 1980; Bockheim 1997) and share similarities with the arid soils of the JRN and SEV LTER sites. The soils found in the MCM are poorly developed, coarse-textured, and often have high salinities. They also have the lowest organic carbon and biological activity of any soils on this planet (Campbell and Claridge 1987; Campbell et al. 1998). The physical and chemical environment of Antarctic soils varies along gradients of temperature, moisture, organic matter content created in part by legacies of past climates (Campbell and Claridge 1987).

No other soil systems are known to exist in which nematodes represent the top of the food chain and where food webs are as simple in structure. The nematode community in MCM soils consists of an endemic species Scottnema lindsayae (Timm 1971), a microbial feeder (bacteria and yeast), Plectus antarcticus, a bacterial feeder, and Eudorylaimus antarcticus, an omnivore-predator (Freckman and Virginia 1991). Because nematodes are aquatic animals, moisture is a more important factor for survival in Antarctica than low temperature (Kennedy 1993). Moisture from melting snow and streams is available to soils only intermittently , so organisms must be capable of prolonged survival with limited moisture and temperatures below freezing (Cameron and Conrow 1969). Nevertheless, nematodes are ideally suited for survival in this extreme environment. They can enter a survival state, anhydrobiosis, for extended periods. This effectively decouples them from the nutrient cycle, which may contribute to the extremely slow rate of nutrient and carbon exchange in the MCM.

C.4. Ecological Modeling
We are aware of no models developed to simulate ecological processes within the Antarctic dry valleys prior to MCM-I. Indeed, few ecological models exist for any polar desert. Moreover, many of the data necessary to develop such models were seldom collected before MCM-I, such as concurrent measurements of biomass (providing state variables), biological activities (relevant equations and parameters) and environmental controls (the driving functions). Thus, ecological models developed during MCM-I represent some of the first efforts to capture the dynamics of polar desert ecosystems in formal mathematical models. As mentioned above, our initial view of the dry valley ecosystems was that biological communities were controlled primarily by temperature and hydrologic regimes, as they determine the availability of liquid water. However, energy availability likewise controls the location, structure and activity of dry valley communities (Moorhead and Priscu 1998). Thus, the presence and activity of biota are restric ted to those times and places with a favorable juxtaposition of energy and moisture regimes.

Results of modeling studies of benthic mats in MCM streams and lakes suggest a positive net annual primary production (Moorhead et al. 1997a, 1997b). Ongoing studies of phytoplankton production are more equivocal (Davis 1998). Patterns of secondary production are even less certain.

Autotrophy in soil systems has not been examined in detail, although soil respiration and populations of soil fauna respond to experimental manipulations of microclimate and additions of substrate. Simulations of population dynamics of soil nematodes demonstrate enormous uncertainty in model behavior due to uncertainties in microclimatic regimes within soils. Moreover, the lack of correlations between physicochemical attributes of soils and the distributions and abundances of soil organisms has made it difficult to verify the primary environmental controls on these communities (Powers et al. 1995; Freckman and Virginia 1997, 1998).

Patterns of secondary production are difficult to determine for MCM communities because of low abundances of organisms comprising food webs. Respiration rates usually are lower than potential photosynthetic rates for benthic mats in streams (Hawes and Howard-Williams 1998) and this may be the case for microbial mats in lakes (Moorhead et al. 1997b). Studies also demonstrate extremely low rates of respiration in plankton communities of the lakes (Lizotte and Priscu 1992), although net annual respiration may exceed photosynthesis. Trophic webs in planktonic and soil communities consist of few taxa, many with low abundances (Powers et al. 1995; Freckman and Virginia 1997, 1998; Laybourn-Parry et al. 1997; James et al. 1998).

C.5. Legacies
Recent syntheses of studies confirm that many MCM landscape do not exist in steady state. These disequilibria appear to be persistent, natural phenomena, manifested in ways that relate to patterns of biological activities. Present work linking climate to glacier mass balance, streamflow, lake ice thickness and lake volume (Fountain et al. 1998) will provide many of the drivers for ecological models, and serve as the basis for interpreting long-term ecological responses to changes in climate over past millennia. In addition, the spatial heterogeneity in current landscape patterns (i.e. distribution of former lake sediments and stream channels) resulting from past climates, influence the modern biotic communities (Miller et al. 1997). For example, one hypothesis of MCM-I was that a principle source of soils organic carbon was from ancient lakes in Taylor Valley, which existed ca. 10,000 yrs BP. Stable isotopic signatures of soil organic matter (Fig. 2.11) confirmed this hypothesis (Burkins et al. 1997). Thus, m odern food webs in the soil are supported by carbon that was fixed long ago, similar to Arctic ecosystems in which peat may fuel a significant component of aquatic food webs. This relationship between past autotrophy and present heterotrophy is controlled by long-term hydrological regime (lake level) driven by fluctuations in streamflow and climate. Thus, the MCM behave in a similar function to the other LTER sites in that major ecological links occur where terrestrial and aquatic systems meet (Swanson et al. 1982). In the dry valley environment, however, an important element of this linkage is temporally separated.

D. MCM-II RESEARCH PLAN

Our hypotheses provide a basis for the direction and organization of the research of MCM-I. A multi-faceted approach is required to address our hypotheses, including monitoring, experiments, and ecological models. The following research plan outlines these components of our proposed research and the hypotheses.

D.1. Overview of Research Plan to Address Hypotheses
Our hypotheses direct and organize MCM-LTER and their relation to the research plan are discussed below and summarized in Table (2.1A).

D.1.a. Hydrology (focus 1). The production, transport, distribution and accumulation of liquid water is the key driving variable of the MCM ecosystem. Hypotheses 1a-1c address the role of meteorology variation (both spatial and temporal) in the distribution of liquid water throughout the various landscape components of the ecosystem. The relationship of climate to glacier mass balance to streamflow to lake-level rise and fall is the key to understanding the current ecosystem as well as legacies left from the past. The monitoring of climate (i.e. meteorological data), glacier mass change, streamflow and lake level are integral to our understanding of ecosystem dynamics. We are therefore investing a considerable amount of resources in understanding the dynamics of water in the system.

D.1.b. Biological Activity and Diversity (focus 2). The distribution of biological activity is related to current and past distribution of liquid water in MCM. Hypotheses 2a and 2b address the importance of present day hydrologic regimes in controlling biomass, primary production and decomposition in the streams and lakes of MCM. Hypothesis 2c focuses on the role of past hydrologic regimes and present-day soil chemistry influencing biological activity and diversity in the soils of MCM. The occurrence and utilization of soil organic carbon today is a legacy (and, hence, a direct linkage) to past hydrologic regimes in the MCM.

