Journal of Applied Ecology (1988), 25,

Transcription

1 Journal of Applied Ecology (1988), 25, LITTER DECOMPOSITION, NITROGEN DYNAMICS AND LITTER MICROARTHROPODS IN A SOUTHERN APPALACHIAN HARDWOOD FOREST 8 YEARS FOLLOWING CLEARCUTTING BY JOHN M. BLAIR* AND D. A. CROSSLEY, JR.' Department of Entomology, Institute of Ecology, University of Georgia, Athens, Georgia 30602, U.S.A. SUMMARY (1) Litter decomposition rates, nitrogen dynamics and litter microarthropods on the xeric slopes of a watershed 8 years after clearcutting (WS 7) and on an adjacent reference watershed (WS 2) at the Coweeta Hydrologic Laboratory were measured using litterbags. Litter of Comusflorida L., Acer rubrum L. and Quercusprinus L. was used as experimental substrates in three plots on adjacent areas of both watersheds. Results from this study were compared with those obtained before and after cutting to assess the longer-term changes induced by canopy removal. (2) Reduced litter decomposition rates were associated with clearcutting. Litter nitrogen dynamics were also affected. Net immobilization of nitrogen in litter substrates was lower on WS 7. Lower net nitrogen immobilization was related to slower increases in nitrogen concentration per unit mass lost in litter on WS 7. (3) Mean annual densities of total litter microarthropods remained 28% lower on WS 7 than on WS 2, 8 years after cutting. Previous studies indicated clearcutting initially reduced mean annual densities of litter microarthropods by>50%. Mesostigmata and Oribatei densities averaged 50 and 54% lower, respectively, than on WS 2. Prostigmata and Collembola densities averaged 20 and 24% lower, respectively, than on WS 2. This differential response changed the relative abundances of major groups. Changes in litter decomposition rates and nitrogen dynamics were consistent with effects associated with lower microarthropod densities and suggest that reduced microarthropod densities may be important. (4) Results of this study differ from those at northern hardwood forest sites where clearcutting caused increased decomposition rates. This suggests that generalizations from northern hardwood forests may not apply to other regions. Instead the effects of canopy I removal depend on both the nature of pre-disturbance processes and on the 'site-specific'! effects of disturbance on the processes studied. j INTRODUCTION \ r Decomposition of litter and mineralization of nutrients contained in it strongly influence \ soil nutrient availability and ultimately ecosystem primary productivity. Disturbances which alter decomposition and other soil processes can allow nutrient losses and a decline in site productivity (Bormann et al. 1974; Swank & Waide 1980). One common disturbance of eastern deciduous forests in the U.S. is clearcutting following timber harvesting. Several studies have examined the effects of forest vegetation removal on hydrologic flows, nutrient fluxes, primary production, litter decomposition and soil ' organic matter turnover. Some have documented increased losses of nutrients in ; * Present address: Department of Entomology, Ohio State University, 1735 Neil Avenue, Columbus, Ohio 43210, U.S.A. 683

2 684 Decomposition following clear cutting streamflow following clearcutting of northern hardwood forests (i.e. Pierce et al. 1972; Bormann & Likens 1981). Increased nutrient output after clearcutting was partly attributable to a large input of litter along with increased decomposition and mineralization rates due to a more favourable microclimate for decomposers (Stone 1973; Bormann et al. 1974). Longer-term (i.e. > 15 years) decreases in surface organic matter following clearcutting of northern hardwood forests may also result from increased decomposition rates (Covington 1981; Federer 1984; Hix & Barnes 1984). Based on these observations, increased decomposition rates and decreases in forest floor mass have been incorporated in models of forest floor dynamics following disturbance (Aber, Botkin & Melillo 1978). Many studies of the effects of forest cutting are based on a chronosequence approach, using measurements of forest floor mass at sites in different stages of recovery following clearcutting (Covington 1981; Federer 1984). Fewer studies have followed changes in soil surface organic matter through time at a single site (but see Johnson, Smith & Burger 1985), and fewer still have measured changes in litter decomposition rates using litterbags or a comparable method. The experimental clearcutting of a forested watershed (catchment) at the Coweeta Hydrologic Laboratory in the southern Appalachians gave the opportunity to measure changes in forest floor processes through time at a single site. Other studies of clearcut forests have indicated that generalizations from northern hardwood forests may not apply to forests in other regions (Whitford et al. 1981; Will, Hodgekiss & Madgwick 1983; Binkley 1984; Mattson 1986). Disturbances of coniferous forests in the north-western U.S. produced only modest hydrologic exports of nutrients (Fredriksen 1975, cited in Swank & Waide 1980) while clearcutting at Coweeta produced the lowest nutrient losses via streamflow recorded in the U.S. (Swank & Waide 1980; Swank 1986). Factors likely to reduce nutrient losses in the southern Appalachians include rapid recovery of net primary productivity (Boring, Monk & Swank 1981) and reduced decomposition rates (Seastedt 1979). Cable-logging and clearcutting of a watershed (WS 7) at Coweeta produced immediate decreases in rates of leaf litter decomposition (Seastedt 1979) and rates of woody litter decomposition on xeric sites (Abbott & Crossley 1982), while surface organic matter has subsequently increased (Mattson 1986). Densities of litter microarthropods also decreased after clearcutting (Abbott, Seastedt & Crossley 1980; Seastedt & Crossley 1981). Microarthropods influence decomposition rates and forest floor nutrient fluxes (Anderson & Ineson 1984; Seastedt 1984), and reductions in their densities could affect decomposition processes following clearcutting. The objectives of this study were (i) to measure decomposition rates, nitrogen dynamics and microarthropod densities in three litter substrates on the xeric slopes of a watershed in its eighth year of regrowth following clearcutting and on an adjacent uncut reference watershed; and (ii) to compare these results with those collected before and immediately after clearcutting to assess its longer-term effects on litter decomposition and litter microarthropods. MATERIALS AND METHODS Site description The U.S. Forest Service Coweeta Hydrologic Laboratory, located in the southern Appalachian Mountains of south-western North Carolina (35 00' latitude, 83 30' longitude), is a 2185-ha forested basin consisting of numerous small watersheds that serve as experimental units. The native hardwood forest of the area is a mixture of deciduous

