Ecology

By Peter Saetre

To quantify C and N mineralization and the response of soil organisms in soil mixtures under controlled conditions, I set up a laboratory soil replacement series with soil from a pure spruce and a pure birch stand. To examine the role of a key species on process rates, the earthworm Aporrectodea caliginosa was added to half of the microcosms. Carbon mineralization and microbial biomass increased with the proportion of birch soil along the experimental mixtures, while N mineralization decreased. Microbial community structure (estimated by examining the phospholipid fatty acid pattern) changed linearly along the soil series. Carbon mineralization, microbial biomass, and microbial community structure in soil mixtures could therefore be predicted from the patterns in pure birch and spruce soils. However, the mineralization of N was lower in mixtures than predicted from the pure soils. Earthworms increased C and N mineralization, decreased microbial biomass, and modified microbial community structure, but they required at least 25% birch content in the soil mixture to be active throughout the experiment. The results suggest that microbial community structure and respiration are predictable in soil mixtures, but that earthworms exhibit thresholds and may respond in a nonlinear manner. Possible mechanisms behind the observed C and N mineralization pattern were explored with a simple model in which N mineralization was linked to the C flow through microbes. The model suggests that nitrogen concentration of assimilated substrate may have been lower in birch soil than in spruce soil, although the opposite was true for bulk soil. Alternatively, the rate of inorganic nitrogen losses, through abiotic fixation to organic matter and/or gaseous losses, may have been higher in birch than in spruce soil.

Key words: Aporrectodea caliginosa; C and N mineralization; microbial biomass; microbial community structure; phospholipid fatty acid pattern; soil mixtures.

INTRODUCTION

Plant species affect patterns of nutrient cycling in natural ecosystems (Hobbie 1992). This effect on nutrient dynamics may either be direct, through uptake, use and loss of nutrients; or indirect, by plants influencing the structure and function of the decomposer community. For example, the quality of litter differs among plant species, and litter quality may directly affect the decomposition rate through the palatability of the substrate to decomposers (Berg and Ekbohm 1991). Substrate quality may also strongly influence the composition of the decomposer community (Swift et al. 1979), which in turn may affect decomposition of plant material (Elliott and Elliott 1993). To what extent these effects depend on the direct influences of the plants or on indirect effects of the plant on decomposers, possibly mediated through the response of a key species or organism groups, needs to be investigated.

In mixed litter of different tree species, observed decomposition rates can rarely be predicted from single-species litters (McTiernan et al. 1997). For example, positive interactions with respect to carbon mineralization may be expected when carbon or nutrient limitations in one substrate are released by the presence of the other substrate, or when key organisms from one soil trigger decomposition of previously unavailable carbon sources in the other soil. Deviations in decomposition rates of soil mixtures and litter mixtures have been ascribed to changes in the abundance and composition of the soil fauna and micro flora (Chapman et al. 1988, Williams and Alexander 1991, Morgan et al. 1992).

Blair et al. (1990) suggest that the deviations in N fluxes in mixed litterbags from those predicted using single-species bags were caused by a nonlinear response of decomposer organisms to litter mixtures. Furthermore, they suggest that invertebrate-microbial interactions would have a greater effect on decay rates and nutrient release in the later stages of decomposition.

Birch has an almost legendary reputation in forestry history as a tree species that improves soil conditions (Gardiner 1968). Dimbleby (1952) envisaged that birch trees on heather moorland would change a raw humus into a mull within 60-100 yr through its influence on pH and soil activity and the associated increase in earthworm populations, known to be very important for soil formation and soil fertility (Darwin 1881). Therefore, discussions on the effects of birch on soil fertility have often been connected with speculations about a positive feedback between birch trees and earthworm populations.

Earthworms can be introduced into coniferous soils if both pH and the calcium content of the soil are increased, for example by liming (Robinson et al. 1992a, b). However, populations of epigeic and endogeic earthworms were also maintained in a Norway spruce stand when a mixture of birch and alder litter was added to the forest floor (Huhta 1979), and it has been suggested that leaving a broad-leaved tree species such as birch in a mixture with spruce may be a cost-effective way of improving soil quality for earthworms, as well as promoting forest growth.

This study has three main objectives. (1) I describe the changes in C and N mineralization along an laboratory birch-spruce soil replacement series, to discover whether there are any synergistic or antagonistic effects of mixing two soils of different qualities. (2) I examine whether the effects of the native endogeic earthworm Aporrectodea caliginosa on these processes differ across the soil mixtures. (3) I analyze whether changes in process rates along the soil replacement series or differences in the effects of earthworms are associated with corresponding changes in microbial community structure or microbial activity. To explore possible mechanisms behind the observed patterns in C and N mineralization pattern along the soil replacement series, I used a simple model, in which N mineralization was linked to the C flow through microbes.

METHODS

Sampling and pretreatment of soil and earthworms

Litter and humus from the uppermost 10 cm of the soil horizon were collected in April 1995 from four randomly selected points in the center of a 64-yr-old Norway spruce stand (Picea abies (L.) Karst.) and an adjacent 59-yr-old silver birch stand (Betula pendula Roth). Samples from each stand were pooled. Earthworms were collected by hand in the birch stand. The soil samples were stored overnight at 4 [degrees] C, and earthworms were stored at 4 [degrees] C until they were identified to species 4 d later.

Humus was passed through a 4-mm sieve, to homogenize the soil and to remove roots. Twigs and mosses were removed by hand from the litter. Humus and litter were frozen (-20 [degrees] C) for 3 d to kill earthworms and earthworm cocoons. After thawing, water content (105 [degrees] C), organic matter (OM) content (600 [degrees] C), and pH ([H.sub.2]O) were determined. Water-saturated humus was left to self-drain on a nylon mesh in 5-cm-tall cylinders. The water-holding capacity (WHC) was defined as the water content remaining in the humus after 24 h.

Microcosms were prepared by placing fresh humus (24 g dry mass) of five different mixtures in cylindrical styrene/nitrile plastic jars (diameter 8 cm), and watered to 45% of WHC. The soil mixtures used were: 100% spruce soil, 75% spruce and 25% birch soil, 50% spruce and 50% birch soil, 25% spruce and 75% birch soil, and 100% birch soil (based on OM content). Moist spruce needles and birch leaves (1.5 g dry mass) were placed on top of the humus in the same proportions as the soil. The result was a litter layer several millimeters thick on top of a humus layer 4 cm thick. In total, 15 microcosms of each soil mixture were prepared and left for 1 wk at 15 [degrees] C to equilibrate.

