SNOW DEPTH, SOIL FROST AND NUTRIENT LOSS IN A NORTH TEMPERATE FOREST
(DEB 96-52678)Peter M. Groffman, Institute of Ecosystem Studies
Timothy J. Fahey, Cornell University
Charles T. Driscoll, Syracuse University
Janet P. Hardy, U.S. Army, Cold Regions Research and Engineering LaboratoryA. PROJECT SUMMARY
While much global change research has focused on direct effects of long term changes in temperature and precipitation on the structure and function of ecosystems, some of the most dramatic consequences of climate change may be the result of indirect effects. We suggest that one of the most dramatic effects of global climate change on northern forests will be a reduction in snow cover, which will lead to increases in soil freezing, nutrient loss and surface water acidification, and changes in soil-atmosphere trace gas fluxes and forest tree species composition.
A lack of snow cover results in colder soil temperatures, more extensive soil freezing, and an increase in freeze/thaw cycles. Previous studies have suggested (but not verified) that these stresses result in root and microbial mortality resulting in an increase in labile organic carbon (C) and nitrogen (N) input to soil (via root and microbial death) and increases in soil moisture and available N (via reduced uptake by trees and microbes). These changes increase net mineralization and nitrification rates, nitrate and cation leaching losses, and the acidification of drainage waters. Moreover, in the long-term, we believe that differential resistance to freezing stress will be a key regulator of species composition in northern forests under a warmer climate.
In this proposal, we request funds to initiate a long-term experiment of decreases in snowpack accumulation at the Hubbard Brook Experimental Forest (HBEF), a northern hardwood dominated forest located in the White Mountains of New Hampshire. We will use 10 m x 10 m shelters to reduce snowpack accumulation and will calibrate an existing model (SNTHERM) that depicts snow depth and soil frost dynamics given past or future climate scenarios for our site. Root dynamics will be studied with minirhizotrons and we will examine relationships between patterns of root mortality and soil microbial C and N content, bioavailable organic C, N mineralization, nitrification and denitrification. Soil solution chemistry will be monitored with zero tension lysimeters and soil- atmosphere fluxes of CO2, N2O and CH4 will be quantified in field chambers. Laboratory studies will examine the effects of length, frequency and temperature of freezing events on microbial and root mortality. To evaluate the long-term implications of our results, we will use the SNTHERM model and our long-term database on streamwater chemistry to examine the effects of past natural freezing events on nutrient loss from hardwood forests at HBEF, and to forecast future events given climate change scenarios. To assess the environmental relevance of our treatments, we will incorporate soil temperature into our existing long-term monitoring program so that we can characterize natural freezing events that occur in the future for comparison with our experimental treatments. In coordination with other ongoing research at HBEF, we will be able to evaluate the importance of changes in snow cover relative to other natural and anthropogenic disturbances experienced by northern hardwood forests.
C. PROJECT DESCRIPTION While much global change research has focused on the direct effects of long term changes in climate on the structure and function of ecosystems (Ojima et al. 1991, Field et al. 1992, Melillo et al. 1993, Pielke et al. 1993), there is widespread recognition that the most dramatic consequences of climate change may occur due to indirect effects. Examples of these indirect effects include changes in fire frequency, invasion by exotic species, changes in hydroperiod, extreme temperature and precipitation events, and changes in snowpack accumulation (Fajer et al. 1989, Smith and Shugart 1993, Vitousek 1994, Schimel 1995, Hornung and Reynolds 1995, Suffling 1995). Characterizing and quantifying indirect effects represents a great challenge to global change research since these responses are ecosystem- specific and difficult to predict.I. INTRODUCTION
In this proposal, we request funds to investigate the consequences of decreases in snow cover on biotic functions and biogeochemical processes in a northern hardwood forest. Previous research (Likens et al. 1977, Edwards and Cresser 1992, Pilon et al. 1994, Boutin and Robitaille 1994, Hardy and Albert 1995) has shown that snow depth is a key regulator of soil temperature, root and microbial dynamics, soil nutrient loss, drainage water acidification and soil-atmosphere trace gas fluxes in northern forests. Several studies have documented ecosystem and regional- scale effects of snowpack variation on forest productivity, nutrient loss and other processes (Auclair et al. 1992, Lydersen 1995, Melloh and Crill 1995, Mitchell et al. 1996, Goulden et al. 1996). We suggest that one of the most significant consequences of global climate change on northern forests will be a reduction in snow cover. This disturbance will lead to increases in soil freezing, nutrient loss and drainage water acidification, and changes in trace gas fluxes and forest composition.
A lack of snow cover results in colder soil temperatures, more extensive soil freezing, and an increase in freeze/thaw cycles (Edwards and Cresser 1992). Previous studies have suggested (but not verified) that these stresses result in root and microbial mortality, releasing labile organic carbon (C) and nitrogen (N) to soil (via root and microbial death) and increasing soil moisture and available N (via reduced uptake by trees and microbes) (Sakai and Larcher 1987, Pilon et al. 1994, Boutin and Robitaille 1994). These changes lead to increases in net mineralization and nitrification rates, nitrate (NO3-) and cation leaching losses and acidification of drainage waters (Skogland et al. 1988, Christensen and Christensen 1991, DeLuca et al. 1992, Brooks et al. 1995, 1996). Over the long-term, we believe that differential resistance to freezing stress will be a key regulator of species composition in northern forests under a warmer climate condition.
In this proposal, we request funds to initiate long-term experiments to examine the consequences of decreases in snowpack accumulation at the Hubbard Brook Experimental Forest (HBEF), a northern hardwood dominated forest located in the White Mountains of New Hampshire. We propose to quantify the effects of decreases in snowpack accumulation on root dynamics of two key tree species in this forest (sugar maple, yellow birch), microbial biomass and activity, NO3- and cation loss, the acid- base chemistry of drainage water, and soil- atmosphere trace gas fluxes. In addition to experimental studies, we will initiate a long-term monitoring program of soil freezing events that will be closely coupled to ongoing long-term measurements of vegetation, root activity, microbial biomass and activity and drainage water chemistry. Long-term monitoring will allow us to evaluate the importance of natural freezing events and to assess the environmental relevance of our treatments.