D.1.c. Biogeochemical Processes (focus 3). Biogeochemical processes and the cycles of nutrients and carbon they produce are the linkages between physical components and the biological ones in the MCM. In our conceptual model, biogeochemical processes are also linked to the past through legacy effects (Fig. 2.1). The transformation of nutrients and carbon is brought about by biological activity. Transport of these chemicals is related to the movement of water, present and past. Therefore, our studies of biogeochemical processes in the MCM represent an integrative activity. The work outlined below addresses these biogeochemical processes and hypothesis 3a and 3b. Streamflow measurements and stream chemistry will be used to determine the flux of important biogeochemicals from the glaciers to the lakes. Biological measurements in the streams will delineate the influences of biotic factors on these fluxes.

D.2. Long-Term Monitoring
The core MCM-I monitoring program will be continued with sampling protocols and frequency adjusted based upon the results obtained so far. We will augment the program to include newly required parameters; these augmentations are highlighted below. Methods are presented in the following section.

D.2.a. Meteorology (Hydrology). All LTER sites acquire continuous meteorological data. For MCM-I, a network of 8 meteorological stations are operated in Taylor Valley (Doran et al. 1995). Valley-floor stations are near the edge of Lakes Bonney, Hoare, and Fryxell, and at the mouth of the valley. Four other stations are located on the Taylor, Howard, Canada, and Commonwealth Glaciers. To achieve a broad climatic perspective of the MCM, stations are located at Lake Vida, Lake Vanda and Lake Brownworth in Victoria and Wright Valleys (Fig. 1.2).

Snowfall measurements. For MCM-II, we need accurate measurements of the frequency and magnitude of precipitation because of its dramatic effects on glacier melt, streamflow, light levels in lakes (by transmission through surface ice and snow) and soil biota. We will make manual measurements of snow fall at each occupied camp throughout the field season and install precipitation gauges on each meteorological station. We will also install a new station at 1000 m elevation on Howard Glacier to obtain a better indication of the vertical gradient in climate within the dry valleys. This work will be the responsibility of Dr. Fountain.

D.2.b. Glaciers (Hydrology). Because glacial melt is the source of water to streams and lakes, we will continue to monitor glacier mass balance. Measurements of surface energy balance will continue because glacial melt is sensitive to small changes in radiant energy, temperature and snowfall. The relation between surface energy balance and glacier mass balance is a driving force for hydrological memory. This work will be the responsibility of Dr. Fountain.

Glacier mass balance. We will measure the surface energy balance to partition mass loss between melt and sublimation-evaporation. Ablation stakes and bamboo poles provide only total mass change. Because sublimation-evaporation accounts for 40-90% of the mass loss and inherent measurement errors occur with typically small magnitudes of change in mass balance (~10 cm/yr), an alternative method is needed to estimate melt. To provide daily estimates of melt needed for comparisons to streamflow data, we will develop models of continuous glacial melt based on energy balance measurements.

Slope and aspect influences on mass loss. Large differences in mass loss from the vertical ice cliffs (~ -35 cm/yr) and the sub-horizontal glacier surface (~ -10 cm/yr) suggest that slope and aspect control mass loss on glacial surfaces. A small-scale network of stakes (placed on Canada Glacier in 1996-97) will be followed to measure variations in ablation caused by differences in slope and aspect. These results will help extrapolate the coarser-scale measurements of glacier mass loss to different areas of a glacier and to different glaciers.

D.2.c. Streams. Streams are the hydrologic links between the melting glaciers and the lakes. Small climate variations affect glacial melt and produce spatial and temporal variations in streamflow and chemistry. In-stream processes, including solute exchange with the hyporheic zone, sediment and chemical modification of the water by the geologic substrate, mosses and algal mats, determine the chemistry of water entering lakes. Moreover, the ecology of the streams is a dynamic component of the MCM ecosystem. Thus, we will continue to monitor streamflow, stream chemistry, long-term changes in the distribution, and abundance and growth of stream biota. This work will be the responsibility of Drs. McKnight, Priscu and Lyons.

Stream gauge network (Hydrology). During MCM-I, a network of 15 recording stream gauges was operated in Taylor Valley. The gauges provide a continuous record of discharge. Continuous measurements of temperature and conductivity are included at half the gauges to allow estimates of annual solute load to the lakes. The stream chemistry has proven to be extremely variable (Lyons et al. 1998), so we will continue this monitoring program for MCM-II. Although we attempted during MCM-I to relate solute concentrations to discharge (Lyons and McKnight unpublished), the years of MCM-I were extremely low-flow years compared to the late 1980s and early 1990s (reference USGS web site). Given inherently high variation in flow rates and stream chemistry, and the relatively slow response of lake chemistry to changes in stream inflow, more data are needed to establish a quantitative relationship between the chemistry and discharge in these stream/lake systems. We have begun to measure both suspended sediment load (SSL) and particular organic matter in the streams. This will continue during MCM-II to quantify sediment and allochthonous carbon inputs to the lakes.

Stream transect network (bioactivity and biodiversity). For MCM-I, stream algal mats were studied for transects at 16 stream sites in Taylor Valley in January 1994 and 1998 (Alger et al. 1997; McKnight et al. 1998). These sites include 10 locations in the Fryxell basin and 3 locations in each of the Hoare and Bonney basins. The mats were characterized for pigment content, algal species counts, particulate organic C, N and P content, and location areal of coverage of streambed and position relative to thalweg and stream margins. Comparison of results from resampling these sites in 1997-98 with results in 1993-94 shows that four years of low flow have promoted growth and development of algal mats, especially in stream reaches which were uninhabitable during previous years of high flow. The lateral extent of algal mat coverage contracted at some sites due to lower flows and mosses moved inward and colonized patches of Nostoc mat. For MCM-II, we will continue monitoring of mats at these transect si tes and will maintain a schedule of every 3-4 years. In the event of a high-flow year, we will resample the following year to document persistence or loss of algal mats. The monitoring program of MCM-I included measurements of primary productivity at transect sites. However, the variation among replicate samples from the same mat type in a transect site was generally equal to the variation for that mat type from different transect sites. These measurements will only be continued for a few sites. Gradual changes in extent of coverage and species distribution is a more useful metric for long term productivity in these streams.

D.2.d. Lake Water Column. In order to have two basins influenced by continental climate and two by the maritime climate, we monitor both the east and west lobes of Lake Bonney, as well as Lakes Fryxell and Hoare. Each lake has distinct physical, chemical and biological characteristics which can change over time in response to streamflow inputs of water, major ions and nutrients (Priscu 1995; Lyons et al. 1998). We have added measurements of microzooplankton to address Hypothesis 2b (at the suggestion of the MCM-I advisory committee). Selected parameters and sampling intervals are listed in Table 2.2 and are collected during routine limno runs at each site. In context of Hypothesis 2b, the lake biota are active throughout the year (unlike stream and soil biota) and upward diffusion of nutrients and DOC continues year round. In the event of opportunities for year around field activity from McMurdo Station, we would propose to conduct a lake monitoring program during a spring/winter/fall period on MCM-II This work is the responsibility of Drs. Priscu, Lyons, Doran and McKnight.

