1 Introduction

Species of actinorhizal Alnus that fix atmospheric nitrogen (N2) through the metabolic activity of the filamentous bacterial symbiont Frankia, play an important role in the nitrogen cycle of temperate forest ecosystems (Tjepkema et al. 1986; Baker and Schwintzer 1990; Huss-Danell 1997; Dawson 2008). Because of their N2 fixation ability, these FrankiaAlnus systems have been utilized for the revegetation of deteriorated wildlife habitats and for the rehabilitation of nitrogen-deficient disturbed areas (Zitzer and Dawson 1992; Enoki et al. 1997; Sharma et al. 2002; Yamanaka et al. 2005; Son et al. 2007). In addition, in the context of responses to elevated atmospheric CO2 concentration, N2-fixing tree species have become a focus of interest as an additional N input to ecosystems (Norby 1987; Houghton et al. 2001; Hungate et al. 2003; Tobita et al. 2005; Koike et al. 2006).

Measurement of nodule biomass is essential for estimating the amount of N2-fixation at any scale. Accuracy of belowground biomass estimates is lower than that aboveground due to difficulties in sampling root system components (Akkermans and van Dijk 1976; Sharma and Ambasht 1986; Rytter 1989; Tateno et al. 2004; Hendricks et al. 2006; Coleman 2007; Helmisaari et al. 2007; Sakai et al. 2007). Several studies have attempted to estimate nodule biomass of Alnus species at the stand level in managed plantations. For young small trees, nodule biomass has been estimated from regression equations that relate tree size to nodule biomass measured in whole excavated plants (Binkley 1982; Bormann and Gordon 1984). Nodule biomass has also been estimated from the amount contained in excavations of various dimensions around trees (e.g., 0.5 × 0.5 m, 0.2 × 0.2 m), especially for large specimens in old plantations (Binkley 1981; Sharma and Ambasht 1986; Binkley 1992; Son et al. 2007). In managed plantation, nodule biomass of Alnus species varies depending on stand age, and species composition (Sharma and Ambasht 1986; Son et al. 2007), tree size and stand density (Bormann and Gordon 1984), and so on. In naturally regenerated stands, such as at floodplains or roadsides, however, the distribution of trees may not be as orderly as those of managed forests. In such naturally regenerated stands, it may be difficult to estimate nodule biomass from area-based data.

The horizontal distribution of nodules of Alnus species as well as coarse and fine roots of other tree species has not been well studied in closed forests. Rytter (1989) plotted the pattern of spatial distribution of nodules between coarse and fine roots among the stems in managed, young (3–7 years) Alnus incana plantations. Several studies have demonstrated that coarse root biomass is highest close to stems (Millikin and Bledsoe 1999) while fine root distribution is not related to distance from stems (Leuschner et al. 2001). On the other hand, Yanai et al. (2006) reported that the horizontal distribution of fine roots biomass is sensitive to the position of trees even in a 20–70 years old northern hardwood forest. Rytter (1989) also demonstrated that there was a tendency toward more homogeneous horizontal nodule distributions with increasing age, though there was a large variation in nodule density with a concentration of nodules near the stems of trees <5 year old. In naturally regenerated vegetation there will be larger variations in tree size than in managed forests, and there may be large variations in the horizontal distribution of nodules, even within even-aged stands.

Nodules of Alnus species are formed by the infection of lateral roots by Frankia (Baker and Schwintzer 1990; Huss-Danell 1997). Nodules can be several years old and grow large size (Akkermans and van Dijk 1976). Nodule size affects nitrogenase activity, because nodules have varying amounts of non-nitrogen-fixing tissue with increasing size (Sharma and Ambasht 1984; Hurd et al. 2001). If tree size influences nodule size structure, estimations of N2-fixation may need to consider nodule size structure.

