During the mild winters of 1997/98 and 1998/99, our snow removal manipulation produced relatively mild freezing events (soil temperatures seldom decreased below -4o C) (Figures 1 and 2, Hardy et al. in press). Previous laboratory studies suggest that significant biogeochemical effects should only occur with a severe freeze that directly causes root and microbial mortality (Groffman et al. in press - A, Nielsen et al. in press, Clein and Schimel 1995). Surprisingly, our treatment caused significant increases in fine root mortality (Figure 3, Tierney et al. in press), nitrification rates, soil NO3- concentrations (Figure 4, Groffman et al. in press), marked acidification of soil solutions (Fitzhugh et al. in press - A), leaching of N, P, Ca and Mg (Tables 1 - 3, Fitzhugh et al. in press - B), and a shift in the timing of fine root production (Tierney et al. in press). No significant increase in rates of N mineralization was detected even though the increase in root mortality represented a significant input of labile N to the soil microbial community (Groffman et al. in press - B).
Our initial hypothesis was that snow is an important insulator of the forest soil so that reduced snowpack induces soil freezing, causing fine root and microbial mortality. This mortality subsequently results in increased hydrologic and gaseous N losses as the dead root and microbial cells are decomposed, mineralized and nitrified. While our treatment increased N losses as expected, the detailed results raised important mechanistic questions about the response of the soil ecosystem to freezing disturbance. Soil temperatures in the treatment plots were less severe than those required to cause mortality of winter hardened fine roots in laboratory experiments, yet overwinter mortality in treated plots was substantially elevated. While the mechanisms by which this mortality occurred can only be elucidated by additional study, several explanations seem plausible. First, a long-duration freeze event, such as occurred in our field plots, may have caused root mortality at temperatures which are sub-lethal during the shorter events typical of lab studies. Second, cycles of freezing and thawing which occur under field conditions may cause mortality at lower temperatures than the sustained freezing typical of lab studies (Sakai and Larcher 1987). Finally, the physical nature of freezing and root anchorage in field soils may cause mechanical damage to roots at sub-lethal temperatures, effects which might not occur in laboratory studies.
Calculation of the N release associated with the elevated fine root mortality induced by freezing indicated that it was a substantial fraction of the measured increase in hydrologic N loss, suggesting that fine root decomposition contributed significantly to the increased loss. However, the effects of the snow manipulation on N loss were likely more complex than a simple increase in mineralization associated with root mortality. Despite the root mortality, no significant increase in soil N mineralization was detected, although nitrification and soil NO3- levels were increased. These results suggest that reduced N uptake by fine roots was also important as a regulator of hydrologic loss. Fine root uptake may have been impaired directly by reduction in fine root length due to elevated mortality, and indirectly by sublethal effects on fine root physiology or severed mycorrhizal connections. Reduced fine root uptake could reduce competition for ammonium, allowing increased nitrification to occur even without a stimulation of mineralization. Thus soil freezing altered fine root dynamics and disrupted the temporal and spatial synchrony between nutrient availability and nutrient uptake in the northern hardwood forest, causing nutrient loss.
Fluxes of N leaching from the soil freezing treatment plots ranged from 1,880 to 4570 mol ha-1 yr-1 in the Oa horizon and from 488 to 1340 mol ha-1 yr-1 in the Bs horizon which are significant in comparison to wet N deposition to the HBEF during 1992-93 (525 mol ha-1 yr-1 ; Mitchell et al. 1996) as well as stream nitrate export at watershed 6 of the HBEF during 1997 (25.1 mol ha-1 yr-1 ). Accelerated mobilization of inorganic P from the forest floor, coupled with retention in occluded forms in the mineral soil horizons could affect the relative roles of N and P as limiting nutrients for forest growth. Amounts of inorganic P loss induced by freezing ranged 14.5 to 31.5 mol ha-1 yr-1 in the Oa horizon and from 0.41 to 7.1 mol ha-1 yr-1 in the Bs horizon, which are significant relative to available P pools (22 mol ha-1) and P mineralization rates in the forest floor (182 mol ha-1) in this ecosystem (Yanai 1992).
Our results have important implications for the function of forest ecosystems under changing climatic conditions. Mild freeze events are likely to become increasingly common if snowpacks develop later and melt earlier, as is predicted to occur as the climate warms due to increased atmospheric CO2 levels (Cooley 1990). Increased freeze frequency may disrupt the spatial and temporal synchrony between nutrient availability and nutrient uptake, and lead to forests with higher root turnover, less ability to retain atmospheric N deposition and higher N2O emissions. These changes could decrease timber production, increase delivery of N to receiving waters, affect competitive relationships among tree species and increase greenhouse forcing and stratospheric ozone depletion (Vitousek et al. 1997).
Our results suggest that forest species composition will be an important regulator of the response of forest ecosystems to increased freeze frequency. Sugar maple treatment plots had higher NO3- concentrations and drainage water acidity than yellow birch treatment plots. This contrasting response to treatment likely derives from differences in litter quality between these species. Litter quality has been shown to have strong effects on organic matter quality and N dynamics in soil (Pastor et al. 1984, Scott and Binkley 1997, Finzi et al. 1998). Differences in mycorrhizal associations (yellow birch is an ectomycorrhizal species, while sugar maple is endomycorrhizal) may also play a role in response freezing stress. Our results are consistent with recent studies that suggest that natural or anthropogenic factors that influence species composition will be critical determinants of ecosystem response to environmental change (Tilman 1998, Lovett and Rueth 1999, Nielsen et al. in press).
In a more basic sense, our study raises questions about the importance of freeze frequency as a regulator of the nature and extent of N cycling and loss in forest ecosystems. Perhaps freeze frequency should be considered in large-scale functional evaluations of ecosystems. Forests with soils that freeze frequently may have inherently different patterns of element cycling and loss than those that do not.
Our results also reinforce the idea that subtle changes in climate may have much more significant effects on ecosystem function than simple changes in mean temperature and precipitation (Vitousek 1994, Watson et al. 1998, Tilman 1998, Walker et al. 1999). In addition to snow cover, changes in factors such as seasonal drought, ice storms and extreme precipitation events likely need to be included in assessments of the effects of climate change on ecosystem function.
