Interactive influences of silvicultural management and soil chemistry upon soil microbial abundance and nitrogen mineralization

Interactive influences of silvicultural management and soil chemistry upon soil microbial abundance and nitrogen mineralization

Fores~xjology Management ELSEVIER Forest Ecology and Management 103 (I 998) 129- I39 Interactive influences of silvicultural management and soil c...

1MB Sizes 3 Downloads 12 Views

Fores~xjology Management ELSEVIER

Forest Ecology

and Management

103 (I 998) 129- I39

Interactive influences of silvicultural management and soil chemistry upon soil microbial abundance and nitrogen mineralization Sherri Jeakins Morris *, R.E.J. Boerner Depurtnwni

of Plant

Biology,

Received

Ohio

Stute

24 February

Unkr.sity,

Columbus.

1997; accepted

19 May

OH 43210,

USA

1997

Abstract The purpose of this study was to determine whether soil acidification, a widespread, chronic mode of disturbance and forest thinning, a site specific acute disturbance, produced interactive effects capable of producing changes in more general ecosystem properties and processes. Two forested sites in the Daniel Boone National Forest, KY which were similar in history, management, and parent material but which differed in the degree of soil acidity were selected for study. In each forest site we sampled two plots that were experimentally thinned in the 1960’s and two adjacent unmanaged plots for soil chemical characteristics, microbial abundances, N mineralization and nitrification. There were significant differences between forest sites and significant effects of management for all soil chemical parameters, fungal biomass, N mineralization (site only) and nitrification. Soils from managed plots were generally higher in pH, nutrient availability and nitrification rates. There were significant interactions between site and management for NO,, pH, Ca, Ca:AI ratio, and nitrification resulting from the greater magnitude of the management effect at the less acidified site. Additionally, there were interactive effects of site and management in organic C, NH, and fungal hyphal length because plots at the two sites showed dissimilar effects of management. Modeling using path analysis determined that N mineralization was most strongly predicted by combinations of organic C, Al, Ca:Al ratio, and fungakbacteriai ratio, while inorganic N availability and Ca:AI ratio were the main factors for nitrification. Fungal hyphal length was most strongly predicted by Al and pH; in contrast, path analysis failed to produce a model for bacteria1 abundance. These results suggest (1) that acute and chronic modes of disturbance have the potential to interact in a significant and ecologically meaningful manner and (2) that research to assess forest health should be based on key ecosystem processes, such as N mineralization and nitrification. 0 1998 Elsevier Science B.V. KcJJ~?rds:

Acidification:

Management;

N mineralization;

Nitrification;

Bacteria

1. Introduction Terrestrial ecosystems are subject to a variety of natural and anthropogenic disturbances which vary

* Corresponding author. [email protected]

Fax:

+ 1-614-292-6345:

0378-l 127/98/$19.00 0 1998 Elsevier PII SO378I I27(97)00183-7

e-mail:

smor-

Science B.V. All rights reserved.

and Fungi;

Path analysis

in frequency, patch size, and intensity. Some are widespread and chronic (e.g. drought, atmospheric deposition) whereas others are site-specific and acute (e.g. fire, forest harvesting) resulting in a wide range of impacts on both ecosystem structure and function. Considerable research has been devoted to the understanding of the effects of single disturbance types on

vegetation and ecosystem functioning, both for natural and anthropogenic disturbances. However, less attention has been directed towards possible interactive effects of multiple disturbance types, especially combinations of chronic. widespread and sitespecific, acute disturbances, except at the physiological level (e.g. Mooney et al.. 199 1). A recent study of the effects of disturbance regimes on forest ecosystem function in mesic oak forests examined the effects of experimental thinning on forest structure and function along an historical atmospheric deposition gradient in the Ohio River valley (Boerner and Sutherland, 1995, 1997). The explicit purpose of that study was to determine whether changes resulting from chronic, widespread acidic deposition and changes caused by site-specific. acute biomass removal (thinning) had long-term effects on mesic oak forests in this region. The results identified longitudinal gradients of soil chemistry and nitrification rates associated with the deposition gradient (Boerner and Sutherland, 1995) and significant, long-term effects of experimental thinning on soil chemical properties (Boerner and Sutherland, 1997). This research did not, however, explicitly address the potential for these two disturbance types to have either interactive effects or impact upon a broader range of ecosystem processes. The focus of this current study was to determine whether such widespread, chronic and site-specific, acute modes of disturbance have interactive effects. To accomplish this, we examined two of the research sites studied by Boerner and Sutherland (1995, 1997) in more detail to determine whether soil acidity and management interacted, and if their independent or interactive effects caused changes in broader, more general ecosystem properties and processes. Our specific objectives were to: (1) quantify the independent and interactive effects of soil acidity and management (experimental stand thinning) on soil chemistry. microbial abundance. N mineralization, and nitrification as measures of ecosystem structure and function; and (2) to use exploratory path analysis, a form of structural modeling, to determine what specific environmental changes induced by acidity and/or management underlay changes in microbial abundance and/or organic N turnover.