Physical and chemical characteristics (Hydrology and Biogeochemical processes). Lake levels are measured both intermittently and continuously. Lake ice measurements include thickness, the piezometric water level, surface ablation and temperature (cf. Chinn 1993). Thickness controls light transmission to underlying waters. Ablation of lake ice is the principle path of mass loss from the lakes because they have no outlets. These measurements will be related to glacier melt and stream discharge data to better quantify the water flux and balance through landscape units (hypotheses 1a,b,c).

Photosynthetically active radiation (PAR) is a primary driver for phytoplankton photosynthesis, so we will continuously monitor underwater radiation at depth. Underwater profiles will be made on monthly intervals (coincident with primary productivity measurements). During limno runs, temperature, conductivity and in situ fluorescence (a measure of chlorophyll a) will be measured with a Seabird model 25 CTD probe. We will collect water samples from the entire profiles for major ion chemistry at least two times each yearearly in the season and in late January. Because the top 10 m of these lakes are so responsive to hydrologic variation, major ion chemistry samples will be collected from the surface 10 m during all limno runs. Ecologically important solutes, such as DIC, nutrients and DOC, will also be collected at all depths during all limno runs.

Plankton abundance and productivity/(Bioactivity and Biodiversity). Because phytoplankton exist in narrow bands in the lakes (Spaulding et al. 1994; Priscu 1995), both discrete and high resolution in situ profiles are necessary for determination of phytoplankton biomass. Phytoplankton species distribution will be monitored to understand variations between years and between lakes. Measurement of bacterial abundance and heterotrophic activity will also be continued. The abundance of protozoa will be added to MCM-II to obtain a complete picture of the changes in the lake ecosystems in response to climate. Detailed studies of grazing by ciliates and nannoflagellates conducted in MCM-I will allow for interpretation of these data to assess the response of the changing conditions.

D.2.e. Lake Microbial Mats (Bioactivity and Biogeochemistry). Microbial mats in Lake Hoare are accumulators of carbon and nutrients. Our studies during MCM-I indicate that mat primary production rates are equal to the rates in the top 15 m of the water column and that mats influence the overall lake biogeochemistry. We will examine changes in C, N and P content of specific mat layers. Because the physiologically active layer contains photosynthetic pigments which may vary with depth in the mat, chlorophyll a, phycoerythrin and accessory pigments analyses will be used to determine layers with potential photosynthetic activity and provide information on the species present (e.g. Lizotte and Priscu 1998). This activity will be the responsibility of Dr. Hawes, Priscu, and Doran.

Lake mat transects. Our monitoring program for the benthic mats will focus on transects at permanent sites in each lake, extending laterally from just under the ice to beyond the maximum depth of colonization. We will study Lake Hoare primarily, but will sample other lakes in Taylor Valley on an every 4 year schedule comparable to the stream site monitoring. Measurements will match those of stream mats: pigment content of upper layers, algal species counts, and long-term record of mat vertical growth by measuring height with reference to fixed points, and particulate organic C, N and P content of specific mat layers. We will focus on static variables and will be able to compute temporal changes, providing net rates of growth or loss. We will measure instantaneous changes in primary productivity and respiration on the cohesive upper layer of cores at selected sites. The role of protist grazing in benthic mats is unknown and may be an important pathway in carbon and nutrient cycling. In MCM-II, we will enumerate protists from representative mats. To place the lake mat community into our overall legacy concept, we will collect samples from each of the lakes for natural isotopic abundance of carbon and nitrogen.

D.2.f. Soils. MCM-II will study soil ecosystems in a comprehensive manner. Continuous measurements of soil microclimate at study sites in the Hoare, Bonney and Fryxell basins will include PAR, soil temperature at several depths and air temperature at 5 cm (inside chambers), and at 1 m above the surface; soil moisture and salinity will be measured at several depths. This activity will be the responsibility of Drs. Wall and Virginia.

Elevational Transects (Bioactivity and Biogeochemistry. Soils in Taylor Valley represent a range of dry to moist habitats, differing soil types and salinities and varying geomorphic features at varying elevations. There are no soil maps. Because material transport, on an annual basis, influences soil chemical and physical properties as well as biotic communities, in MCM-I, we established a transect at three elevations near Lake Hoare to monitor soil properties and biological communities on a long-term temporal scale. The three 10m² plots on this transect were selected based on geomorphological features and they have distinct differences in climate, soil moisture, soil properties and soil nematode diversity. We will continue to monitor this transect every three years, looking for long term patterns of soil community change as a function of climate. Results from MCM-I indicate that biodiversity and attributes of the soil environment are more spatially variable in the MCM than in most other ecosystems (Freckman and Virginia 1998). To extend our understanding of spatial variation in soils and biota essential for regionalization, we will examine spatial variation in soil properties and biota along replicated 70 m transects located at sites in the Bonney, Hoare and Fryxell basins. Samples (n = 36) will be collected in a nested scheme along each line to minimize the number of samples required to minimize spatial autocorrelation from cm to m scales (Robertson and Gross 1994; Cavigelli et al. 1995). Semivariagrams for nematodes and soil properties (organic C, NO3(-), pH, conductivity, moisture) will be used to determine if biota and soil environmental factors vary over similar spatial scales in the dry valleys. Our results will be compared with similar studies conducted on another arid LTER site (JRN), where the impact of desertification on soil heterogeneity is a central focus (Schlesinger et al. 1990).

Material Transport (Bioactivity and Biogeochemistry). Aeolian transport of organic material and viable biota, such as anhydrobiotic invertebrates (nematodes, rotifers and tardigrades), protozoa and microbial propagules, may be an important link between streams, lakes and soils, as well as being a means of establishing soil and stream invertebrate communities. We propose to continue monitoring wind transport of organisms and material using 2 methods tested during the past two years by Wall and Virginia. Tube pans weighted with marbles will be placed at various locations throughout the valley and on soil near streams, and on lake ice, and left over winter. In the spring, the pans will be collected, contents placed in a sterile bag and subsamples of materials examined for organisms using sterile methods. Material collected will be analyzed for chlorophyll a to assess production. Remaining subsamples will be frozen for other investigators to use for identification. The second approach uses air part icle collectors (made of Frisbees attached to a collecting bottle) successfully employed by the British Antarctic Survey and ourselves, which will allow cross-Antarctic comparisons of wind inocula to soils and aquatic systems throughout Taylor Valley (BAS; BIOTAS Handbook, 1997).

Biotic Effects on Soil Decomposition, Carbon and Nutrient Bioavailability (Bioactivity and Biogeochemistry. Increases in soil temperature and moisture may change the bioavailability of essential elements by altering solubility and diffusion rates in soils, or by changing the amounts of organic compounds (responses to increased productivity or increased grazing of soil invertebrates on bacteria). We will establish a long-term experiment in the Bonney, Hoare and Fryxell basins, that consists of a control and 3 treatments 1) increased moisture, 2) soil warming (ITEX chambers), and 3) soil warming + increased moisture. The potential rate of decomposition processes will be determined by measuring the tensile strength of 100% cotton strips placed vertically in the soil and sampled periodically over six years. In some plots we will bury fine mesh litter bags containing dried lake algae to determine if this traditional method can be used in Antarctic soils. Biotic influences will be assessed by measuring organic carbon fractions in soil from experimental plots. Available ions in the soils will be assessed using ion exchange membrane sheets and treatment differences in the amount of absorbed ions (NO3(-), PO4(3-)) on the resins will be compared to the biodiversity and biomass of organisms on the plot and soil.