Alnus hirsuta var. sibirica is widely distributed in northern districts and highlands of Japan. Stands have regenerated naturally and developed into low density populations in areas degraded by road construction in Takayama, central Japan. We investigated nodule biomass in order to estimate N2 fixation in this naturally regenerated stand. We propose that there will be large horizontal variability in nodule density per unit area depending on distance from the stems of individual trees. We also propose that there will be large variations in nodule size structure among tree sizes. To test these propositions, we collected descriptive information on the distribution and size structure of root nodules, and made quantitative estimates of nodule biomass for Alnus hirsuta var. sibirica. Though the turnover of root nodules could not be estimated in this study, a structural description at a single point in time is a necessary prerequisite for the estimation of most rate processes in natural stands.

2 Materials and methods

2.1 Study site

The study site was at approximately 1,100 m altitude on Mt. Norikura in the eastern part of Takayama City, Gifu Prefecture, central Japan (36°9′N, 137°15′E). The mean annual temperature was 7.1°C, and mean annual precipitation was 2,190 mm during the period 1990 to 1995 at 1,340 m altitude near the study site (data from Institute for Basin Ecosystem Studies, Gifu University, Gifu, Japan). A study plot of 30 m × 35 m was set up in an alder (Alnus hirsuta Turcz. var. sibirica (Fischer) C.K. Schn.) stand. The plot faced approximately southward on bearing 141°12′. The elevation difference between the highest and lowest positions within this plot was 5.2 m. Before road construction work, the area was a small valley. When a road was built through the site in 1975, the valley was buried. Alder regenerated naturally after the disturbance. In our study site, all of the canopy trees were alders. Tree height was about 15 m, and the canopy of this stand was almost closed (Hasegawa and Takeda 2001). Tree stem diameter at 1.3 m above ground-line (dbh) was measured from April to November in 1995. The mean (±SD) dbh of trees was 12.4 (±3.8) cm in April, and the frequency distribution of dbh was unimodal (Fig. 1), indicating that this stand comprised trees of similar age. Stand density was 1114 ha−1 and total basal area was 14.8 m2 ha−1 in April. Although the site floor was covered densely with herbaceous plants, regenerating specimens of several species of trees and shrubs, non-N2 fixer, were also present.

Fig. 1
figure 1

Alnus hirsuta var. sibirica: Size class distribution of dbh (diameter at breast height); of live trees in April 1995 (open columns) and trees dead before 1995 (black columns)

2.2 Spatial distribution, size structure, and biomass of nodules

The spatial pattern of live alders was analyzed using the function L(t), a transformation of Ripley’s K(t) function (Besag 1977; Ripley 1977; Diggle 1983; Nanami et al. 1999). Nodules of alder were measured on 4 occasions from June to October, 1995 to clear the seasonal variation. Five trees from 9 cm to 20 cm dbh were selected on each occasion according to the 20th, 40th, 60th, and 80th percent of dbh (Hasegawa and Takeda 2001) to represent the range of the dbh of alder in the study site. Root systems within 1 m from the outer edge of each individual stem were excavated carefully. In this study, “within 1 m” starts beyond the outer edge and somehow goes beyond 1 m from the stem axis. Root systems of other individual were identified carefully as much as possible, and their nodules were excluded. All clusters of nodules were collected within intervals of 0–20, 20–40, 40–60, 60–80, and 80–100 cm from the outer edge of stems. On the three occasions (July, September, and October) except June, nodules were collected in each compartment divided by four directions; north, south, east, and west. The diameters of subtending roots at the bases of all collected nodules (diameter of the subtending root) were measured. There was a positive correlation between nodule size or weight and diameter of the subtending root for little decayed nodules. Size of nodules was represented by the diameter of the subtending root because the degree of nodule decay of nodules varied greatly. After the clinging soil and debris were removed by hand, the nodules were washed and dried at 80°C. Nodules were counted and nodule biomass calculated for the areas from the outer edge of stems for each sampling occasion. In this report, we show nodule number data for 18 individuals on four occasions and nodule biomass data for 12 individuals on three occasions. Our estimates of the total number and biomass of nodules within 1 m from the edge of each tree stem were treated as though there were no other nodules associated with each individual alder. Nodules of an individual tree of A. hirsuta var. sibirica are reported to distribute over the crown area, and to reach >8 m from the stem for almost the maximum tree size class (Okabe 2002). In our study, the largest tree in dbh had a crown extending up to 4 m from the stem. Since we sampled within only 1 m from the outer edge of individual tree stems, calculations of nodule biomass for larger trees are likely underestimates. For larger A. hirsuta var. sibirica trees, there are few nodules below around 20 cm depth (Okabe 2002). In other studies, the root systems of Alnus incana were shallow, with more than 90% of the biomass in the uppermost 9–10 cm of the soil (Rytter 1989). In this study, all nodules occurring within the upper 20 cm of soil were collected.