2. Methods 2.1. Study sites We selected two forested sites in the Daniel Boone National Forest. KY. from the suite of forested sites used in an earlier study of N mineralization along an historical deposition gradient (Boerner and Sutherland, 1995, 1997): McKee Experimental Forest. Jackson County (lat 37”27’N, long 83”59’W) and Robinson Experimental Forest, Laurel County (lat 37”1S’N, long 83”2O’W). The two sites were chosen because they were similar in history. soil texture. deposition history, management regimes. and prrmanagement vegetation, but differed in the degree ot. soil acidity. The McKee Experimental Forest was undedain by interbedded sandstone. siltstone, shalt. and coal of the Pennsylvanian-aged Lee and Breathitt Formations (Hayes, 1989). The soils formed from the colluvium derived from these bedrocks were predominantly clay loam ultisols of the Shelocta-Cilpin complex (typic hapludults) (Hayes, 1989). Although the Robinson Experimental Forest was also underlain by bedrocks of the Breathitt Formation, at Robinson the sandstones and shales were interhedded with marine limestones (Hinrichs, 1978). The soils at the Robinson Experimental Forest were mapped as belonging to the Shelocta-Gilpin-Haelton complex (typic hapludults) (Randy Smallwood. Soil Conservationist, U.S.D.A. Natural Resources Conservation Service, Jackson, KY, personal communication). The region containing both the Robinson and McKee forests has a temperate, continental, humid climate with maximum and minimum yearly temperatures of 20°C and 6°C. and mean annual precipitation of I 17 cm (Hayes, 1989). Based on analysis of soil samples taken in unmanaged portions of these sites in 1992 by Boemer and Sutherland (1997) differences in parent materials produced soil chemical characteristics which varied significantly between the two forests (Table 1). Soils from the Robinson Experimental Forest were significantly higher in pH, Ca, Mg, K, P, and Ca:Al molar ratio than were soils from the McKee Experimental Forest. In contrast, McKee soils were higher in Al and NH, than were Robinson soils.

S.J. Morris,

R.E.J. Boemer/

Forest Ecology

Table 1 Soil chemical characteristics of McKee and Robinson Experimental Forests describing location, soil types and chemistry Characteristic

PH NO, w/kg NH, m/kg Al mg/kg Ca w/kg Ca:AI ratio Mg w/kg K m/kg P mg/kg

McKee (standard

Robinson (standard

error)

4.40 (0.04) 1.22 (0.10) 1.96(0.14) 161.37 (8.48) 122.88 (14.21) I .48 (0.62) 26.99 (2.55) 69.36 (2.73) 1.26 (0.05)

error)

5.10 (0.07) 1.38 (0.22) 1.30(0.10) 58.69 (9.41) 582.64 (50.09) 90.82 (17.15) 125.38 (7.90) 101.87 (2.45) 1.52 (0.05)

Significant differences at P I 0.05 are indicated by * and were taken from the original statistical analysis of Boerner and Sutherland f 1997) which included three additional forests.

To determine the recent deposition histories of these two sites, we analyzed the records of deposition of NO, and NH, between 1984-1995 for 18 National Atmospheric Deposition Program (NADP) collection sites in the central hardwood forest region (including sites in Arkansas, Kentucky, Illinois, Indiana, Missouri, North Carolina, Ohio, Pennsylvania, Tennessee, and West Virginia west of the Appalachian Mountains), including three NADP sites located within 50-80 km of our study sites (NADP/NTN, 1996). Although there was no significant linear relationship between longitude and NH, deposition, there was a significant negative relationship between longitude and NO, deposition (r2 = 0.683, P I 0.0001). These regressions suggested that

Table 2 Tree species composition

and basal area (ma/ha)

Site plot # (treatment)

Quercus

McKee 3 (control) McKee 10 (control) McKee 9 (managed) McKee 13 (managed) Robinson 9 (control) Robinson 11 (control) Robinson 4 (managed) Robinson 18 (managed)

20.4 22.1 20.9 15.0 16.6 21.0 1.5 6.3

“Includes bIncludes ‘Includes echinata.

spp.”

and Management

0.9 1.6 0.2 0.6 0.0 0.3 6.4 4.4

~pp.~

Fraxinus

129-139

131

Robinson was receiving slightly more NO, deposition per year than McKee (although the means were not significantly different). During 1984-1995, precipitation pH for the region immediately surrounding the study sites ranged from 4.27 to 4.56, and total N deposition ranged from 2.83 to 6.08 kg/ha/yr (NH, = 0.77 to 3.18 kg/ha/yr, NO, =9.85 to 17.19 kg/ha/yr). Based on these data, we concluded that the two sites had similar N deposition histories, at least over the period 1984- 1995. Four plots were chosen in each forest based on management history. Two of the four plots in each forest were acutely thinned in 1958-1960 as part of a larger study of oak regeneration and growth. Approximately 70% of the woody biomass was removed from McKee, whereas biomass removal in the Robinson plots was > 90% (M. Dale, U.S.D.A. Forest Service, personal communication). The remaining two plots in each forest were unmanaged control plots, wherein the woody vegetation of control and managedplots within a forest were similar prior to the experimental thinning. Details of the management treatments are given by Boerner and Sutherland (1997). As of 1994, oaks (Querclns spp.1 accounted for > 80% of the basal area of the study plots in the McKee Experimental Forest, and there were no obvious differences in species composition or relative abundancebetween control and managedplots (Table 2). The vegetation of the control plots at Robinson was also dominated by oaks. In contrast. the managed plots in Robinson had relatively lower

of the study sites in the McKee Acer

103 (1998)

americana

0.0 0.0 0.0 0.1 0.0 0.0 1.2 2.2

Quercus alba, Q. rubra, Q. prinus, Q. celutina. and Q. coccinea. Acer saccharum and A. rubrum. Prunus serotina, Nyssa syltatica, Juglans nigra, Cercis canadensis, Magnolia acuminata, Cornus florida, Sassafras albidum, Betula nigra,

and Robinson

Experimental

Liriodendron

tulipifera

0.0 0.0 0.1 2.0 0.2 0.3 9.1 2.6

Fagus grandifolia. and Oxydendrum

Asimina arboreum.