3 J. M. BLAIR AND D. A. CROSSLEY JR 685 TABLE 1. Initial N concentrations, C:N ratios, lignin concentrations, lignin :N ratios, and water- and ethanol-soluble concentrations of Cornus florida, Acer rubrum and Quercus prinus litter used in this study* % Water % Ethanol Species %N C:N % Lignin Lignin:N soluble soluble C. florida A.rubrum Q. prinus * For details on methods of analysis see Blair (1988). trees dominated by Quercus, Gary a and Acer spp. Mean annual rainfall in the basin varies from 1700 mm at lower elevations to 2500 mm on some ridges. Rainfall is evenly distributed throughout the year. Mean annual temperature is 13 C. The two watersheds studied are in the north-eastern portion of the Coweeta basin. Watershed 7 (WS 7) is a 59-ha, south-facing catchment with elevations of m and slopes of 20-80%. Prior to 1977, WS 7 was covered by uneven-aged Quercus-Carya forest. Watershed 7 was divided into three areas based on previous vegetational studies and site preparation (Boring & Swank 1986). Pre-cut vegetation consisted of a mesic hardwood cove association dominated by Liriodendron tulipifera, Tsuga canadensis and Acer rubrum (approximately 10% of the area), mesic slopes dominated by Carya and Quercus spp. (approximately 36% of the area) and more xeric slopes and upper ridges dominated by Q. coccinea, Pinus rigida and Kalmia latifolia (approximately 54% of the area) (Boring, Monk & Swank 1981). In the summer of 1977, merchantable timber was ] cut and removed by cable-logging to minimize forest floor disturbance and compaction. i Following removal of merchantable boles, all remaining trees were felled and the slash left I in place. There was rapid regrowth from stump and root sprouts along with early successional herbaceous species (Boring, Monk & Swank 1981; Boring & Swank 1986). l Watershed 2 (WS 2) is an adjacent 14-3-ha catchment, with similar slope, elevation, aspect and climate, which was used as an unmanipulated reference. The vegetation was dominated by uneven-aged Quercus, Carya and Acer spp. (Berish & Ragsdale 1985). Decomposition dynamics and litter microarthropods Litter decomposition rates, nitrogen dynamics and litter microarthropod densities 8 I years after clearcutting (1985) were measured using litterbags. Litterbags with an inside I area of 10 x 10 cm were made of fibreglass window-screen material with a 1-6 x 1-8 mm! mesh. Senesced leaves of Cornus florida L. (flowering dogwood), Acer rubrum L. (red j maple) and Quercus prinus L. (chestnut oak) were collected in October 1984 and air-dried! at 22 C. These species were abundant on both watersheds, comprising about 45 % of leaf I litter inputs in (Risley 1987), and were chosen to give a range of resource quality i (Table 1). Newly senesced litter was collected from several sites at similar elevations in the Coweeta basin, pooled by species and mixed prior to filling litterbags to ensure that homogenous litter substrates were used on both watersheds. Approximately 2-5 g of airdried leaves were placed in pre-weighed bags. Five bags of each species were oven-dried at 50 C to develop equations relating air-dry to oven-dry mass. On 4 January 1985 fifty-two litterbags of each species were placed in each of three plots (1-5 x 1-5 m) arranged in a transect and spaced m apart on adjacent areas of both watersheds. All six plots were located on similar elevations, slopes and aspects on both watersheds. Plots on WS 7 i

4 686 Decomposition following clearcutting \ were located on the xeric sites formerly occupied by the Quercus-Pinus-Kalmia I community because it was the predominant type and because we expected the effects of clearcutting to be longer lasting there. Five bags of each species were immediately! retrieved to determine mass loss due to handling. These values were used to adjust initial masses of litterbags before determining subsequent mass loss. Litterbags were retrieved randomly from within each plot at 2-week intervals from January to November and in mid-december 1985, and every 2 months in 1986 for a total of twenty-nine collection dates over 2 years. The last collection date was 6 January Litterbags were weighed before processing to determine percentage moisture. Maximum and minimum temperatures at the soil-litter interface between collection dates were obtained using max-min thermometers with remote probes at one plot on each watershed. Microarthropods were extracted from litterbags using Tullgren funnels and separated into five broad taxonomic groups: Prostigmata, Mesostigmata, Oribatei, Collembola and 'others'. Densities were expressed as numbers of individuals g~' of litter remaining. Seasonal densities were calculated by grouping collection dates during the first year as follows: winter (January-March), spring (April-June), summer (July-26 September) and autumn (27 September-December). Litter moisture and temperature data were also pooled to calculate seasonal averages. Following microarthropod extraction, intact litterbags were oven-dried at 50 C, the litter reweighed and ground, and subsamples ignited at 500 C for 4 h to determine percentage ash-free dry mass (% AFDM). Any obvious extraneous material was removed before weighing. However, after 6 months in the field, soil contamination was evident in some litterbags on each collection date. Therefore, litter masses were corrected for soil infiltration before determining % mass loss using the following soil correction equation (Walther 1983): FLi = (SaAFDM-SlAFDM)/(LiAFDM-SlAFDM) where FLi is the fraction of litterbag content that is actually litter; SaAFDM is the % AFDM of the entire litterbag sample; S1AFDM is the average % AFDM of the soil at the site; Li AFDM is the initial % AFDM of the litter substrate. The underlying assumption is that organic matter content (% AFDM) of the litter and the soil remain constant throughout the study. The equation calculates the fraction of litterbag content that is actually litter based on any reduction in % AFDM of the sample caused by soil contamination. The equation is preferable to simply calculating litter mass loss as %. AFDM remaining when soil moving into the litterbags has a high organic matter content, since organic matter from the soil would contribute to the apparent. AFDM of the litter. H First-year decay rate constants were calculated from percentage mass remaining data, using a single negative exponential decay model X/X 0 =e~ kt, where X/Xo=percentage mass remaining at time t, *=time elapsed in years, and fc=the annual decay constant (Olson 1963). The single negative exponential model was fitted to the data by least squares regression of the natural logarithm of mean percentage mass remaining over time. Nitrogen concentrations in the residual litter were measured monthly in 1985 and bimonthly in Nitrogen concentrations were determined with a Technicon Model II Autoanalyzer following micro-kjeldahl digestion (Technicon 1970). Nitrogen concentrations for litterbags contaminated with soil (as indicated by reductions in % AFDM of the sample) were corrected using the following equation: LiN = [SaN - (FS1 x SIN)] ~ FLi