D.3. Continuing Long-Term Experiments (Bioactivity and Biodiversity)
In MCM-I, we initiated 3 long-term experiments designed to elucidate processes controlling the persistence and biodiversity of biota in the streams, lake benthos, and soils. These questions continue to be a focus in MCM-II and the results from the experiments have been valuable in formulating our hypotheses on Bioactivity and Biodiversity. These experiments will be continued with some adjustments in either sampling frequency or expansion of measured parameters.

D.3.a. Long-Term Relict Channel Reactivation Experiment. To investigate developmental patterns of stream algal communities, during January 1995 we routed meltwater to a relict stream channel. Based upon aerial photography, the last period of sustained flow in this channel was in 1969. Soon after wetting, relict algal mats began growing. These mats grew more rapidly than mats in streams with regular summer flow, possibly due to greater solute and nutrient concentrations in the relict channel (Fig. 2.12). Clearly, algal mats can survive desiccation for long periods and respond rapidly to hydration. We will continue this experiment during MCM-II, to follow changes in algal mats and solute concentrations as the relict streambed and hyporheic zone is leached. Further, we will follow the development of invertebrate communities with the downstream advance of flow each year, and along transects across the stream channel to the adjacent soils as the moisture gradient evolves. The reactivated channel has, so far, only experienced low-flow conditions. We anticipate that the relict channel biota will change more dramatically during a high flow year than biota in streams with regular flow.

D.3.b. Long-Term Sediment Dump Experiment. One perturbation to the lake benthos environment is sudden deposition of sediment from the overlying ice-cover as it melts. This is a spatially heterogeneous process and the magnitude of the disturbance may be related to climate influences on the ice-cover. During MCM-I, a sediment addition experiment was begun to assess rates of benthic microbial mat growth as well as the mechanisms by which sediment recolonization occurs. This experiment provides insight into how the lake benthos may respond to small changes in climate.

We placed sediments of known organic matter concentration on the bottom as pre-formed, frozen, sediment patties with a total of 9 treatments plus controls. Six of these treatments were set up in triplicate and three of the treatments have had five experimental units put in place. This allows for uniform experimental units with no loss of sediment-associated organic matter. The initial design of the experiment calls for retrieving four patties every two years. These results will allow quantification of both the rate and principal modes of recolonization of sediments deposited from the overlying ice-cover.

D.3.c. Long-Term Soil Manipulation Experiment. We have hypothesized that climate (water and temperature) control the activity of soil organisms, but that energy supply determines population density and biodiversity of soil communities. One energy source may be wind blown algal material from lakes and streams, representing a linkage between soils and aquatic systems. During MCM-I we found that aeolian organic carbon has characteristics indicating an algal source and that there are significant variations in deposition among basins. In MCM-I, a long-term experiment was started at Lake Hoare. This incorporates a randomized block design of eight replicates of eight treatments, representing combinations of increased water, soil warming and carbon added as either sucrose or mannitol solutions. Soil is warmed approximately 2°C by placing 0.83 m diameter polycarbonate International Tundra Experiment chambers (ITEX, Marion et al. 1997) over the soil. A related experiment compares additions of lake algae as a supplemental C source and its interaction with soil warming. Soils are analyzed for soil temperature, soil moisture, soil respiration, protozoa diversity and abundance, nematode diversity, life stage and abundance, other invertebrate diversity and abundance, total bacteria, and soil physical and chemical properties. Changes in nematode populations are useful indicators of disturbance to soil and sediment communities (Freckman 1982; Freckman 1988; Bongers 1990). Because this long term experiment will help us to understand linkages between landscape units, we will continue application of soil treatments through MCM-II with decreased sampling frequency (once every 2-3 years) to extend the experiment through MCM-II (limited soil is available for sampling).

D.4. New Experiments--Testing Our Conceptual Model
For MCM-II, we propose to conduct two new experiments which will further test in an integrated manner our hypotheses relating to 1) the legacy concept of the McMurdo Dry Valleys and 2) the dependence of ecosystem function on small changes in annual climate. The "paleoclimate" experiment examines the potential response of a valley comparable to the Taylor Valley in the past and involves transplant and manipulative experiments in soils and stream channels. The snow fence experiment is a long-term manipulation experiment in which changes from increased snow accumulation will be followed in all environments in all basins of Taylor Valley. Both of these experiments will test our conceptual model that climate interacts with landscape legacies and linkages to determine functioning of the present day ecosystem. All PIs will participate in both of these experiments. The paleoclimate experiment will be lead by Drs. Doran and Wall and the snow fence experiment will be lead by Drs. Fountain and Moorhead.

D.4.a. Paleoclimate Study. The notion of temporal disturbance and recovery has been a major theme in landscape ecology (Turner et al. 1993). We can examine this notion by comparing the modern ecosystems in Taylor Valley to a paleoclimatic analog when temperatures and precipitation were lower, the nematode community was less diverse, and streamflow was more irregular. In so doing, we will be able to establish the rate of recovery for soil and stream ecosystems. This comparison is termed the "paleoclimate experiment" because it allows us to "look" backwards in time to observe the MCM ecosystem prior to the last warming event that began about 103 years before present. Within 45 km of the MCM site, the Arena and Beacon Valleys (Fig. 2.13) are thought to have undergone little disturbance for at least the past 8.1 and 4.3 M years, respectively (Marchant et al. 1993). These valleys are colder than Taylor Valley with a mean annual temperature -30° to -35°C and precipitation about 10 mm (Bockheim 1982; Marchant et al. 1993). These current environments may be analogs to Taylor Valley at ~1500 yrs BP during the lowest recorded lake levels in both Taylor and Wright Valleys (Matsubaya et al. 1979; Lyons et al. in press).

Surveys of Soil and Stream Communities. In Years 3 and 4 of MCM-II, we will establish soil and stream transects in Arena and Beacon Valleys to compare species composition and abundance between these environments and our ongoing work in Taylor Valley. We will exchange 0.5 m² x 10 cm deep boxes of carefully excavated soil from Taylor Valley and Arena or Beacon Valley. Cotton strips will be placed in each plot to determine decomposition rates. Soil samples will be removed periodically to assess the physical, chemical, and biologic characteristics. Meteorological stations, glacier ablation stakes, and stream gauges will be established to compare the meteorological and hydrological environments with Taylor Valley during the study period. Stream biota in current active stream channels will be documented at stream transect sites comparable to those in Taylor Valley.