Forests in 1994 Others’

Total

0.5 1.o 0.8 0.6 3.3 3.4 4.5 5.2

21.8 24.7 22.0 18.3 20.1 25.0 19.3 20.7

triloba.

Caga

spp., Pinus

abundance of oaks and greater abundance of maples ( Acer spp.), white ash (Fraxinus americana), and yellow-poplar (Liriodendron tulipifera) than did either Robinson control plots or any of the plots in McKee (Table 2). 2.2. Field sampling In June 1994, two random transects were established perpendicular to slope contours in each plot. Along each transect, 10 samples were taken to a 15cm depth with a soil corer at random intervals from 3 to 15 m for determination of soil chemical parameters. Samples were also taken at six of these intervals for microbial analysis. As we observed no significant differences between the transects in soil chemical parameters, we pooled the data from the two transects within a plot; thus there were 20 samples per plot for soil chemical parameters with 12 of those samples also analyzed for microbial properties. Independence of samples for the microbial analyses was maintained through the use of alcohol washes of the soil corer between samples. Samples were stored and transported to the laboratory under refrigeration.

2.3. Microbial

clbund~mce

Bacterial density and fungal hyphal length were determined using fluorescent staining procedures with europium chelate and fluorescent brightener. followed by analysis with tluorescent microscopy (Con ners et al., 19941. All soils were stained for microbial biomass within 24 h of removal from the field. Fungal to bacterial biomass ratio (F:B ratio) was calculated on a mass basis using conversion factors for fungi and bacteria from Sakamoto and Oba ( 19941. 2.5. N minerulization

and nitrijkation

The fifth set of soil subsamples was placed in incubation chambers with moisture maintained at 50-70s of field capacity for 35 days. At day 35, soils were extracted with 2 M KC1 for NH, and NOi and analyzed as above. Initial and final concentrations of NH, and NO, were used to calculate daily rates of N mineralization and nitrificanon. Relative nitrification was calculated by determining the amount of NO, produced from the total amount of inorganic nitrogen available for nitrification.

2.3. Soil chemical parameters 2.6. Experimental design und data analwis Each soil sample taken for chemical analysis was air dried, sieved and divided into five subsamples. Soil pH was determined on one set of subsamples using a soil slurry of 0.01 M CaCl? (Hendershot et al., 1993) on six randomly chosen samples per transect. For the second set of subsamples, 15 g of soil was extracted with 30 ml of 3 M KC1 and analyzed for NH,, NO,, PO, and Al. Extractable NH, and NO, were determined from these extracts using colorometric techniques on a Lachat AE Autoanalyzer. PO, was determined using the SnCl, Method (APHA, 1976) and Al was determined using the Bamhisel and Bertsch (19821 method developed for hydroxy = aluminum. The third set of samples were extracted with 1.0 M NH,OAc and Ca was determined on an ILS 951 Atomic Absorption Spectrophotometer. Ca:Al ratio was calculated on a molar basis. Percent organic carbon was determined on the fourth set of subsamples using the Walkley-Black method (Allison, 196%.

The overall experimental design was a 2 X 2 X 2 completely randomized, factorial design. with two experimental forests (hereafter ‘sites’). two experimental treatments (managed vs. control), and two plots per treatment per experimental forest. All response variables were found to be normally distributed (PROC UNIVARIATE; SAS, 19851, then were analyzed by analysis of variance. To clarify the interactions between site and treatment, the data was sorted by site and analyzed as a one way ANOVA. All significant differences are at P I O.OS, except as otherwise noted. To further determine what soil factors might be responsible for differences among sites or treatments. we developed regression models for soil microbial biomass components, N mineralization, and nitritication using the system of structural modeling and causal analysis referred to as path analysis (Arbuckle. 1995) using Amos 3.51 (SmallWaters, Chicago. IL).

Xl.

Morris.

R.E.J. Boemer/

Forest Ecology

We chose to use path analysis rather than other regression procedures in order to incorporate the many, significant covariances among our independent variables (soil chemical properties) into the model.

and Management

I6 ;

103 (1998)

133

129-139

6, pH

NO,-N

s-

3. Results 3.1. Soil chemical parameters Two-way analysis of variance demonstrated both a significant difference between forest sites (P < 0.001) and a significant effect of management ( P < 0.001) for all eight soil chemical parameters tested. Furthermore, for six of the eight parameters (i.e. all but Al and PO,), there was a significant interaction between forest site and management. Extractable Al was significantly greater and extractable PO, significantly lower in soils from McKee than Robinson (Fig. 1). In addition, there was significantly more extractable Al in soils from control plots and significantly more extractable PO, in soils from managed plots in both forest stands.

