Over the past four years, we have investigated the consequences of decreases in snow cover on biotic functions and biogeochemical processes at the Hubbard Brook Experimental Forest (HBEF), a northern hardwood-dominated forest in the White Mountains of New Hampshire (www.hbrook.sr.unh.edu). The HBEF has been the site of numerous biogeochemical studies, primarily focused on whole northern hardwood forest watershed ecosystem element budgets. While the importance of snowpack dynamics and winter climate on these budgets has been noted (Likens and Bormann 1995), there have been few detailed studies or experimental manipulations focused on overwinter processes. The objective of our experiment was to quantify the effects of decreases in snowpack accumulation on soil freezing, root dynamics of two key tree species (sugar maple, yellow birch) at HBEF, microbial biomass and activity, the chemistry of drainage water, and soil-atmosphere trace gas fluxes. In addition to the experimental work, Jordan et al. (in preparation) developed a model (SOILTHERM) of snow depth and soil frost dynamics that we use to analyze field data and to evaluate climate scenarios. A long-term (36 yr) database on streamwater chemistry is used in concert with this model to examine the effects of past natural freezing events on nutrient loss and to forecast possible future response under changing climatic conditions.

The HBEF is located near W. Thornton in central New Hampshire, USA (lat. 43°56 N, long. 71°45'W). The experimental forest covers an area of 3,138 ha and ranges in altitude from 222 to 1,015 m above sea level (Likens and Bormann 1995). Weather data has been collected at the HBEF Headquarters site since 1955. The mean air temperature is 19°C in July and -9°C in January. Average annual precipitation at HBEF is 140 cm. A continuous snowpack develops each year to a depth of approximately 1.5-m.

The forests of HBEF are a combination of deciduous and coniferous species typical of the northern hardwood forest ecosystem. American beech (Fagus grandifolia), sugar maple (Acer saccharum) and yellow birch (Betula alleghaniensis) dominate vegetation at HBEF. The forest was selectively cut in the 1880's and 1910's and some of the older stands were damaged by a hurricane in 1938. The soils at HBEF are characterized as mostly well drained, acidic (pH 3.9), spodosols (haplorthods) of sandy loam texture developed from unsorted basal tills. Soils are shallow (75 - 100 cm) with a thick (0.03-0.15-m) organic layer at the surface. The typically continuous snowpack insulates the soils, which usually remain thawed during the overwinter period (Likens and Bormann, 1995).

The specific sites for this study included two sites dominated (> 80%) by sugar maple and two dominated (> 80%) by yellow birch. The four sites are referred to as Sugar Maple 1 (SM1), Sugar Maple 2 (SM2), Yellow Birch 1 (YB1), Yellow Birch 2 (YB2) and vary in elevation, aspect and slope. Two 10 x 10 m plots were located in each stand, with one randomly designated as a treatment plot and one designated as a reference plot. In the fall and winter of 1996, we cleared minor amounts of understory vegetation from both treatment and reference plots for plot installations and to facilitate shoveling. Soil properties specific to our study sites were measured both in the field and in the laboratory.

From the first autumn snowfall until early February, the designated treatment plots were kept snow free to simulate the effects of a late accumulating snowpack on the soil temperature regime, soil freezing, and below ground biogeochemical processes. This treatment was applied during two consecutive winters: 1997-98 and 1998-1999. As soon as practical after each snowfall, shovels were used to clear the treatment plots of the new snow. We manually compacted 5 to 10-cm of snow from early-winter storms to protect plot installations and the forest floor from shovel damage and to increase the albedo of the forest floor to aid in soil freezing. We used the smooth, backside of the shovels to carefully compact the snow and protect the soil from disruption prior to its freezing. This compacted snow layer was maintained throughout the entire treatment period and observations of the forest floor each spring confirmed that the protective compact layer of snow was effective. The reference plots accumulated snow at natural rates all winter and the treatment plots accumulated snow at natural rates after shoveling stopped in early February.

Data on snow and frost depths, snow and soil temperature, volumetric soil moisture and soil water volume were collected from fall 1997 through spring 2000. This period is referred to as the "measurement period" which included two winter seasons of treatment and one winter season of "recovery."

Snow and frost depths were measured throughout the winter. We measured snow depths using a metal meter stick at least every other week, but not always at every site. Each measurement represents the mean of 10 to 100 randomly selected measures of snow depth. We took depth measurements immediately adjacent to the reference plot to avoid disturbing snow in the plot. During winter 1997-98, we calculated frost depths from hourly soil temperature data described below. This technique provided information on when soil temperatures, in 10-cm intervals, dropped below 0°C. To improve the accuracy and resolution of frost depth measurements, we installed frost tubes in the fall of 1998 (Ricard et al., 1976). We measured frost depth using the frost tubes every one to two weeks during the winters of 1998/99 and 1999-2000. During the thaw period we noted the timing and extent of soil thaw from the surface downward.

Figure 3a. Winter sampling of “N-Min Cores” at shoveled Freeze Project Sugar Maple 1 “Freeze” Plot. This is a midwinter sampling event: the cores were taken in late fall, and left to incubate for several months. At this point the soil is frozen solid, so a chisel is necessary to remove the cores. (Click for larger image.)