Reciprocal Soil and Mat Transplant Experiments. Biodiversity of soils is hypothesized to vary with soil moisture, temperature, organic matter and other soil factors. To examine climatic controls on biodiversity, we propose to exchange soils and their communities from the Fryxell basin, where soils have a younger geologic age, higher organic carbon content, and a more diverse nematode community, with soil from Arena Valley, which is colder, has older soils and a less diverse nematode community. Because biological activity peaks between 2-10 cm depth (Powers et al. in press), we will remove sufficient soil to fill ten, 0.5 m² x 10 cm deep boxes at each site, and transplant it to the comparison site where soil was removed. Care will be taken to minimize disturbance to the soils, and to layer the soil so as to simulate the original soil 'structure'. Cotton strips will be placed in each plot in sufficient numbers to allow decomposition to be measured over a 6-year period (BIOTAS). Soil samples will be removed for initial analyses of soil chemistry and annual analyses of total bacteria and nematodes. Relict channels may be present in Arena and Beacon Valleys as in Taylor Valley. If so, we will look for stone pavement habitats and collect samples of sediments on dried mat material. We will then attempt to culture algae from these samples and examine species distribution.

D.4.b. Snow Fence Experiment. One of our major findings in MCM-I is that changes in snow cover exhibit profound effects across all components of the ecosystem. We propose an experiment that increases snow accumulation on all key components of the dry valley ecosystems and measures subsequent ecosystem response. We will increase snow accumulation by either transporting snow from the high elevations to the valley floor or by erecting barricades (i.e. snow fences) that collect wind drifted snow (Fig. 2.14). We prefer the latter as they better simulate natural accumulations of snow on the valley floor. A snow fence experiment at NWT led to an increase in snow accumulation which had a large effect on the carbon balance and associated biogeochemical cycles in the soils (Brooks et al. 1995). Similarly, we will evaluate the role of increased moisture on the physical and chemical conditions of the underlying substrate which, in turn, effects ecosystem dynamics.

We anticipate that increased snowpack will increase soils biomass and increase atmospheric chemical and particle fluxes to all landscape types. It will, however, decrease biologic activity in the lakes because of reduced light, and will decrease water and nutrient fluxes from the glaciers and streams to the lakes. Once the snow has ablated, biologic activity and water flux will soon return to or exceed previous levels. We will carefully obtain snow cores to examine vertical gradients in snow density, chemistry, sediment, and organisms. Over soils or stream algal mats, if a sufficient plug of soil is collected, sub-snow moisture and biologic activity can be determined. Thermistors will be embedded in snow-covered and adjacent snow-free soil to monitor temperature changes.

We will install 2 fences over each landscape component (soils, lake ice, streams and glaciers) in the Lake Bonney and Lake Fryxell watersheds. Installing fences in different watersheds will take advantage of the different climatic regimes in the valley and help test climatic sensitivity. Snow accumulation on the leeward side of the fence will decrease with distance from the fence, providing a gradient of moisture. Thus with one experiment we can address all MCM-II hypotheses and linkages. Our ability to conduct this work will depend on review for environmental impacts under NSF guidelines and the Antarctic Conservation Act.

Snowpack and Soil Samples. Snowpack measurements will be taken at intervals determined by the weather. At minimum, measurements will be taken in the spring and fall to determine the seasonal variation in snow accumulation, and after every snow storm. Measurements will be taken during periods of rapid melting to assess rate of snowpack depletion. Melt water flux from the snowpack will be estimated from thermistor measurements and the nearest meteorological stations and calculations of snow sublimation. Nutrients and chloride will be measured in the snow cores during early season, the most important period, and immediately after major snowstorms. Soils or stream substrate will be sampled at each location along the fence in late season when a gradient of snow melt has provided moisture for the activity of the biota. Soil samples will be analyzed for the nematodes, chlorophyll a, total bacterial biomass, organic carbon and nitrogen and total soluble salt.

Material Transport. Accumulation of aeolian sediment and biota will be determined using collection pans weighted with marbles and placed in front of the fence and along the accumulation gradient. They will be sampled in spring and in fall to capture the seasonal variation in material transport. Based on the mass collected, we will prioritize analysis of sediment and biotic factors. If possible, samples will be analyzed for total suspended sediment load and size distribution, and for nematodes, algal species, chlorophyll-a and total bacterial biomass.

D.5. Integrative Studies
D.5.a. GIS--Mapping and Remote Sensing (Hydrology). A significant limitation to current research within the McMurdo Dry Valley LTER program is the lack of accurate maps of the region. This situation makes it extremely difficult to express data within a spatially explicit context and restricts development of landscape models. A priority task at the International Centre for Antarctic Information and Research (ICAIR; Christchurch, New Zealand) is the production of an accurate base map and Geographical Information System (GIS) of the McMurdo Dry Valley region. Work at ICAIR has focused on two GIS elements 1) ensuring that a new datum network is established for the dry valleys and 2) digitization of the relevant USGS 1:50K sheets from the mylar originals for every layer of data for the series. A comprehensive, complete and integrity-checked set of layers has been combined with ICAIR's original topo data for this series. These data layers now are being converted to the new datum so that the entire area will be registered before year's end. Once a corrected 1:50K base map is prepared, other existing thematic layers (hydrology, glaciology, etc.) will be added. Field verification of active stream channels is underway in MCM-I. These maps are critical for casting MCM research in a spatial context.

D.5.b. Carbon Budget for Taylor Valley (Biogeochemical Processes). Results of MCM-I studies indicate that the carbon cycle of the dry valleys integrates soil, lake and marine systems, linked across space by the physical transport of organic materials, and linked through time by fluctuating lake levels. Developing a comprehensive budget of carbon dynamics in MCM minimally requires working knowledge of in situ carbon dynamics of modern soil and lake communities, as well their spatio-temporal linkages. During MCM-II, Drs. Doran, Virginia, Wall and Lyons will continue to develop a map of d13C distributions throughout Taylor Valley. Modern mat communities in lakes and streams at various water depths and locations will be compared to soil carbon values. The goal of the work is to understand the sources of carbon throughout the modern landscape and to project future patterns from current biological activities and transport mechanisms. Aspects of this integrative project have been previously described, but additional studies are proposed to aid in this integration, as follow.

D.5.c. Soil Carbon Pools (Biogeochemical Processes). Soil carbon cycle in the McMurdo region has importance as an extreme "end-member" of global soil ecosystems. A soil carbon budget for Taylor Valleys is required to integrate our studies of the carbon legacy with current rates of NPP and nutrient cycling. Soil carbon pools will be determined based upon systematic regional sampling of soil profiles (Brown et al. 1996; Burkins et al. 1997) located at varying distances and elevations from lakes, streams, and the ocean. The possible origins of this carbon (soil derived, lacustrine, marine) will be determined by measuring the natural abundance of 13C and 15N in the soil organic matter. Measurements of lake mat d13C abundance will constrain the isotopic variation of lacustrine carbon sources (see below). The quality of soil organic carbon will be assessed by separating light fraction carbon (labile, rapidly cycling C) and heavy fraction carbon (Connin et al. 1997a, b). Our goal is to develop a coar se-scaled soil carbon map for Taylor Valley to relate soil carbon quantity, quality, and source to soil NPP, respiration, and biodiversity.