300

McKee

C

McKee

Robinson

Fig. 2. Extractable NO,-N and Ca (in mg/kg), pH, molar Ca:AI ratio, organic C (in 5%) and extractable NH,--N (in mg/kg) for soils from McKee and Robinson Experimental Forests, and control and managed plots. Each histogram bar represents the mean of 40 samples for NO,-N and Ca, 24, 23, 24, 24 respectively for pH. 40, 39, 39, 40 respectively for Ca:AI ratio. 40 for NH,-N. and 40, 40, 39, 37 samples for organic C with the standard error of each mean shown. Significant differences at P 5 0.05 are indicated by an asterisk.

AI ’ (w&)

200 ;

* I

100

0

+ ; Organic (%I

rla

‘50 1.PO,-P j Wk)

Fig. I. Extractable Al and PO, (in mg/kg and *g/kg respectively) for soils from McKee and Robinson Experimental Forests, and control and managed plots. Each histogram bar represents the mean of 79 samples with the standard error of each mean shown. Significant differences at P I 0.05 are indicated by an asterisk.

There were two general patterns among those six soil parameters for which there was a significant interaction between forest site and management. In the first pattern, extractable NO,, Ca, Ca:Al ratio, and pH were all significantly higher in managed than control plots in both forests and greater in Robinson soils than McKee soils (Fig. 2). The interaction for these four variables was the result of the difference in the magnitude of the management effect at the two sites (Fig. 2). In contrast, in the second pattern, organic carbon was significantly greater in soils from the managed plots than from control plots at Robinson but not McKee, and extractable NH, was significantly more available in soils from managed plots than control plots at McKee but not Robinson (Fig. 2). An interaction resulted for both variables because plots at the two sites failed to show similar effects of management. Overall, with the exception of Al, the avail-

134

S.J. Morris,

R.E.J.

Boerner/

Forest

Ecology

und

Mcmagement

JO.3 f IYYN) 12%139

ability of most nutrients was greater in soils from managed plots than control plots. 3.2. Microbial

I===== Coptrol Mttnaged Lr.- _~~_.

abundance

The density of bacteria averaged 9.73 X 10’ cells/g ( & 3.66 X 10’) and did not vary significantly between sites or treatments. Fungal hyphal length, in contrast, was significantly greater in soils from McKee (8.31 i 0.60 m/g) than from Robinson (4.57 + 0.42 m/g). There was also a significant interaction between site and treatment in fungal hyphal length: management resulted in increased fungal hyphal length at McKee and decreased fungal hyphal length at Robinson, at least at P < 0.10 (Fig. 3). The fungal to bacterial biomass (F:B) ratio was significantly greater in soils from McKee (0.36 i 0.02) than in soils from Robinson (0.17 k 0.021, but was not affected significantly by management. 3.3. Nitrogen mineralization

McKee

cE

tkltrid

--.____.

Managed

0.25 -

9.0 ! Relative 1 Nitrification 6.0 ! (“/I

0.0 I.-. McKee

and nitrifcation

Under laboratory conditions, rates of N mineralization were significantly greater in soils from managed plots (0.616 mg N/kg soil/day f 0.027) than in soils from control plots (0.399 mg N/kg soil/day + 0.023). Laboratory N mineralization rates did not differ significantly between sites, nor was the interaction between sites and management significant. Net N mineralization and relative nitrification, the amount of nitrate produced as a function of the Fungal HyphalLength

0.50 1



Robinson

Fig. 3. DFS active fungal hyphal length (m/g) for soils from McKee control and managed plots, and Robinson control and managed plots. Each histogram bar represents the means of 23, 24. 22 and 23 samples respectively with the standard error of each mean shown. Significant differences at P 5 0.10 among treatments within forests are indicated by an asterisk.

Robinson

Fig. 3. Nitrate accumulation (mg/kg/day) and relative nitrification (T&‘/c)for soils from McKee control and managed plots. and Robinson control and managed plots. Each histogram bar represents the means of 80 (whole forests) and 40 samples (within forests) with the standard error of each mean shown. Significant differences at P < 0.05 among treatments within forests are indicated by an asterisk.

amount of inorganic N available for nitrification. were both affected significantly by site, management. and their interaction. The rates of N mineralization and relative nitrification were greater in Robinson soils than McKee soils, and greater in soils from managed plots than control plots at both sites (Fig. 4). Again, there was a significant interaction between sites and treatments resulting from difference in the magnitude of the management effect between sites. 3.4. Regression modeling We used path analysis to determine which of the soil chemical factors (if any> underlay the influence of acidity and management practices on microbial abundance and nitrogen mineralization. The first step was to determine which of the soil chemical compo-

S.J. Morris,

R.E.J.

Boerner/

Forest

Ecology

and

Al

PO,

pH

Ca

Ca:Al

NO,

NH,

PO, PH Ca Ca:Al NO, NH, Org c

“C-j “(6) “(6) “(-) V) ns ns

“(+) “(+) “(+) Y+) ns “(+)

Y+) “(+) “(+) “C-j ns

“(+) Y+) “(-) “(+)

“(+) ns Y+)

“C-j Y+)

“(+)

McKee

Al

PO,

pH

Ca

Ca:Al

NO,

NH,

PO, PH Ca Ca:AI NO, NH, Org c

9-j “(-) “(6) “(-) ‘G) “(-) Y+)

Y+) Y+) “(+) ns “(+) ns

“(+) “(+) ns Y+) “(4

‘Y+) ns Y+) ns

L ‘II+) ns

ns nh

Y+)

Robinson

Al

PO,

pH

Ca

Ca:M

NO?