II. OBJECTIVES
1. Establish long-term experimental treatments of decreased snowpack at HBEF. We will establish experimental treatments in single-species stands of sugar maple and yellow birch. Plots will be instrumented with thermistors and time domain reflectometry (TDR) probes to characterize snow and soil temperature and moisture dynamics. 2. Examine the effects of decreases in snowpack accumulation and increases in soil freezing on fine root dynamics. We will evaluate root dynamics using minirhizotrons and will quantify nutrient uptake and litterfall. 3. Examine the effects of decreases in snowpack accumulation and increases in soil freezing on microbial biomass C and N dynamics. We will characterize changes in the size and activity of the soil microbial biomass and will quantify N mineralization, nitrification, denitrification and respiration rates throughout the year. 4. Examine the effects of decreases in snowpack accumulation and increases in soil freezing on soil solution chemistry. We will use zero tension lysimeters to sample soil solutions. 5. Quantify the effects of decreases in snowpack accumulation and increases in soil freezing on soil-atmosphere fluxes of methane (CH4) nitrous oxide (N2O) and carbon dioxide (CO2) using in-field chambers. 6. Conduct laboratory and greenhouse studies to evaluate the specific effects of soil freezing and very low soil temperatures on plant root and soil microbial dynamics. 7. Evaluate the potential regional scale effects of snow reduction on ecosystem processes. We will calibrate an existing model (SNTHERM) of soil freezing using experimental data. This model will then be used to evaluate previous soil freezing events that have occurred over the last 33 years at HBEF and to predict changes in snow depth and soil frost given climate change scenarios. Experimental data will then be used to evaluate the long-term effects of climate change-induced snow reduction on community composition, N losses, drainage water acidification and trace gas fluxes in northern hardwood forests.
III. HYPOTHESES
1. Reductions in snow cover will increase the duration and depth of soil freezing and the frequency of freeze/thaw events in a northern hardwood forest. 2. Reductions in snow cover will result in increases in root mortality, with differing response among tree species. Sugar maple will suffer greater freeze damage to roots than yellow birch in both field and greenhouse settings. 3. Rates of C and N mineralization, nitrification and denitrification and levels of available C, soil moisture and inorganic N will increase, while pools of microbial biomass C and N will decrease in response to soil freezing events. These changes will be associated with inputs of labile C and N from root and microbial mortality, and decreases in root uptake of water and nutrients. The increase in denitrification will be less than the increase in mineralization and nitrification because while mineralization and nitrification rates will be enhanced for a large part of the year, denitrification activity will be restricted to early spring and fall.. 4. Nitrate, cation and dissolved organic carbon concentrations (DOC) will increase in drainage water, resulting in an acidification event due to disturbance of microbial and root activity by freezing. 5. Soil-atmosphere fluxes of CO2 and N2O will increase and uptake of CH4 will decrease due to the changes in C and N dynamics associated with soil freezing. 6. In the long-term (3-5 years), plots dominated by species that are highly susceptible to soil freezing (sugar maple) will undergo severe decline (tree mortality, nutrient loss), while in plots dominated by less susceptible species (yellow birch), the effects will be less dramatic and will disappear or attenuate over time.
IV. RATIONALE AND JUSTIFICATION There has been recent interest in biogeochemical processes during winter in temperate ecosystems. Studies have demonstrated that root and microbial processes are surprisingly active in cold (0 - 5 °C) and even frozen soils (Vogt et al. 1986, Coxson and Parkinson 1987, Taylor and Jones 1990, Sommerfeld et al. 1994, Clein and Schimel 1995, Melloh and Crill 1995, Brooks et al. 1996). A significant portion (20 - 50%) of ecosystem C and N cycling and soil-atmosphere trace gas fluxes can occur during winter. However, the nature, extent and regulation of many biogeochemical processes during winter are poorly characterized (Edwards and Cresser 1992).A. Snow and ecosystems
Snow is a critical regulator of soil processes in temperate forests during winter, acting as an insulator, preventing soil from freezing in many cases and providing moisture to support biological processes (Marchand 1987, Brooks et al. 1995). Snowpack dynamics and ice features also influence the exchange of gases, water and solutes overwinter and are a major regulator of nutrient outputs at spring melt (Likens et al. 1977, Lewis and Grant 1980, Rascher et al. 1987, Williams and Melack 1991, Hardy et al. 1995, Winston et al. 1995, Davis et al. 1995). Although the physics and chemistry of snow have been well studied in relation to water quantity and quality issues, the role of snow as a regulator of soil and ecosystem processes is poorly understood.
Interest in snow as a regulator of ecosystem biogeochemical processes has accelerated in recent years due to concerns about global climate change, and more importantly, due to observations of marked biogeochemical changes at the regional scale associated with low snow years. Snow "drought" has been linked to sugar maple decline in Canada (Auclair et al. 1992, Pilon et al. 1994) and to dramatic changes in watershed nitrate (NO3-) outputs in the northeastern U.S. (Mitchell et al. 1996). A lack of mechanistic understanding of these events, coupled with predictions that global warming is expected to be most dramatic at high latitudes, during winter, has heightened interest about ecosystem processes during winter and how they respond to environmental change. While there are no detailed predictions of snow depth changes with global warming for the northeastern U.S., existing studies and models have suggested that snowpacks respond very dynamically to climate change (van Katwijk et al. 1993, Giorgi et al. 1994, Williams et al. 1996, Moore and McKendry 1996).
B. Effects of soil frost and freeze/thaw on roots Soils in most temperate forests do not generally freeze and there is no evidence of accelerated root mortality during normal winters (Hendrick and Pregitzer 1992, Fahey and Hughes 1994). Unfortunately, observations of root mortality have not been made during the rare winters when extensive soil freezing has occurred in temperate hardwood forests (e.g. 1969-70, 1973-74 and 1988-89 at HBEF). Minirhizotron observations in the boreal conifer forest of Alaska suggest that root mortality is concentrated in the winter months as concrete frost permeates to the permafrost layer (R. Reuss, personal communication).