Aporrectodea caliginosa Savigny was the most common earthworm species among those collected, comprising more than 50% of the individuals. Juvenile individuals of this species were picked out, rinsed in tap water, and left for 48 h at 15 [degrees] C on wet filter paper to empty their guts. Thereafter the mass of each earthworm was recorded. Two earthworms (total mass 150 [+ or -] 30 mg, mean [+ or -] 1 sp) were introduced to six microcosms of each of the five soil mixtures, and worms were left to acclimatize for 24 h at 15 [degrees] C.

Site and soil description

The Norway spruce stand and the Silver birch stand had been planted on formerly arable land, near Mankarbo (60 [degrees] 14 [minutes] N, 17 [degrees] 28 [minutes] E), central Sweden. The site is 40 m above sea level, the temperature sum of the growing season (above 5 [degrees] C) is 1310 degree-days, and average precipitation is 400 mm during the growing season with a yearly average of 590 mm.

The canopies were closed in both stands. The Norway spruce stand almost lacked a herbaceous layer, but a few individuals of Vaccinium myrtillus L. and Rubus saxatilis L. persisted. Dicranum spp. and Hylocomium splendens (Hedw.) B.,S.&G., dominated the sparse bottom layer, and Mnium spp. and Pleurozium schreberi (Brid.) Mitt. were also common. The birch stand was more open to light penetration and had a rich herbaceous layer, dominated by Ranunculus repens L., Lactuca muralis L., Rubus idaeus L., and Athyrium filix-femina L. The bottom layer was poorly developed. In the birch stand, Hypnum cupressiforme, Rhytidiadelphus triquetrus, and Climacium dendroides were common in addition to the moss species found in the spruce stand. The soil at the site was a 60-90 cm deep Histosol (FAO) with an OM content of 80%, overlying a sandy sediment. The humus form was mull. In the Norway spruce stand, soil pH was 5.4, C:N ratio 17.5, cation exchange capacity (CEC) 119 [cmol.sub.c]/kg dry mass, [Ca.sup.2+] concentration 80 [cmol.sub.c]/kg dry mass, and base saturation (BS) 70%. Corresponding values in the birch stand were: pH 6.3, C:N ratio 14.7, CEC 144 [cmol.sub.c]/kg dry mass, [Ca.sup.2+] 125 [cmol.sup.c]/kg dry mass, and BS 91%.

The experiment

The experiment ran for 14 wk at 15 [degrees] C in darkness. The microcosms were watered weekly to maintain constant mass and soil moisture. C[O.sub.2] evolution from the microcosms was measured twice each week during the first seven weeks and thereafter once a week. Partial pressure of C[O.sub.2] in the air of the microcosms was determined before and after incubating microcosms for 2 h with airtight lids. A gas chromatograph (Hewlett-Packard 5890, Avondale, Pennsylvania), equipped with a thermal conductivity detector, was used. The rate of C[O.sub.2] accumulation was calculated, taking into account the C[O.sub.2] absorbed in soil water (Persson et al. 1989). For this calculation, pH between actual measurements was estimated by linear interpolation.

A subset of microcosms was destructively sampled at the start of the experiment, after 7 wk, and after 14 wk. At the start of the experiment, three replicate microcosms of each soil mixture were analyzed with respect to soil pH, soil C:N ratio, [K.sub.2]S[O.sub.4]-extractable N[H.sub.4]-N, N[O.sub.3]-N and total N, microbial biomass C and N, and phospholipid fatty acids (PLFAs). On the two following occasions, the mass of earthworms (after 48 h of starvation) in the destructively sampled microcosms was also recorded. Soil C:N ratio was measured only at the start.

On the second sampling occasion, 16 of the 30 starved earthworms, representing four different size classes ([less than]0.10, 0.10-0.12, 0.12-0.14, and [greater than]0.14 g fresh mass), were selected for determination of basal respiration. Four earthworms in the same size class were placed in one plastic jar with sterilized, moistened sand (70% [H.sub.2]O by mass). They were left for 24 h at 15 [degrees] C before C[O.sub.2] evolution was measured as described above.

Soil chemical and biochemical analysis

To determine pH ([H.sub.2]O), 7 g soil (fresh mass) was extracted for 2 h in 50 mL distilled water on a reciprocal shaker. The carbon and nitrogen content of freeze-dried soil was analyzed on a Carlo Erba NA 1500 (Carlo Erba Strumentazione, Milan, Italy). To determine extractable N[H.sub.4]-N, N[O.sub.3]-N, and total N, 10 g soil (fresh mass) was extracted in 100 mL 0.5 mol/L [K.sub.2]S[O.sub.4] for 1 h on a reciprocal shaker. The extract was filtered through a 0.2-[[micro]meter] cellulose acetate filter, then analyzed photometrically for N[H.sub.4]-N, N[O.sub.3]-N, and total N (after persulphate oxidation), using a flow injection analyzer (Tecator FIAstar 5010; Tecator AB, Hoganas, Sweden).

Microbial C and N were measured using the fumigation-extraction (FE) method (Martikainen and Palojarvi 1990), with the modification that 10 g soil (fresh mass) was fumigated for 20 h at 25 [degrees] C. Then the soil was extracted in 100 mL 0.5 mol/L [K.sub.2]S[O.sub.4] for 1 h, filtered through a 0.2-[[micro]meter] cellulose acetate filter, and analyzed for total N (as above) and dissolved organic carbon (DOC), on a total carbon analyzer (Shimadzu TOC-5000; Polynom, Solna, Sweden). To convert the amount of extracted N to microbial N, an extraction efficiency ([k.sub.EN]) of 0.54 was used (Brookes et al. 1985, Joergensen and Mueller 1995). The corresponding extraction efficiency used for carbon ([k.sub.EC]) was 0.45 (Vance et al. 1987, Joergensen 1995).