5 J. M. BLAIR AND D. A. CROSSLEY JR 687 TABLE 2. A summary of first-year decomposition rates on pre-cut WS 7 and other unmanipulated watersheds in the Coweeta basin k (% remaining, t= 1 year)* Watershed Comusflorida A. rubnan Quercus prinus WS (30-8) -0-83(43-6) -0-55(57-8) WS (26-0) -0-72(48-6) -0-66(51-4) WS (47-8) (57-9) (72-2) WS (27-8) -0-53(49-0) -0-34(69-3) * Rates are expressed as annual decay constants (k) derived from a single negative exponential model (Olson 1963). Percent mass remaining at the end of one year is indicated in parentheses. WS 18 data are from Cromack (1973). WS 2 and WS 7 data are from J. B. Waide, R. L. Todd, J. R. Webster, and D. A. Crossley, Jr, unpublished. where LiN is the nutrient concentration in the residual litter; SaN is the nutrient concentration of the entire sample; SIN is the average nutrient concentration in the soil; FS1 is the fraction of the sample that is soil (1 FLi from the above soil correction equation). Net nitrogen immobilization or release was calculated as the product of percentage mass remaining and nitrogen concentration in the residual material divided by the initial nitrogen concentration of that litter type. Pre-cut and first-year post-cut decomposition rates and litter microarthropod densities were based on similar litterbag studies. For further details on methods see Seastedt (1979) and Seastedt & Crossley (1981). Statistical comparisons of data from clearcut and reference plots are presented, where appropriate, to indicate significant differences. However, because only one watershed was clearcut, reliance on inferential statistics to imply treatment effects at the watershed level would be inappropriate (Hurlbert 1984). Instead, comparisons of pre-cut and post-cut differences between the clearcut and reference watersheds and long-term changes in relevant parameters are used to assess the effects of clearcutting. RESULTS Decomposition rates Litter decomposition rates on WS 2 and WS 7 were believed to be similar before the clearcutting of WS 7 in 1977 (J. B. Waide, R. L. Todd, J. R. Webster & D. A. Crossley Jr, unpublished data). Decomposition rates of litter of selected species on pre-cut WS 7 were similar to rates obtained on other unmanipulated watersheds (WS 2 and WS 18) in the Coweeta basin (Table 2). Decomposition rates on WS 18, a north-facing, more mesic watershed were expected to be somewhat higher than rates on watersheds 2 and 7. Measurements of decay rates on WS 2 and WS 7 were not taken in the same year before clearcutting. The available data do indicate, however, that litter decayed at least as fast on WS 7 as on WS 2 before the clearcut. Decay rates measured in successive years with similar climates were somewhat higher on WS 7 (Table 2). Similar litter inputs and standing stocks of surface organic matter on comparable areas of WS 2 and WS 7 before clearcutting indicate similar decomposition rates. Given similar litter inputs and decomposition rates, standing mass of surface organic matter should be similar (Olson 1963). Pre-cut measurements of surface organic matter (L and F layers) were 816 g m~ 2 on

6 688 Decomposition following clear cutting TABLE 3. A comparison of annual decay rates (k) of Cornusflorida, Acer rubrum and Quercus prinus litter on clearcut WS 7 and uncut WS 2 (1985 data)* Clearcut Uncut WS 7 WS 2 Species Slope r 2 Slope r 2 C.florida P< A.rubrum P<0-025 Q.prinus P<0-01 * Slopes (k) were calculated by fitting the data to a single negative exponential model (Olson 1963). Coefficients of'determination (r 2 ) indicating goodness-of-fit are presented. Probability values are based on a one-tailed t- test of the hypothesis that rates obtained on WS 7 plots are slower than those obtained on WS 2 plots (Zar 1984). TABLE 4. Comparisons of litter decomposition rates on WS 2 and WS 7 prior to clearcutting and in the first and eighth years following cutting of WS 7* Pre-cut Post-cut (1 year) Post-cut (8 years) Species WS 7 WS 2 WS 7 WS 2 WS 7 WS 2 Cornusflorida (27-8) (47-8) (55-6) (34-3) (51-3) (45-8) Acer rubrum (49-0) (57-9) (70-5) (53-7) (62-4) (52-9) Quercus prinus (69-3) (72-2) (76-7) (65-2) (78-1) (73-1) * Decompositions rates are expressed as first year annual decay rates (k) derived from a negative exponential model (Olson 1963). Percentage mass remaining at the end of 1 year is indicated in parentheses. Pre-cut and first year post-cut rates are composite values based on collection of litterbags from several sites on both watersheds (J. B. Waide, R. L. Todd & D. A. Crossley Jr, unpublished data). WS 7 and 880 g mr 2 on WS 2 (Seastedt & Crossley 1981), indicating similar decomposition rates on the two watersheds. Decreased litter decomposition rates were associated with clearcutting. The three litter substrates examined decayed in the following order on both watersheds: C.florida>A. rubrum > Q. prinus. However, annual decay rates were lower on the WS 7 plots relative to rates obtained for the same substrates on WS 2 plots (Table 3). Measurements of decay rates in the first year following clearcutting also indicated lower decomposition rates on WS 7 relative to WS 2 and to pre-cut rates on WS 7 (Table 4). This reduction was initially most pronounced for the more rapidly decaying C.florida and A. rubrum litter. In the first year after clearcutting, annual decay rates (k) of C.florida, A. rubrum and Q. prinus litter on WS 7 were 46,44 and 33% lower, respectively, than on WS 2. Decomposition rates on WS 7 remained lower on the xeric WS 7 sites 8 years after clearcutting, although the magnitude of the difference had decreased substantially for C.florida and A. rubrum litter (Table 4). Annual decay rates of C.florida, A. rubrum and Q. prinus in the eighth year of regrowth on WS 7 were 17, 14 and 32% lower, respectively, than on WS % 2. Nitrogen dynamics Litter nitrogen dynamics were also altered on the xeric WS 7 plots. Absolute increases in nitrogen content of litterbags (net nitrogen immobilization) were reduced on WS 7