Soil NPP and its spatial variation will be examined by determining soil chlorophyll concentrations (an index of soil algal biomass; Metting 1994), by gas exchange and by 14C (Warembourg and Kummerow 1991). For soil NPP, closed chambers will be placed over the soil, incubated in light or darkness, and the CO2 concentration in headspace determined. Soil respiration also will be measured using an open gas exchange system and soil cuvette (LICOR 6200). Our intent is to develop a relationship between chlorophyll, which is relatively easy and inexpensive to measure, and a smaller number of more intensive measurements of carbon fixation. The data are important for the carbon cycle model.

D.5.d. Paleocarbon in Lakes (Biogeochemical Processes). In order to better understand the legacy of the lake organic matter in the soil ecosystem and the processes leading to the production and preservation of lacustrine carbon today, a series of investigations will be undertaken. Dr. Doran has performed experiments on the control of d13C in organic matter buried in deltaic and lake bottom regions in coordination with the work of Drs. Lyons and Priscu on d13C and DIC dynamics in the lakes. The results show 1) that there is significant variation in d13C of mat organic carbon and 2) that these differences may be associated with water column CO2 depletion (Neumann et al. in review). In the CO2 saturated bottomwater of Lake Hoare, mats fractionate carbon isotopes by values of 30%, as predicted from theory. In contrast, shallow Lake Hoare mats display enriched d13C values, consistent with the CO2 undersaturation of these waters (Fig. 2.15). Therefore, the d13C of mat material reflects the depth zone of formation of that material. It is clear that the ancient lacustrine carbon is not uniformly distributed throughout the landscape. This patchiness is due to the location of features of the ancient high stand lake. During MCM-II, we plan to coordinate the carbon/nitrogen isotope work in soils and lakes. This will provide better understanding of the source and location of the "legacy" organic carbon now in the soils.

D.5.e. Lake Nutrient Balances (Biogeochemical Processes). The hydrology and geochemistry program initiated in MCM-I provided the preliminary data for us to begin the construction of nutrient balances for dissolved inorganic nitrogen and phosphorus species within each lake. We will use (and refine) existing (MCM-I) and proposed (MCM-II) streamflow and nutrient chemistry to develop detailed budgets for each of our study lakes. Ratios and concentrations of these "new" nutrients will be related to the relict pools within each lake to determine the turnover rates of each pool. Upward diffusion of relict nutrients will be computed as described by Priscu (1995). These data, in concert with existing nutrient bioassay data (Priscu 1995) and routinely collected primary productivity data, will provide us with information on the roles that new and relict nutrient pools have on phytoplankton growth and vertical structure. The relative roles of these two nutrient pools will depend upon interannual variation in glacial melt, resultant streamflow, nutrient transformations within the streams and weathering processes associated with the stream channels.

D.6. Ecological Modeling (Biogeochemical Processes)
Bioactivity and development of ecological models for interactions within ecosystem components are important means to test and develop our hypotheses. The modeling effort is conducted by Dr. Moorhead in collaboration with other PIs.

D.6.a. Primary Production. In MCM-II, our conceptual model (Fig. 1.5) will be applied to the cyanobacterial communities embedded within permanent lake ice (Spigel and Priscu 1998). It also will be used to simulate primary production in soils at the narrow interface where light penetration from above and water vapor movement from underlying permafrost overlap to provide a suitable microhabitat for algae. Higher resolution of light and nutrient regimes in lakes will permit estimating patterns of primary production at higher spatial and temporal resolution for phytoplankton and benthic mats, a priority of MCM-I (Moorhead et al. 1997b, 1998).

D.6.b. Secondary Production. Food webs in soils are limited largely to algae, yeasts, bacteria, protozoans and nematodes (Freckman and Virginia 1997, 1998), and aquatic food webs are limited to algae, bacteria, protozoans and rotifers (Laybourn-Parry et al. 1997; James et al. 1998). Investigations of MCM-I are providing species identities, estimates of abundance and biomass of trophic groups, as well as measures of community respiration, which will be used to develop food web models of energy and nutrient flow in MCM-II (see below).

Studies also are planned to improve resolution of temperature and light responses of respiration and photosynthesis. Models of production and turnover for particular groups of heterotrophs in dry valley ecosystems will be possible when more detailed data of abundance and distributions of taxa through time and space are available (see below).

D.6.c. Population Dynamics. Few of the species present in MCM ecosystems have been examined sufficiently to support a population model incorporating developmental stages of the organism. The primary exception is the soil nematode, Scottnema lindsaye (Overhoff et al. 1993). A model that previously was used to simulate population dynamics of the nematode, Acrobeloides nana, in a warm desert environment (Moorhead et al. 1987), was modified to estimate development of Scottnema in dry valley soils (Fig. 2.16). Soil temperatures recorded during MCM I were used to drive the model. Results suggested that in situ development of this species may require several Antarctic summers due to short summer seasons and slow nematode development at the prevailing low temperatures. A priority of MCM-II is to gain more detailed soil climate data, including moisture levels. This information will permit a more accurate assessment of development, activity and respiratory patterns of Scottnema, and may permit application of the model to other species.

D.6.d. Food Webs. A model was developed to simulate energy and nutrient flux through the food webs of MCM soils (Fig. 2.17), utilizing the approach of Hunt et al. (1987, 1989). In essence, biomass estimates of the various groups are assumed to represent steady-state values (a particularly valuable approach when temporal resolution of data is low). General metabolic characteristics of these types of organisms (cf. Hunt et al. 1987) are used to estimate ingestion, egestion, assimilation and mineralization of carbon and nitrogen, driven by climate regime. Flows of material between trophic groups are assumed to balance demand, resulting in a cascade of cumulative top-down demands for energy and nutrient supply throughout the food web. Current efforts focus on the initial formulation of this model and application to sites where soil climate and trophic structure are known. Continued and expanded monitoring of soil climate, especially soil moisture regimes, and structure of soil food webs will suppor t further model development and application to other sites during MCM-II. Moreover, such a modeling approach may be suitable to address energy and material flow through food webs existing in microbial mats and planktonic communities in aquatic environments, as supporting data become available.

D.6.e. Productivity, Complexity and Stability. A suite of recent modeling studies by Moore et al. 1993, have explored potential relationships between productivity and structural complexity of simple food chains. In essence, they propose that fluctuations in energy supply may limit the complexity of food webs, especially in simple systems. The central premise of this hypothesis is that probabilities of extinction are related to energy supply, with vulnerability to extinction increasing with trophic distance from the source of energy that drives the system. The soil communities of the dry valleys are a model system in which to test this notion because they consist of short food chains and few species (Freckman and Virginia 1998). Moreover, the number of species varies between locations in a manner that at least partly is independent of physicochemical characteristics of the site, random extinction could account for this pattern. A proposed modeling activity of MCM-II is to explore the possible im pacts of extinction processes on structuring food webs in MCM soils. Moore's approach is based on the Lotka-Volterra equations of predator-prey interactions, which may be developed from baseline estimates of energy flows through components of the food web model (see above).