The ability of the shoot tissues of temperate plants to survive sub-freezing temperatures is well known, and the mechanisms of shoot tissue hardening have been extensively catalogued (Sakai and Larcher 1987). Less research has been conducted on frost hardiness of root tissues, partly because temperature fluctuations are less pronounced for soil than for the atmosphere. Root tissues do exhibit a considerable capacity to survive sub-freezing temperatures, and like shoots the process of cold hardening is a conditioned response. Not surprisingly, for the species that have been studied, lethal minimum temperatures for roots are higher than for shoots (Sakai and Larcher 1987). Moreover, the onset of hardening is later for roots than shoots, while de-hardening begins earlier in roots. That the processes are somewhat independent is illustrated by the observation that shoot tissue hardening is conditioned in part by daylength whereas that of roots seems to depend only upon temperature (Bigras and D'Aoust 1993).
The question of whether the distribution and abundance of plants is influenced by cold soils and freezing injury to roots has not been addressed conclusively, despite a variety of studies and evidence on the subject. In short-term exposure tests, the soil temperatures that are lethal to the fine roots of temperate and boreal zone trees appear to be much lower than those routinely observed under natural conditions (Sakai and Larcher 1987, von Fircks 1992, Calme et al. 1994). Moreover, a comparison of the frost hardiness of roots of subalpine conifers in the Pacific Northwest (Coleman et al. 1992) revealed no correlation between soil temperature regimes of the sites of origin and laboratory tests of cold hardiness. However, studies of tissue damage to roots of willow suggested that incomplete acclimation rather than absolute temperature was connected with injury (Von Fircks 1992). Similarly, Braathe (1995) observed that root injury and tree dieback of birch was associated with spring thaw-freeze sequences. The duration of root exposure to sub-freezing temperatures probably also affects injury responses (Sakai and Larcher 1987). For example, extensive damage to roots and consequent die-back of shoots of sugar maples (Boutin and Robitaille 1994) occurred as a result of season-long snow removal that produced temperatures that were considerably higher than the lethal temperatures observed for short-term exposures (Sakai and Larcher 1987). Whether more realistic exposures to soil freezing will injure fine root tissues of temperate zone trees is not known.
There is some reason to expect differential sensitivity of the fine roots of northern hardwood species to soil freezing. Calme et al. (1994) compared the frost hardiness of three northern hardwood species and concluded the following order of increasing cold hardiness: red oak < sugar maple < yellow birch. The elevation range of yellow birch normally exceeds that of sugar maple, and birch is a common associate of red spruce and balsam fir in the lower subalpine zone of northeastern mountains (Battles et al. 1995). These conifer-dominated forests appear to be more routinely subjected to concrete soil frost than deciduous forests (Fahey and Lang 1975) due to lower snow retention by conifer canopies, the insulating characteristics of the litter, and radiation absorption and reflection by the evergreen canopy.
The extreme soil freezing conditions associated with the experimental snow removal treatment of Boutin and Robitaille (1994) induced serious decline symptoms in mature sugar maple, illustrating a mechanism that may alter relative species abundance in northern hardwood forests. An alternative mechanism could be the effects of soil freezing on seedling survival; because their root systems are more concentrated in the surface soil layers where thermal fluctuations are most pronounced, seedlings may be more sensitive to soil freezing than mature trees. If soil freezing affects the survival of seedlings of various species differentially, and if soil freezing occurs at high frequency in response to climate change, then the composition of northern hardwood forests could change, as windows of opportunity for seedling recruitment are increasingly restricted.
In sum, regional climate changes that influence snowpack accumulation might affect the structure and function of northeastern forests. Even if there are no differential effects of soil freezing on root mortality of northern hardwood tree species, winter mortality will undoubtedly alter the magnitude and timing of substrate availability to the decomposer community (see below). Fine roots are a labile resource for decomposers (Fahey et al. 1988), and the large increase in NO3- leaching following cutting of northern hardwood forests is linked to the mineralization of fine roots and consequent increases in substrate availability to nitrifying bacteria (Vitousek et al. 1982, Aber et al. 1989, Fahey and Arthur 1994). Although mesh bag studies suggest that fine roots are unexpectedly resistant to decay (McClaugherty et al. 1982, Fahey et al. 1988), direct observations (Hendrick and Pregitzer 1993, Fahey and Hughes 1992) indicate very rapid decomposition of fine roots that die in their rhizosphere; mesh bag studies probably result in artifacts associated with removal of roots from the rhizosphere community. No direct observations of fine root decay have been conducted following massive mortality events (e.g., after cutting or perhaps soil freezing).
C. Effects of soil frost and freeze/thaw on soil microbial biomass and activity and trace gas fluxes While it is generally recognized that soil frost and freeze/thaw events have important effects on soil microbial biomass and activity, this disturbance is not well characterized due to the complexity of the soil ecosystem and to the multiple effects that freezing has on different biological, physical and chemical variables. Many studies have suggested that freeze/thaw cycles stress and/or kill off a portion of the soil microbial biomass (Ross 1972, Morely et al. 1983, Skogland et al. 1988, Christensen and Christensen 1991, DeLuca et al. 1992, Clein and Schimel 1995). This partial sterilization leads to a marked increase in C and N mineralization as the growth and activity of the surviving microbes are stimulated by the input of substrate ( i.e. the killed microbial cells). While this response has been well demonstrated in laboratory studies, its relevance to field conditions remains a critical question. There is also uncertainty relating to the duration of enhanced mineralization and the response to repeated freeze/thaw events. It is likely that the response of mineralization to freezing will vary with soil temperature, the size of the microbial biomass and other soil conditions (Edwards and Cresser 1992). Freeze/thaw events also affect soil structure, which influences C availability, aeration and water and nutrient movement, all of which have important effects on microbial biomass and activity.