2.3 Data analysis

Statistics were performed with JMP (SAS Institute 2003). Analysis of covariance (ANCOVA) was calculated to estimate the effect of tree size (dbh). In all analysis, tree size of continuous variable was used as a covariate. In the analysis of spatial distribution of nodules, nodule number per area was used as a dependent factor. Firstly, the direction and distance from the outer edge of stems were used as independent factors (categorical variables). If no differences among directions were detected, directions were combined in following analysis. Next, distance from the outer edge of stems and sampling occasions were used as independent factors (categorical variables). If no differences among dates were detected, dates were combined in further analysis. When interactions between tree size and distance from the stem were detected, correlation analysis between tree size and nodule number per area were conducted for each distance from the stem separately. In the analysis of spatial distribution of nodule biomass, nodule biomass per area was used as a dependent factor, and the same procedure was applied as nodule number per area.

In the analysis of size structure of nodules, firstly, frequency of nodule number at each size class of nodules (diameter of subtended root) within 1 m from the edge of individual tree was used as dependent factors, and nodule size (10 classes from 0–1 mm to 9–10 mm with 1 mm interval) and sampling occasions were used as independent factors (categorical variables). If no differences among dates were detected, dates were combined in further analysis. Next, nodule number per unit area was used as dependent factors, and distance from the outer edge of stems and nodule size were used as independent factors (categorical variables). If interactions between tree size and other independent factors were detected, correlation analysis between tree size and nodule number per area were conducted separately.

Nodule size structure was assessed using the two-parameter Weibull function (Weibull 1951; Bailey and Dell 1973; Tanouchi and Yamamoto 1995). The difference between the observed and a fitted distribution was examined by the Kolmogorov-Smirnov test. Correlation analysis between tree size and the Weibull scale parameter, b, and the Weibull shape parameter, c (Weibull 1951) were conducted. The dates were combined in this analysis. The scale parameter b is related to the mean of the distribution, while the shape parameter c indicates the asymmetry of the distribution. The value of the shape parameter c describes the inverse J-shape for c < 1 and the exponential distribution for c = 1. The Weibull distribution approximates a normal distribution for 3.25 < c < 3.61, and becomes increasingly positive skewed below this range and negative skewed above this range.

3 Results

3.1 Distribution pattern of Alnus hirsuta var. sibirica

In the study site, there were several suppressed alders (Fig. 1) and trees that had died before April, 1995 (Figs. 1 and 2). Live alders were significantly clumped at spatial scales of 0.5–1 m, 5.5–6 m, 8 m (P < 0.05) and 8.5–10 m (P < 0.01) (Fig. 3).

Fig. 2
figure 2

Alnus hirsuta var. sibirica: Distribution pattern of live and dead trees in April 1995. Black symbols: live trees in April 1995; open symbols: trees dead before April 1995

Fig. 3
figure 3

Alnus hirsuta var. sibirica: L(t) values for the population in April 1995. The solid line shows actual L(t) values of extant plants, dashed and dotted lines show 95% and 99% confidence envelopes derived from 1000 simulations of random point processes in the study plot, respectively. Values outside the envelopes indicate significant departures from randomness

