REPRINTED FROM Forest Soils and Land Use. Proc. Fifth North Am. For. Soils. Conf. Colo. State Univ. (Aug.,"1978.)

Transcription

1 REPRINTED FROM Forest Soils and Land Use. Proc. Fifth North Am. For. Soils. Conf. Colo. State Univ. (Aug.,"1978.) IMPACTS OF TIMBER HARVEST AND REGENERATION SYSTEMS ON STREAM FLOW AND SOILS IN THE EASTERN DECIDUOUS REGION E. L. Stone, W. T. Swank and J. W. Hornbeck In any view of the vegetation of North America, the eastern deciduous forest appears as a natural physionomic unit. It stands in contrast to the boreal and mixed conifer forests to the North, the pine-rich forests of the South and Southeast, and the prairie transitions westward, though nowheres sharply separated from these. Merely bounding the region, however, suggests the great range in growing season, precipitation, and snow cover. The region is also characterized by a wide variety of soil materials, slopes, and age and depth of weathering. Superimposed on this diversity are the impacts of 1 1/2 to 3 1/2 centuries of agricultural settlement. A large fraction of the present forest is secondary on soils formerly tilled or cleared for pasture. Most of the continuous forest has been harvested one or more times and some portions have been grazed or destructively burned. Apart from some notable exceptions, the apparent "naturalness" of the deciduous forest is a tribute to its recuperative abilities rather than forbearance on the part of the past users. This review draws together our present understanding of how cutting in eastern deciduous forests for any of a variety of purposes affects soil and streamflow. Our aims are to indicate principles that now seem well established, and to suggest magnitudes of expected changes and recovery rates. A major limitation to generalization is that most of the data on streamflow response comes from only five experimental areas Coweeta (N.C.), Fernow (W. Va.), Walker Branch (Tenn.), Leading.Ridge (Penn.) and Hubbard Brook (N.H.) all of which are sited in moderately steep terrain and, excepting Leading Ridge, in per-humid climates. Thus, our discussion is dominated by results from the most humid and mountainous portions of the region, with only broad inferences about the remainder. These experimental results and interpretations are not representative of landscapes dominated by lateral movement of water through ill-drained soils with perched water tables. Neither intensive site preparation, fire, broadcast application of herbicides, nor fertilization are commonly used in regenerating native deciduous forests, although Respectively, Charles Lathrop Pack, Professor of Forest Soils, Cornell University, Ithaca, NY 14853; Program Director, Ecosystem Studies, National Science Foundation, Washington, DC 20550; and Principal Forest Hydrologist, Northeastern Forest Experiment Station, Forestry Sciences Laboratory, Durham, NH

2 they may be in its conversion to pine, and so we have left their consideration to the companion papers. Attempts to generalize and predict the ecological consequences of present management activities are greatly aided by comparisons with undisturbed reference areas. Since none of the watershed research areas are actually pristine, we, like others, use the terms "undisturbed" or "uncut" in a relative sense, with a time frame of several decades rather than centuries. A particular need is to develop sets of baseline or benchmark values from such relatively undisturbed areas as gauges with which to compare the nature, magnitude and duration of changes induced by treatment. These baselines may be single numbers, ratios, regressions or distribution curves, as required by complexity of properties and relationships. Application of present baseline values must be tempered by awareness of the short period and imperfections of data collection, the long response times of forest ecosystems (see, for example, Leak, 1974) and, especially, the great physical diversity of the region. Water Yield Major effects of forest cutting, or any other reduction in live canopy, are reduced transpiration and interception. The consequences are greater availability of soil water for remaining plants, greater water movement downward or laterally and, eventually, greater addition.to groundwater or streamflow. Of these, streamflow from water-tight basins proves to be the most readily and accurately measured. Hence volume, flow rate, and chemical composition of stream water before and after treatment have come to be major indicators of, net impacts and ecosystem response. Of the standard harvest and regeneration methods, complete clearcutting obviously produces the maximum reduction in overstory canopy and thus in evapotranspiration. Elimination of residual or redeveloping leaf surface, as through herbicide use, burning or intensive site preparation, further increases amount and duration of the yield increase. On the other hand, the "commercial" clearcuts of the recent past often left a sparse residual canopy of small or unwanted trees with appreciable capacity to transpire even in the first season after cutting. Data from experimental watersheds indicate that increased water yield from completely clearcut areas allowed to regenerate naturally will commonly range around 25 to 40 cm in the first year. These amounts are normal maxima for more humid areas. Actual amounts will vary downward on shallow soils with limited moisture storage capacity, in climates with low streamflow rates, and wherever an appreciable transpiring canopy remains. The yield decreases in subsequent years i.e., return to base streamflow levels often approximate a logarithmic decay curve (Fig. 1), although there are exceptions. The time to return depends both on amount of firstyear increase and on a rate term related to rapidity of canopy redevelopment, Slow restoration of a full canopy, and perhaps the time required for full root occupancy of deep soils, reduce the rate of decrease, as appears to have been illustrated for Coweeta watershed No. 13 (Swank and Helvey, 1970). 517