The partial sterilization effect, combined with disruption of soil structure and input of labile organic matter from root mortality may enhance N mineralization and nitrification in ecosystems affected by freeze/thaw events (DeLuca et al. 1992, Edwards and Cresser 1992, Boutin and Robitaille 1994, Clein and Schimel 1995, Brooks et al. 1996). In many temperate forest soils, nitrification is regulated by intense competition for ammonium (NH4+) between plants, heterotrophic microbes and nitrifiers (Aber et al. 1989). Disruption of this competition at the ecosystem scale by clear cutting, blowdowns and other major disturbances stimulates nitrification and NO3- loss (Vitousek et al. 1982, Aber et al. 1989). Freeze/thaw events can be viewed in this context (i.e. as disturbances of the soil/plant system that allow for nitrification to proceed). Changes in freeze/thaw events could thus be an important disturbance influencing ecosystem N loss, acidification of drainage waters (driven by nitrification), N2O fluxes (nitrifiers can produce N2O) and CH4 fluxes (there are links between soil NH4+, nitrifiers and methanotrophs). Increased nitrification following freeze/thaw events may not result in increased NO3- availability and loss if denitrification increases as a NO3- sink. Several studies have demonstrated increases in denitrification following freeze/thaw events in forest soils (Edwards and Killham 1986, Groffman and Tiedje 1989, Christensen and Christensen 1991). In temperate forest ecosystems, denitrification tends to be most vigorous in the late winter/early spring period when soils are wet (anaerobic) and soil NO3 - concentrations are relatively high (Groffman and Tiedje 1989). However, it is likely that stimulation of nitrification will overwhelm enhanced denitrification because denitrifiers are more sensitive to freezing stress than nitrifiers (Cooke 1990) and more importantly, denitrification occurs only during the period of very wet soil conditions immediately following soil thaw, while enhanced nitrification persists for several months following soil disturbance (Aber et al. 1989, Boutin and Robitaille 1994).
Microbially-mediated fluxes of trace gases in soil are strongly linked to soil C and N dynamics and thus respond strongly to freeze/thaw events. The increase in N mineralization and nitrification following freeze/thaw enhances N2O fluxes (Bremner et al. 1980, Goodroad and Keeney 1984a,b, Cates and Keeney 1987, Christensen and Tiedje 1990) and decreases CH4 uptake (Steudler et al. 1989). Increases in C mineralization should increase CO2 production (Clein and Schimel 1995). In addition to effects on the biology of trace gas fluxes, freezing alters gas diffusion into and out of the soil (Goodroad and Keeney 1984a, Burton and Beauchamp 1994, Melloh and Crill 1995).
D. Effects of soil frost and freeze/thaw on soil solution chemistry and nutrient loss Acidic deposition has impacted soil and drainage waters in forested regions of the northeastern U.S. and eastern Canada (Jeffries et al. 1986, Landers et al. 1988). Chronic acidification of surface waters is largely due to elevated inputs of SO42-. Surface waters also exhibit episodic acidification associated with hydrologic events. In the Northeast U.S., episodic acidification of surface waters is generally associated with increases in concentrations of NO3- and dilution of concentrations of basic cations during spring snowmelt and storm events (Schaefer et al. 1989, Wigington et al. 1990). Emissions of SO2 peaked in 1970 and concentrations of SO42- in precipitation and surface waters have decreased since that time (Likens et al. 1996). However, these changes have not been accompanied by increases in acid neutralizing capacity (ANC) in drainage or surface waters (Dillon et al. 1988, Driscoll et al. 1989, Morgan 1990, Driscoll and van Dreason 1993). The recovery of surface waters from inputs of acidic deposition has been restricted due to depletion of basic cations in soil (Bailey et al. 1996a, Likens et al. 1996) and increases in concentrations of NO3- (Driscoll and van Dreason 1993, Stoddard and Murdoch 1993). Concern has been expressed that forests in the Northeast U.S. may be susceptible to "nitrogen saturation" (Aber et al. 1989, Kahl et al. 1993, Stoddard 1994). Due to elevated inputs of N from atmospheric deposition and decreasing uptake of N by vegetation associated with increasing stand age, forested watersheds are exhibiting increased leaching losses of NO3- (Hedin et al. 1995). Unfortunately, due to large pools of N in forest ecosystems and the strong biotic control of the terrestrial N cycle, it is difficult to discern "real" long-term trends in N loss.
In addition to atmospheric deposition, climate and biotic disturbances such as soil freezing may influence acidification and N loss in terrestrial ecosystems. For example, Mitchell et al. (1996) recently reported a pattern of elevated NO3- concentrations and loss from forested watersheds across the northeastern U.S. in response to a soil freezing event. Very low air temperature and limited snow cover during early winter in 1989 resulted in a soil freezing event that likely contributed to very high concentrations of NO3- during the spring of 1990. For sites in the Adirondack and Catskill regions of New York, the period of elevated NO3- loss persisted during summer baseflow conditions. Given the effects of soil freezing on root and microbial processes described above, it is likely that this regional soil freezing event stimulated nitrification, resulting in elevated NO3- losses. This event also resulted in short-term (2-yr) increases in the acidity of surface waters. It is likely that soil freezing events also result in elevated concentrations of DOC in drainage waters. Elevated export of DOC from soil enhances acidification and the transport of trace metals to downstream surface waters (Driscoll et al. 1993, Driscoll et al. 1995).
V. EXPERIMENTAL PLAN 1.
Snow manipulation
We will establish new plots for these proposed studies at the HBEF in the White Mountain National Forest in New Hampshire. Hubbard Brook has been the site of a wide range of ecological and biogeochemical studies, and is a NSF Long Term Ecological Research (LTER) site (see Results from Prior NSF Support section). Vegetation at HBEF is dominated by American beech (Fagus grandiflora), sugar maple (Acer saccharum) and yellow birch (Betula lutea), soils are acidic (pH 3.9) Typic Haplorthods. Our experimental design calls for 12 plots:
2 species x 2 snow treatments x 3 replicates = 12 plots (birch) (reduction) (maple) (reference) We are using maple and birch because of differences in cold hardiness between these species, i.e. the elevation range of yellow birch normally exceeds that of sugar maple, and birch is a common associate of red spruce and balsam fir in the lower subalpine zone of northeastern mountains. We are not using beech because this species is currently afflicted with beech bark disease at HBEF.