Figure 3b. Winter sampling of “N-Min Cores” at shoveled Freeze Project Sugar Maple 1 “Freeze” Plot. This is a midwinter sampling event: the cores were taken in late fall, and left to incubate for several months. At this point the soil is frozen solid, so a chisel is necessary to remove the cores. In this photo, we have removed the snow crust that we have allowed to form over the forest floor in the “freeze” plots. (Click for larger image.)

Figure 3c. Winter sampling of “N-Min Cores” at shoveled Freeze Project Sugar Maple 1 “Freeze” Plot. This is a midwinter sampling event: the cores were taken in late fall, and left to incubate for several months. At this point the soil is frozen solid, so a chisel is necessary to remove the cores. (Click for larger image.)

Figure 3d. Location of soil core (Click for larger image.)

For overwinter incubation, approximately 25 cores were collected in late November. Five of these cores were returned to the laboratory for immediate extraction and 20 were left to incubate in situ. Five cores were then harvested at approximately 4 wk intervals (January, February, March, April). Rates of N mineralization and nitrification were calculated from the month-to-month accumulation of total inorganic N and NO₃^- in the incubated cores. The November samples served as the "initial" extractions for all overwinter months (December through March) due to the difficulty of sampling frozen soil (Stottlemyer and Toczydlowski 1996).

Soil solutions were stored at ~ 4 °C in a constant temperature room until analysis. Samples were analyzed for the following solutes: pH; acid neutralizgin capacity, dissolved inorganic carbon (DIC), dissolved organic carbon (DOC), ammonium (NH₄⁺), nitrate (NO₃^-), soluble reactive phosphorus (SRP), total nitrogen (TN), sodium (Na⁺), potassium (K⁺), magnesium (Mg⁺⁺) and calcium (Ca⁺⁺). As concentrations of DIC, DOC, NH₄⁺, NO₃^-, and SRP may change during storage, these analyses were completed as soon as possible (typically within 2 weeks of sample collection). Analyses of SRP were initiated in February 1998, while analyses of the other solutes were completed on all samples, sample volume permitting.

Analysis of DIC was via phosphoric acid addition to convert DIC to CO₂, followed by infrared detection (Dohrmann 1984). Analysis of DOC occurred after filtration by persulfate and ultraviolet enhanced oxidation, followed by infrared detection of CO₂ (McDowell et al. 1987). Ammonium was analyzed with an autoanalyzer via phenate colorimetry (APHA 1981). Nitrate was analyzed by ion chromatography (Tabatabai and Dick 1983). Soluble reactive phosphorus was measured through the formation of a blue antimony-phospho-molybdate complex and measurement on a UV-VIS spectrophotometer at 880 nm. This technique was believed to primarily measure ortho-phosphate but due to hydrolysis, some polyphosphate, organic phosphate, and metal phosphates may be detected. Inorganic P (P_i) was assumed to equal SRP. Total nitrogen (TN) was analyzed by persulfate oxidation and analysis of NO₃^- on an autoanalyzer via hydrazine reduction (Ameel et al. 1993). Dissolved organic N (DON) was calculated as the difference between TN and inorganic N (NH₄⁺ + NO₃^-).

Soil solution pH was determined at the HBEF potentiometrically with a glass electrode within 24 hours of sample collection. Acid neutralizing capacity was measured by autotitration with strong acid and Gran plot analysis (Gran,1950). Acid neutralizing capacity could not be measured on samples collected from September 1999 onward as the result of failure of the autotitrator. Acid neutralizing capacity values during the last three months of the experiment were therefore missing. Sodium, K⁺, Mg₂⁺, and Ca₂⁺ were analyzed by ion chromatography (IC) on samples collected through January 1998 and by atomic absorption (AA) thereafter (Slavin 1968). Comparisons between results from IC and AA indicated that there were no significant differences in estimated concentrations between these methods.

The hydrological fluxes predicted by the simple model were compared with those predicted by a more complex, process-based model of catchment hydrology developed for the HBEF, BROOK90 version 3.24 (Federer 1995). Catchment parameters for W6 of the HBEF were used in BROOK90 simulations. Input data (daily values of solar radiation, minimum and maximum air temperatures, vapor pressure, and wind speed) were obtained from the HBEF web page (www.hbrook.sr.unh.edu). BROOK90 was run beginning January 1996, allowing 23 months for soil water storage to "equilibrate." Model output from December 1997 through December 1998 was used to calculate hydrological fluxes through the Oa and Bs soil horizons. BROOK90 predicted that 105 and 101 cm of water flowed through the Oa and Bs horizons, respectively, while the simple model predicted that 132 and 108 cm of flux occurred through the respective horizons. While estimates of the hydrological flux through the Bs horizon were similar between BROOK90 and the simple model, the flux through the Oa from BROOK90 seemed unreasonably small (i.e., ET occurring below the Oa horizon was likely greater than 4 cm). The variation in hydrological flux among soil solution sampling dates was strongly correlated (p < 0.005) between predictions by the simple model and BROOK90, indicating that the temporal patterns of hydrological flux were similar between the simple model and BROOK90. We therefore concluded that the simple mass balance model was a reasonable tool for calculating hydrological fluxes through the soil horizons.