3 YEARS AFTER CUTTING OR HERBICIDE Figure 1. The return of increased streamflow towards base level in the years following cutting or herbicide use follows a logarithmic decay curve. Numbers refer to designation of experimental watersheds. Large year-to-year variations from the generalized curve may be associated with marked departures of growing season precipitation from long-term means. Otherwise, with prompt and full restocking the average duration is commonly about 6 to 10 years. Such generalized patterns for clearcutting studies aid in understanding how partial or sequential cutting affects yield. Douglass and Swank (1972) suggested that first year increases following partial cutting in merchantable forests were proportional to the reduction in basal area hence in canopy coverage above a threshold value of about 13% basal area (Fig. 2). The existence of a threshold is plausible when cutting is confined to scattered individual trees or narrow strips, because much of the radiant energy and soil moisture formerly absorbed by the removed trees is intercepted by their neighbors. Such border effects have less consequence when the same total removal is concentrated in larger units. Hence the data of Fig. 2 can also be represented as a continuous function incorporating an additional variable, the energy available for evapotranspiration, resulting from the range in slope, aspect and latitude of the experimental watersheds. This model (Fig. 3), combined with a logarithmic decay curve over subsequent years, affords a plausible basis for predicting the expected effects of partial cutting and subsequent return to baseline values. 518

4 u LJ (t O COWEETA O^DFERNOW B LEADING RIDGE AHUBBARD BROOK o o o A 20 o z tr 10 cc U- 0 "0 20 4O KX) REDUCTION IN BASAL AREA (%) Figure 2. Streamflow increases in the first year after cutting are related to the reduction in stand basal area (Douglass and Swank, 1972). 50 r Insolotion Index REDUCTION IN BASAL AREA (%) Figure 3. Strearaflow increases in the first year after cutting are related to isolation index of the watershed, as well as basal area reduction (R =.89) (Douglass and Swank, 1975). 519

5 An exception to estimating yield increases from canopy reduction alone likely will arise when streamside bands of trees or tall shrubs are left for temperature control or other protection. The margins of such "leave strips" receive additional energy, and have free access to any subsurface flow at times when moisture stress reduces transpiration of upland trees. With the per-humid climate and deep soils or Coweeta, the cost of such strips in terms of yield increases may simply equal that of an equal strip of upland forest (Helvey and Hewlett, 1962), but elsewhere it is likely to be appreciably greater (Mader et al., 1972; Anderson et al., 1976). Similarly, certain cutting configurations on certain soils may alter the expected yield increases. For example, at Hubbard Brook, 1/3 of a small, south-facing watershed was clearcut in the form of east-west strips 25 m wide, separated by 50 m uncut strips. Water yield increases for the first and second years after cutting were only about 1/10, rather than 1/3, of those obtained from a nearby block clearcut (Hornbeck et al., 1975). A small fraction of the difference probably was due to an uncut shade strip along the stream channel, which was lacking in the block clearcut. A more important cause must have been moisture use by the residual stand. Halfwidth of the 25 m strips was less than the maximum rooting radius of adjacent trees; moreover, the strips averaged 800 m of E-W border per hectare cut. Equally important, perhaps, a dense fragipan over most of the area forced excess moisture to flow laterally through the rooting zone. Successive cutting of additional 25 m strips at 2-year intervals produced relatively larger increases in streamflow, although total yield increase for the first 6 years was only 58% of that from the paired block clearcut (Hornbeck and Likens, 1978). Suppression of residual and regenerating foliage by herbicides after clearcutting increases first year water yields and delays return to base level (Douglass and Swank, 1975; Hornbeck, 1975; Kochenderfer and Aubertin, 1975). Conversely, replacement of a deciduous forest by a dense evergreen cover reduced annual streamflow below the original base level. This is demonstrated most strikingly by direct replacement studies at Coweeta (Fig. 4) (Swank and Miner, 1968; Swank and Douglass, 1974), although predictable from the soil moisture studies of Urie (1959); Holstener-Jorgensen (1961); and Levy (1969). Soil moisture supply seldom limits transpiration at Coweeta and substitution of white pine for hardwoods here reduced annual water yield by about 20 cm below base level 15 years after planting. By comparison, Urie's (1977) estimates for Michigan, where moisture stress is common, are about 10 cm less for pine on well drained sands and 13 cm less on imperfectly drained soils. Greater apparent use by pine is due in part to year-around interception loss and a longer season for transpiration, rather than any major difference in daily transpiration rates when the hardwoods are in full leaf (Zahner, 1955). Thus, experimental manipulations of vegetation have increased streamflow by as much as 40 cm and reduced it by about 20 cm, compared with the base level of undisturbed deciduous forest. This range especially affects relative stream volume during the "low flow" stage, during and after the period of active transpiration, as discussed later. Such differences in stream flow also affect absolute loss rates of nutrients from the system 520