Each plot will include two canopy trees of the target species and will be centered in a grove of that species. Small, largely monospecific groves of sugar maple and yellow birch (typically 6-10 canopy trees) are common in the study area (Fahey and Hughes 1994). Encompassing two canopy trees will permit treatment plots of 100 m2.
We will use snow shelters to manipulate the snow depth at six of the 12 plots. Shelters will be deployed in early November, before snowfall begins, and will be left in place through mid-January. This treatment will simulate the shortened snow season that is likely to occur under global warming. Water (artificial precipitation), equivalent to the volume of snowfall, will be added to the reduction plots at three week intervals to prevent a water difference between the treatments, i.e. to simulate global warming that results in less snow and more rain.
During the first year of the project, we will design, construct, and test our snow shelters and instrument all plots for temperature, moisture, root, solute chemistry and trace gas measurements. Snow manipulation treatments will be applied in the winters of 1997/98 and 1998/99.
Snow shelters will consist of a two 5 m x 10 m wood frames with a post and beam type construction and corrugated fiberglass (or steel) roofing. The shelters will not be enclosed on the sides. The lumber for the frames will be pre-cut at U.S. Army Cold Regions Research and Engineering Laboratory (CRREL) and assembled on site (HBEF) during the first year of funding. Corner and mid-support posts will be buried at least 0.6 m in the ground to avoid frost heaving. During the winters of treatment (project years 2 and 3), roofs will be placed on the structure to prevent snow accumulation in the treatment plots. The shelter design allows snow to shed from the roof. The shelter design (including roof height above the snow surface) will take into consideration the need to minimize the effects of the roof on the energy balance components (primarily radiation). This tentative design will be tested and possibly modified prior to final installation. We are considering the feasibility of removable roofs, so that roofs are on the shelter during precipitation events only therefore minimizing the effects on the energy balance at the snow surface. This tentative design will be tested and possibly modified prior to final installation.
To accommodate the snow shelters it will be necessary to remove small trees from the plots. Because the Hubbard Brook Forest is largely even-aged and most stands are in the stem-exclusion stage of development (Oliver 1981), small trees are not abundant. In fact, overstory trees comprised an average of 94% of the total basal area during our most recent survey. Saplings will be cut at the base and removed from the plot in late summer 1996, thereby allowing about 18 months for stabilization following the disturbance.
In each plot a vertical string of temperature sensors will be installed to measure the snow and soil temperature profile at 0.1 m intervals. Temperatures will be measured from a depth of -0.5 m below, to +1.0 m above the ground surface. Small, waterproof, thermistors (accurate to +/- 0.1 °C) will be installed in the soil from -0.1 m to -0.5 m depth. Snow temperatures will be measured using 30-gage thermocouples (accurate to +/- 0.5°C) to minimize the capacity for solar loading, which is a major problem with thermistors. Thermocouples will be used to measure snow temperatures from 0 to +1.0 m above the ground. A small and simple support will allow the thermocouple string to hang, independently, away from any potential temperature altering source. The above design has been thoroughly and successfully tested in forests of northern Vermont.
Campbell Scientific data acquisition technology is well tested in winter environments and will be used for this work. A CR10x datalogger with storage module will acquire, process, and store all data. A multiplexer will be necessary to accommodate the additional channels for the thermistors and thermocouples. The datalogger and battery will be housed in an insulated enclosure to minimize temperature extremes on the reference thermistor. Soil and snow temperatures will be scanned every minute with hourly averaged values stored. One datalogger system will be used for 2 plots (we will purchase a total of 6 datalogger systems).
During the winters of testing (project years 2 and 3), snowpack characteristics (primarily depth, layering, density, and grain size) will be measured in each of the treatment plots. This information will allow a comparison of snowpack properties and processes in the different plots and provide input data for the SNTHERM model (Jordan 1991). SNTHERM is an internationally recognized, one-dimensional mass and energy balance model for predicting temperature profiles within snow and frozen soil (Jordan 1991). Additionally it calculates energy exchange at the snow surface and predicts many properties (density, liquid water content, layering, grain size) important to snow processes. Data input requirements include meteorological data (available from HBEF) and initial snow and soil temperatures and properties. SNTHERM will be initialized for Hubbard Brook forests and used to model effects of reduced snow cover on soil properties. Snow shelter construction, snow and soil temperature instrumentation, and application of the SNTHERM model will be supervised by Janet Hardy from CRREL.
2.
Root responses to treatments will be measured using minirhizotrons. We will employ a standard approach to quantify fine root dynamics as used for the past two years at HBEF, and employed successfully at other forested sites (Hendrick and Pregitzer 1992). The minirhizotron is used to obtain images at monthly intervals during the snow-free season, and the morphology and growth of individually-identified roots is recorded for each image. This information allows us to calculate fine root production, longevity and demography and their responses to treatments.
Our previous experience indicates that four minirhizotron access tubes per plot for mineral soil and two per plot for forest floor will yield sufficient root numbers to provide good estimates of longevity. In each plot, four plexiglass tubes will be installed at a 45o angle to about 50-70 cm depth (typical depth of densipan at HBEF) to measure mineral soil roots and two tubes will be inserted in the organic horizon to measure the quantitatively important forest floor roots. Tubes will be anchored to discourage image shifts owing either to loose fit (especially for forest floor tubes) or to soil frost heaving. Based upon the experience of R. Reuss (pers. comm.) for frozen soils at Bonanza Creek, Alaska, we do not anticipate problems from soil freezing, but our first-year field trials will provide experience to correct any problems that may arise. Tubes will be scribed with sampling frames (15x10 mm) to provide "permanent plots" for resampling roots, and a specially- designed indexing handle will be used to assure rapid relocation of each frame on the tubes. These methods permit efficient sampling and digitization of root cohorts using the minirhizotron technique. During the winter, tubes will be insulated in an attempt to match the snow treatments so that the tube surfaces do not depart significantly from the bulk soil temperatures. Thermocouples will be positioned on the surface of selected tubes during the trial season to assure that these procedures are effective.