NH,

PO, PH Ctl Ca:Al NO, NH, Org c

ns F) “(-) “C-j +I ,I(-) “(+)

“(+) “(+) Y+) Y+) ns “(+)

“(+) “(+) ns “(+)

Y+) “(+)

“(+)

F+)

‘F+)

ns “(+)

ns

“P i 0.05. hP i 0.10.

nents covaried. Twenty-three out of 28 of the possible pairwise combinations of soil chemical components covaried significantly for data pooled from

Table 4 Results of exploratory turnover

path analysis

of the effects of various

103 (1998)

135

129-139

both sites (Table 3). The five combinations of parameters that did not covary (i.e. that were truly independent variables) included either NH, or organic C. Nineteen out of 28 of the possible pairwise combinations of soil chemical components covaried significantly for data from the McKee site alone (Table 3). The nine combinations of parameters that did not covary included either NO, or organic C. Twenty-one out of 28 of the possible pairwise combinations of soil chemical components covaried significantly for data from the Robinson site alone (Table 3). The seven combinations of parameters that did not covary included either NH, or PO,. We then modeled microbial abundance using that covariance matrix, first for the entire data set with the two sites pooled, then for each site alone. We had little success in developing path regression models for bacterial abundance. The models for the two sites pooled and for Robinson explained < 11% of the variance (Table 4). The path model for bacterial abundance at McKee was somewhat stronger (1.’ = 0.279). but had only one significant regression component: available NH,. The models for fungal abundance were somewhat more robust, especially in the soils from Robinson. In the pooled and individual site models, fungal abundance was negatively related to pH and positively related to extractable Al. The path models for F:B ratio at the two individual sites each explained > 80% of the variance in F:B ratio among samples. Not surprisingly, Al and the abundances of bacteria and fungi were the most important regression coefficients for modeling F:B ratio.

Table 3 Covariance matrix for soil chemical characteristics of McKee and Robinson Experimental Forests pooled and individually. Signs within parentheses describe direction of relationship Pooled

Management

soil chemical

parameters

on microbial

abundance

and pathways

Model

Sites pooled

Robinson

McKee

Bacteria Fungi Fungi:Bacteria ratio N mineralization

r’ = 0.108 (none) r’ = 0.329 (Al, pH) r2 = 0.334 (Al) r2 = 0.527 (Al, F:B ratio, NH,,pH, PO,) r’ = 0.855 (Ca: Al ratio, NO,, NH,.fungi)

r’ = 0.007 (none) rz = 0.558 (Al) rz = 0.927 (Al, fungi, bacteria)

r’ = 0.279 r’ = 0.185 rz = 0.803 r’ = 0.476

Net nitrification

For each model, the total model determined by their critical ratio),

r’ = 0.664

(Organic

(NH,) (pH) (bacteria, (none)

N

fungi)

C. Ca:Al ratio, NO,) r2 = 0.826

r’ = 0.888

(NO,.

of organic

Ca:Al ratio, pH, NH,.

rz is given, followed by those factors in descending order of significance.

which

Ca, N min.)

were significant

(NO,.

regression

Ca,bacteria)

components

at P < 0.05 (as

For the exploratory path modeling for N mineralization and NO, accumulation. we added the abundances of bacteria and fungi and the F:B ratio to the matrix of soil chemical properties. The path models for N mineralization ranged in r2 from 0.476 to 0.664, with the model using the Robinson data being the strongest (Table 4). In the pooled data set, Al and the F:B ratio were the strongest regression components, whereas in the soils from Robinson alone, organic C content and the Ca:Al ratio were strongest. Although the model for N mineralization at McKee accounted for almost half of the variance among samples, no one regression component was significant at P < 0.05. Bacterial abundance and PO, availability entered this model at P < 0.10. The three path models for NO, accumulation each explained at least 82% of the variance among samples (Table 4). Overall, the availability of inorganic N and the Ca:Al ratio were the strongest regression components for NO, accumulation. 4. Discussion Our first objective was to quantify the independent and interactive effects of soil acidity and management on soil chemistry, microbial abundance, N mineralization. and nitrification. Soils from the more-acid McKee Experimental Forest had significantly lower pH, NO,, PO,. organic C. Ca, and Ca:Al ratio and significantly greater extractable Al and NH, than did the soils from Robinson. These differences were consistent with differences found in soil chemistry between sites with and without limestone influence in the parent material (e.g. Boerner and LeBlanc, 199.5) as well as with changes produced by acidic deposition in acidification-sensitive sites (Aber et al., 1989; Foster et al., 1989: Boemer and Sutherland. 1995). While there were no significant differences in bacterial abundance between McKee and Robinson soils, soils from the more-acid McKee site had significantly greater fungal abundance and fungal to bacterial biomass (F:B) ratio. Previous studies have also reported greater fungal biomass in more acid soils (Alexander, 1978), although others have demonstrated lower fungal biomass in experimentally-acidified soils (e.g. Baath et al., 1980) or no effect of acidification on fungi (Kytoviita et al..