7 J. M. BLAIR AND D. A. CROSSLEY JR mass III Days in the field FIG. 1. Mean percentage of initial mass and nitrogen remaining over time in (a) Cornus florida, (b) Acer rubrum and (c) Quercusprinus litter on uncut WS 2 ( ) and clearcut WS 7 ( ) from January 1985 to January relative to WS 2. Mass loss curves and net nitrogen fluxes in litter on WS 7 and WS 2 are presented in Fig. 1. The pattern of net nitrogen immobilization/release varied among litter species in a similar way on both watersheds. Initial net nitrogen losses in A. rubrum (Fig. Ib) and Q. prinus (Fig. Ic) litter were followed by increased net immobilization. Cornus florida litter (Fig. la), did not exhibit an initial loss phase but did immobilize nitrogen. All species appeared to begin a net release phase by the end of the study. Although general patterns of net nitrogen immobilization and release were similar on both watersheds, less nitrogen was immobilized in each substrate on WS 7 relative to WS 2 (Fig. 1). Greater net nitrogen immobilization on WS 2 was due to greater increases in nitrogen concentration in the residual litter (Fig. 2). A way to integrate mass loss and nitrogen concentration data is to plot percentage mass remaining as a function of nitrogen concentration in the residual litter (Aber & Melillo 1980). Data from C. florida litter on both watersheds are presented in this format in Fig. 3. Regressions of mean percentage mass remaining over mean nitrogen concentration in the residual litter were calculated for

8 690 Decomposition following clear cutting 2-0 (a ) Cornus f/orida 1-0 (b) Acer rubrum 1-0 z 5? (c) Quercus prinus Days in the field FIG. 2. Changes in the mean nitrogen concentration of decomposing (a) Cornus florida, (b) Acer rubrum, and (c) Quercus prinus litter on uncut WS 2 ( ) and clearcut WS 7 ( ). Bars represent S.E. (n = 6 on days and «= 3 on days ). each litter type on both watersheds. Slopes, 7-intercepts and coefficients of determination are presented in Table 5. The slope describing the relationship between percentage mass remaining and nitrogen concentration is a measure of the rate of increase in nitrogen concentration per unit mass lost. For all three litter types this slope is 'steeper' for the WS 7 data, indicating slower increases in nitrogen concentration per unit mass lost in the xeric WS 7 plots. Litter microarthropods Microarthropod densities on WS 2 and pre-cut WS 7 are assumed to be similar. Soil/ litter cores taken in clearcut (WS 7), partly-cut (WS 7), and control (WS 2) plots in the summer of 1977 revealed no significant differences in microarthropod densities, high coefficients of similarity between partly-cut and control plots and lower coefficients of similarity between the clearcut plot and the others (Abbott, Seastedt & Crossley 1980; Seastedt & Crossley 1981).

9 100 J. M. BLAIR AND D. A. CROSSLEY JR 691 D).E WS2, WS % Nitrogen FIG. 3. Slopes of the inverse relationship of mean percentage mass remaining and mean nitrogen concentration in the residual litter for Comus florida on uncut WS 2 (O) and clear cut WS 7 ( ). The solid line is derived from the WS 2 data and the dashed line from the WS 7 data. TABLE 5. Slopes, Y-intercepts and coefficients of determination (r 2 ) of the inverse linear relationships of percentage mass remaining and nitrogen concentration in residual litter on uncut WS 2 and clearcut WS 7; regressions are based on data collected over a 464-day study period (January 1985-April 1986) Species (WS) Slope r-intercept r 2 Comusflorida (WS 2) *** (WS7) Acerrubrum (WS 2) ** (WS7) Quercusprinus (WS 2) * (WS7) * 0-05 <P< 0-10; **P<0-025; *** /><0-0025; values indicate the probability level at which slopes are statistically different based on a one-tailed r-test (Zar 1984). In the first year following clearcutting the mean annual density of total litter microarthropods in C. florida and Q. prinus litterbags on WS 7 was reduced > 50% relative to WS 2 (Seastedt & Crossley 1981). There was also a change in the pattern of seasonal abundance of microarthropods. Microarthropod densities on WS 2 approximated a unimodal curve, with maximum densities occurring in the summer. The pattern of seasonal abundance on WS 7 was bimodal, with peak densities occurring in the spring and autumn (Seastedt 1979). Summer densities were reduced, presumably due to highly elevated temperatures and reduced litter moisture (Abbott, Seastedt & Crossley 1980). Results in the eighth year of regrowth indicated partial recovery of litter microarthropod densities. However, mean annual densities of total microarthropods remained significantly lower on WS 7 relative to WS 2 (Table 6). Mean annual densities of total microarthropods in all litterbags (C. florida, A. rubrum and Q. prinus) on WS 7 were approximately 28% lower than on WS 2 (Table 6). The seasonal distribution of microarthropods had resumed a normal curve and the greatest differences occurred in the autumn (Fig. 4).