Minirhizotron images will be digitized and analyzed using the efficient ROOTS computer program (Hendrick and Pregitzer 1992). The length, diameter and standard morphological criteria of each root will be recorded at monthly intervals. Based upon our previous experience, we anticipate classifying roots into several morphology classes: white roots, tan roots, brown roots, woody roots, black roots, and transparent (disappearing) roots. We recognize that root mortality is difficult to detect on the basis only of visual criteria, but a certain sign of the death of a root is its eventual disappearance from an imaged frame. The date of mortality can then be estimated by reviewing the sequence of morphology changes in the images leading up to a root's disappearance.
In each plot we will measure separately the production, mortality and longevity of roots from three depth classes: forest floor, upper mineral soil (0-15 cm depth) and deep mineral soil (15-60 cm depth). For each depth class the roots from all the tubes will be pooled into a single sample. This procedure is necessitated by the erratic occurrence of roots on the tubes; thus, our indexes of root response will be pooled values on a plot-wide basis. Annual fine root production can be calculated in two ways using minirhizotron information, both methods relying on an independent measurement of fine root biomass. First, assuming fine root biomass is at steady state, production is calculated as the ratio of biomass to median root longevity. The weighted median longevity is estimated from the longevity of each monthly root cohort (assuming birth date as the mid-point between successive images at which a root segment first appears) and the relative length production for each cohort. In the second approach, the ratio of annual production of root length to initial root length is measured for the rhizotron images. This ratio is multiplied by the initial fine root biomass to calculate production (Hendrick and Pregitzer 1996). Similarly, annual mortality is estimated using the ratio of annual length mortality to initial root length. For the purposes of this experiment we will rely upon annual measurements of fine root biomass from forests adjacent to our experimental site (using funds from NSF-LTER) because it would be impossible to precisely measure fine root biomass on each of the plots (Fahey and Hughes 1994).
Perhaps more revealing than the annual estimates of the responses of fine root production and mortality will be the seasonal responses of root phenology and demography. We expect the treatments to disrupt the normal seasonal patterns of root growth and mortality in northern hardwoods (Fahey and Hughes 1994, Hendrick and Pregitzer 1996) and we expect differential effects on the roots of sugar maple and yellow birch. Our minirhizotron data will provide quantitative information on the demographic transition probabilities and phenological timing of root birth, death and morphological changes. Thus, we will be able to evaluate both overall production and mortality responses as well as any treatment effects on the nature and timing of life history events in the root systems of mature sugar maple and yellow birch.
Litterfall and Foliage Chemistry. Treatment effects on nutrient uptake by the trees will be evaluated by measuring litterfall and foliage chemistry in each plot for two years before and two years after treatment. Litterfall will be collected with our standard litter traps (0.1 m2 baskets). Based upon our detailed litterfall studies at HBEF (Hughes and Fahey 1994) we expect that a random network of 5 traps per plot will be capable of detecting a 20% change in litter biomass at 95% confidence level. Canopy foliage will be collected in midsummer with the aid of a shotgun and samples will be pooled for analysis of foliage chemistry, area per leaf and specific leaf weight. Macronutrient concentration in foliage and leaf litter will be measured after dry combustion (N,S) or wet ashing (P, Al, basic cations) by standard methods (Mou et al. 1993).
3.
The objective of this work will be to characterize changes in microbial biomass C and N content, net N mineralization and nitrification rates, denitrification rates, concentrations of available soil C and trace gas fluxes in the different treatments. We expect that microbial biomass will decrease and that pools of available C and rates of mineralization, nitrification and denitrification will increase in plots with reduced snow cover. We expect that N2O and CO2 production will increase and that CH4 uptake will decrease in plots with reduced snow.
We expect that these changes in microbial biomass and activity will be strongly coupled to changes in root mortality (e.g. input of labile organic matter) discussed above. Minirhizotrons will provide some information on changes in substrate availability and organic matter (root) mineralization. Mortality data from the minirhizotrons can be used to calculate organic matter inputs on a monthly basis. Although mineralization cannot be measured directly with minirhizotrons, root disappearances from the images provide information on decay rates (time from death to complete decay). Together with mortality information, the root disappearance observations will indicate roughly the timing and magnitude of C and N release from roots.
Microbial biomass C and N will be measured six times per year, as the CO2 and inorganic N released in a 10 day aerobic incubation of a fumigated and re-inoculated sample (the chloroform fumigation-incubation method, Jenkinson and Powlson 1976). Our previous work at Hubbard Brook (Weintraub et al. 1995), as well as other studies (Holmes and Zak 1994) suggest that this sampling regime is adequate to characterize temporal variability in microbial biomass at HBEF.
Iin situ net N mineralization and nitrification will be measured using a "closed core" procedure (Raison et al. 1987, Robertson et al. in press). At monthly intervals, paired intact soil cores (O, A and B horizons to 20 cm will be separated) will be taken in the field. The ends of one of the cores will be covered with a polyethylene bag and returned to the soil while the other will be transported to the laboratory for extraction (2N KCl) followed by colorometric analysis of NH4+ and NO3-. The field-placed cores will be left in place for one month and will then be extracted and analyzed. Five replicate cores will be incubated in each plot. Net mineralization will be quantified as the accumulation of NH4+ plus NO3- in the field-placed cores minus the levels of NH4+ and NO3- in the initial cores. Net nitrification will be quantified as the accumulation of NO3- alone.