1990). These apparent inconsistencies may be due to differences in the source of acidification (H,SO, vs. HNO, vs. mixtures of acids). experimental conditions (field vs. laboratory). rate of acidification, and methods of quantifying fungi. Although we did not observe a decrease in bacterial abundance with increasing acidification, as did Baath et al. ( 1980) and Kytoviita et al. (1990), there may still have been differences in bacterial community composition hetween our two sites that could not have been resolved by our staining methods. The rates of N mineralization reported here were similar to those reported for sandstone-derived soils in southern lndiana (Matson and Vitousek. 1981: Boerner and LeBlanc. 1995) and southern Ohio (Plymale et al.. 1987). hut somewhat lower than those reported from richer, till-derived soils in central Ohio (Boerner and Koslowsky, 1989). Nef N mineralization did not differ between sites; however. net NO, accumulation was significantly greater in soils fi-om the less-acidic Robinson Experimental Forest. Again these results Were consistent with those from studies in similar sites (Boemer and LeBlanc. 1995) and with those reported by Boemer and Sutherland ( 1995) in a study of N mineralization in unmanaged control plots at Robinson and McKci:. Most of the soil chemical parameters we measured were also affected significantly by management: pH, organic C, NO <, NH,&. PO,,. Ca, and Ca:Al ratio were all significantly greater in managed plots than control plots. and extractable Al was greater in the control plots than in the managed plots. Reports of increased pH as a result of thinning or harvesting are common in the literature (c.g. Matson and Vitousek, 198 I; Hombeck, 1992; Vesterdal et al., 199S), and Boemer and Sutherland f 1997) report parallel results for N. Ca. Ca:Al. Mg. and K in ‘i larger set of thinned versus control plots in oak forests of eastern North America. As both the degree of thinning and the differences between managed and control plots at Robinson were grealer than those at McKee. our data also suggest a relationship between the intensity of thinning and the magnitude 06 the differences in soil chemistry that result. However, the magnitude of difference may also have been affected by the difference in parent material and pre-management soil chemical conditions between our two sites (cf. Vesterdal et al.. 1995).

S.J. Morris,

R.E.J. Boerner/Forest

Ecology

There were no significant effects of management on either bacterial abundance or F:B ratio, and the pattern of change in fungal abundance as a consequence of management was in opposite directions in the two sites. Baath (1980) reported a decrease in fungal abundance as a result of clear-cutting in Sweden, and attributed the decrease to the cessation of root growth and exudation in the clear-cut area. Although we did not observe a similar decrease in our managed sites, our managed plots were probably small enough so that roots from resprouting and newly-established plants would quickly colonize the managed plots. Thus, the magnitude of effect on fungi may be related both to the intensity and overall area of harvest. Rates of N mineralization and nitritication were significantly greater in soils from managed plots than in soils from control plots. Matson and Vitousek (1981) demonstrated greater rates of N mineralization and nitrification in soils from clear-cuts than undisturbed forest sites in southern Indiana, and these differences were still detectable nine years after clear-cutting. They suggested the greater N turnover in clear-cuts may have been related to changes in substrate quality, soil temperature, and soil moisture, all of which may have also varied between our control and managed plots. Many of the soil chemical parameters we measured exhibited strong interactions between acidity and management. Organic C, pH, NO,, Ca, and Ca:Al ratio were all significantly greater in soils from Robinson than McKee, and greater in soils from managed plots than control plots. At Robinson, where the influence of limestone in the parent material produced colluvium with greater mineral availability, there was the potential for thinning-induced reductions in root uptake and proton production to result in large differences in soil chemistry between managed and control plots. In contrast, the soils at McKee were probably sufficiently nutrient poor prior to thinning so that any subsequent change in root uptake and proton exudation was unlikely to produce as large a change in soil chemistry as at the more nutrient rich Robinson. Thus, the sensitivity of forest soils to specific harvesting strategies may depend on specific soil factors, including cation exchange capacity and weatherable mineral availability (Binkley and Richter, 1987).

and Management

103 (1998)

129-139

137

There were no interactive effects of acidity and management on bacterial abundance, F:B ratio, or net N mineralization, and no consistent interactive effects on fungal biomass. There were, however, significant interactive effects on net nitrification. N mineralization is the result of the metabolic activities of a diverse guild of soil organisms, including bacteria, fungi, and soil animals. As a result, changes in microbial community composition, if they actually occur as a result of acidity and management, would not necessarily be expected to result in changes in the rate of N mineralization. In contrast, most nitrification is accomplished by a small and specialized guild of bacteria whose activity is sensitive to changes in environmental conditions, especially pH and moisture (Robertson, 1982). Thus, effects of soil acidity and management on soil properties, such as pH and moisture, should be expected to impact ecosystem processes performed by small, specialized guilds more rapidly than processes performed by generalists. Our second objective was to use exploratory path analysis to determine what changes induced by soil acidity and/or management underlay the variations in microbial abundance and organic N turnover we observed. We had only modest success in modeling microbial abundances based on the parameters we measured. We were unable to construct a model for bacterial abundance which explained more than 28% of the variance. However, this result was, in retrospect, not surprising. The abundance and diversity of bacteria present in a given forest soil is probably as closely linked to substrate diversity and lability as it is to total organic C or N content. Furthermore, important species compositional differences in the bacterial community may exist between these two forest sites or between managed and control plots, but would not be resolved by the methods we used. Our models for fungal abundance for the pooled data set and Robinson were somewhat stronger, explaining approximately 33% and 56% of the variance among samples, respectively. The path models for F:B ratio at the two individual study sites were stronger yet, with each accounting for at least 80% of the variance. The models for fungal abundance and F:B ratio indicated that of the soil variables we measured pH and extractable Al were the strongest predictors for fungi. Raubuch and Beese (1995) found