Unlike soil organisms, plankton live in a more predictable and less variable environment in MCM lakes, but MCM-I studies have shown that the relative abundance of particular species fluctuates over time in a manner that is not clearly related to changes in physicochemical characteristics. Thus, controls over population dynamics are not apparent. Indeed, May (1974) was one of the first to demonstrate that complex patterns of population dynamics could result from relatively simple interactions among species. Another modeling goal of MCM-II is to explore the population dynamics of planktonic species using Lotka-Volterra competition equations to simulate interactions among the numerically dominant taxa. The bases for defining the relative strengths of competition coefficients among different species (as required for this method) are not clear. However, we propose to calculate potential competition on the basis of (1) light absorption, expressed as a function of average cell diameter for each species, assuming con stant optical densities of cellular constituents, and 2) nutrient uptake, as a function of average cell diameter for each species, assuming uptake to be proportional to surface area. This approach, albeit crude, will define competition according to the relative abilities of cells to compete for energy and nutrients, and provide a basis of comparison to more traditional "chemostat" type models of population dynamics for planktonic organisms driven by abiotic factors.

D.7. Intersite Comparisons within LTER Network
Our cross-site efforts over the past year included the following: participation in the LTER comparison of stream productivity, a contribution to the Freshwater Biology special issue from the LTER-NTL sponsored workshop, Cross-site Comparisons, Syntheses and Interactions, and our participation in the LTER Intersite Hydrology project at H. J. Andrews experimental Forest. Dr. Doran attended the Archaeoclimatology Workshop in Santa Barbara hosted by PAL-LTER. In addition, Lyons has been working with Dr. G. Kling (ARC) in an attempt to compare chemical weathering rates between the two sites. It has traditionally been thought that polar conditions should inhibit chemical weathering; however, recent work in Siberia has suggested that this is not the case. Chemical weathering controls the CO2 composition of the atmosphere over geologic time scales (Berner 1995). Lyons has proposed to the Network office the beginning of an intersite comparison that would include all sites with surface water or ground water flux measur ements. The goal is to utilize hydrologic and chemical data to calculate weathering rates and then evaluate the roles that climate, presence and absence of vegetation and land use patterns have on chemical weathering. This concept was initially approved by the Network office in late 1997.

E. FIELD AND EXPERIMENTAL METHODS

E.1. Monitoring

E.1.a. Meteorology. Stations query sensors at 30 s intervals and store data every 20 minutes. Stations measure air temperature and humidity at 3 m above the surface, soil (or ice temperature) at multiple depths, wind speed/direction, and incoming and outgoing solar flux. Lake stations measure PAR, and glacier stations measure longwave radiation. Barometric pressure is measured at Lake Hoare and Canada Glacier. Snowfall is measured using Belfort gauges, by a salt dilution method.

E.1.b. Aqueous Geochemistry. Glacier ice/snow, meltwater, streams, and lakes are analyzed for parameters in (Table 2.3). Major ion chemistry is analyzed using ion chromatography (Welch et al. 1996). Samples are filtered through Whatman GF/F filters and stored at 20°C. Ammonium, nitrite, nitrate and soluble reactive phosphate are determined with a LACHAT autoanalyzer. Reactive silicate is determined colorimetrically (Toxey et al. 1997). DOC samples are analyzed with an O-I carbon Analyzer. Oxygen is measured using Winkler titrations. Dissolved inorganic carbon (DIC) is by infrared spectrometry on acid sparged samples.

E.1.d. Streams. Discharge is measured manually at different stages and a rating curve developed for each gauge (VonGuerard et al. 1995). Gauges with a Parshall flume and a rectangular weir in the cutoff wall are used for the most accurate data at low flow. Other gauges use natural channel controls and provide data with 10% accuracy for flows above 10 L/s. Each lake basin has 2 gauges with flumes. Discharge for ungauged streams is measured 3-4 times per season.

Algal mats are characterized according to color. Five samples of each color type are collected using a #13 cork borer. Sample location is mapped on small contour maps. Samples are preserved in 10% formalin for species analysis and filtered onto GF/C filters for chlorophyll determination (Strickland and Parsons 1972). Species composition and relative abundance are determined using a Nikon Diaphot phase contrast microscope. Subsamples are placed in a 2 ml settling chamber with a 26 mm diameter (Utermohl, 1958) and examined at 400x magnification. Seven random fields (a total area of 0.44 mm²) are tallied. Descriptions of morphotypes are presented in McKnight et al (1998). Percent cover is determined by measuring the length and width of all specimens. Cell "depth" is measured to estimate biovolume.

E.1.e. Lakes. Lake sampling is monthly from October to January. Water samples are collected at 2 m intervals using a Teflon coated Niskin bottle. Methods are summarized in Table 2.3 and details are provided below.

Manual measurements of lake level are made in spring and fall using a survey level from a Bench Mark calibrated to those of the New Zealand Antarctic Program. Continuous measurements (20-min. intervals) are recorded using a pressure transducer suspended between a buoy and anchor. Ice temperatures are measured using a thermistor string frozen into the ice-cover.

Spherical sensors (LICOR 193) are deployed at 10 m beneath the ice surface. Sensor output is logged continuously at 20 min (summer) or hourly (winter) intervals. This instrument has been specifically designed for under-ice use and produces high resolution (~5 cm) profiles (Spigel and Priscu 1998). Density is obtained from CTD measurements using modified equations of state for seawater (Spigel and Priscu 1996).

Phytoplankton productivity is measured as 14CO2 uptake in light and dark bottles suspended in situ for 24 h (Priscu 1995). Discrete and high resolution in situ profiles are used to determine phytoplankton biomass. Discrete samples are filtered through Whatman GF/F filters and chlorophyll a (phaeophytin corrected) is measured by fluorometry (Lizotte and Priscu 1998) and used to convert in situ fluorescence profiles to absolute chlorophyll values. Samples for algal enumeration are preserved in Lugol's solution, settled and counted with an inverted microscope.

Bacterial productivity is measured as tritiated thymidine incorporation with a protocol which is linear for 20 h (Takacs and Priscu, in review). Conversion of thymidine incorporation rates to in situ rates and to carbon-based, production-use factors determined in laboratory bacterial growth experiments (Takacs and Priscu in review). Bacterial (>0.2 um) numbers are determined by epifluorescent spectrometry of acridine orange stained cells (0.01% final concentration) preserved with 5 % formalin. Biomass is determined with geometric measurements of fluorescent cells.