3.2 Distribution pattern of nodule number

There was no consistent difference in nodule density among four directions outward from the stems (data not shown). The distribution pattern of nodules around the stem (calculated as density per unit area at intervals of 20 cm from the outer edge of stems) was not affected by occasion and varied with tree size, as confirmed by the tree dbh × distance interaction (P < 0.0001) (Table 1) . In small trees, nodules tended to be highly concentrated in the area nearest the stem (0–20 cm), and decreased drastically with increasing distance (Fig. 4). As tree size increased, the difference in nodule density between samples near and far from the stem decreased. Nodule number in the area nearest the stem (0–20 cm) decreased with increasing tree size (P = 0.003). In contrast, nodule number far from the stem tended to increase with tree size (60–80 cm; P = 0.0002, 80–100 cm; P = 0.054). The median points of the distribution of nodule numbers per unit area within 1 m from the outer edge of stems occurred further from the stems with increasing tree dbh (P < 0.0001) (Fig. 5). There was no significant relationship between tree size and total number of nodules within 1 m from the outer edge of stems (data not shown; P = 0.301).

Table 1 Analysis of covariance of nodule number per area (number m−2) related to the distance (0–20 cm, 20–40 cm, 40–60 cm, 60–80 cm, 80–100 cm) from the outer edge of stem at four sampling occasions (June, July, September, and October)
Fig. 4
figure 4

Alnus hirsuta var. sibirica: Spatial distribution of nodule numbers per unit area by tree size (dbh, cm) and distance from the outer edge of stem. Values are summed for each 20 cm interval within 1 m from the outer edge of stem. Dates were combined at each tree dbh classes. Tree size (dbh, cm) was classified into five classes according to the 20th, 40th, 60th, and 80th percent of dbh in the study site. The range of dbh of sample trees in each size class: Dbh 1st; 9.2–10.3 cm (n = 4), Dbh 2nd; 11.2–13.0 cm (n = 4), Dbh 3rd; 13.9–14.9 cm (n = 4), Dbh 4th; 15.6–16.0 cm (n = 3), Dbh 5th; 18.4–20.9 cm (n = 3)

Fig. 5
figure 5

Alnus hirsuta var. sibirica: Median point of distribution for nodule number within 1 m from the outer edge of stem. Nodule number per unit area calculated from the data for 20 cm intervals within 1 m from the outer edge of stems. The dates were combined in this Figure

3.3 Size class structure of nodules

The size class structure of nodules (represented by diameter of the subtending root) ranged up to a maximum of 9.0 mm (Fig. 6). About 70% of alder nodules were <3 mm. Frequency of nodule number of each nodule size class on all sampling dates varied with tree size, as confirmed by the tree dbh × nodule size class interaction (P = 0.0029) (Table 2). The smaller size tree had a lager proportion of nodules in smaller size classes. The Weibull scale parameter, b, and the Weibull shape parameter, c, of the Weibull distribution fitted to size class structure of nodules tended to increase with increasing tree dbh (Fig. 7). The shape parameter c values ranged from 1.42 to 2.47 among the rage of tree dbh from 9.2 cm to 20.9 cm, suggesting that the distribution approximated gentle bell-shaped distributions. The scale parameter b is related to the mean of the distribution and ranged from 1.85 to 3.03. These results indicate that size structure of nodules differed among tree dbh, shifting from smaller to larger size class with increasing tree dbh.

Fig. 6
figure 6

Alnus hirsuta var. sibirica: Size class structure of nodules from sampled trees. Nodule size is represented by diameter of the subtending root. Dates were combined at each tree dbh classes. Tree size (dbh, cm) was classified into five classes according to the 20th, 40th, 60th, and 80th percent of dbh in the study site. The range of dbh of sample trees in each size class: Dbh 1st; 9.2–10.3 cm (n = 4), Dbh 2nd; 11.2–13.0 cm (n = 4), Dbh 3rd; 13.9–14.9 cm (n = 4), Dbh 4th; 15.6–16.0 cm (n = 3), Dbh 5th; 18.4–20.9 cm (n = 3)