significant negative correlations between microbial biomass and Al in highly acidified soils. However, Hung and Trappe (1983) postulate that increases in fungal dominance are due to decreases in pH. We feel that our positive correlations of microbial biomass with extractable Al may be really be an indirect effect of pH on Al mobility, rather than a direct Al effect on fungi. Whether it is pH or Al availability that actually drives the increase in fungal hyphal length and F:B ratio in our sites, it is clear that either the direct or indirect result of increased acidity is an increase in fungi relative to bacteria. Our attempts to model net N mineralization and net nitrification using exploratory path analysis were more successful. With the data from the two study sites combined, net N mineralization was best predicted by extractable Al, and F:B ratio, with soils having the lowest F:B ratio and extractable Al showing the greatest rate of N mineralization. The N mineralization model for Robinson soils was even stronger, and indicated that soils highest in inorganic N and Ca:Al ratio (i.e. the soils from the managed plots) had the greatest N mineralization rates. Although exploratory path analysis produced a model explaining almost half of the variance in N mineralization rates among samples from McKee, no single factor we measured was a significant regression component at P < 0.05. The abundance of bacteria and available P were the strongest components in this model, though they were only significant at P < 0. IO. In sites where environmental variables vary relatively little, such as in our plots at McKee and in the Indiana plots studied by Boemer and LeBlanc (1995), the lack of variance in independent variables will result in an inability to construct predictive models for this generalized process. In contrast, the three path models we constructed for net nitrification each accounted for > 82% of the variance, and indicated that rates of nitrification were most strongly and positively correlated with the availability of inorganic N and the Ca:Al ratio. Other studies have also linked nitrification rates to inorganic N availability, soil acidity (as measured by pH or Ca:Al ratio). and soil moisture (e.g. Robertson. 1982; Plymale et al.. 1987; Boemer and LeBlanc, 1995). Because most nitrification is accomplished by a small guild of bacteria with well-defined and similar environmental constraints, robust modeling of

nitrification is an easier task to accomplish than is the modeling of a more diffuse process such as N mineralization. We draw two major conclusions from this study. First, acute and chronic modes of disturbance have the potential to interact in a significant and ecologically meaningful manner. Studies which address the impact of chronic disturbances or processes along putative gradients without considering the potential impact of differences in silvicultural history (e.g. Raubuch and Beese, 1995) have the pot.ential to draw spurious inferences concerning the effects of the chronic process. Our second conclusion relates to the search for indicators for the rather nebulous concept of a ‘forest health’ indicator. N mineralization and nitrification are key processes in most terrestrial ecosystems, and are subject to change as the result of changes it) environmental conditions. As a result, they meet several of the characteristics one would require for an indicator property or process. However, one must realize that the quantification of these processes requires multiple sampling, either on site or in the laboratory. It would be preferable to use indicators which can be quantified rapidly and with a single sampling visit. such as microbial biomass or abundance. However, as our results indicate that neither N mineralization nor nitrification was well correlated with either microbial abundance or biomass, the utility of microbial abundance or biomass as ecosystem health indicators appears weak at best. We advocate developing indicators based on key ccosystern processes, such as N mineralization and nitrification even if they are more time and effort consumptive. Acknowledgements This study was funded by the U.S.D.A. Forest Service Northern Global Change Program. We thank Kelly Decker. Robert Ford. Jennifer Brinkman, and Michael Fisher for assistance with field sampling and laboratory analysis, Elaine Kennedy Sutherland for stimulating the intellectual basis for this study and for editorial assistance, Michael Allen and Scott Subler for reviews of earlier drafts, and Jeff Stringer and Randy Smallwood for information on the soils and history of the study sites.