Protozoa are studied in samples preserved with buffered glutaraldehyde. Nannoflagellates are counted by staining 30-50 ml with DAPI filtering onto 2 mm polycarbonate membranes, and counting 20 Whipple grids under both UV and blue filters with epifluorescence microscopy (to distinguish phototrophic and heterotrophic cells). Ciliate samples (500 ml or 1-liter) are preserved in 1% Lugol's solution, concentrated to 1 ml and counted in a Sedgewick-Rafter chamber. Cell sizes are determined with a line graticule and biovolume calculated using the nearest geometric shape. Biovolume to biomass conversions are obtained using a carbon conversion factor for nannoflagellates from Borsheim and Bratbeck, 1987 and for ciliates from Putt and Stoeker, 1989. Growth rates are determined using the method of McManus (1993). One unscreened sample is a control; the second sample is screened through a 10 mm mesh for nannoflagellates, or a 100 mm mesh for ciliates, to remove predators. Because there are no mesozooplankton predators on ly one unscreened incubation is required. Water samples are poured into 4 L containers and incubated at in situ conditions. At 8 time intervals, 3 subsamples are preserved and protist abundance determined.

E.1.f. Benthic Mats. Cores are collected with SCUBA with a polycarbonate core tube (5 cm dia.). The cohesive surface layer is removed and analyzed for species composition (epifluorescence microscopy), chlorophyll a (spectrophotometric analysis of DMSO/acetone extracts), phycobilin pigments (spectrophotometric analysis of water extractable material remaining from the DMSO/acetone extracts) and accessory pigments using High Performance Liquid Chromatography (Lizotte and Priscu 1998). Vertical growth is measured at three permanent transects established using horizontal aluminum bars driven into the sediment. The horizontal bar has a series of holes to accommodate calibrated needles dropped from the bar to measure distance to the mat surface. Cores for elemental analysis will be frozen and sectioned. Sections will be dried and analyzed for particulate organic C and N using a Carlo Erba elemental analyzer. Particulate phosphorus is measured after wet oxidation (Solorzano and Sharp 1980). Elemental results will be normalized to sample area and sediment weight.

PAR measurements are made using a LICOR 6200. Mat samples (upper 3 cm) will be examined for ciliates by a method of Finlay et al. (1979). The mat surface is washed with filtered lake water and examined. The ciliates are counted until 150 ciliates have been recorded. Mat samples are examined microscopically to verify ciliate removal.

E.1.g. Soils. Soil and air temperature are monitored using thermocouples. Soil moisture and salinity are measured at 2, 5 and 10 cm depths with the Hydra Soil Moisture Probe (Campbell 1990). Decomposition measurements use the loss of tensile strength in cotton strips as a measure of mass loss, as used by the BAS at Antarctic sites. Because small treatment plots cannot be sampled repeatedly, available ions will be measured using thin rigid (2.5 x 2.5 x 0.1 cm) ion exchange membrane sheets placed in the soil (Abrams and Jarrell 1992).

All soils and biota samples are placed in sterile Whirl-Pak bags and returned to McMurdo Station laboratories. Soil chemical and physical properties are determined using standard procedures (Freckman and Virginia 1997, Virginia and Jarrell 1983): pH of a saturation paste; Organic N and organic C by dry combustion following removal of carbonates with acid (Carlo-Erba Model 1500 C/N Analyzer, Nelson and Sommers 1982), KCl-extractable NO3(-) and NH4(+) and NaHCO3(-) extractable P on a Lachat autoanalyzer (Olsen and Sommers 1982); water extractable (1:5 soil:water) cations (Ca(2+), Mg(2+), Na(+), SO4(2-) and K(+) by ICP optical emission spectrometry; conductivity (EC) using a conductivity meter; and texture by hydrometry.

Microbial biomass is determined by epifluorescence microscope counting on DAPI stained samples, with image analyses for enumeration. Biovolume measurements are used for total and bacterial biomass (van Veen and Paul 1979; Bloem, et al. 1995). Cellular shape and sizes is recorded as a measure of diversity.

Soils in 4 cm diameter polycarbonate tubes will be examined for ciliates (Finlay et al. 1979). Water will be added to soil samples until ciliate species can be discriminated from soil particles (usually involves a dilution of 10x). Stream mat samples will be washed with a known volume of lake water and one drop (5ul) examined using a magnification of 100-200x. Ciliates in each drop will be counted until 150 ciliates have been recorded. Protozoa will be analyzed using the Most Probable Number identification outlined in Bamforth (1991, 1995). Nematodes, rotifers and tardigrades are counted and taxa preserved for further species identification (Freckman and Virginia 1993). Nematodes are identified to species, trophic group and life stage, and tardigrades and rotifers (and any other biota) will be enumerated (Table 2.4).

E.2. Methods for continuing and new long term experiments

E.2.a. Relict Channel Reactivation Experiment. A 1-ft high sand bag wall was constructed in January 1995 at the base of the upper reach of Von Guerard Stream, directing about 50% of the flow to channels to the west. During the subsequent four years, sampling sites have been established as flow advances downstream. In January 1997, the flow in the relict channel reached the confluence with Harnish Creek for a few days. A gauging station is in place above this confluence, but only temperature and conductivity have been recorded due to the very low flow. The summer of 1997/98 was very cold and flow did not reach the gauge. Sampling and mapping of stream mats and chemical analysis are conducted in the monitoring program.

E.2.b. Sediment dump experiment. Simulated sediment dumps were deployed by SCUBA dives on the lake bottom in January 1997. Resampling is planned every two years. Sediment was collected from Lake Hoare ice (by melting). Sediment had an average moisture content of 13.4% and an OM content of 0.48% (SD = 0.02%, n = 10).

E.2.c. Paleoclimate experiment. The methods for the soil and stream transplant and manipulation experiments are the same as in the monitoring program.

E.2.d. Snow fence experiment. Fences about 1.5 m high x 5 m long made of spaced vertical wooden slates wired together will be installed perpendicular to prevailing winds on the glaciers and lakes, over stream channels and on soils. Although other fence styles may be more efficient at collecting snow, the reduced cost, ease of transport and installation outweigh the disadvantages. To minimize disturbance, measurements will be made from a portable aluminum ladder that bridges the snow by leaning on the fence. Snow depth will be measured using a thin rod. Snowpack density and water equivalent will be measured using a snow-tube sampler (e.g. Mt. Rose sampler). The ladder will be repositioned over the snow at several locations to replicate measurements. Snow samples will be melted at 2°C and filtered for analysis following procedures in the monitoring program. Sediment will be analyzed for nematodes, chlorophyll a, total bacterial biomass, organic carbon and nitrogen and total soluble salt. Over soi ls or stream algal mats, if a sufficient plug is collected, sub-snow moisture and biologic properties will be determined.

In the MCM, the bands marking the high lake stands of 20,000-10,000 years ago are unmistakably apparent on the valley walls. From our research in MCM-I, we have become aware that this and other legacies from past climate regimes strongly overprint the modern MCM ecosystem, in conjunction with the modern interannual climate variability. Significant linkages between landscape units are active both in the modern day and through legacies from past climate regimes. Here, we present hypotheses relating to the "legacy effects" in our conceptual model of MCM. We propose to examine these hypotheses by continuing the MCM monitoring program with some modifications, by continuing and expanding our long-term manipulative field experiments, and by integrative and modeling studies. The new long-term experiments are the paleoclimate experiment and the snow fence experiment, both of which will integrate studies of glaciers, streams, lakes and soils.