10 692 Decomposition following clear cutting TABLE 6. Mean annual densities+ S.E. (H=138) (individuals g~' litter) of major microarthropod groups in Cornus florida, Acer rubrum and Quercus prinus litterbags on uncut WS 2 and clearcut WS 7 (1985 data) Total microarthropods Prostigmata Cornus florida (WS2) * (WS7) 28-34±2-46 (WS2) (WS7) Acer rubrum *** ** ±1-04 Quercus prinus ±1-96** ± * 8-12±0-83 Mesostigmata (WS2) (WS7) *** *** Oribatei (WS2) (WS7) *** ±2-06*** 4-99 ± *** Collembola (WS2) (WS7) Other (WS2) (WS7) * 2-51 ± ± ± * * P<0-05; ** P<0-01; *** P<0-001; values indicate the probability level at which densities are significantly different based on a one-tailed f-test (Zar 1984). 80 Total microarthropods 20 - Oribatei " Mesostigmata n I? ' 5 } }1 10 \ L 5. I 1 *J i 5 _T 51 - \ i _i 1 1. i n 5. i i, g 20 (D Prostigmata 15 - j h ' 1 II ll 5 4 } ^, J 51,i 0 i 1 1. J 1 n i i i Winter Spring SummerAutumn Winter Spring SummerAutumn 20 Collembola T Other Winter Spring SummerAutumn FIG. 4. Mean seasonal densities of major microarthropod groups in Cornus florida litterbags on uncut WS 2 (O) and clearcut WS 7 ( ). Bars represent S.E. (n = 36). Patterns of seasonal densities of microarthropods in chestnut oak and red maple litterbags were similar, although densities were higher in red maple litterbags.

11 J. M. BLAIR AND D. A. CROSSLEY JR TABLE 7. Percentage composition of the total microarthropod fauna from litterbags of Cornusflorida, Acer rubrum and Quercusprinus on uncut WS 2 and clearcut WS 7 (1985 data) Group Prostigmata Mesostigmata Oribatei Collembola Others Percentage of total fauna Cornusflorida Acer rubrum Quercus prinus WS2 WS7 WS2 WS7 WS2 WS : 3-8 ' TABLE 8. Mean + S.E. seasonal maximum and minimum temperatures ( C), based on six recording intervals per season, recorded at the soil-litter interface on uncut WS 2 and clearcut WS 7 Uncut WS 2 Clearcut WS 7 Season Maximum Minimum Maximum Mininum Winter 31-0± Spring 31-8± ±l Summer ±l ±0-7 Autumn TABLE 9. Seasonal means ( + S.E.) of percentage moisture of Cornus florida litterbags at the time of collection on uncut WS 2 and clearcut WS 7, with six bags per watershed and six collection dates per season (n = 36) Season Uncut WS 2 Winter ±7-91 Spring ±27-93 Summer Autumn ±6-48 Clearcut WS ± ± ± ±6-24 Various taxa of microarthropods responded differently to clearcutting, changing the relative abundances of major groups (Table 7). Similar patterns were observed in litterbags of all species. The groups most reduced by clearcutting, on a percentage basis, were mesostigmatid and oribatid mites (Table 6). Mean annual densities of mesostigmatid and oribatid mites in all litterbags on WS 7 were 50 and 54% lower, respectively, than on WS 2. Mean annual densities of prostigmatid mites and collembola in all litterbags on WS 7 were 20 and 24% lower, respectively, than on WS 2. The greater reduction in populations of oribatid and mesostigmatid mites increased the relative abundance of prostigmatid mites and collembola on WS 7 (Table 7). The mean annual density of the 'other' group, which included insects, spiders, chilopods, diplopods and other assorted invertebrates was higher on WS 7 by an average of 34%. DISCUSSION Clearcutting appeared to slow litter decomposition rates on WS 7 at Coweeta. Decreased decomposition rates on the xeric portion of WS 7 remained apparent 8 years later.