S.J. Morris,

R.E.J. Boerner/

Forest Ecology

References Aber, J.D.. Nadelhoffer, K.J., Steudler, P., Melillo, J.M., 1989. Nitrogen saturation in northern forest ecosystems. Bioscience 39, 378-386. Alexander, M., 1978. Effects of acidity on microorganisms and microbial processes in soil. In: Hutchinson, T.C., Havas, M. (Eds.), Effects of Acid Precipitation on Terrestrial Ecosystems. Plenum, New York, pp. 363-374. Allison, L.E., 1965. Organic carbon. In: Black, C.A., Evans, D.D., White, J.L.. Ensminger, L.E., Clark, FE. (Eds.), Methods of Soil Analysis: Part 2. Chemical and Microbiological Properties. American Society of Agronomy, Madison, WI, pp. 13671396. APHA, 1976. Standard Methods for the Examination of Water and Wastewater, 14th edn., American Public Health Association, New York. Arbuckle. J.L., 1995. Amos Users’ Guide. SmallWaters. Chicago. IL. Baath, E., 1980. Soil fungal biomass after clear-cutting of a pine forest in central Sweden. Soil Biol. Biochem. 12, 495-500. Baath, E., Berg, B., Lohm, U., Lundgren, B., Lundkvist, H., Rosswall, T., Soderstrom, B.E., Wiren, A., 1980. Effects of experimental acidification and liming on soil organisms and decomposition in a Scats pine forest. Pedobiologia 20, 85-100. Bamhisel, R.. Bertsch, P.M., 1982. Aluminum. In: Page, A.L., Miller. R.H., Keeney, D.R. (Eds.), Methods of Soil Analysis: Part 2. Chemical and Microbiological Properties. American Society of Agronomy/Soil Science Society of America, Madison, WI, pp. 275-300. Binkley, D., Richter, D., 1987. Nitrogen cycles and H+ budgets of forest ecosystems. Adv. Ecol. Res. 16, l-51. Boerner, R.E.J., Koslowsky, S.D., 1989. Microsite variations in soil chemistry and nitrogen mineralization in a beech-maple forest. Soil Biol. Biochem. 21. 795-801. Boerner, R.E.J., LeBtanc, D.C., 1995. Landscape position, substrate quality. and nitrate deposition effects on forest soil nitrogen dynamics in the Hoosier National Forest. Appl. Soil Ecol. 2. 243-251. Boerner. R.E.J., Sutherland, E.K., 1995. Nitrogen dynamics in oak forest soils along a historical deposition gradient. In: Gottschalk, K.W., Fosbroke, S.L.C. (Eds.1. Proceedings of the 10th Central Hardwood Forest Conference. U.S.D.A. Forest Service General Technical Report NE-197. Broomall, PA, pp. 523-533. Boerner, R.E.J., Sutherland, E.K., 1997. Soil chemical characteristics of control and experimentally-thinned plots in mesic oak forests along a historical deposition gradient. Appl. Soil Ecol., in press. Conners. K.. Zink, T., Bainbridge. D., Allen. M., Morris. S.,

and Management

103 f 1998) 129-139

139

1994. Europium staining for soil ecosystems disturbance evaluation (California). Restor. Manage. Notes 12. 21 l-212. Foster, N.W., Hazlett, P.W., Nicolson, J.A., Morrison, I.K., 1989. Ion leaching from a sugar maple forest in response to acidic deposition and nitrification. Water, Air, Soil Pollut. 48. 25l261. Hayes, R.A., 1989. Soil Survey of Jackson and Owsley Counties, KY. U.S.D.A. Soil Conservation Service, Washington DC. Hendershot, W.H., Lalande. H.. Duquette, M.. 1993. Soil reaction and exchangeable acidity. In: Carter, M.R. (Ed.), Soil Sampling and Methods of Analysis. Lewis Publishers, Boca Raton, FL, pp. 141-145. Hinrichs, E.N., 1978. Geologic map of the Noble Quadrangle. Eastern Kentucky. U.S. Geological Survey Geology Quad. Map GQ 1476. Washington DC. Hornbeck, J.W., 1992. Comparative impacts of forest harvest and acid precipitation on soil and streamwater acidity. Environ. Pollut. 77, 151-155. Hung. L.L., Trappe, J.M., 1983. Growth variation between and within species of ectomycorrhizal fungi in response to pH in c,itro. Mycologia 75, 234-241. Kytoviita, M., Fritze. H., Neuvonen, S., 1990. The effects of acidic irrigation on soil microorganisms at Kevo, northern Finland. Environ. Pollut. 66, 21-3 I. Matson, P.A., Vitousek. P.M., 1981. Nitrogen mineralization and nitrification potentials following clearcutting in the Hoosier National Forest, Indiana. For. Sci. 27, 78 l-79 I. Mooney, H.A.. Winner, WE.. Pell, E.J. (Eds.), 1991. Response of Plants to Multiple Stresses. Academic Press, San Diego, CA. NADP/NTN. 1996. NADP/NTN Annual Data Summary. http://nadp.nrel.colostate.edu/NADP. Plymale, A.E., Boemer. R.E.J., Logan, T.J., 1987. Relative nitrogen mineralization and nitrification in soils of two contrasting hardwood forests: effects of site microclimate and initial soil chemistry. For. Ecol. Manage. 21, 21-36. Raubuch. M.. Beese, F., 1995. Pattern of microbial indicators in forest soils along an European transect. Biol. Fertil. Soils 19. 362-368. Robertson, G.P., 1982. Nitrification in forested ecosystems. Phil. Trans. R. Sot. Lond. B 296, 445-457. Sakamoto, K., Oba, Y.. 1994. Effect of fungal to bacterial biomass ratio on the relationship between CO? evolution and total soil microbial biomass. Biol. Fertil. Soils 17, 39-44. SAS, 1985. Statistical Analysis System User’s Guide: Statistics. Version 5. SAS Institute. Cat-y. NC. Vesterdal, L., Dalsgaard, M., Felby. C., Raulund-Rasmussen. K., Jorgensen, B.B.. 1995. Effects of thinning and soil properties on accumulation of carbon. nitrogen, and phosphorus in the forest floor of Norway spruce stands. For. Ecol. Manage. 77, l-10.