Table 2 Analysis of covariance of the frequency of nodule number related to nodule size (0–1 mm, 1–2 mm, 2–3 mm, 3–4 mm, and >4 mm) at four sampling occasions (June, July, September, and October)
Fig. 7
figure 7

Alnus hirsuta var. sibirica: The relationships between tree size (dbh, cm) and the two parameters, Weibull scale parameter b (closed circle) and Weibull shape parameter c (open circle), of the Weibull distribution fitted to size class structure of nodules. Line (for parameter b) and broken line (for parameter c) represent linear regressions displaying a statistical tendency (P < 0.1). The regression equations found are, [parameter b] = 0.044 [dbh] + 1.84 (n = 18, R2 = 0.171, P = 0.088); [parameter c] = 0.045 [dbh] + 1.36 (n = 18, R2 = 0.160, P = 0.100)

The relationships between tree size and nodule number per unit area on all sampling dates varied among nodule size classes and distances from tree stems, as confirmed by the tree dbh × nodule size class × distance interaction (P < 0.0001) (Table 3). Within the area 0–20 cm from the outer edge of stems, the nodule number of all size classes tended to decrease with increasing tree size (Table 4, uppermost row). In nodule size classes >1 mm at 20–40 cm from tree stems (Table 4, second row from top) and all size classes at positions further from tree stems (Table 4, third to fifth row from top), there was no clear relationship with tree size, or there was a tendency for nodule number to increase with tree dbh.

Table 3 Analysis of covariance of the nodule number per area (number m−2) related to nodule size (0–1 mm, 1–2 mm, 2–3 mm, 3–4 mm, and >4 mm) and the distance (0–20 cm, 20–40 cm, 40–60 cm, 60–80 cm, 80–100 cm) from the outer edge of stem
Table 4 Correlation analysis between tree dbh (cm) and nodule number per area (m−2) for each distance from stem (cm) and each nodule size (mm) separately

3.4 Distribution pattern of nodule biomass

There was no consistent difference in nodule biomass per unit area among four directions outward from the stems (data not shown). On three sampling occasions, the distribution pattern of nodule biomass per unit area at intervals of 20 cm from the stem varied with tree size (Fig. 8), as confirmed by the tree size (dbh) × distance interaction (P = 0.0005) (Table 5). The relationships between tree size and nodule biomass per unit area at each intervals of 20 cm from the outer edge of stem on all sampling dates showed a pattern similar to those of nodule number; nodule biomass in the area nearest the stem (0–20 cm) tended to decrease with increasing tree size (P = 0.096). In contrast, nodule biomass far from the stem increased with tree size (20–100 cm; P < 0.05). The median points of the distribution of nodule biomass per unit area within 1 m form the edge of stems occurred further from the stems with increasing tree dbh (Fig. 9a). Unlike nodule number, nodule biomass within 1 m from the edge of individual stems increased with tree size (Fig. 9b; [nodule biomass] = 0.442 [dbh]2.01, R 2 = 0.747, P < 0.001). Nodule biomass per area within 1 m from the edge of individual stems also showed significant relationships with tree size (data not shown; [nodule biomass per area] = 0.168 [dbh]1.90, R 2 = 0.723, P < 0.001). By using the relationship in Fig. 9b, the average nodule biomass per tree within 1 m from the outer edge of stems in April 1995 was estimated to be 77.5 g (± 3.8) (5.6 to 199.2 g). When nodule biomass was calculated for the area of this study plot, nodule biomass in April 1995 was estimated to be 84.1 kg ha−1.

Fig. 8
figure 8

Alnus hirsuta var. sibirica: Spatial distribution of nodule biomass per unit area by tree size (dbh, cm) and distance from the outer edge of stem. Values are summed for each 20 cm interval within 1 m from the outer edge of stem. Dates were combined at each tree dbh classes. Tree size (dbh, cm) was classified into five classes according to the 20th, 40th, 60th, and 80th percent of dbh in the study site. The range of dbh of sample trees in each size class: Dbh 1st; 9.2–10.3 cm (n = 3), Dbh 2nd; 12.2–13.0 cm (n = 3), Dbh 3rd; 13.9–14.9 cm (n = 2), Dbh 4th; 15.6–16.0 cm (n = 2), Dbh 5th; 18.4–19.4 cm (n = 2)

Table 5 Analysis of covariance of nodule biomass per area (g m−2) related to the distance (0–20 cm, 20–40 cm, 40–60 cm, 60–80 cm, 80–100 cm) from the outer edge of stem at three sampling occasions (June, September, and October)
Fig. 9
figure 9

Alnus hirsuta var. sibirica: Median point of distribution for nodule biomass (a), and total nodule biomass (b) within 1 m from the outer edge of stem. Nodule biomass per unit area (a) and per tree (b) calculated from the data for 20 cm intervals within 1 m from the outer edge of stems. The dates were combined in these Figures. Line shows fitted least-squares regression (b): [nodule biomass] = 0.442 [dbh]2.01, (n = 12, R2 = 0.747)

4 Discussion

In this naturally regenerated Alnus hirsuta var. sibirica even-aged stand (17–18 years old), horizontal distribution of nodules around each tree varied greatly among tree sizes (Figs. 4, 5, 8, and 9a). In particular, for trees with smaller dbh, including suppressed trees, there was a large variation in nodule density, and a concentration of nodule density near the stem. In contrast, there was less variation in nodule distribution with distance from the stem for bigger trees (mean dbh = 12.4 cm) (Fig. 4). In managed young Alnus incana plantations, similar changes in nodule horizontal distribution variation were observed among different stand ages. Although, in young stands <5 year old, there is a concentration of nodules near the stem, there is tendency toward more homogeneous horizontal nodule distributions with increasing age (Rytter 1989). Unlike managed plantations, the clumped distribution pattern of alders (Figs. 2 and 3) and the large variation in tree size (as dbh, Fig. 1) may cause that the large variation in nodule spatial distribution we observed in this even-aged naturally regenerated stand. In this study, nodules of other individuals of alder appearing within 1m from the outer edge of target trees were excluded as much as possible, but this was not the case in Rytter’s (1989) work. It unclear whether the distribution of nodules would become more homogeneous around stem was we to include the nodules of trees other than the targets. Yanai et al. (2006) reported that fine root biomass as well as coarse root biomass was still greatest close to tree stems even in 20–70 years old northern hardwood forest, and they proposed that root measurements should be made with attention to patterns of tree distribution. Nodules of Alnus species have a horizontal distribution intermediate between those of coarse and fine roots. Fine root distribution is sometimes dependent on the positions of individual trees, hence nodule distribution may be also affected by tree distribution (as we have shown here). Accordingly we suggest that patterns of tree distribution and tree size should also be taken into account in measurements of root nodules in naturally regenerated stand.

Information on root diameter distribution is a fundamental to understanding root systems and soil functioning, although current data are limited (Blouin et al. 2007; Zobel et al. 2007). In this study, nodule size was represented using the diameter of the subtending root, because nodules of Alnus species are formed by the infection of lateral roots by Frankia and can be several years old with partly decay (Baker and Schwintzer 1990; Huss-Danell 1997). We found that nodule size class structure also showed large variations among tree sizes in this naturally regenerated stand (Figs. 6 and 7). Size structure of nodules shifted from smaller to larger size classes with increasing tree dbh. Since many studies have demonstrated that age of nodules affects nitrogenase activity (Sharma and Ambasht 1984; Hurd et al. 2001), we proposed that variations in size structure of nodules among tree sizes be taken into account when estimating N2-fixation rates in a naturally regenerated stand.

Nodule density in each size class varied depending on the distance from the stems and tree size (Table 3). Though the nodule density in the larger size class (>2 mm) tended to increase with increasing tree size, nodule density at the nearest site from stems (0–20 cm) was lower in larger trees than smaller trees for all size class of nodules (Table 4). There are two possible reasons for these results. The first is nodule senescence. Partial decay of larger nodules exposed at the soil surface occurred nearest the stem (a result of root enlargement in large trees, unpublished observation), suggesting high rates of nodule senescence led to relatively low nodule density closest to the tree trunks. The second possibility is that there is inhibition of nodule formation by increasing nitrogen (N) availability in the soil beneath large trees. In a pure Alnus hirsuta stand, there were at least two fold increases in total inorganic nitrogen concentrations in the presence of the alders (Son et al. 2007). Alnus crispa also increased inorganic soil N production in a northern boreal forest (Rhoades et al. 2001). High soil N concentration limits nodule formation in Alnus rubra stands (Martin et al. 2003). In this study, however, we could not detect significant relationships between tree size and inorganic N availability in the soil measured on two sampling occasions (data not shown).

Nodule biomass of Alnus species varies with stand age, stand density, and tree size (Bormann and Gordon 1984; Sharma and Ambasht 1986). In this study, nodule biomass within 1 m from the outer edge of individual stems increased with tree size (Fig. 9b), and nodule biomass per area was estimated to be 84.1 kg ha−1 in April 1995. These values are within the range of 30 to 480 kg ha−1 reported for several Alnus species (Zavitokovski and Newton 1972; Binkley 1981, 1982; Bormann and Gordon 1984; Sharma and Ambasht 1986; Rytter 1989; Binkley 1992; Lee and Son 2005; Son et al. 2007), including pure Alnus hirsuta plantation in Korea (estimated to be 179.3 kg ha−1 for a 27-year-old stand [Son et al. 2007] and 220 kg ha−1 for a 38-year-old stand [Lee and Son 2005] using soil pit sampling methods).

Allometric relationships between tree dbh and leaf mass per tree occur in stands of several Alnus species (Binkley 1982; Bormann and Gordon 1984; Tadaki et al. 1987), suggesting that trees with larger leaf mass may form larger numbers of root nodules in the alder stand we studied. The leaf litter biomass in this study site was 3.07 t ha−1 in 1995, values which are not larger than those of other stands (Sharma and Ambasht 1987; Rytter 1989). The ratio of nodule biomass to leaf litter biomass in the alder stand we studied was 2.7% in 1995, values smaller than the previously reported range of 5.6% to 9.8% (Akkermans and van Dijk 1976; Tripp et al. 1979; Binkley 1981; Bormann and Gordon 1984). Accordingly, we may have likely made underestimates of nodule biomass in this alder stand through some margin of procedural error.

We calculated nodule biomass in the Alnus hirsuta var. sibirica stand as total nodule biomass per plot area. Total nodule biomass of individual trees was estimated using the regression of nodule biomass on tree size (Fig. 9b). When the nodule biomass calculation is based on stand density, there is likely an underestimation because the total area within 1 m from the outer edge of stems covers only about one-third of study plot. Another reason is that tree root systems will extend beyond 1 m from the outer edge of stems, especially in large trees. Nodules of Alnus hirsuta var. sibirica with 46 cm dbh, which is almost maximum size of this species, distribute over the crown area, and reach >8 m from the stem (Okabe 2002). In our study, the largest tree in dbh (22 cm dbh) had a crown extending up to 4 m from the stem. Since we sampled within only 1 m from the outer edge of tree stems, calculations of nodule biomass per individual for larger trees would be underestimates. Though root systems of Alnus hirsuta var. sibirica develop horizontally to only a limited extent and tend to concentrate to around the trunks (Karizumi 1979), investigations of nodules within the crown area may be needed for more precise estimates.

We conclude that horizontal distribution of nodules around each tree of A. hirsuta var. sibirica in naturally regenerated stand varied greatly by tree size even though the stand was even-aged. There were also large variations in nodule size class structure around each tree among tree sizes. Though these results are a structural description at a single point in time, nodule biomass could be estimated by taking into consideration tree size and position of individual tree in this naturally regenerated stand, and may be utilized for evaluation of the N2 fixation rate.