6 e.g. the above range of 60 cm water represents a 6.0 kg/ha range in transport of an element present at a constant concentration of only 1 ppm. CALIBRATION PERIOD OQQ 1 I5-1 5z -j ' 10 5 Q UJ 10 LU o o 1 o: O -j t O -5 ^~H_r^ i-p-^ L YEAR Figure 4. Replacement of native hardwood forest by white pine plantation decreases streamflow. Bars indicate departure of measured streamflow from that predicted for the uncut condition. Coweeta Watershed No. 1. (Swank and Douglass, 1974). Perched Water Tables Large areas of deciduous forest, including much secondary forest on formerly cultivated land, occur on nearly level to gently rolling landscapes with fragipans or slowly permeable subsoils. Both rooting depth and moisture storage are restricted. Temporary, or perched water tables within the soil mantle are widespread, however, at least during winter and early spring (Lyford, 1964; Fritton and Olson, 1972). Excess water moves laterally to streams through an elaborate mosaic of soil drainage classes (Lyford, 1974), the morphological features of which reflect the height and duration of those water tables. Though data on the effects of forest cutting in such landscapes are few (Fletcher and McDermott, 1957), water movement must be more complex than in the steeply sloping landscapes discussed thus far (Hewlett and Hibbert, 1967; Helvey et al., 1972). Complete clearcutting would increase height and duration of the temporary soil water tables as excess water moved slowly streamward. Seep areas would remain wet longer. There would be opportunity for direct evapotranspiration from wet surfaces and for maximum transpiration from developing vegetation, probably reducing yield increases below those expected on hydrologically deep soils. The longer duration of soil saturation, or even ponding in very poorly drained areas, would discriminate against or favor 521

7 species according to their tolerance of poor aeration. The return to base level, however, should follow a decay curve, much as in Fig. 1. Where perched water tables are common, partial cuttings would affect streamflow much less than anticipated from Figs. 2 and 3, and especially so in climates where growing season potential ET greatly exceeds precipitation. A portion of the excess water made available by cutting would be utilized by uncut areas downslope. Thus in such landscapes streamflow may be a less sensitive indicator of hydrological events and recovery in the sub-ecosystems comprising the watershed than in more steeply sloping basins. Substituting conifers for deciduous cover in landscapes with poorly or imperfectly drained soils would reduce duration and height of perched water tables (Levy, 1969), although the actual magnitude of change would be influenced by amount of fall and winter precipitation. Conversely, practices that eliminate an appreciable conifer component from mixed stands would tend towards an opposite effect. Where actual ET was already well below potential, however, water formerly transpired by conifers early in the growing season might well be used by their deciduous replacements (Urie, 1959) leaving little excess available for streamflow. Low Flow Reduced transpiration after heavy forest cutting obviously means higher soil moisture levels and less storage for current precipitation. These consequences are manifested not only in greater total yield but in greater discharge during and immediately after the growing season when streamflow from uncut forest is normally lowest (Fig. 5), and sometimes nonexistent (Kochenderfer and Aubertin, 1975). Results from Fernow and Coweeta show increases to be on the order of 1-3 cm per month after clearcutting (Douglas and Swank, 1975; Kochenderfer and Aubertin, 1975). Although often representing only a small part of the total yearly runoff in more humid areas, increased low flow volumes have evident consequences for stream biota and downstream users. Low flow increases in the first and subsequent years after cutting can be expected to roughly parallel those in total yield and to be affected similarly by type and intensity of harvest. Within this framework, however, low flow increases are likely to be proportionately less important or consistent on very shallow soils with either very high or very low summer rainfall. Peak Flow The reduced drain on stored soil water after cutting means less available storage; hence a greater percentage of summer and fall storms appears as streamflow, including higher storm peaks and quickflow volumes (Hewlett and Helvey, 1970). Peak flow differences between harvested and uncut areas due to this mechanism disappear as soon as the soils of both are fully recharged (Lull and Sopper, 1965; Anderson et al., 1976). Prior to recharge, peak differences are least and of shortest duration for hydrologically shallow soils, and greatest for deep friable soils in areas of low summer 522

8 SO 100 PERCENT OF TIME FLOW IS EQUALLED OR EXCEEDS) Figure 5. Partial forest removal increases low flow volumes. Measured flow from partially cut watershed (solid line) lies outside 95% confidence limits (horizontal bars) of flow-duration curve predicted for uncut condition (dashed line). Leading Ridge Watershed No. 2 (Lynch et al., 1975). rainfall. The return to pre-harvest base level follows the reduction in increased total yield. The differences in peak flows between harvested and uncut are seldom influential during periodic major floods, because usually they occur after previous rainfall has recharged any soil moisture deficits (Hornbeck, 1973). Peak flows resulting from rapid snow melt or rain-on-snow affect only part of the region. Here the effect of reducing even the leafless crown canopy is to hasten snow melt in early spring. The earlier and more rapid melt on cutover areas causes peak flow volumes to increase in the early part of the melt season, and decrease in the latter part (Verry, 1972; Hornbeck, 1973). The desynchronization of peak flows between cutover and uncut areas lowers the total peak over the whole of a large watershed. Road and skid trail systems sometimes increase peak flows, notably when designed poorly or with no regard to future consequences. Even well designed roads have some effect where deep side hill cuts intercept subsurface storm flow and channel it rapidly to streams. If either design or maintenance is inadequate erosion may entrench ruts and ditches, converting logging roads