Phospholipid fatty acids (PLFAs) were used to characterize the microbial community structure and as an estimate of microbial biomass. PLFAs were extracted and analyzed by the method of Frostegard et al. (1991). Nomenclature of fatty acids follows that used by Tunlid and White (1992). In this study, 10 g soil (fresh mass) was frozen in liquid nitrogen, freeze-dried, and milled, after which a 0.50-g (dry mass) subsample was used in the extraction. Methyl esters derived from the phospholipids were analyzed on a gas chromatograph (HP 5890) equipped with a flame ionization detector, following Frostegard et al. (1993b). Thirty-one fatty acids (indicated in [ILLUSTRATED FOR FIGURE 6b OMITTED]) were identified using the retention times previously determined for soil PLFAs by gas chromatography/mass spectrometry. A conversion factor of 340 [[micro]mol] PLFAs/gram biomass C was used to estimate microbial biomass, assuming that there was no difference between the concentration of fatty acids and lipid phosphate, as suggested by data from organic soils (Frostegard et al. 1991). PLFAs considered to be of bacterial origin only (Frostegard et al. 1993a) were summed to give an index of bacterial biomass. The amount of the fatty acid 18:2[Omega]6 was used as an indicator of soil fungi (Federle 1986), and the ratio between 18:2[Omega]6 and bacterial PLFAs was used as an index of fungal/bacterial biomass, as suggested by Frostegard and Baath (1996).

Statistical analysis

Results were analyzed with a three-factor ANOVA: the factor Soil having five levels, the factor Earthworm having two levels, and the factor Time having two levels. Since the initial observations did not include the earthworm treatment, these were excluded from the ANOVA. Normality and independence of residuals were checked visually, and heterogeneity of variances tested with Cochran's test (Dixon and Massey 1983). Power analysis was performed with the help of operating characteristic curves (Montgomery 1991).

Soil respiration was assumed to decrease exponentially with time, and in the time period of this study it was assumed to approach a constant value. C[O.sub.2] evolution from each microcosm was therefore fitted to the nonlinear regression model dC[O.sub.2]/dt = C + [R.sub.0] x [e.sup.-rt], where C (in milligrams of C[O.sub.2]-C per day) is the constant that C[O.sub.2] evolution approaches with time, [R.sub.0] (in milligrams of C[O.sub.2]-C per day) is the initial contribution of the labile carbon pool to C[O.sub.2] evolution, r is the rate of exponential decrease in soil respiration with time (in days), and [R.sub.0]/r (in milligrams of C[O.sub.2]-C) is the total amount of C[O.sub.2] that can evolve from the labile carbon pool. Cumulative C[O.sub.2] evolution was calculated assuming a linear change between measurements. A two-factor ANOVA was used to analyze the effect of soil and earthworms on the derived variables C, r, [R.sub.0]/r, and cumulative C[O.sub.2] evolution.

If there are no interactions between the soils in the mixtures with respect to process rates and microbial variables, a linear relationship between these variables and the contents of each of the soils in the mixture can be expected. To detect whether the change in C and N mineralization, microbial biomass, and microbial community structure along the soil gradient differed from linearity, birch content in soil was treated as a continuous variable in the statistical model. The linearity of the regression was tested by the F ratio between the mean square of nonlinearity and the mean square error from the original model, in which birch content was treated as a class variable. The mean square of nonlinearity was calculated by subtracting the original model's residual sum of squares from the regression model's residual sum of squares, and thereafter dividing by the corresponding difference in degrees of freedom (Mead et al. 1993).

To examine whether the experimental treatments affected the microbial community structure, patterns of PLFAs were analyzed using principal component analysis (PCA). Each sample was represented by a vector of PLFAs (expressed as log-transformed percentage of total PLFA content, based on moles per liter). The sample scores along the two first ordination axes of the PCA were used as derived variables and analyzed with the three-way ANOVA. All statistical analyses were performed using SYSTAT (1992).

A simple model to link microbial dynamics to C and N mineralization

To explain the observed patterns in C and N mineralization, I formulated a simple model in which these processes were linked to microbial dynamics, earthworm grazing and excretion, and inorganic nitrogen losses [ILLUSTRATION FOR FIGURE 1 OMITTED]. This model is based on the concept that microbes are carbon limited, which means that nitrogen mineralization and immobilization by microbes passively follow the flow of carbon, and that microbial assimilation is limited by the amount and quality of substrate only. Thus microbial carbon assimilation is assumed to be resource- or donor-controlled, and not influenced by the microbial biomass (Zheng et al. 1997). The model is a special case of a general model of decomposition of soil organic matter (SOM) (Bosatta and Agren 1991, Agren and Bosatta 1996).

The change in microbial biomass carbon ([C.sub.m]), microbial biomass nitrogen ([N.sub.m]), and the inorganic nitrogen pool ([N.sub.inorg]) is described by three differential equations. Since microbes are carbon-limited, we can write these three equations:

d[C.sub.m]/dt = A - D - P - R (1)

d[N.sub.m]/dt = [f.sub.s]A - [f.sub.m]D - [f.sub.m]P - [M.sub.m] (2)

d[N.sub.inorg]/dt = [M.sub.m] + [M.sub.ew] - L (3)

where variables are defined as in Table 1.

If we assume that microbial respiration (R), mortality (D), and predatory losses from earthworm grazing (P) are proportional to microbial biomass carbon ([C.sub.m]), this can be written:

R = [q.sub.C[O.sub.2]][C.sub.m] (4)

P = p[C.sub.m] (5)

D = d[C.sub.m] (6)

where [q.sub.C[O.sub.2]] is the metabolic quotient of microbes, p is the fraction of microbes eaten by earthworms per unit time, and d is the relative nonpredatory mortality rate of microbes.