12 694 Decomposition following clear cutting However, the degree of difference between clearcut and control plots had decreased for C. fiorida and A. rubrum litter, suggesting that decay rates are returning to pre-cut levels. Decreased decomposition rates may contribute to increased surface organic matter storage following clearcutting. There were significant increases in surface organic mass following clearcutting of WS 7 (Mattson 1986). Pre-cut surface organic mass (L and F layers) was estimated at 816 g m~ 2 and in the first year after clearcutting increased to 970 gm~ 2 (Seastedt & Crossley 1981). By the sixth year surface organic mass had increased to 1158 g m~ 2 (Mattson 1986). Increased forest floor mass is partly due to inputs of debris during clearcutting,! but reduced decomposition rates may also be important. In general, two factors changes in microclimate and in litter quality may potentially reduce litter decomposition rates. Microclimatic changes included increased temperature extremes and more frequent and intense wetting and drying cycles. Both temperature extremes and variability recorded at the soil-litter interface on WS 7 were greater than on WS 2 (Table 8). Litterbags collected from WS 7 were also generally drier (Table 9). This may be due to slower recovery of canopy cover on the xeric southwest-facing slopes (Boring & Swank 1986). Although increased temperatures and actual evapotranspiration (AET) following clearcutting should increase decomposition rates (Meentemeyer 1978), this was not the case (Whitford et al. 1981). The combination of greater temperature extremes and moisture variability created a microclimate that was suboptimal for the decomposer community, affecting both primary (bacteria and fungi) and secondary (microarthropods, nematodes, etc.) decomposers. Forest floor microbial biomass, estimated from ATP measurements, decreased in the first year after cutting (Waide et al. 1987) and both shorter-term (1-3 years, Waide et al. 1987) and longer-term (6 year, Mattson 1986) decreases in COa evolution from the forest floor occurred. Maximum temperatures near 50 C recorded at the soil-litter interface after clearcutting (Abbott, Seastedt & Crossley 1980) were sufficient to kill oribatid mites or inhibit their reproduction (Seastedt & Crossley 1981). Rapid canopy development may ameliorate the harsher microclimate, but it may take considerably longer for the decomposer community to recover completely. Although the soil-litter fauna play an important role in decomposition and nutrient cycling, few North American studies have considered the effects of clearcutting on the fauna. Our results indicated reduced total microarthropod densities and changes in the relative abundance of microarthropod groups on WS 7. Similar results were reported following cutting of a mixed conifer-hardwood forest in Canada (Bird & Chatarpaul 1986) and both cutting alone and cutting followed by slash burning in a British Columbia forest (Vlug & Borden 1973). Microarthropods can affect decomposition processes directly and indirectly (Seastedt 1984). These effects include comminution and modification of the substrate, selective grazing and microbial stimulation (Crossley 1977). Exclusion or reduction of microarthropods usually decreases litter decay rates (see review by Seastedt 1984). Previous studies at Coweeta indicated a 33% reduction in annual decay rates of C.florida litter treated with naphthalene to reduce microarthropods (Seastedt & Crossley 1983). Lower microarthropod densities on WS 7 may contribute to reduced decomposition rates. Decay rates were most reduced for Q. prinus litter, whichjvas the slowest decomposing species. This agrees with the observation that microarthropods have a relatively greater effect on the decomposition of more recalcitrant litter (Seastedt 1984). Changes in relative abundance of microarthropod groups may also be important due to differences in the trophic/functional roles of these groups. The Prostigmata include a

13 J. M. BLAIR AND D. A. CROSSLEY JR 695 variety of forms, although most associated with the forest floor are piercing fungivores, predators or generalized saprovores. The Mesostigmata found in this study were mostly predacious forms. Many Mesostigmata are predacious on nematodes (Mankau & Imbriani 1978) and may affect populations of microbial-feeding nematodes which may, in turn, affect decomposition. Both the Oribatei and Collembola are primarily fungivorous. Reduced oribatid densities may be particularly important in lowering decomposition rates. Oribatid mites are fungivores with chelate mouthparts capable of fragmenting and comminuting litter (Seastedt 1984; Norton 1985). Oribatids showed the most extreme immediate response to the cleareut (Abbott, Seastedt & Crossley 1980; Seastedt & Crossley 1981) and they remain most reduced (Fig. 4). Oribatid populations may be more susceptible to microclimatic changes and may also take longest to recover due to low fecundity and long generation times (Norton 1985). Litter nitrogen dynamics were altered on WS 7. Nitrogen concentrations in the residual litter increased at a slower rate per unit mass lost and this resulted in lower levels of net nitrogen immobilization. This suggests that clearcutting affects the underlying processes controlling nitrogen fluxes in forest litter and that the effects may be relatively long lasting. Several hypotheses may account for the slower increases in nitrogeilconcentration per unit mass lost on WS 7. (i) Differences in availability of exogenous nitrogen (i.e. throughfall, litter leachate, etc.) between the watersheds may mean that more carbon must be respired to incorporate a given amount of nitrogen in microbial biomass. (ii) Differences in microclimate and resource quality may affect the composition of the primary decomposers community. Different decomposers have different protoplasmic C:N ratios and assimilate carbon and nitrogen with different efficiencies (Alexander 1977). (iii) The microclimate on WS 7 may result in alternating periods of microbial growth, during which carbon is respired and nitrogen immobilized, and microbial senescence when nitrogen that was incorporated in microbial biomass can be leached. Alternating wet and dry periods increased nutrient leaching in laboratory microcosms (Witkamp & Frank (1970). (iv) Lower microarthropod densities may cause lower nitrogen immobilization. Microarthropods can influence microbial activity and affect nutrient immobilization and release in litter (Seastedt 1984). The presence of microarthropods generally increases nitrogen concentrations in decaying litter (Seastedt 1984). Microarthropod effects on net nitrogen fluxes depend on effects on both concentration and mass loss. In this study, slower decay rates, slower increases in nitrogen concentration and lower levels of net immobilization were observed on WS 7, where microarthropod densities were also significantly lower. This suggests that reduced microarthropod densities may contribute to changes in decomposition processes following clearcutting. However, caution must be used in implying a cause and effect relationship. Lower microarthropod densities may simply be correlated with the same factors that result in lower decay rates and may not themselves be a significant causal factor. Further studies are needed to clarify the relative importance of microarthropod reductions in relation to abiotic factors. Site-specific effects of clearcutting Forests in the south-eastern U.S., and elsewhere, may respond differently to clearcutting than north-eastern U.S. forests. Clearcutting in north-eastern U.S. forests increased decomposition rates and caused long-term decreases in surface organic matter (Bormann & Likens 1981; Covington 1981; Federer 1984). Clearcutting also increased