9 and skid trails into a new drainage net. Among other consequences, such a net hastens the flow of excess water towards streams and can markedly increase peak flow. Changes of this nature tend to be progressive, preventing return to preharvest base levels. The quality of road and skid trail layout and their subsequent fate have no necessary relationship to choice of harvest system in the managed forest. Exploitive logging, however, often takes the form of commercial clearcut or diameter limit selection systems, coupled with little or no concern for the long run impact of the casually located transport system. Such a combination can result in drastic increases in peak flow. On the other hand, heavy harvests coupled with carefully engineered and maintained transport systems result in peak flows that are little different from those of uncut forests (Lull and Reinhart, 1972). Soil Moisture Those concerned with development of forest regeneration, wildlife foods, or other vegetation after cutting have more direct interest in soil moisture availability than in streamflow. The amount of available moisture stored within the rooting depth varies enormously with soil characteristics and sometimes with completeness of dormant season recharge from only a few cm to over 50 cm. This is replenished in some degree by growing season rainfall, Complete removal of the overstory leaves the stored moisture at the disposal of any residual and redeveloping vegetation, although surface evaporation losses increase sharply, as does temperature, with full sunlight. The foregoing synposis is now conventional wisdom, documented by numerous studies in this region and others (Fletcher and McDermott, 1957; Helvey and Hewlett, 1962; Patric, 1973; Anderson et al., 1976; Sartz and Knighton, 1978). Moisture availability declines as the new vegetation develops, but often irregularly because of patchiness in kind and density of cover, and variable penetration of roots. Streamflow patterns (Fig. 1) reflect the greater availability and gradual return to base level for the watershed as a whole. Small patches, however, will deviate markedly from this average pattern as, for example, when they lag behind others in redevelopment of deep-rooted cover, or are situated to receive unused water from upslope. The. persistence of early succession or mesic vegetation in such areas affects wildlife habitats and vegetative diversity. Only local observation provides a basis for estimating such lags and their consequences. The contrast in soil moisture between complete, and partial or sequential cutting depends very much on degree and pattern of the latter. Soil moisture conditions in the center of moderately wide strip or group selection cuts resemble those described above except that the likelihood of augmented lateral flow in downslope positions is less. Average soil moisture content will diminish with proximity to the uncut forest, becoming least in openings created by single tree selection. Nevertheless, even single tree removals leave some small soil volumes only sparsely occupied by roots of adjacent trees, and thus as islands of greater moisture availability for a brief time.. 524

10 St r e am Temperature Removal of the overhead canopy increases'maximum summer temperatures of streams by several degrees, up to 4-6 C (Swift and Messer, 1971; Hornbeck and Federer, 1975; Kochenderfer and Aubertin, 1975; Lynch et al., 1975). In the climates of the eastern deciduous forests such increases often exceed the favorable upper limit, and sometimes the lethal limit for cool water fish. Past concern in this region has been chiefly with clearcutting in the vicinity of streams valued for trout fishing. Experimental work on small watersheds demonstrate that adverse increases in stream temperatures can be avoided by maintaining a cover of high shrubs or trees along much or all of the live stream channel. Such "leave strips" have no other values as well, but may pose problems such as windthrow, riparian water consumption, and lower wood yields. Their effectiveness in temperature control, however, means that elevated stream temperature need not be regarded as an inevitable consequence of harvest method but is subject to analysis and tradeoff considerations in planning forest uses. Sedimentation The primary sources of any additional sediment resulting from timber harvesting activities are roads, skid trails and landings, as numerous reviews indicate (Lull and Reinhart, 1972; Stone, 1973; Anderson et al., 1976). This is notably so in deciduous forests, where post-harvest fire or intensive site preparation is seldom used, regrowth is rapid, and mass erosion is a rare consequence of logging (Patric, 1976). Small and temporary increases in stream turbidity are almost inevitable consequences of any product removal but can be made minimal by careful layout and construction of roads, caution in wet weather logging, and post-harvested care. When such precautions are followed, damage to aquatic resources and water quality from sedimentation is essentially nil, as many investigators have concluded (Reinhart et al., 1963; Hornbeck and Federer, 1975; Kochenderfer and Aubertin, 1975). Unfortunately, however, the procedures now so effective on well managed forests (Kochenderfer, 1970) are still not being applied on large areas of forest land. Although many operations have little effect on stream turbidity or bed load, the number that do is sufficient to increasingly generate regulatory action. The form of harvest or regeneration method of itself has little direct influence on sedimentation hazard. It does, however, determine the total length of roads and skid trails being constructed or in active use to harvest the same annual volume from a forest property under sustained yield. Roads and trails are most vulnerable to erosion at these stages. Obviously, the length in active use must be greater where harvest is diffused widely, as under selection systems, than where concentrated in a few heavily cut areas. Deterioration of debris dams in high gradient streams, or of root systems supporting over-steepened banks after clearcutting may allow temporary small increases in turbidity or bed load transport. These can be expected to return to base level as forest cover re-establishes and, of course, would be largely avoided where protective strips were left.