When the microbial biomass is in equilibrium with available substrate, (i.e., d[C.sub.m]/dt = d[N.sub.m]/dt = 0), we can use Eq. 1 to eliminate the microbial assimilation (A) from Eq. 2. Then Eqs. 2, 4, 5, and 6 can be used to express the microbial nitrogen mineralization/immobilization rate ([M.sub.m]) in terms of microbial biomass carbon ([C.sub.m]):

[M.sub.m] = [C.sub.m] ([q.sub.C[O.sub.2]] [f.sub.s] - [[f.sub.m] - [f.sub.s]][d + p]). (7)

The rate of microbial carbon assimilation is expected to be the same in the control and in the earthworm treatment for each substrate quality. This is based on the assumptions that microorganisms are carbon limited and that earthworms do not affect substrate quality. Thus, we can use Eq. 1 when microbes are in equilibrium to express the fraction of microbes eaten per unit time (p) as a function of the ratio between microbial biomass in the control and the earthworm treatment:

p = ([C.sub.m(control)]/[C.sub.m(earthworm)] - 1) ([q.sub.C[O.sub.2]] + d). (8)

[TABULAR DATA FOR TABLE 1 OMITTED]

Thus, by using Eqs. 7 and 8 it is possible to express the change in inorganic N (Eq. 4) as a function of four microbial parameters ([C.sub.m], [q.sub.C[O.sub.2]], [f.sub.m], and d), the substrate N:C ratio ([f.sub.s]), the contribution of earthworm mineralization ([M.sub.ew]), and abiotic fixation or gaseous losses of N (L). Earthworm contribution to net mineralization can be thought of as being composed of two components, one originating from grazing on microbes, and another from feeding on other food sources. For simplicity, I have assumed that all microbial nitrogen consumed by earthworms is mineralized and that mineralization from feeding on other food sources (c) is fairly constant. With these two assumptions, the contribution of earthworms to N mineralization ([M.sub.ew]) can be written:

[M.sub.ew] = [f.sub.m] P + C. (9)

RESULTS

Initial differences in soils

The soil from the birch stand had initially higher pH, higher amounts of extractable inorganic nitrogen, higher microbial biomass, and higher soil respiration as compared to the adjacent spruce stand (Table 2). The [TABULAR DATA FOR TABLE 2 OMITTED] initial differences in soil pH, microbial biomass, microbial biomass N, and soil respiration were retained throughout the incubation.

Earthworm growth and respiration

In all soil mixtures containing birch soil, earthworms increased in mass by [approximately]45% during the 14-wk incubation. In spruce soil, earthworm biomass increased by only 20%. By the end of the experiment, these worms had curled up and appeared to be inactive. Earthworms mainly grew during the first 7 wk of incubation, and there was no difference in relative growth at 7 wk and at 14 wk ([ILLUSTRATION FOR FIGURE 2 OMITTED], Table 3), although after 14 wk earthworms tended to grow better with more birch soil.

The basal respiration of earthworms in moist sand was 36 [+ or -] 4.0 [[micro]gram] C[O.sub.2]-C[center dot][d.sup.-1][center dot][(g earthworm fresh mass).sup.-1] (mean [+ or -] 1 SE). Basal respiration was not correlated to size class (P = 0.41).

C and N mineralization

The content of birch soil in the soil mixture and the presence of earthworms both had a clear, positive effect on carbon mineralization ([ILLUSTRATION FOR FIGURE 3A, B OMITTED], [ILLUSTRATION FOR FIGURE 4A OMITTED], Table 3). Birch soil had a persistent, positive effect on C[O.sub.2] evolution, clearly visible in the long-term behavior of carbon mineralization (C in the regression), which increased linearly with birch content in soil [ILLUSTRATION FOR FIGURE 3A OMITTED]. The positive effect of earthworms, on the other hand, was limited to the early, exponential phase of decomposition. The rate of decrease in soil respiration over [TABULAR DATA FOR TABLE 3 OMITTED] time (r) was lower in the earthworm treatments than in the control, and earthworm presence thus increased the total amount of C[O.sub.2] evolution during the exponential phase ([R.sub.0]/r) [ILLUSTRATION FOR FIGURE 3B OMITTED].

An exception to this pattern was the effect of earthworms in pure spruce soil, where the initial positive effect of earthworms was offset by an [approximately]20% lower long-term C[O.sub.2] evolution (P [less than] 0.01, pairwise comparison between the asymptotic C value in earthworm and control in spruce soil). This resulted in a cumulative C[O.sub.2] evolution from the pure spruce soil that was higher in the control than in the earthworm treatment at the end of the experiment. In the earthworm treatment the change in cumulative C[O.sub.2] evolution along the soil gradient differed significantly from linearity (P [less than] 0.05; [ILLUSTRATION FOR FIGURE 4A OMITTED]). Carbon mineralization was fairly well described by the regression model dC[O.sub.2]/dt = [R.sub.0][e.sup.rt] + C [ILLUSTRATION FOR FIGURE 3 OMITTED], and fitting data from each microcosm to this model gave [r.sup.2] values of 0.86 [+ or -] 0.08 (mean [+ or -] 1 SD).

Net nitrogen mineralization (i.e., accumulation of inorganic nitrogen), decreased with increasing birch content in soil mixtures, and was higher in earthworm treatments than in the control ([ILLUSTRATION FOR FIGURE 4B OMITTED], Table 3). Earthworms increased the net rate of N mineralization by [approximately]140 [[micro]gram] N[center dot][(g fresh mass).sup.-1[center dot][d.sup.-1]. However, the effect of earthworms was lower in 100% birch soil. The change in mineralization rate along the soil gradient was not linear in the control microcosms (P [less than] 0.05), and I measured lower mineralization rates in soil mixtures than predicted. The mineralization rates appeared to differ from linearity in the earthworm treatments also, but this deviation was not statistically significant. The rate of nitrogen mineralization increased moderately with time in all treatments, the rate being [approximately]14% higher at week 14 vs. week 7 (Table 3). An exception to this pattern was the earthworm-spruce treatment, where nitrogen mineralization tended to decrease with time. Inorganic nitrogen occurred almost exclusively as N[O.sub.3]-(98 [+ or -] 0.2%, mean [+ or -] 1 SD)[greater than]), and the [N[H.sub.4].sup.+]; [N[O.sub.3].sup.-] ratio was affected neither by soil mixture nor by earthworm treatment (data not shown).

Microbial biomass and activity

Microbial biomass (as indicated by the sum of PLFAs) and microbial biomass N increased linearly with birch content in soil and decreased uniformly in the presence of earthworms ([ILLUSTRATION FOR FIGURE 5A, B OMITTED], Table 3).

There was a small decrease in microbial biomass (sum of PLFAs) during the first 7 wk of the experiment, followed by a build-up during the next 7 wk. The quotient between microbial respiration (total respiration minus earthworm respiration) and microbial biomass ([q.sub.C[O.sub.2]]) thus decreased continuously throughout the experiment. The specific activity of microbes was not affected by soil mixture or earthworms, but at the end of the experiment [q.sub.C[O.sub.2]] was lower in pure spruce soil with earthworms than in pure spruce control soils (P [less than] 0.01).

N in microbes followed the dynamics of microbial biomass. However, in microcosms lacking earthworms, microbial biomass N increased more than microbial biomass between weeks 7 and 14. This resulted in a higher microbial N:C ratio in the control than in the earthworm treatments at the end of the experiment (Table 3).

The estimates of microbial C with the fumigation-extraction (FE) method were highly variable, and were affected neither by soil mixture nor by earthworm treatment. However, a power analysis also revealed that the pattern in the sum of PLFAs could well be contained within the noise of the FE method. That is, when using the FE method the probability of not detecting a true difference of 10% in microbial C between the control and the earthworm treatment (the observed difference in the sum of PLFAs), was [greater than]55%, and the probability of not detecting a true difference of 30% between two soil mixtures (the difference between birch and spruce soil in the sum of PLFAs) was at least 15%. Therefore these measurements were not used to estimate microbial biomass.

Microbial community structure

Microbial community structure, as revealed by the PLFA pattern, changed linearly along the soil gradient and was modified by earthworm presence and by time [ILLUSTRATION FOR FIGURE 6A, B OMITTED]. The soil gradient was reflected in the first PCA axis (Table 3), which explained most of the variation in PLFAs (58.1%). Of the 31 PLFAs studied, 27 changed linearly in relative concentration (percentages, based on moles per liter) along the soil mixture gradient (data not shown). Earthworm presence and time significantly influenced the scores along the second PCA axis (Table 3), which explains an additional 14.1% of the variation in PFLAs. The effects of time and of earthworms along this axis were in opposite directions; the PCA scores increased with time, but were lower in earthworm treatments than in the control.

Only a few of the PLFAs influenced the second PCA axis; 18:2[Omega]6 (an indicator of soil fungi) decreased in the presence of earthworms, while 10Me18:0 (an indicator of actinomycetes) was higher in the earthworm treatment than in the control. Of all the PLFAs, 18:2[Omega]6 had by far the greatest impact along the second PCA axis (see arrow in [ILLUSTRATION FOR FIGURE 6B OMITTED]). Thus the main effect of earthworms and time on microbial community structure could be reduced to a change in the relative concentration (percentages, based on moles per liter) of 18:2[Omega]6. This was also reflected in the fungal/bacterial ratio, which was significantly lower in the earthworm treatment than in the control, and which increased with time (Table 3).

After 14 wk the effect of earthworms in spruce soil differed from the pattern described above. At this time the relative concentration of 18:2[Omega]6 was higher in earthworm treatments than in the control (P [less than] 0.05), [TABULAR DATA FOR TABLE 4 OMITTED] resulting in higher scores along the second PCA axis (P [less than] 0.05) and a higher fungal/bacterial ratio (P [less than] 0.05). Thus the effect of earthworms in spruce soil observed at week seven was similar to that in the other soil mixtures, but was reversed by the end of the experiment.

Parameterization and fit of the N mineralization model

When analyzing the pattern of C and N mineralization with the model described in Fig. 1, I assumed that each parameter could be represented by an average value over time. This served as a first approximation of parameter values and simplified the analysis substantially.

I regarded initial values of soil N:C ratio ([f.sub.s]) to be representative for the entire incubation time, since only a minor part of the substrate carbon ([approximately]3%) was mineralized during the experiment. To estimate microbial parameters, I used data from week 7. At this time, the metabolic quotient ([q.sub.C[O.sub.2]]) was 0.027 [+ or -] 0.001 [d.sup.-1] (mean [+ or -] 1 SD) and the microbial N:C ratio 0.094 [+ or -] 0.008, and they were not significantly affected by treatment (P [greater than] 0.1). Therefore, these parameters were set constant for all treatments. I used a value of [10.sup.-4] [h.sup.-1] for the relative nonpredatory mortality rate (d) (Anderson and Domsch 1990), and assumed this rate to be independent of treatment. The predation parameter (p) was solved for each soil mixture using Eq. 8, observed values of microbial C, and the above estimates of [q.sub.C[O.sub.2]], and d (Table 4).

Since the variation in rate of nitrogen mineralization over time was small, the rate was assumed to be constant for modeling purposes, and was estimated using the mean of weeks 7 and 14. From the results (Table 3) it was apparent that both microbial C and [q.sub.C[O.sub.2]] changed over time. However, 85-90% of carbon mineralization could be explained by the constant long-term rate of C[O.sub.2] evolution (C). Therefore, setting microbial C and [q.sub.C[O.sub.2]] to constant values is an acceptable approximation of the late phase of the experiment, and provides a rough estimate for the entire period.

Three versions of the model d[N.sub.inorg]/dt = [M.sub.m] + [M.sub.ew] - L (Eq. 2) were fitted to the data, using one set of observations per treatment (n = 10). In version A of the model, nitrogen losses (L) due to abiotic fixation or volatilization were excluded. In version B, nitrogen losses were excluded, and the N:C ratio of the substrate utilized by microbes (f.sub.s]) was allowed to change linearly along the soil gradient, instead of being set to the observed values. Version B thus represents the case when microbes in birch soil utilize substrate with lower nitrogen content than do the microbes in spruce soil. In version C, nitrogen losses were allowed to change linearly along the soil gradient. Version A could not be fitted to the data. However, the other two versions gave an equally good fit ([r.sup.2] = 0.92 and 0.94, respectively) and almost identical estimates of the contribution of earthworms to nitrogen mineralization (c = 1.62 N[center dot][d.sup.-1][center dot][[g OM].sup.-1]). Fitted values for nitrogen losses and soil N:C ratios from versions B and C of the model are shown in Table 4. Observed and fitted values of net nitrogen mineralization for version B of the model are shown in Fig. 4b.

DISCUSSION

C and N mineralization

The microcosm experiment showed that C mineralization rate increased with increasing birch content in the soil mixture, while net N mineralization rate decreased. C and N mineralization changed regularly along the soil replacement series. However, there were small and significant deviations from linearity that indicated both synergistic and antagonistic soil mixture effects.

C and N mineralization of litter mixtures and soil mixtures can rarely be predicted from the behavior of pure litters and soils (Chapman et al. 1988, Blair et al. 1990, Williams and Alexander 1991, Morgan et al. 1992). In this study, C[O.sub.2] evolution from soil mixtures without earthworms was as predicted, while N mineralization in soil mixtures without earthworms was lower than expected. My results are consistent with those of McTiernan et al. (1997), who found that C[O.sub.2] evolution from mixed spruce-birch litter was predictable from the behavior of pure litters, but that release of inorganic nitrogen during the first 15 wk of incubation was lower in mixtures than expected. Nitrogen retention or gaseous losses in birch soil may have increased in mixtures with spruce soil, which had higher rate of net N mineralization.

In microcosm experiments, earthworms almost always increase N mineralization (Haimi and Huhta 1990, Haimi and Einbork 1991, Binet and Trehen 1992), whereas soil respiration is usually less affected (Scheu 1994, Wolters and Ekschmitt 1995). My results are consistent with earlier findings. Earthworm N excretion, in connection with feeding activity, probably caused the high rate of N mineralization throughout the experiment. The effects of earthworms on C mineralization was, on the other hand, transient and I suggest that earthworms mobilized a limited pool of labile carbon for decomposers in the early phase of the experiment.

Earthworms had unexpected effects on mineralization processes in the pure soils. The comparatively small effect of earthworms on N mineralization in pure birch soil may be explained by birch soil having been exposed to a higher degree of earthworm processing prior to the experiment. Earthworms may have shifted the characteristics of the mineralizable stores of SOM in the birch stand, which had a much higher earthworm abundance than the spruce stand (P. Saetre, personal observation). If there were nitrogen-rich carbon sources available for earthworms to feed upon in the spruce soil, the contribution of earthworms to N mineralization would be expected to be larger in microcosms containing spruce soil.

The negative effect of earthworms on C mineralization in pure spruce soil can probably be explained by the fact that earthworms did not thrive in pure spruce soil. The earthworms may have suffered from the low pH and [Ca.sup.2+] concentration. They may also have depleted their food sources. The negative effect of earthworms was only apparent toward the end of the experiment, when a relaxed grazing pressure in the spruce soil was associated with a change in microbial community structure and a lower specific microbial activity ([q.sub.C[O.sub.2]]) resulting in even lower rates of C mineralization.

Microbial community structure

The physical and chemical properties of the soils had major effects on microbial biomass and microbial community structure along the experimental gradient (Table 3). The phospholipid fatty acid composition of a soil sample reflects the lipid composition of all intact biological membranes in that sample. Most of the PLFAs are shared by many groups within the soil community (Vestal and White 1989). Therefore it is difficult to relate changes in PLFA pattern to specific groups of organisms.

However, comparison of the pattern found in this study with patterns from earlier work allows several conclusions to be drawn. PLFA 16:1[Omega]5 showed the highest relative increase with birch soil content, and it is also the most sensitive PLFA associated with increasing pH in limed soils and soil fertilized with wood ash (Frostegard et al. 1993a, Baath et al. 1995). Also, the relative increases in PLFAs 16:1[Omega]9, 18:[Omega]7c, and a15:0 with increasing pH, and the decreases in i15:0 and Cy19:0 are consistent among this and previous studies. Baath et al. (1995) concluded that the above changes in PLFA pattern were closely correlated with increased bacterial activity, which corresponds well with the increased soil respiration with the proportion of birch soil in this study. Furthermore, the microbial community appeared to be structurally dominated by bacteria in both the spruce and birch soil, since the fungal: bacterial ratio in these soils was approximately one-fourth that previously found in spruce forest, and was more similar to the ratio found in grassland and arable soils (Frostegard and Baath 1996).

The observed decrease in microbial biomass in the earthworm treatment was most probably due to earthworm grazing. The change in microbial community structure reflected a relative decrease of soil fungi and a relative increase of actinomycetes. Soil fungi are known to be destroyed during passage through the earthworm gut (Anderson 1988), and actinomycetes have been reported to be selectively stimulated in the earthworm gut (Brown 1995).

There was no sign of an indirect positive effect of earthworms on microbial specific activity ([q.sub.C[O.sub.2]]) - in contrast to Zhang and Hendrix (1995), for example. However, it is likely that stimulatory effects of earthworms are most pronounced during the early phases of decomposition. Fig. 3b suggests that during the exponential phase of decomposition, earthworms had an overall positive effect on soil respiration, while the effect of grazing became predominant after day 40. Between days 5 and 40 I did not measure microbial biomass; therefore, a passing stimulatory effect on metabolic quotient may have been overlooked in this study.

Modeling C and N mineralization

Cumulative carbon mineralization increased with increasing content of birch soil, while cumulative net nitrogen mineralization decreased. This is somewhat surprising, since the nitrogen content of the bulk soil was higher in the birch than in the spruce soil. This result was explored with a simple model where nitrogen passively follows the flow of carbon through microbes. The N mineralization rate will thus be a function of the carbon assimilation rate, microbial properties, and substrate nitrogen concentration. This approach is conceptually simple and it has been used successfully in describing the dynamics of carbon and nitrogen mineralization from decomposing soil and litter (Agren and Bosatta 1996). From the modeling exercise it became apparent that with the assumptions used, this simple model could not yield a negative correlation between C and N mineralization. However, if one of several auxiliary assumptions were included in the model it was possible to fit the model to observed data. Two of these were: (1) a systematic deviation of the N:C ratio of the microbially assimilated carbon from that observed in the bulk soil, resulting in a nitrogen concentration of available carbon sources that decreased with increasing proportion of birch soil; and (2) a loss of inorganic nitrogen due to abiotic nitrogen fixation or volatilization that increased with the proportion of birch in soil.

When soil or litter decomposes, nitrogen concentration increases with time, whereas decomposition rate and substrate quality decrease (e.g., Agren and Bosatta 1996). The higher nitrogen concentration in the birch soil compared with spruce soil may be due to a greater portion of that soil being comprised of more humified and nitrogen-rich compounds, contributing little to C mineralization. Bulk soil may poorly reflect the nitrogen concentration in the OM assimilated by microorganisms. Thus, the nitrogen content of microbially assimilated carbon may actually have been lower for birch than for spruce soil (see fitted nitrogen to carbon ratio of substrate [f.sub.s] in Table 4).