Denitrification rates will be measured using the acetylene-based intact core technique described by Tiedje et al. (1989) and Groffman et al. (in press). Intact soil cores (0-15 cm depth) will be collected using a 2-cm diameter punch auger fitted with plexiglass inserts. A total of 10 core samples will be collected from each plot for a total of 120 cores per sample date. We will sample at monthly intervals during summer and fall and weekly periods during spring (15 dates total).
The core samples will be stored overnight at in-field soil temperatures. On the following day, the cores will be removed from the incubation unit and sealed with rubber stoppers. Acetylene (to at least 10 kPa final concentration) will be added to the headspace of each core and mixed into the soil pores by repeated pumping with a 30 mL syringe. The cores will then be incubated at in-field soil temperatures for 6 hours. Gas samples will be removed from the headspace after 2 and 6 hours. Gas samples, blanks and standards will be stored in evacuated glass "autosampler" vials and will be analyzed for N2O by electron capture gas chromatography (GC). Annual N loss will be calculated by extrapolating measured rates over the intervals between sampling dates.
Groffman has just received funding from the USDA NRICGP to develop a new, non C2H2-based soil core method for measuring denitrification at HBEF. Depending on the progress of that project, the new method may be used for the studies proposed here.
To assess soil C availability, we will quantify levels of water extractable and biodegradable DOC. six times per year (Boyer and Groffman, in press). To quantify biodegradable DOC, we will isolate water extractable organic carbon by combining soil and distilled water and shaking overnight at 4°C, followed by centrifugation and filtration through GF/F glass fiber filters. The filtrate will be inoculated with 0.5 mL of raw soil extract and incubated at 25°C for 14 days. Biodegradable DOC will be quantified as the decrease in DOC concentration over time. Levels of DOC will be measured 4 times during the incubation using a Shimadzu 5050 DOC Analyzer.
Trace gas fluxes will be measured at least 12 times per year at each site, including overwinter. Three chamber bases will be permanently installed 4 cm into the soil at each site. The chambers will be constructed from 16.5 cm wide by 20 cm long pieces of PVC pipe fitted with a septum. An air-tight well cap will be placed on the chambers to initiate flux measurements. Samples will be taken with an air-tight syringe every 10 min for 0.5 h and added to evacuated vials. The vials will be returned to the laboratory and analyzed for CH4 by flame ionization GC for CO2 by thermal conductivity GC and N2O by electron capture GC. Standards, spikes and field blanks will be prepared along with chamber samples.
Plots will also be instrumented with soil gas sampling probes at three depths. Soil gas concentrations in freezing and thawing soils often show marked spatial and temporal variation due to ice dynamics (Fernandez and Kosian 1987, Burton and Beauchamp 1994, Melloh and Crill 1995). Soil gas measurements will thus be useful for interpreting patterns of trace gas fluxes measured during winter. Soil gas samplers will be of the design described by Burton and Beauchamp (1994), where a 1.25-cm i.d. Schedule 80 PVC tube is fitted with sampling wells constructed from 10-mL disposable syringes fitted with lengths of "spaghetti" tubing. The spaghetti tubing will be connected to double-ended sampling needles at the soil surface.
4.
Zero-tension lysimeters, similar to the design of Driscoll et al. (1988), will be installed in duplicate beneath the Oa and Bh horizons and within the lower Bs horizon of each plot by excavating a pit at the base of each of the experimental plots. Soil pits will be backfilled after installation to allow for soil water collections during the winter months. TDR probes, for soil moisture monitoring will be installed at the same depth as the lysimeters. The TDR system will be calibrated for frozen soil conditions (Spaans and Baker 1995, Seyfried and Murdock 1996). Lysimeters will be installed during the summer of 1996 and monitored for 18 months prior to the initiation of the experimental treatments. Soil solutions will be monitored at monthly intervals throughout the year, with collections weekly during spring snowmelt (March-May). Solutions will be analyzed for all major solutes including solutes of particular interest NO3-, NH4+, dissolved organic N, DOC, Al fractions and acid neutralizing capacity using the methods described in Driscoll and van Dreason (1993). The analysis of major solutes and Al fractions will enable us to characterize changes in the acid base chemistry of drainage water that occurs in response to freezing events.
In long-term studies of soil solution chemistry at the HBEF, we have utilized zero-tension lysimeters to evaluate linkages between soil solution and stream chemistry (Driscoll et al. 1988, Likens et al. 1994). These lysimeters are positioned at three elevations (750m, 730m 600m) just west of the biogeochemical reference watershed (w6) and have been used to collect soil water beneath the Oa, Bh and within the Bs horizons since 1984. These plots are adjacent to other ongoing root and microbial studies at HBEF. As part of this study we will install thermisters in these plots at 0.1 m intervals from the surface of the forest floor to the C horizon (as in the snow reduction plots) to establish a long-term record of soil temperature. This record will be used to evaluate the response of roots, microbial biomass and activity, and soil water chemistry to future natural soil freezing events.
5.
We will conduct laboratory studies to answer several specific questions that will be useful for interpreting our field results. In particular, we need to determine if soil freezing alone causes root and microbial mortality, or are very low (< -10 oC) temperatures also required. We will also determine the effects of multiple freeze thaw events, of different duration, at different soil moisture contents, on seedling, root and microbial mortality.
In the first year, we will examine the interactive effects of soil frost duration and minimum soil temperature. The experimental design will include 2 species (maple and birch) x 2 temperatures (-2 °C, -15 °C) x 2 frost durations (15 and 60 d), producing a total of 40 experimental units (including unfrozen controls and 4 replicates). Samples will be obtained from 2-year-old clearcuts near HBEF. In late summer 1996 we will excavate 0.1 m2 soil blocks containing several seedlings of the subject species. Other vegetation will be removed from the blocks. Each block will include forest floor horizons and about 5 cm of upper mineral soil. The blocks will be placed in greenhouse flats and transported to IES where they will be maintained in greenhouses until December. Treatments will be applied in growth chambers during the winter and the flats will be returned to the greenhouses in early April. The suite of microbiological and soil measurements will be similar to the field experiments, and the shoots and the root systems of the individual seedlings will be inspected and measured after careful excavations from the soil in mid-June.