11 Nutrients Nutrient losses resulting from harvest and regeneration activities carry the implied threat of reduced productivity unless long-term nutrient availability is maintained by natrual replacement or artificial measures. A further concern is that nutrients leached into streams from treated areas may adversely affect downstream values and uses. Direct measurement of nutrient losses in the soil is difficult, however, and productivity changes may not be fully apparent for many years. Since water is the major vehicle of nutrient transport from the landscape (apart from product removal and, possibly, fire) analysis of stream concentrations is our present best indicator of loss rate and the return to pre-harvest levels. The concern for nutrient removal in harvest is a century-old issue (Rennie, 1955) that has now reappeared in more serious form. In the deciduous forest removals in conventional harvests of only the usable stemwood and the attached bark range from less than one (P) up to 20 (Ca) kg/ha/yr for the various major nutrients. These amounts are generally believed less than the rate of replacement by atmospheric additions and mineral weathering or, for nitrogen, biological fixation. Intensive harvest practices that remove all stems as well as branches and twigs (which are higher in element content), increase nutrient removal by a factor of 2 to 4. Numerous summations of this sort have been made or can be derived from published values (Boyle et al., 1973; Hornbeck, 1977; Likens et al., 1978). The capacity of sites to tolerate such losses without depletion depends on sustained "plant effective" availability from both atmosphere and soil sources, rather than gross totals, and cannot be estimated accurately from our present methods. Since soils of the region vary greatly in availability and "supplying power" of various nutrients under agricultural cropping, the same soils under forest will certainly not respond uniformly to intensive harvest. Results from long-term agricultural experiments of continuous cropping without nutrient replenishment (Anon., 1960) suggest the likely responses of forests: Wherever net annual rate of removal exceeds replacement of "available" nutrient pools, yields decrease, either gradually or steeply, towards a new level determined by the annual supply of the most critical element. In forests, reduced growth would be reflected in longer rotations or cutting cycles. In all likelihood the first stages of depletion would bring shifts in overstory and understory species as "nutrient demanding" species (Mitchell and Chandler, 1939) were placed at a competitive disadvantage. Presumably, depletion would be more rapid in secondary forests on lands that had been cultivated without net replacement. Nutrient concentrations and total export in stream water represent integrated values for the entire watershed rather than simple averages from different points. Nutrients released at one point may be absorbed downslope, as suggested later. Another avenue of loss, although its magnitude is unknown, is denitrification; its occurrence would reduce the measured outflow of both nitrate and associated cations. The influence of substrate lithology sometimes outweighs that of the ecosystem above it, as at Walker Branch 526

12 where Ca export from the underlying limestone is two to threp times greater than the entire annual uptake by the forest (Table 1) and no indication of the tight cycling in the highly weathered surface soils. Nevertheless, nutrient cycles constructed for three eastern deciduous forests, representing a range of substrates and climates, show a number of similarities in component magnitudes (Table 1). This gives some hope of Table 1. CONTENT OR FLUX IN THREE UNDISTURBED HARDWOOD FORESTS. (Henderson et al., 1978; Likens et al., 1977) NITROGEN Walker Hubbard Unit Coweeta Branch Brook Coweeta CALCIUM Walker Hubbard Branch Brook Vegetation Litter Fall Forest Floor Mineral Soil Watershed Loss Watershed Loss kg/ha kg/ha/yr kg/ha kg/ha * cm kg/ha/yr % Amount Cycling * cm depth. eventually establishing generalized baselines that will be applicable to many other forests. The stream losses of N at Hubbard Brook are high relative to the other two areas, despite similar inputs in precipitation (5-7 kg/ha/yr) but amount to only about 0.4% of the total content in the forest floor, the site of most rapid turnover. Generalities about dissolved nutrient losses after forest cutting and their magnitudes are now becoming clear, although explanations for these and for marked differences among locations still involve many assumptions or unknowns. Two difficulties In translating research results.into ron.1. 1:1 fc applicability are, first, that many experimental treatments were focused on water and not intended to simulate commercial forest practices. Although these experiments revealed previously unappreciated avenues of loss, the magnitudes of response may not be applicable to commercial harvests. Second, published results embody a variety of methodologies, analytical precision, and units of measure for example, concentration, conductivity, and annual mass loss.

13 The greater stream discharge after cutting necessarily means proportionately greater nutrient loss if concentration of elements are unchanged. Increased losses of elements from surface soils through this mechanism alone nre relatively modest. Increased flow becomes progressively more important when concentrations also rise above base levels. Although small changes in concentration probably arise from a variety of causes, the mechanism driving large increases is formation of nitrate. Increased temperature and moisture in the upper soil layers after cutting lead to greater decomposition or organic matter, with mineralization of some part of the organic nitrogen. The subsequent conversion of amtndhiacal nitrogen to the soluble nitrate anion is the critical step. Formation of the anion in turn brings exchangeable bases into solution. Further microbial uptake, root absorption or denitrification may reduce the amount of nitrate and bases in solution; otherwise precipitation moves the excess into groundwater or streamflow. The same coupling of cation loss with nitrate is well demonstrated in agricultural lysimeters (Raney, 1960). Hence, streamflow concentrations are affected not simply by the on-site physical changes induced by cutting but by the amount of nitrogen accessible for release, by the proportion actually converted to nitrate, and by whatever events occur along its pathway to the channel. This detail is given because experimental watersheds of the region fall into two distinct groups in respect to magnitude of losses after clearcutting, Losses of NCL-N and associated bases are relatively large at Hubbard Brook, but smaller at Leading Ridge and small or non-apparent at Fernow, Coweeta and Marcell (Corbett et al., 1978). One conspicuous difference between the two groups is the quantity of nitrogen contained in the forest floor (Fig. 6 and Table 1) where it is subject to rapid transformation after harvest cutting or comparable disturbance. The forest floor at Hubbard Brook is high both in total weight and in nitrogen concentration, relative to most other deciduous forests and many coniferous ones also (Fig. 6). The Hubbard Brook location is by no means unique, however, suggesting that similar losses may be expected wherever comparable forest floors occur in climates of equal leaching potential. Losses from the completely denuded watershed at Hubbard Brook attracted widespread attention. They totaled 165 kg N and 236 kg Ca per ha in the first two years after cutting (Likens et al., 1978), with maximum streamflow concentrations rising to 18 ppm NO -N during the second years. Loss rates remained high for two additional years but, as the authors strikingly demonstrate, fell rapidly after herbicide application ceased and vegetation redeveloped. Nitrate losses fell to base level of the uncut forest, and Ca and K to near base level in the same season that regrowth attained a net above-ground production of only 2 mt/ha. Losses from clearcut areas allowed to regenerate naturally after harvest have been much lower, but nevertheless reveal the same pattern of large NO increases in the second year. Excess loss, that is, above the predicted baseline value, totaled 45 kg N and 24 kg Ca per ha for the first two years, 528