14 696 Decomposition following clearcutting litter decay rates in a Ponderosa pine forest in central Arizona (Klemmedson, Meier & Campbell 1985). Effects on decomposition of cellulose squares at three elevations in Vancouver, B.C., Canada varied among sites and with depth in the forest floor profile (Binkley 1984). Clearcutting reduced decay rates of Pinus radiata litter in New Zealand (Will, Hodgekiss & Madgwick 1983). Our results indicate reduced decomposition rates on south-facing slopes in the southern Appalachians. Litter decay rates in the mesic areas of WS 7 were reduced for at least the first 3 years after clearcutting (J. B. Waide, R. L. Todd & D. A. Crossley Jr, unpublished data) and decay rates remain lower 8 years later on the xeric sites. i The different responses of'southern Appalachian and north-eastern hardwood forests to clearcutting are partly due to pre-cut differences in forest floor processes. Baseline decomposition rates are apparently lower in north-eastern hardwood forests. Standing mass of surface organic matter (L, F and H layers) at Hubbard Brook in the White Mountains of New Hampshire was estimated to be 4680 g m~ 2 (Gosz, Likens & Bormann 1976) as opposed to 1144 g m~ 2 on WS 2 at Coweeta (Berish & Ragsdale 1986). Estimates of standing stocks of nutrients in the forest floor at Coweeta were also much lower than at Hubbard Brook (Swank & Waide 1980). Slower decomposition rates and larger accumulations of organic matter and nutrients are thought to be due to periods of unfavourable climate (i.e. long cold winters with a heavy snow cover) (Swank & Waide 1980). Canopy removal at these sites may increase solar insolation and lead to more optimal conditions for decomposition. Forests at more southern latitudes, such as Coweeta, do not experience harsh winters and have adequate precipitation throughout the year. Here, conditions for decomposition may be more favourable under a closed canopy. Canopy removal may decrease decomposition rates by increasing temperatures and reducing moisture levels. In addition to these regional effects, changes induced by clearcutting may also depend on such site-specific attributes as aspect, harvesting and site preparation techniques. Therefore we suggest that the effects of forest disturbance depend not only on the disturbance itself, but on the nature of pre-disturbance processes and on 'site-specific' effects of the processes studied. ACKNOWLEDGMENTS We thank L. R. Boring, B. L. Haines, R. W. Parmelee, T. R. Seastedt, W. T. Swank, J. B. Waide and two anonymous reviewers for criticism and comments on earlier drafts of this manuscript. Thanks to the U.S. Forest Service for their cooperation in the use of the Coweeta facilities and to J. B. Waide for allowing us to use some of his unpublished data. This research was supported in part by NSF grant BSR to the University of Georgia Research Foundation. REFERENCES Abbott, D. T. & Crossley, D. A. Jr (1982). Woody litter decomposition following clear cutting. Ecology, 63, Abbott, D. T., Seastedt, T. R. & Crossley, D. A. Jr (1980). The abundance, distribution and effects of clearcutting on oribatid mites (Acari: Crytostigmata) in the southern Appalachians. Environmental Entomology, 9, Aber, J. D. & Melillo, J. M. (1980). Litter decomposition: measuring state of decay and percent transfer into forest soils. Canadian Journal of Botany, 58, Aber, J. D., Botkin, D. B. & Melillo, J. M. (1978). Predicting the effects of various harvesting regimes on forest floor dynamics in northern hardwoods. Canadian Journal of Forest Research, 8,