Abiotic fixation of [N[H.sub.4].sup.+] to SOM is strongly pH dependent (Nommik and Vahtras 1982). High rates of abiotic fixation ([approximately]20 [[micro]gram] N[center dot][[g OM].sup.-1][center dot][d.sup.-1]) has been reported for litter and humus in short-term experiments (Axelsson and Berg 1988, Schimel and Firestone 1989). However, fixation rates were two orders of magnitude lower in a 20-d incubation of agricultural soils ([approximately]0.7 [[micro]gram] N[center dot][[g OM].sup.-1][center dot][d.sup.-1]; Trehan 1996). Gaseous losses of nitrogen may occur because N[O.sub.2] is produced parallel to nitrification, and nitrogen may also be lost through biological denitrification (Paul and Clark 1989). When soil moisture is at field capacity or lower, N[O.sub.2] production is usually [less than]1% of net nitrification rates, whereas [N.sub.2] losses may be an order of magnitude higher (Martikainen et al. 1993, Maag and Vinther 1996). Denitrification rates may be up to 5 times higher from earthworm casts as compared to uningested soil (Elliott et al. 1990, 1991).

Thus, rates of nitrogen losses due to abiotic fixation or gaseous losses can be expected to be lower in spruce soil than in birch soil, which had a higher pH and had been exposed to a higher earthworm abundance prior to the experiment. However, expected rates from literature appear to be an order of magnitude lower than rates fitted in the model (L in Table 4). Therefore, I propose that a decreasing nitrogen concentration in available carbon sources with increasing proportion of birch soil is likely to have caused the negative correlation between carbon and nitrogen mineralization along the soil replacement series.

Conclusions

Without earthworms there were linear changes in microbial biomass and microbial community structure along the experimental spruce-birch soil replacement series, but the function of the microbial community with respect to decomposition processes did not change substantially, (e.g., [q.sub.C[O.sub.2]] and microbial N:C ratio were approximately constant in the soil mixtures). Therefore, microbial respiration and thus C mineralization in the soil mixtures without earthworms could be predicted from rates in pure soils. However, this was not the case with earthworms. Earthworms required at least 25% birch content in the soil mixture to be active throughout the experiment, and their effect on N mineralization was lower in pure birch soils than in other soil mixtures. Empirical and modeling results suggest that differences in earthworm exposure prior to the experiment may have affected the nitrogen concentration of the carbon pools available for earthworms and microorganisms, and thus the rate of N mineralization.

ACKNOWLEDGMENTS

I thank Jan Bengtson for discussions and encouragement throughout the work with this paper. I also thank Erland Baath, Goran Agren, Jon Norberg, and Michael Sjoberg for discussions and comments on the manuscript; Jeremy Flower-Ellis for greatly improving the language and structure of this paper; and Pal Axel Olsson for technical help with the analysis of PLFAs. The study was financed by the Swedish Council for Forestry and Agriculture Research (grant to Jan Bengtsson and Helene Lundkvist).

LITERATURE CITED

Agren, G.I., and E. Bosatta. 1996. Theoretical ecosystem ecology. Cambridge University Press, Cambridge, UK.

Anderson, J. M. 1988. Spatiotemporal effects of invertebrates on soil processes. Biology and Fertility of Soils 6: 216-227.

Anderson, T.-H., and K. H. Domsch. 1990. Application of eco-physiological quotients ([q.sub.C[O.sub.2]] and qD) on microbial biomasses from soils of different histories. Soil Biology and Biochemistry 10:251-255.

Axelsson, G., and B. Berg. 1988. Fixation of ammonia (15N) to Pinus sylvestris needle litter in different stages of decomposition. Scandinavian Journal of Forest Research 3: 273-279.

Berg, B. and G. Ekbohm. 1991. Litter mass-loss rates and decomposition patterns in some needle and leaf litter types. Long-term decomposition in a Scots pine forest. VII. Canadian Journal of Botany 69:1449-1456.

Binet, E, and P. Trehen. 1992. Experimental microcosm study of the role of Lumbricus terrestris (Oligochaeta: Lumbricidae) on nitrogen dynamics in cultivated soils. Soil Biology and Biochemistry 24:1501-1506.

Blair, J. M., R. W. Parmelee, and M. H. Beare. 1990. Decay rates, nitrogen fluxes, and decomposer communities of single- and mixed-species foliar litter. Ecology 71:1976-1985.

Bosatta, E., and G.I. Agren. 1991. Dynamics of carbon and nitrogen in the organic matter of the soil: a generic theory. American Naturalist 138:227-245.

Brookes, P. C., A. Landman, G. Pruden, and D. S. Jenkinson. 1985. Chloroform fumigation and the release of soil nitrogen: a rapid direct extraction method to measure microbial biomass nitrogen in soil. Soil Biology and Biochemistry 17:837-842.

Brown, G. G. 1995. How do earthworms affect microflora and faunal community diversity? Plant and Soil 170:209-231.

Baath, E., A. Frostegard, T. Pennanen, and H. Fritze. 1995. Microbial community structure and pH response in relation to soil organic matter quality in wood ash fertilized, clear cut or burned forest soils. Soil Biology and Biochemistry 27:229-240.

Chapman, K., J. B. Whittaker, and O. W. Heal. 1988. Metabolic and faunal activity in litters of tree mixtures compared with pure stands. Agriculture, Ecosystems and Environment 24:33-40.

Darwin, C. R. 1881. The formation of vegetable mould through the action of worms with observation on their habits. Murray, London, UK.

Dimbleby, G. W. 1952. Soil regeneration on north east Yorkshire moors. Ecology 40:331-341.

Dixon, W. J., and F. J. J. Massey. 1983. Introduction to statistical analysis. McGraw-Hill, New York, New York, USA.

Elliott, P. J., D. Knight, and J. M. Anderson. 1991. Variables controlling denitrification from earthworm casts and soil in permanent pastures. Soil Biology and Biochemistry 11:24-29.

Elliott, P. W., D. Knight, and J. M. Anderson. 1990. Denitrification in earthworm casts and soil from pastures under different fertilizer and drainage regimes. Soil Biology and Biochemistry 22:601-606.

Elliott, W. M., and N. B. Elliott. 1993. Relative effect of litter and forest type on rate of decomposition. American Midland Naturalist 129:87-95.

Federle, T. W. 1986. Microbial distribution in soil - new techniques. Slovene Society for Microbiology, Ljubljana, Slovenia.

Frostegard, A., and E. Baath. 1996. The use of phospholipid fatty acid analysis to estimate bacterial and fungal biomass in soil. Biology and Fertility of Soils 22:59-65.

Frostegard, A., E. Baath, and A. Tunlid. 1993a. Shifts in the structure of soil microbial communities in limed forests as revealed by phospholipid fatty acid analysis. Soil Biology and Biocheomistry 25:723-730.

Frostegard, A., A. Tunlid, and E. Baath. 1991. Microbial biomass measured as total lipid phosphate in soils of different organic content. Microbiological Methods 14:151-163.

Frostegard, A., A. Tunlid, and E. Baath. 1993b. Phospholipid fatty acid composition, biomass, and activity of microbial communities from two soil types experimentally exposed to different heavy metals. Applied and Environmental Microbiology 59:3605-3617.

Haimi, J., and M. Einbork. 1991. Effects of endogeic earthworms on soil processes and plant growth in coniferous forest soil. Biology and Fertility of Soils 13:6-10.

Haimi, J., and V. Huhta. 1990. Effects of earthworms on decomposition processes in a raw humus soil: a microcosm study. Biology and Fertility of Soils 10:178-183.

Hobbie, S. E. 1992. Effects of plant species on nutrient cycling. Trends in Ecology and Evolution 7:336-339.

Huhta, V. 1979. Effects of liming and deciduous litter on earthworm (Lumbricidae) populations of a spruce forest, with an inoculation experiment on Allolobophora caliginosa. Pedobiologia 19:340-345.

Joergensen, R. G. 1995. The fumigation-extraction method to estimate soil microbial biomass: calibration of the [k.sub.EC] value. Soil Biology and Biochemistry 28:25-31.

Joergensen, R. G., and T. Mueller. 1995. The fumigation-extraction method to estimate soil microbial biomass: calibration of the [k.sub.EN] value. Soil Biology and Biochemistry 28:33-37.

Maag, M., and F. P. Vinther. 1996. Nitrous oxide emission by nitrification and denitrification in different soil types and at different soil moisture contents and temperatures. Applied Soil Ecology 4:5-14.

Martikainen, P. J., M. Lehtonen, K. Lang, and W. D. Boer. 1993. Nitrification and nitrous oxide production potentials in aerobic soil samples from the soil profile of a Finnish coniferous site receiving high ammonium deposition. FEMS Microbiology Ecology 13:113-121.

Martikainen, P., and A. Palojarvi. 1990. Evaluation of the fumigation-extraction method for determination of microbial N in a range of forest soils. Soil Biology and Biochemistry 22:797-802.

McTiernan, K. B., P. Ineson, and P. A. Coward. 1997. Decomposition of tree leaf litters in mixtures: C[O.sub.2] and N release. Oikos 78:527-538.

Mead, R., R. N. Curnow, and A.M. Hasted. 1993. Statistical methods in agriculture and experimental biology. Chapman and Hall, London, UK.

Montgomery, D.C. 1991. Design and analysis of experiments. John Wiley and Sons, New York, New York, USA.

Morgan, J. L., J. M. Campbell, and D.C. Malcolm. 1992. Nitrogen relations of mixed-species stands on oligotrophic soils. Pages 65-85 in M. G. R. Cannell, D.C. Malcolm, and P. A. Robertson, editors. The ecology of mixed-species stands of trees. Blackwell, London, UK.

Nommik, H., and K. Vahtras. 1982. Retention and fixation of ammonium and ammonia in soils. Pages 123-171 in F. J. Stevenson, editor. Nitrogen in agricultural soils. American Society of Agronomy, Madison, Wisconsin, USA.

Paul, E. A., and F. A. Clark. 1989. Soil microbiology and biochemistry. Academic Press, San Diego, California, USA.

Persson, T., H. Lundkvist, A. Wiren, R. Hyvonen, and B. Wessen. 1989. Effects of acidification and liming on carbon and nitrogen mineralization and soil organisms in mor humus. Water, Air, and Soil Pollution 45:77-96.

Robinson, C. H., P. Ineson, T. G. Piearce, and A. P. Rowland. 1992a. Nitrogen mobilization by earthworms in limed peat soils under Picea sitchensis. Journal of Applied Ecology 29:226-237.

Robinson, C. H., T. G. Piearce, P. Ineson, D. A. Dickson, and C. Nys. 1992b. Earthworm populations of limed coniferous soils: field observations and implications for forest management. Forest Ecology and Management 55:117-134.

Scheu, S. 1994. Effects of earthworms on nutrient dynamics, carbon turnover and microorganisms in soils from cool temperate forests of the Canadian Rocky Mountains - laboratory studies. Applied Soil Ecology 1:113-125.

Schimel, J.P., and M. K. Firestone. 1989. Nitrogen incorporation and flow through a coniferous forest soil profile. American Journal of Soil Science 53:779-784.

Swift, M. J., O. W. Heal, and J. M. Anderson. 1979. Decomposition in terrestrial ecosystems. Blackwell, Oxford, UK.

SYSTAT. 1992. Statistics. Version 5.2 edition. SYSTAT, Incorporated, Evanston, Illinois, USA.

Trehan, S. P. 1996. Immobilization of 15[[NH.sub.4].sup.+] in three soils by chemical and biological processes. Soil Biology and Biochemistry 28:1021-1027.

Tunlid, A., and D.C. White. 1992. Biochemical analysis of biomass, community structure, nutritional status, and metabolic activity of microbial communities in soil. Dekker, New York, New York, USA.

Vance, E. D., P. C. Brookes, and D. S. Jenkinson. 1987. An extraction method for measuring soil microbial biomass-C. Soil Biology and Biochemistry 19:703-707.

Vestal, J. R., and D.C. White. 1989. Lipid analysis in microbial ecology. BioScience 39:535-541.

Williams, B. L., and C. E. Alexander. 1991. Interactions on mixing litters from beneath sitka spruce and Scots pine and the effects on microbial activity and N-mineralization. Soil Biology and Biochemistry 23:71-75.

Wolters, V., and K. Ekschmitt. 1995. Earthworm effects on the use of C sources by microorganisms: non-linear response to temperature alteration. Biology and Fertility of Soils 19:109-114.

Zhang, Q. L., and P. F. Hendrix. 1995. Earthworm (Lumbricus rubellus and Aporectodea caliginosa) effects on carbon flux in soil. American Journal of Soil Science 59:816-823.

Zheng, D., J. Bengtsson and G. Agren. 1997. Soil food webs and ecosystem processes: decomposition in donor-control and Lotka-Volterra systems. American Naturalist 149:125-148.