14 NITROGEN CONTENT, kg/ha 300O FOREST FLOOR, VOLATILE MATTER ONLY, mt/ho Figure 6. Well-developed forest floors under hardwoods (open triangles and star) usually have higher N/OM ratios than those under conifers (all other points 1000 kg N). Where OM/C = 1.85, solid line represents a C/N ratio of 36 and the Hubbard Brook value (star) is 26 (Literature data compiled by Stone, 1975). and only 60 and 36 kg, respectively, for the first five years. For both nutrients, an approximate return to baseline occurred in the 5th year after cutting, slightly sooner than for water yield (Hornbeck and Likens, 3978, and unpubl. ). Streamwater samples from several commercial clearcuts in the vicinity of Hubbard Brook confirm the generality of these nitrate loss patterns (Pierce et a.l., 1972). For reasons that are not apparent, the estimated annual losses from one location, Gale River, substantially exceeded those from the others as well from the complete clearcut on the Hubbard Brook Experimental Forest. The progressive strip clearcut watershed at Hubbard Brook 1/3 cut in successive 25 m wide strips each second year yielded only 58% as much additional water above baseline as the complete block clearcut. Similarly, it yielded only 49% as much excess NO -N and 70% as much Ca as the block clearcut (Hornbeck et al., 1975; Hornbeck and Likens, 1978). It is not clear whether the lower yield of nitrate was due to a lesser rate of formation, or to plant uptake as percolating waters moved through adjacent uncut or regenerating strips. In any case, it appears that on such soils partial or progressive cutting can effectively reduce loss of dissolved nutrients from the watershed (Hornbeck and Likens, 1978).

15 An aspen clearcutting study on the Marcell Experiment Forest, Minnesota, showed no difference in streamwater nutrient concentrations before and after treatment (Verry, 1972). Hence excess nutrient losses were low and due solely to greater water yield-some 31% in the first year. Dissolved nitrogen losses averaged only about 1.5 ppm, with over half in organic form. This study may not be wholly comparable with others cited, however, because both treated and control watersheds contained central bogs that might have affected concentrations of waters passing through them. At Coweeta, chemical analyses of streamflow began too late to characterize losses immediately after cutting. Average annual concentrations of NO -N in these streams, treated or control, are exceedingly low, ranging from to 0.02 ppm (Swank and Douglass, 1975, 1977) as compared with a mean of 0.42 for undisturbed watersheds at Hubbard Hrook (Likens ct al., 1977). The high precision of analysis, however, allows detection of an apparent increase in NO -N from watersheds cut, wholly or in part, 10 to 17 years previously and allowed to regenerate without interference. The actual mass loss is small, about 2 kg/ha/yr more than the undisturbed controls. These observations suggest small persistent differences in biological processes controlling output of this important element. As might be expected, decreased water yield following conversion of native hardwood to white pine plantations at Cdwefeta likewise decreases net annual loss of cations, resulting in a 1.5 to 3.5 kg/ha/yr greater retention than the undisturbed hardwood control plots. Tentative Generalizations Douglass and Swank (1972) have commented on the difficulty in generalizing from experimental watersheds to others within similar rainfall regimes. As our cautionary preface indicates, such difficulties are magnified when a larger region and variety of vegetative treatments are considered. Nevertheless, the need to extend research findings beyond their immediate site is evident. -In Fig. 7, we generalize, in a familiar format, our present imperfect understanding of how several streamflow and soil parameters should respond to the spectrum of harvest and regeneration treatments in the deciduous forest, away from its subhumid margins. This mode of expression does not pretend to be quantitative, and the ordinate axes should npt be read as equal, nor even necessarily linear. Further, the information available is too scanty to warrant separate rating of the various forms of partial cutting designed to produce even-age stands, and these must be viewed as intermediates between shelterwood and clearcutting, though usually similar to the former in effect. These diagrams indicate that most of the adverse impacts considered are either small to moderate, or can be made so by well-tested specific practices, as in respect to treatment of roads. It does not follow, however, that all pronounced impacts are necessarily of the same consequence in all landscapes, or are to be avoided at all costs: only that their magnitude, environmental costs, and means of mitigation be recognized and weighed in the forest planning process. 530