During years 2 and 3 additional laboratory experiments will be conducted. The exact design of treatments will depend in part upon first-year results, but we anticipate examining the interactive effects of soil moisture with either minimum temperature or duration of freezing.
6.
Our snow manipulations will help determine if decreases in snow cover, that are likely to occur in response to global warming, have important effects on root and microbial dynamics, nutrient loss, the acid-base chemistry of drainage waters and trace gas fluxes in northern hardwood forests. While it is difficult to evaluate long-term dynamics in a three year study, we have several opportunities that will allow us evaluate the long-term significance and consequences of snowpack reduction and soil freezing on ecosystem processes.
First, our treatment is relatively extreme. Soil freezing events currently have a return frequency of 10-15 years at Hubbard Brook. By conducting experimental treatments of snowpack reduction for two consecutive years, we will accelerate the frequency of soil freezing events well above the rate expected to occur in response to global warming. Essentially, we will be simulating 20-30 years of global climate change in our three year study.
Second, by calibrating the SNTHERM model to the HBEF, we will be able to simulate snow and soil freezing using various scenarios of past and future climate conditions. We will use the model to "hindcast" previous soil freezing events such as occurred during the winters of 1969-70 and 1973-74. Results of model hindcasts will be compared with our long-term records of stream chemistry to evaluate the role of soil freezing events on element leaching and acidification of surface waters at HBEF. We will also use SNTHERM to predict how snowpack levels and freeze frequencies will change under various scenarios of future global climate change.
Finally, the long-term research effort at Hubbard Brook provides a venue for evaluating the significance of snow reductions and soil freezing in several ways. First, if this project is funded, we will incorporate the snow reduction treatments that we establish into the next cycle of the HBEF LTER project. Our plan would be to terminate the snow reduction treatments for a period of time to study plot recovery from two years of snow reduction and then re-apply the treatments at a more realistic, but still accelerated return interval. Second, as part of the project proposed here, we will instrument our long-term soil solution monitoring plots with thermistors so that we will be able to characterize the "natural experiment", i.e. a natural soil freezing event, when it occurs. The natural experiment will be a valuable assessment of the environmental relevance of our snow reduction treatments. Finally, the large body of past and ongoing work at HBEF (see Results from Prior NSF Support) will provide a context for evaluating the importance of snow and soil freezing relative to other natural and anthropogenic drivers of ecosystem processes.
VI. RESULTS FROM PRIOR NSF SUPPORT Timothy J. Fahey, Charles T. Driscoll and Peter M. Groffman
Title: Long-Term Ecological Research at the Hubbard Brook Experimental Forest
NSF Awards: BSR87-002331, 9211768; $2,500,000, $3,500,000; Date 1 Oct 87 - 31 Sept 92, 1 Oct 92 - 30 Sept 98. Driscoll and Fahey, principal investigators, Groffman, collaborator. Summary: The overall goal of the LTER study at HBEF is to develop a better understanding of the responses of northern hardwood ecosystems to natural and anthropogenic disturbances. We are conducting research on sites within the HBEF with contrasting histories of disturbance, as well as continuing the collection and analysis of long-term data sets. Our strategy has been to build upon previous and ongoing work at the HBEF through the use of long-term records and manipulated watersheds. Thus, we have used LTER funds to help support monitoring activities, initiate process-level research projects, and improve data management to facilitate integration of past and current research at the HBEF. A few of our recent initiatives include:
The Effects of Whole-Tree Clear-Cutting on Soil Processes. Whole-tree clear-cutting represents a severe ecosystem disturbance, and leads to leaching losses of nutrients from the soil profile, increased acidification, and elevated concentrations of Al in soil solutions and streamwater (Dahlgren and Driscoll 1994, Romanowicz et al. 1996). The process of nitrification, resulting in the production of nitric acid in both the forest floor and mineral soil horizons, was the principal mechanism driving these changes. The acidity generated through nitrification was largely neutralized by release and leaching of basic cations and inorganic monomeric Al (Ali). The major source of nutrient loss was from the forest floor.
Biogeochemistry of Basic Cations: Ca and K. Synthesis and integration of biogeochemical cycles at Hubbard Brook has provided insights into the mechanisms causing long-term changes, landscape-level patterns and responses to disturbance. Efforts in the past three years have focused on Ca and K (Likens et al. 1994). The combination of naturally base-poor soils, high deposition of anthropogenically-derived acids and periodically accelerated removal of bases associated with forest harvest would be expected to make the Hubbard Brook ecosystem susceptible to excessive depletion of soil base cations, especially Ca. In fact, a quantitative interpretation of our records of precipitation, soil and streamwater chemistry strongly suggests that ecosystem recovery in response to decreases in acid deposition may be delayed significantly (Likens et al. 1996). Reductions in pollution emissions of SO2 and consequent decreases in deposition of strong acids followed the passage of the Clean Air Act in 1970, but at HBEF these reductions have been accompanied by declining deposition of strong-base cations, especially Ca. Coincident reductions in streamwater concentrations of strong acids and strong bases since 1970 have resulted in a lack of response of pH and ANC of surface waters across the Northeast U.S.
Fine Root Dynamics in Northern Hardwood Forests. We have examined the demography of fine roots in forest floor horizons in four consecutive years using in situ screens (Fahey and Hughes 1994). Results have been consistent among years: median root longevity is about seven months and a high proportion of roots disappears very quickly as a result of consumption or rapid decomposition. Dynamics of roots in the Oa layer are comparable to the Oe layer. In 1993 we began minirhizotron measurements to quantify the same demographic features for mineral soil.
Publications: In addition to the studies cited above, 35 publications have resulted from this study in the past two years. Additional major publications include: Bailey et al. (1996b), Driscoll et al. (1994), Johnson et al. (1994).
D. LITERATURE CITED