15 J. M. BLAIR AND D. A. CROSSLEY JR 697 Alexander, M. (1977). Introduction to Soil Microbiology. Wiley, New York. Anderson, J. M. & Ineson, P. (1984). Interactions between microorganisms and soil invertebrates in nutrient flux pathways of forest ecosystems. Invertebrate-Microbial Interactions (Ed. by J. M. Anderson, A. D. M. Rayner & D. W. H. Walton), pp Cambridge University Press, Cambridge. Berish, C. W. & Ragsdale, H. L. (1985). Chronological sequence of element concentrations in wood of Carya spp. in the southern Appalachian Mountains. Canadian Journal of Forest Research, 15, Berish, C. W. & Ragsdale, H. L. (1986). Metals in low-elevation, Southern Appalachian forest floor and soil. Journal of Environmental Quality, 15, Binkley, D. (1984). Does forest removal increase rates of decomposition and nitrogen release? Forest Ecology and Management, 8, Bird, G. A. & Cbatarpaul, L. (1986). Effect of whole-tree and conventional forest harvest on soil microarthropods. Canadian Journal of Zoology, 64, Blair, J. M. (1988). Nitrogen, sulfur and phosphorus dynamics in decomposing decidous leaf litter in the southern Appalachians. Soil Biology and Biochemistry. In press Boring, L. R., Monk, C. D. & Swank, W. T. (1981). Early regeneration of a clear-cut Southern Appalachian forest. Ecology, 62, Boring, L. R. & Swank, W. T. (1986). Hardwood biomass and net primary production following clearcutting in the Coweeta basin. Proceedings of the 1986 Southern Forest Biomass Workshop (Ed. by R. Brooks), pp Tennessee Valley Authority, Norris, U.S.A. Bormann, F. H. & Likens, G. E. (1981). Pattern and Processes in a Forested Ecosystem. Springer-Verlag, New York, U.S.A. Bormann, F. H., Likens, G. E., Siccama, T. G., Pierce, R. S. & Eaton, J. S. (1974). The export of nutrients and recovery of stable conditions following deforestation at Hubbard Brook. Ecological Monographs, 44, Covington, W. W. (1981). Changes in forest floor organic matter and nutrient content following clear cutting in northern hardwoods. Ecology, 62, Cromack, K. Jr (1973). Litter production and decomposition in a mixed hardwood and white pine watershed at Coweeta Hydrohgic Station, North Carolina. Ph.D. Dissertation, University of Georgia, Athens. Crossley, D. A. Jr (1977). The roles of terrestrial saprophagous arthropods in forest soils: current status of concepts. The Role of Arthropods in Forest Ecosystems (Ed. by W. J. Mattson), pp Springer-Verlag, New York. Federer, C. A. (1984). Organic matter and nitrogen content of the forest floor in even-aged northern hardwoods. Canadian Journal of Forest Research, 14, Fredrikson, R. L. (1975). Nitrogen, phosphorous, and paniculate matter budgets of five coniferous forest ecosystems in the Western Cascade range, Oregon. Ph.D. Dissertation, Oregon State University, Corvallis, U.S.A. Gosz, J. R., Likens, G. E. & Bormann, F. H. (1976). Organic matter and nutrient dynamics of the forest floor in the Hubbard Brook Forest. Oecologia, 22, Hix, D. M. & Barnes, B. V. (1984). Effects of clear-cutting on the vegetation and soil of an eastern hemlock dominated ecosystem, western Upper Michigan. Canadian Journal of Forest Research, 14, Hurlbert, S. H. (1984). Pseudoreplication and the design of ecological field experiments. Ecological Monographs, 54, Johnson, J. E., Smith, D. W. M. & Burger, J. A. (1985). Effects on the forest floor of whole-tree harvesting in an Appalachian oak forest. American Naturalist, 114, Klemmedson, J. O., Meier, C. E. & Campbell, R. E. (1985). Needle decomposition and nutrient release in Ponderosa pine ecosystems. Forest Science, 31, Mankau, R. & Imbriani, J. L. (1978). Studies on the mite Lasioseius scapulatus, a predator on soil nematodes. Nematropica, 8, 15. Mattson, K. G. (1986). Detrital carbon budget following clear-cutting hardwood forests in the southern Appalachians. Ph.D. Dissertation, University of Georgia, Athens, Georgia, U.S.A. Meentemeyer, V. (1978). Macroclimate and lignin control of litter decomposition rates. Ecology, 59, Norton, R. A. (1985). Aspects of the biology and systematics of soil arachnids, particularly saprophagous and mycophagous mites. Quaestiones Entomologicae, 21, Olson, J. S. (1963). Energy storage and the balance of producers and decomposers in ecological systems. Ecology, 44, Pierce, R. S., Martin, C. W., Reeves, C. C., Likens, G. E. & Borman, F. H. (1972). Nutrient loss from clearcutting in New Hampshire. Proceedings Symposium on Watersheds in Transition. American Water Resources Association Proceedings Series, 14, Risley, L. S. (1987). Acceleration of seasonal leaf fall by herbivores in the Southern Appalachians. Ph.D. Dissertation, University of Georgia, Athens, Georgia, U.S.A. Seastedt, T. R. (1979). Microarthropod response to clear-cutting in the southern Appalachians: effects on decomposition and mineralization of litter. Ph.D. Dissertation, University of Georgia, Athens, Georgia, U.S.A.

16 698 Decomposition following clear cutting Seastedt, T. R. (1984). The role of microarthropods in decomposition and mineralization processes. Annual Review of Entomology, 29, Seastedt, T. R. & Crossley, D. A. Jr (1981). Microarthropod response following cable logging and clear cutting in the southern Appalachians. Ecology, 62, Seastedt, T. R. & Crossley, D. A. Jr (1983). Nutrients in forest litter treated with naphthalene and simulated throughfall: a field microcosm study. Soil Biology and Biochemistry, 15, Stone, E. L. (1973). The impact of timber harvest on soils and water. Report of the President 's Advisory Panel on Timber and the Environment, pp U.S. Govt Printing Office, Washington, D.C. Swank, W. T. (1986). Biological control of solute losses from forest ecosystems. Solute Processes (Ed. by S. T. Trudgill), pp Wiley and Sons, New York. Swank, W. T. & Waide, J. B. (1980). Interpretation of nutrient cycling research in a management context: evaluating potential effects of management strategies on site productivity. Forests: Fresh Perspectives From Ecosystem Analysis. Proceedings of the 40th Annual Biology Colloquium (Ed. by R. H. Waring), pp Oregon State University Press, Corvallis, U.S.A. Technicon Industrial Systems (1970). Operations manual for the Technicon Autoanalyzer II system. Technical publication TAI , Technicon Industrial Systems, Tarryton, New York. Vlug, H. & Borden, J. H. (1973). Soil Acari and Collembola populations affected by logging and slash burning in a coastal British Columbia coniferous forest. Environmental Entomology, 2, Waide, J. B., Caskey, W. H., Todd, R. L. & Boring, L. R. (1987). Changes in soil nitrogen pools and transformations following forest clearcutting. Forest Hydrology and Ecology at Coweeta (Ed. by W. T. Swank and D. A. Crossley, Jr), pp Springer- Verlag, New York. Walther, P. B. (1983). Decomposition processes across a flooding gradient, with special reference to earthworm populations. Ph.D. Dissertation, University of Georgia, Athens, Georgia, U.S.A. Whitford, W. G., Meentemeyer, V., Seastedt, T. R., Cromack, K. Jr, Crossley, D. A. Jr, Santos, P., Todd, R. L. & Waide, J. B. (1981). Exceptions to the AET model: deserts and clear-cut forests. Ecology, 62, Wffl, G. M., Hodgkiss, P. D. & Madgwick, H. A. I. (1983). Nutrient loss from litterbags containing Finns radiata litter: influences of thinning, clearfelling, and urea fertiliser, New Zealand Journal of Forestry Science, 13, Witkamp, M. & Frank, M. (1970). Effects of temperature, rainfall, and fauna on transfer of cesium-137, K, Mg, and mass in consumer-decomposer microcosms. Ecology, 51, Zar, J. H. (1984). Biostatistical Analysis. Prentice-Hall, Inc., Englewood Cliffs, New Jersey. (Received 18 May 1987; revision received 23 January 1988 )