16 VH H M LOW 4 FLOW DlSSOLVED NUTRIENT LOSS (THICK ORGANIC LAYER) i i VL VH WATER YIELD NUTRIENT REMOVAL IN HARVEST- AVE. ANNUAL VL VH l SEDIMENTATION POOR ROADS SKID-TRAILS MAXIMUM STREAM TEMP. VL VH _L PEAK FLOW IF,. (SNOW / MELT) / GOOD CONTROL OF ROADS 8 SKID-TRAILS X POOR ROAOS SKID-TRAILS 'GOOD CONTROL REDUCTION IN SURFACE SOIL WITH SHADE STRIP I l O.M. VL UNCUT SELECTION SHELTER- CLEAR-CUT INTENS. UNCUT SELECTION SHELTER- CLEAR-CUT INTENSIVE WOOD CLEAR-CUT WOOD CLEAR-CUT Figure 7. A schematic synopsis of how harvest methods may affect several environmental parameters in the Eastern Deciduous Forest. The horizontal axis represents harvest methods as a continuous range. The vertical axis scale of impacts ranges from very low (VL) to moderate (M) to very high (VH) but is non-quantitative, non-linear and relative. It is suggestive for the region as a whole, rather than exact for any location. Note that nutrient removals represent yearly averages over an entire rotation period, and so do not differ by harvest method in the managed forest, except as intensive harvesting removes more and finer material. 531

17 LITERATURE CITED Anderson, H. W., M. D. Hoover and K. G. Reinhart t Forests and water: effects of forest management on floods, sedimentation, and water supply. USDA Forest Serv. Gen. Tech. Rep. PSW p. Anonymous (Agronomy Department) The Morrow plots. Univ. Illinois Coll. Agr. Circ Boyle, J. R., J. J. Phillips and A. R. Ek "Whole tree" harvesting: Nutrient budget evaluation. J. Forest. 71: Corbett, E. S., J. A. Lynch and W. E. Sopper Timber harvesting practices and water quality in the eastern United States. J. Forest. 76: Douglass, J. E. and W. T. Swank Streamflow modification through management of Eastern Forests. USDA Forest Serv. Res. Pap. SE p. Douglass, J. E. and W. T. Swank Effects of management practices on water quality and quantity: Coweeta Hydrological Laboratory, North Carolina, p Jn_Municipal Watershed Management Symp. Proc., USDA Forest Serv. Gen. Tech. Rep. NE-13. Fletcher, P. W. and R. E. McDermott Moisture depletion by forest cover on a seasonally saturated Ozark ridge soil. Soil Sci. Soc. Am. Proc. 21: Fritton, D. D. and G. W. Olson Depth to apparent water table in 17 New York soils from 1963 to Cornell Univ. Agr. Exp. Sta. Food Life Sci. Bull p. Helvey, J. D. and J. D. Hewlett The annual range of soil moisture under high rainfall in the southern Appalachians. J. Forest. 60: Helvey, J. D., J. D. Hewlett and J. E. Douglass Predicting soil moisture in the southern Appalachians. Soil Sci. Soc. Am. Proc. 36: Henderson, G. S., W. T. Swank, J. B. Waide and C. C. Grier Nutrient budgets of Appalachian and Cascade region watersheds: A comparison. Forest Sci. 24: Hewlett, J. D. and J. D. Helvey Effects of forest clear-felling on the storm hydrograph. Water Resour. Res. 6: Hewlett, J. D. and A. R. Hibbert Factors affecting the response of small watersheds to precipitation in humid areas, p In W. E. Sopper and H. W. Lull (eds.) Forest Hydrology. Pergamon Press, New York. Holstener-Jorgensen, H [An investigation of the influences of various tree-species and the ages of the stands on the level of the groundwater-table in forest tree stands at Bregentred] (Danish, Eng. Summ.) Forstl. Forsfigs. Danmark 27:

