175
Factors influencing early vegetation establishment following soil scarification in a mixed forest in northern Japan Toshiya Yoshida, Yoko Iga, Megumi Ozawa, Mahoko Noguchi, and Hideaki Shibata
Abstract: Scarification is widely conducted in northern Japan to remove understory dwarf bamboo species in degraded forests for replacement with tree species. To explore ways to enhance species diversity and restoration of mixed forest at the treated site, we clarified the mechanisms that lead to compositional heterogeneity of plant species. We evaluated the relative importance of environmental factors (scarification properties, soil properties, light conditions, litter cover, and presence of canopy trees) for the demography of tall tree species (emergence, mortality, and growth) and whole vegetation structure (species diversity and composition) over the two growing seasons immediately following scarification. Of tall tree species, Betula spp. were dominant (60% in total density), followed by Abies sachalinensis (Fr. Schm.) Masters, Acer mono Maxim., and Phellodendron amurense Rupr. Light intensity was an important factor, having mostly negative effects on the demography of these species. Soil factors (e.g., nitrogen content, moisture) affected the demography mainly of shade-intolerant or hygrophilous species. In general, extreme environmental conditions led to the dominance of grasses, forbs, and lianas rather than tall trees. Maintenance of canopy cover, which limits light and supplies seeds as well as litter, proved to be most important in promoting plant species diversification on the scarification site. Résumé : Dans les forêts dégradées du nord du Japon, le scarifiage est souvent utilisé pour enlever un sous-étage composé d’une espèce de bambou nain et la remplacer par des espèces arborescentes. Pour explorer les façons d’améliorer la diversité des espèces et la restauration d’une forêt mixte sur les stations traitées, les auteurs ont clarifié les mécanismes qui mènent à une composition hétérogène en espèces végétales. Ils ont évalué l’importance relative de facteurs environnementaux (propriétés du scarifiage, propriétés du sol, conditions lumineuses, couverture de litière et présence d’un couvert forestier) sur les caractéristiques démographiques d’espèces d’arbre de grande taille (émergence, mortalité et croissance) et de toute la structure de la végétation (diversité des espèces et composition) au cours des deux saisons de croissance qui ont suivi le scarifiage. Parmi les espèces d’arbre de grande taille, les Betula spp. étaient l’espèces dominantes (60 % de la densité totale) suivie de l’Abies sachalinensis (Fr. Schm.) Masters, l’Acer mono Maxim. et le Phellodendron amurense Rupr. L’intensité lumineuse a constitué un facteur important en affectant surtout négativement les caractéristiques démographiques de ces espèces. Les propriétés du sol (p. ex. contenu en azote, humidité) ont principalement affecté les caractéristiques démographiques des espèces intolérantes à l’ombre ou des espèces hygrophiles. En général, des conditions environnementales extrêmes mènent à la dominance de graminées, de plantes herbacées et de lianes au détriment des arbres de haute taille. Le maintien d’un couvert, qui diminue l’intensité lumineuse et qui fournit des semences et de la litière, s’est avéré être le facteur le plus important pour favoriser la diversité végétale sur les stations scarifiées. [Traduit par la Rédaction]
Yoshida et al.
188
Introduction Soil scarification is surface soil displacement with civil engineering machinery (such as a bulldozer) for silvicultural site preparation. In general, it aims to improve the substrate (usually mineral soil) and remove understory competitors for successful regeneration. Scarification greatly enhances the germination and survival of tree species relative to control areas
or treatments without scarification (Mäkitalo 1999; Karlsson and Orlander 2000; Wurtz and Zasada 2001; Béland et al. 2003), suggesting that it is an effective practice for forest regeneration and restoration of degraded forests. Scarification is conducted widely in northern Japan to get rid of understory dwarf bamboo species before replacement with tree species. Dwarf bamboo (Sasa kurilensis (Rupr.) Makino & Shibata and Sasa senanensis (Franch. & Savat.)
Received 6 February 2004. Accepted 13 September 2004. Published on the NRC Research Press Web site at http://cjfr.nrc.ca on 12 February 2005. T. Yoshida.1 Uryu Experimental Forest, Field Science Center for Northern Biosphere, Hokkaido University, Moshiri, Horokanai 074-0741, Japan. Y. Iga, M. Ozawa, and M. Noguchi. Graduate School of Agriculture, Hokkaido University, Nayoro 096-0071, Japan. H. Shibata. Northern Forestry and Development Office, Field Science Center for Northern Biosphere, Hokkaido University, Nayoro 096-0071, Japan. 1
Corresponding author (e-mail:
[email protected]).
Can. J. For. Res. 35: 175–188 (2005)
doi: 10.1139/X04-156
© 2005 NRC Canada
176
Rehd.) frequently occurs, and it produces dense undergrowth in forests, especially at open sites (Toyooka et al. 1981; Noguchi and Yoshida 2005). This region includes more than 1100 ha of unstocked forest land that is mostly covered with dwarf bamboo (Hokkaido Prefecture 2003), possibly from past windthrow events or partial cuttings. The dominance of dwarf bamboo causes a marked decline in seedling and sapling density (Nakashizuka 1998; Noguchi and Yoshida 2004), and managers have worked for years to figure out a way to get rid of dwarf bamboo species to promote tree regeneration. Aboveground cutting was found to be inadequate, because vigorous propagation can resume from rootstocks. Therefore, from the 1970s, mechanical site preparation (scarification) has been used to remove the whole body of dwarf bamboo and enhance the natural regeneration of tree species. Several studies have examined the consequences of such treatments. In the early successional stage following scarification, species composition is diverse in and among sites (Matsuda and Takikawa 1985; Hayashida et al. 1991; Sato 1998, 1999). However, in the advanced stage of development (older than 20 years), treated sites are likely to be overwhelmingly dominated by shade-intolerant species (e.g., Betula spp.: Matsuda and Takikawa 1985; Wurtz and Zasada 2001; Umeki 2003). In view of the attention given to complex composition and structure in current forestry (Kohm and Franklin 1997), a variety of plant species should be established nowadays. It is therefore important to determine the factors that control the patterns and dynamics of vegetation recovery at scarification sites. Previous studies have indicated the importance of light conditions (Kurahashi et al. 1999) and seed supply (Sato and Tsukada 1996; Sato 1998), but a comprehensive analysis of Japanese forests, taking multiple factors into account, has not so far been undertaken. In particular, the effects of soil properties should be considered (Fleming et al. 1994; Mäkitalo 1999; de Chantal et al. 2003; Frey et al. 2003). The present study aims to clarify the detailed mechanisms by which compositional heterogeneity of plant species arises following scarification in a natural mixed forest in northern Japan. We use multiple regression and canonical correspondence analysis to evaluate the relative importance of environmental factors in determining the demography (emergence, mortality, and growth) of major tree species and the whole vegetation structure (species diversity and composition) in the two growing seasons immediately following scarification. The explanatory variables examined were scarification properties, soil chemical and physical properties, light conditions, litter cover, and existence of canopy (retained trees). We then examine the implications for improving the current scarification practice so as to enhance species diversity and restore a mixed forest on the site.
Materials and methods Study site The study was conducted in the Uryu Experimental Forest, Hokkaido University, located on Hokkaido Island in northern Japan (44°24′N, 142°07′E; elevation, 370 m above sea level). The mean annual air temperature is 3.1 °C, and the mean annual precipitation is 1390 mm (1969–1998; roughly half the precipitation fell in June–September). Snow cover
Can. J. For. Res. Vol. 35, 2005 Table 1. Composition of overstory trees with diameter at breast height ≥10 cm in the study plot. Sum of basal area
Density
Species
m2/ha
Quercus crispula Abies sachalinensis Fraxinus mandshurica Betula ermanii Betula platyphylla Ulmus davidiana Phellodendron amurense Acer mono Alnus japonica Sorbus commixta Others Total
13.1 8.0 5.1 3.8 2.1 1.4 1.1 1 0.4 0.1 0.4 36.6
Stems/ha 28 39 25.5 32.5 23.5 2.5 13.5 12 10 6 5.5 198
% 35.9 21.7 14.1 10.4 5.8 3.8 2.9 2.7 1.1 0.4 1.2 100
% 14.1 20 12.9 16.4 11.9 1.3 6.8 6.1 5.1 3 2.8 100
usually extends from late November to early May, with a maximum depth of about 3 m. The predominant soil is an Inceptisol (acidic brown forest soil). The predominant bedrock is Tertiary andesite. In September 2000, a 1.96-ha (140 m × 140 m) study site was established in a conifer–hardwood mixed forest on a gentle (5%) slope. We measured the tree diameter at breast height (DBH) and drew a crown projection diagram for all overstory trees with DBH ≥ 10 cm. The forest consisted of Quercus crispula Blume, Abies sachalinensis (Fr. Schm.) Masters, Fraxinus mandshurica Rupr. var. japonica Maxim., Betula ermanii Cham., and several other hardwood species, with a total density of 198 stems/ha and a total basal area of 36.6 m2/ha (Table 1). The low density, which is normal in regions of heavy snowfall in northern Japan, may be the result of a few severe windthrow events or past partial cutting (Noguchi and Yoshida 2004). The understory consisted mainly of dwarf bamboo species (S. senanensis). A preliminary vegetation survey (twenty-four 1 m × 1 m quadrats randomly located in the area) confirmed the overwhelming dominance of this species; the mean height was 1.6 m; the mean density was 36.6 stems/m2; and the mean aboveground dry mass was 681 g/m2 (96% of the total biomass). The mean density of seedlings and saplings of tall tree species (defined as species found in the canopy layer) was 2.7 stems/m2, most of which were F. mandshurica younger than 5 years old. Just after the preliminary survey (September 2000), the forest floor was scarified with a bulldozer (D60P11; Komatsu, Ltd., Tokyo, Japan). An attached rake was inserted into the soil surface, and most of the surface soil with understory vegetation (including many of the small trees) was pushed off the whole experimental area. All overstory trees (with DBH ≥ 10 cm) were retained, and scarification was omitted in the area around them (2–4 m in radius, depending on size and distribution of the stems). Data collection The nested 1.0-ha (100 m × 100 m) area was divided into 10 m × 10 m squares (Fig. 1); 60 intersection points, randomly selected from the scarified points, were used in subsequent investigations. We examined patterns and dynamics of vegetation (demography of tall tree species and species © 2005 NRC Canada
Yoshida et al. Fig. 1. Arrangement of the 60 study quadrats (filled squares) in the 1.96-ha plot. The 1-m2 quadrats were randomly selected from the intersection points at 10-m spacing in the nested 1-ha area. Dots and circles indicate remaining trees in the overstory (diameter at breast height ≥ 10 cm) and their projected crown area (the latter are displayed only for trees influencing the nested 1-ha area). The area around the trees (generally within radius of 2–4 m) was not scarified.
177
where Pi indicates the relative dominance of each species (ratio of the percent cover to the total). Measurements of environmental factors Scarification properties A bulldozer with a rake was used for scarification, so the route of travel gave rise to considerable spatial variation in scarification properties. We therefore calculated several different indices to specify each quadrat. First, using a surveyor’s level, we measured ground height (i.e., soil surface) just before and after scarification. We subtracted the postscarification height from the original height, assuming that the difference corresponded to the scarification depth. Second, we calculated an index for bumpiness (which was also created by machinery trails) by measuring the topmost and bottommost ground height in each quarter of the quadrat. The index was calculated according to the following formula: [2]
diversity and composition) during the first two growing seasons following scarification, from spring 2001 to autumn 2002. Measurements of seedling and vegetation In a 1 m × 1 m quadrat, centered on the 60 intersection points, all tall tree seedlings that emerged were marked, and their survival was assessed. Assessments were made in early July, mid-August, and mid-September in 2001 (first growing season) and in late June, mid-August, and mid-September in 2002 (second growing season). The first yearly census was conducted immediately after leaf flush; and the last yearly census, immediately before leaf fall. Below, we define “growing season” as the period between the first yearly census and the last yearly census. We calculated demographic parameters for major tall tree species. “Emergence” was defined as the number of newly confirmed seedlings in a year. “Mortality” was defined as the proportion of dead seedlings to total confirmed seedlings. Mortality during the nongrowing season was considered, as well as that during the growing season. “Growth” was represented as the mean seedling height, measured in the last yearly census in each quadrat. All the vascular plants in the quadrat were recorded by species, stem density, and percent cover in mid-August in the first and second growing seasons. We could not identify Betula, Salix, Carex, or Viola seedlings to the species level at this stage. The cover of all species was summed for each quadrat to give the total vegetation. We also calculated the species richness (number of species) and the Shannon index (H′) for each quadrat: [1]
H′ = ΣPi ln Pi
Bumpiness =
ΣDs Dp
where Ds and Dp, respectively, denote the maximum difference in height in each quarter and in the total 1-m2 quadrat. We also measured ground hardness (kilograms per square centimetre) with a soil hardness tester (No. 351; Fujiwara Ltd., Tokyo, Japan) to assess the degree of soil compaction caused by the machinery; five random repeated measurements were averaged. Finally, we counted the number of stems (including dead ones) of dwarf bamboo in the quadrat in August of the first growing season. More stems were expected to remain in sites with weaker scarification. Soil sampling and analyses Soil sampling was conducted in a 1-m2 area adjacent to the quadrat. The choice of sample locations depended on whether they had vegetation similar to that in the quadrat. We collected surface soil (to 15-cm depth), using an auger, in late July the first growing season and in early August the second growing season. Two samples were collected from the 1-m2 area at each sampling date and mixed prior to analysis. Dissolved inorganic nitrogen (sum of NO3-N and NH4-N contents in soil), which is potentially available to vegetation, was used as a parameter representing soil fertility in the present study. For soils extracted by water (soil:water = 1:5), we determined the NO3 concentrations in the extracted water by ion chromatography (DX-500; Dionex Corp., Sunnyvale, California, USA). With soils extracted by KCl (soil:KCl = 1:5), we determined NH4 concentrations in the extracted solution by an indophenol blue method using a spectrophotometer (U-1500; Hitachi, Ltd., Tokyo, Japan). We measured soil pH and electrical conductivity (EC) (microsiemens per centimetre) using a pH meter (B-212; Horiba, Ltd., Kyoto, Japan) and an EC meter (B-173; Horiba, Ltd., Kyoto, Japan), respectively. Ground temperature and moisture were measured directly in the field at the time of soil collection. Ground temperature (degrees Centigrade, 0–5 cm below surface) was measured with a thermometer (CT-07; CUSTOM, Tokyo, Japan). Ground moisture at the surface soil layer (percentage, 0–11 cm below surface) was measured by time-domain reflectometry (TRIMEComo; Tohoku EIC Corp., Sendai, Japan). For both parame© 2005 NRC Canada
178
Can. J. For. Res. Vol. 35, 2005 Table 2. Pearson correlations for the principal components of (A) scarification properties, (B) soil properties, and (C) light conditions for each component variable measured in the first and second growing seasons. (A) Scarified properties. First year Scarification depth Bumpiness in ground height No. of dwarf bamboo stems Ground hardness
Second year
SCAR1 (40.6%) 0.765* 0.365* –0.738* 0.600*
SCAR2 (26.9%) 0.253 0.853* 0.441* –0.300*
SCAR1 (42.7%) 0.652* 0.254 –0.844* 0.711*
SCAR2 (28.8%) 0.487* 0.827* 0.299* –0.385*
First year SOIL1 (35.4%)
SOIL2 (20.6%)
SOIL1(38.5%)
SOIL2 (25.2%)
0.844* –0.546* 0.587* –0.222 0.722*
–0.167 0.629* –0.639* 0.517* 0.398*
(B) Soil properties.
NO3-N + NH4-N pH EC Ground temperature Ground moisture
Second year
0.825* –0.093 0.857* –0.356* 0.471*
0.246 0.911* 0.001 0.371* 0.026
(C) Light conditions.c
Openness rPPFD
First year
Second year
LIGHT (85.9%) 0.927* 0.927*
LIGHT (86.8%) 0.927* 0.927*
Note: *, p < 0.05. Percent variance explained by principal component analysis axis is shown in parentheses. A description of the variables is given in Table 3. EC, electrical conductivity; rPPFD, photosynthetically active photon flux value relative to that of the top canopy. SCAR 1 and SCAR 2, axes 1 and 2 scarification properties, respectively; SOIL1 and SOIL2, axes 1 and 2 soil properties, respectively; LIGHT, light axis.
ters, five random repeated measurements in the quadrat were averaged. Light conditions Hemispherical photographs were taken at a height of 1.5 m in each quadrat in mid-August in the first and second growing seasons. A digital camera with a fisheye lens was used. The photographs were read into WINPHOT 5 (ter Steege 1996) to calculate the openness and the photosynthetically active photon flux density (PPFD). The estimated PPFD values were converted into values relative to those above the top canopy during the growing season (rPPFD; June–September). Litter cover We measured litter cover in the vegetation quadrat in midAugust in the first growing season and in late July in the second growing season. The percent litter cover for each quarter of the quadrat (0.25 m2) was estimated by observation, and the mean was used as a standard. Snow depth We measured snow depth in each quadrat so as to include a factor representing the nongrowing season. Measurement was conducted in early spring (early May in 2002; snowmelt is usually complete by late May or early June). We presume that midwinter environmental conditions on the forest floor are fairly homogeneous among the quadrats, as the site is completely covered by snow to a maximum depth of 2–3 m. It is the timing of snowmelt that seems to be significant for seedling demography. We believe that a shallow depth of
snow corresponds to early snowmelt in a quadrat. Early snowmelt may increase the risk of frost damage, but it also gives a longer growing season. The timing also greatly influences soil temperature and moisture in the early spring season. Statistical analyses Definition of parameters We used regression analysis and canonical correspondence analysis (CCA) to explain the patterns and dynamics of vegetation in the two growing seasons. The explanatory variables represented scarification properties, soil properties, light conditions, litter cover, and existence of canopy trees. Some of these factors were categorized with principal component analysis (see Table 2) to simplify the models. Results for the 2 years were similar. Two axes were found for factors related to scarification. Axis 1 (named SCAR1) showed a positive correlation with scarification depth and with ground hardness and a negative correlation with the number of remaining dwarf bamboo stems, suggesting that this axis related to scarification intensity. Axis 2 (SCAR2) had a positive correlation with the bumpiness of the quadrat. Two axes were also found for soil properties. Axis 1 (SOIL1) had a positive correlation with nitrogen content (NO3-N + NH4-N; soil fertility), EC, and ground moisture. Axis 2 (SOIL2) had a positive correlation with pH and ground temperature. Factors representing light conditions were categorized on an axis (LIGHT) that had a positive correlation with both openness and rPPFD. © 2005 NRC Canada
Yoshida et al.
179
Overstory conditions were used to evaluate the effects of seed supply or of the canopy itself. We used a binomial variable (CANOP), which describes the existence of a canopy of trees with DBH ≥ 10 cm above the quadrat. A supplementary experiment with seed traps, which were established beside 30 quadrats in the first growing season, showed that the seed supply of major tall tree species (Betula spp., A. sachalinensis, Q. crispula, and F. mandshurica) was positively correlated with the sum of the basal area of conspecific mature trees (r = –0.45 to –0.34, p < 0.05). Instead of using the direct measure of seed supply, we used the sum of basal area within a 20-m radius of the quadrat center (TOTBA: mature trees were defined as trees with DBH ≥ 20 cm: Koike et al. 1992). LITTER indicated the litter cover, and SNOW indicated the snow depth in early spring.
[7]
Species diversity = a0 + a1(SCAR1) + a2(SCAR2) + a3(SOIL1) + a4(SOIL2) + a5(LIGHT) + a6(LITTER) + a7(CANOP)
where CANOP accounts for the canopy’s function of seed supply. The forward stepwise method was used in the multiple regression analyses. The values of the F statistics to enter and remove an independent variable were both set at 1.5. The variance inflation factors for the models showed relatively low values, indicating the absence of multicollinearity. All analyses were carried out with statistical software (SPSS 10.0; SPSS Japan Inc. 1999).
Multiple regression analyses Multiple linear regression analysis was used to explain the demography of major tall tree species (emergence, mortality, and growth) and species diversity of total vegetation. We assumed regression models with the variables described above. For the demographic models, only species that emerged in more than 45 quadrats (75%) were analyzed. Seven variables were fitted to a formula for seedling emergence, according to
Canonical correspondence analysis The software PC-ORD 10.0 (McCune and Mefford 1999) was used to perform a CCA to reveal differences in species response over sites. Second-year vegetation data (stem number) were used in this analysis. The following six variables were employed: principal components of the scarification intensity (SCAR1, SCAR2), soil properties (SOIL1, SOIL2), light condition (LIGHT), and litter cover (LITTER). A Monte Carlo test with 99 permutations was used for testing the significance of the canonical axis.
[3]
Results
Seedling emergence = a0 + a1(SCAR1) + a2(SCAR2) + a3(SOIL1) + a4(SOIL2) + a5(LIGHT) + a6(LITTER) + a7(TOTBA)
where a0–a7 are regression coefficients. For seedling mortality in a growing season, the regression formula is similar to eq. [3]: [4]
Seedling mortality (growing season) = a0 + a1(SCAR1) + a2(SCAR2) + a3(SOIL1) + a4(SOIL2) + a5(LIGHT) + a6(LITTER) + a7(CANOP)
where CANOP is able to account for the hypothesis that high mortality is shown under conspecific adults (Janzen 1970). We used a different model for mortality in the nongrowing season. We calculated simple correlation coefficients for the snow depth at early spring (SNOW) to find the effect of snowmelt timing on mortality: [5]
Seedling mortality (nongrowing season) = a0 + a1(SNOW)
For regression of seedling growth (height) we used six variables, excluding that related to the existence of canopy trees: [6]
Seedling growth = a0 + a1(SCAR1) + a2(SCAR2) + a3(SOIL1) + a4(SOIL2) + a5(LIGHT) + a6(LITTER)
For the mortality and growth models, quadrats with no seedlings of the relevant species were not used. Finally, for species diversity, seven variables were again used:
Environmental variables Table 3 summarizes the environmental variables measured in the present study. The mean scarification depth was 9.4 cm (with a range of –16.5 to 26.5 cm), corresponding approximately to the normal depth of dwarf bamboo roots. Because the scarified soil was mostly pushed off the entire experimental area, only a few quadrats showed an increase in ground height (i.e., accumulation of scarified soil) after the treatment. The hardness of the surface soil was on average 6.6 kg/cm2 in the first growing season and was lower (2.7 kg/cm2 on average) in the second growing season. The bumpiness in the ground height displayed fairly low variation over the quadrats. Changes in soil properties during the two growing seasons seemed to be strongly affected by weather conditions on the sampling day and beforehand, although we carefully selected the sampling days to ensure similar conditions. In fact, both the daily average atmospheric temperature and the total precipitation in the 5 days prior to sampling were higher in the first growing season than in the second (temperature, 20.2 vs. 14.8 °C; precipitation, 5.7 vs. 3.6 mm). Average ground temperatures corresponded to the atmospheric temperature (23.4 °C in the first growing season and 18.2 °C in the second growing season). The ground moisture showed a pattern contrary to that of ground temperature (38.5% in the first growing season and 53.4% in the second growing season, with fairly high variation), probably as a result of higher atmospheric temperature in the first growing season. The average nitrogen content (NO3-N + NH4-N) also showed wide variation over the quadrats (19.5 and 15.6 mg N/kg dry soil on average in the first and second growing seasons, respectively). The EC also displayed fairly high variation (24.4 and 33.2 µS/cm on average in the first and second growing © 2005 NRC Canada
180
Can. J. For. Res. Vol. 35, 2005 Table 3. Summary of environmental variables measured in this study.
Group
Variable
Units
Scarification properties
Scarification depth Bumpiness in ground height No. of dwarf bamboo stems Ground hardness N03-N + NH4-N pH EC Ground temperature Ground moisture Canopy openness rPPFD Litter cover
cm — stems/m2 kg/cm 2 mg N/kg dry soil — µS/cm °C % % % %
Soil properties
Light conditions Litter accumulation
First year
Second year
Mean 9.4 2.2 8.2 6.6 19.5 5.0 24.4 23.4 38.5 38.2 64.2 10.2
Mean
SD
— — — 2.7 15.6 5.0 33.2 18.2 53.4 37.3 59.6 27.5
— — — 1.2 16.1 0.2 7.7 1.2 16.8 7.4 19.0 20.1
SD 8.9 0.5 8.1 5.9 14.6 0.2 10.0 2.5 8.8 7.6 14.8 20.6
Note: Values of the variables for 60 quadrats are shown. Scarification depth, bumpiness in ground height, and number of dwarf bamboo stems were measured only in the first growing season. EC, electrical conductivity; rPPFD, photosynthetically active photon flux value relative to that of the top canopy; SD, standard deviation.
Table 4. Mean vegetation cover of the 60 quadrats in the summer of the first and second growing seasons.
Tall trees Shrubs Lianas Dwarf bamboo Grasses Forbs All
First year (%)
Second year (%)
Mean
Mean 9.9 1.2 0.3 2 1.8 3.3 18.9
2.8 0.3 0.2 1 0 1.3 5.6
SD 3.4 1.4 1.4 3.8 0 5.1 11.4
Fig. 2. Mean densities of tall tree seedlings per 1-m2 quadrat at the end of the first and second growing seasons.
SD 10.1 4.8 0.9 5.4 4.2 7.6 24.1
Note: SD, standard deviation.
seasons, respectively); in contrast, pH showed the lowest variation (5.0, on average, in both growing seasons). The light intensity was quite high at nearly 40% for the canopy openness and 60% of the relative illuminance (rPPFD) on average (Table 3). The mean percent cover of leaf litter increased from nearly 10.2% in the first growing season to 27.5% in the second growing season. Vegetation development Development of vegetation in the two growing seasons is summarized in Table 4. The mean vegetation cover increased from 5.6% in the first growing season to 18.9% in the second growing season, and there was considerable variation between the quadrats (0.2%–141.6%). In the first growing season, tall tree species had the highest mean cover, followed by tall forbs and dwarf bamboo, which regrew from the residual rootstocks. In the second growing season, recruited tall tree species considerably increased their cover (especially in Betula spp.: 3.5%, on average); five out of the major seven species (those having percent cover exceeding 1%) were tall tree species. In all, there were 47 species in the first growing season and 54 species in the second growing season. The average species richness (the number of species observed in a quadrat) was 7.5 in the first growing season, with a slight increase to
9.5 in the second growing season. Species diversity indexes (H′) also showed an increase during the two growing seasons, from 1.13 to 1.27. Demography of dominant tall tree species Figure 2 shows the number of established seedlings (surviving at the latest census) in the two growing seasons. The average density was nearly 50 individuals/m2 in the first growing season and 75 individuals/m2 in the second growing season; the ranges of the values were wide over the quadrats (0–214 and 3–283 individuals/m2 in the first and second growing seasons, respectively). The most dominant species were B. ermanii and Betula platyphylla Sukatchev var. japonica (Miq.) Hara, followed by A. sachalinensis, Acer mono © 2005 NRC Canada
Yoshida et al.
181
Fig. 3. Changes in mean density, number emerging, and number of dead stems of dominant tall tree species in the two growing seasons. Note the differences in scale for the vertical axes.
Maxim., Phellodendron amurense Rupr., Salix spp. (mainly Salix bakko Kimura), Q. crispula, and F. mandshurica. Figure 3 shows 2-year changes in density, number emerging, and mortality for the dominant tall tree species. Betula spp. and Salix spp. maintained a continuous appearance throughout both years, although with relatively high mortality (41.7% and 35.1% in total for Betula spp. and Salix spp., respectively). In contrast, appearance tended to be concentrated in a particular period for A. sachalinensis, P. amurense, A. mono, and Q. crispula (July in the first growing season) and for F. mandshurica (June in the second growing season). These last five species had lower mortality than Betula spp. and Salix spp., resulting in moderate density changes. The mean seedling height at the end of second growing season was >10 cm in Q. crispula and F. mandshurica and >5 cm in P. amurense and A. mono. In contrast, the mean height remained below 2 cm in A. sachalinensis, Betula spp., and Salix spp. Relations between environmental and demographic variables Seedling emergence Table 5 shows results of the multiple regression analyses for the emergence of tall tree seedlings. With data from the first growing season, significant models were obtained for Betula spp., A. sachalinensis, P. amurense, and Salix spp., with LIGHT showing a strong negative correlation. For Betula
spp. and P. amurense, LITTER also had a negative effect. The emergence of Salix spp. was related to SOIL2 (positive); and the emergence of A. sachalinensis, to TOTBA (positive). No significant model was obtained for A. mono or Q. crispula. With data from the second growing season, a significant model was obtained for Betula spp., Salix spp., and F. mandshurica. For Betula spp., LIGHT and LITTER were negatively correlated, and SOIL1 and TOTBA were positively correlated. The emergence of Salix spp. was related to LIGHT (negative), as well as to SOIL1, SOIL2, and SCAR1 (positive). Emergence of F. mandshurica was affected negatively by LIGHT and positively by SOIL1. No significant model was obtained for A. mono. Seedling mortality Table 6 shows the results of multiple regression analyses to explain mortality during the growing season, and Table 7 shows correlations between mortality during the nongrowing season and snow depth in early spring. Mortality of Betula spp. was influenced by CANOP (positive) and SOIL1 (negative) during the first growing season but by LIGHT (positive) and LITTER (positive) when they regenerated during the second growing season after scarification. The mortality of A. sachalinensis was random during the first growing season but was affected by SNOW (positively) and by SOIL2 (negatively) in its second growing season. The mortality of A. mono was correlated with LIGHT (positively) and SOIL2 (negatively) in the first growing season. Phellodendron amurense © 2005 NRC Canada
182
Can. J. For. Res. Vol. 35, 2005
Table 5. Results of stepwise multiple regression analyses, explaining the number of emerging seedlings of dominant tall tree species. R2 Emergence in the first year Betula spp. 0.146 Abies sachalinensis 0.155 Acer mono — Phellodendron amurense 0.111 Salix spp. 0.142 Quercus crispula — Emergence in the second Betula spp. Acer mono Salix spp. Fraxinus mandshurica
(