18 Hornbeck, J. W, Storm flow from hardwood-forested and cleared watersheds in New Hampshire. Water Resour. Res. 9: Hornbeck, J. W Streamflow response to forest cutting and revegetation. Water Resour. Bull. 11: Hornbeck, J. W Nutrients: A major consideration in forest management, p In Intensive culture of northern forest types, Symp. Proc., USDA Forest Serv. Gen. Tech. Rep. NE-29. Hornbeck, J. W. and C. A. Federer Effects of management practices on water quality and quantity: Hubbard Brook Experimental Forest, New Hampshire. p _In_Municipal Watershed Management Symp. Proc., USDA Forest Serv. Gen. Tech. Rep. NE-13. Hornbeck, J. W. and G. W. Likens Chemical content of streams draining two cutover forests in New England. Trans. Am. Geophys. Union. 59:282 (Abstract). Hornbeck, J. W., R. S. Pierce and C, A. Federer Streamflow changes after forest clearing in New England. Water Resour. Res. 6: Hornbeck, J. W., R, S. Pierce, G. E. Likens and C. W. Martin Moderating the impact of contemporary forest cutting on hydrologic and nutrient cycles, p Jfa Int. Assoc. Hydrol. Sci. Publ. 117, Tokyo. Kochenderfer, J. N Erosion control on logging roads in the Appalachians. USDA Forest Serv. Res. Pap. NE p. Kochenderfer, J. N. and G. M. Aubertin Effects of management practices in water quality and quantity: Fernow Experimental Forest, W. Virginia, p In_ Municipal Watershed Management Symp. Proc., USDA Forest Serv. Gen. Tech. Rep. NE-13. Leak, W. B Some effects of forest preservation. USDA Forest Serv. Res. Note NE p. Levy, G Premiers resultats d'etude comparee de la nappe temporaire des pseudogleys sous resineaux et sous feuillus. Ann. Sci. Forest 26: Likens, G. E., F. H. Bormann, R. S. Pierce, J. S. Eaton and N. M. Johnson Biogeochemistry of a forested ecosystem. Springer-Verlag, New York; Heidelberg, Berlin. 146 p. Likens, G. E., F. H. Bormann, R. S. Pierce and W. A. Reiners Recovery of a deforested ecosystem. Science 199: Lull, H. W. and W. E. Sopper How harvesting forest products affect water yields in Appalachia. p In Proc. Annual Meet. Soc. Am. Foresters, Lull, H. W. and K. G. Reinhart Forests and floods in the eastern United States. USDA Forest Serv. Res. Pap. NE p.

19 Lyford, W. H Watertable fluctuations in periodically wet soils of Central New England. Harvard Univ. Forest Pap". 8 p. Lyford, W. H Narrow soils and intricate soil patterns in southern New England. Geoderma 11: Lynch, J. A., W. E. Sopper, E. S. Corbett, and D. W. Aurand Effects of management practices on water quality and quantity: The Penn State experimental watersheds. p _Ln_ Municipal Watershed Management Symp. Proc., USDA Forest Serv. Gen. Tech. Rep. NE-13. Mader, D. L., W. P. MacConnell and J. W. Bauder The effect of riparian vegetation control and stand density reduction in the riparian zone. Univ. Mass. Agr. Exp. Sta. Res. Bull p. Mitchell, H. L. and R. F. Chandler, Jr The nitrogen nutrition and growth of certain deciduous trees of northeastern United States. Black Rock Forest Bull p. Patric, J. H Deforestation effects on soil moisture, streamflow and the water balance in the central Appalachians. USDA Forest Serv. Res. Pap. NE p. Patric, J. H Soil erosion in the eastern forest. J. Forest 74: Pierce, R. S., C. W. Martin, C. C. Reeves, G. E. Likens, and F. H. Bormann Nutrient loss from clearcuttings in New Hampshire, p In Watersheds in Transition, Am. Water Resour. Assoc. Proc. Ser. 14, Urbana, 111. Raney, W. A The dominant role of nitrogen in leaching losses from soils of humid regions. Agron. J. 52: Reinhart, K. G., A. R. Eschner and G. R. Trimble, Jr Effect on streamflow of four forest practices in the mountains of West Virginia. Forest Serv. Res. Pap. NE p. USDA Rennie, P. J The uptake of nutrients by mature forest growth. Plant Soil 7: Sartz, R. S. and M. D. Knighton Soil-water depletion after four years of forest regrowth in southwestern Wisconsin. USDA Forest Serv. Res. Note NC p. Stone, E. L The impact of timber harvest on soils and water. Appendix M, p In Report of the President's Advisory Panel on Timber and the Environment. U.S. Gov't. Printing Office, Washington, D.C. Stone, E. L Nutrient release through forest harvest: A perspective. Am. Soc. Civ. Eng. Symp. Watershed Management, Logan, Utah, Aug p. (processed). 534

20 Swank, W. T. and J. E. Douglass Streamflow greatly reduced by converting hardwo6d stands to pine. Science 185: Swank, W. T. and J. E. Douglass Nutrient flux in undisturbed and manipulated forest ecosystems in the mountains of North Carolina, p Jjn Vol. 1, Watershed research in eastern North America. Smithsonian Institution, Chesapeake Bay Center for Environmental Studies. Edgewater, MD. Swank, W. T. and J. D. Helvey Reduction of streamflow increases following regrowth of clearcut hardwood forests. In Symp. results of research on representative and experimental basins. Wellington, N.Z., Int. Assoc. Hydrol. Sci. Publ. 96: Swank, W. T. and N. H. Miner Conversion of hardwood-covered watershed to white pine reduces water yield. Water Resour. Res. 4: Swift, L. W., Jr. and J. B. Messer Forest cuttings raise temperatures of small streams in the southern Appalachians. J. Soil Water Conserv. 26: Urie, D. H Pattern of soil moisture depletion varies between red pine and oak stands in Michigan. USDA Forest Serv. Lake States Forest Exp. Sta. Tech. Note p. Urie, D. H Groundwater differences on pine and hardwood forests of the Udell Experimental Forest in Michigan. USDA Forest Serv. Res. Pap. NC p. Verry, E. S Effect of aspen clearcutting on water yield and quality in northern Minnesota, p _In_ Watersheds in Transition. Am. Water Resour. Assoc. Proc. Ser. 17, Urbana, 111. Zahner, R Soil water depletion by pine and hardwood stands during a dry season. Forest Sci. 1: