In the present study, we analysed the habitat association of tree species in an old-growth temperate forest across all life stages to test theories on the coexistence of tree species in forest communities. An inventory for trees was implemented at a 6-ha plot in Ogawa Forest Reserve for adults, juveniles, saplings and seedlings. Volumetric soil water content (SMC) and light levels were measured in 10-m grids. Relationships between the actual number of stems and environmental variables were determined for 35 major tree species, and the spatial correlations within and among species were analysed. The light level had no statistically significant effect on distribution of saplings and seedlings of any species. In contrast, most species had specific optimal values along the SMC gradient. The optimal values were almost identical in earlier life stages, but were more variable in later life stages among species. However, no effective niche partitioning among the species was apparent even at the adult stage. Furthermore, results of spatial analyses suggest that dispersal limitation was not sufficient to mitigate competition between species. This might result from well-scattered seed distribution via wind and bird dispersal, as well as conspecific density-dependent mortality of seeds and seedlings. Thus, both niche partitioning and dispersal limitation appeared less important for facilitating coexistence of species within this forest than expected in tropical forests. The tree species assembly in this temperate forest might be controlled through a neutral process at the spatial scale tested in this study.

Introduction

How do multiple tree species coexist in a forest community? This is one of the major issues in ecological studies of forests. Recently, there has been debate from different perspectives: niche assembly (habitat association) versus stochastic processes, such as dispersal limitation. The former emphasises heterogeneity in abiotic and biotic environments, which enable niche segregation among species along gradients of multiple environmental variables (Rüger et al. 2009; Brown et al. 2013); until now, many studies have presented data that support this hypothesis. For example, some species have biased distribution under a specific topography or soil condition, suggesting mitigation of competition with other species (Comita & Engelbrecht 2009; Vieilledent et al. 2010; Chuyong et al. 2011; Metz 2012; Baldeck et al. 2013; Brown et al. 2013). In addition to soil-related factors, light availability makes an important contribution to species composition in the early life stages (Abe et al. 1995; Webb & Peart 2000; Rüger et al. 2009).

In contrast to the niche assembly perspective, the stochastic perspective has recently been highlighted (Hubbell 2001; Rosindell et al. 2011). This theory places less emphasis on environmental heterogeneity and related niche partitioning and suggests that a stochastic process, such as seed dispersal limitation, can explain the local species assembly pattern. From this perspective, some studies, based on data obtained from large-scale forest plots, argue about the limited role of local habitat specialisation and instead emphasise dispersal limitation in the organisation of forest communities (Seidler & Plotkin 2006; Lin et al. 2011; Shen et al. 2013).

Recent studies have suggested that these two hypotheses are not mutually exclusive (Li et al. 2009; Lin et al. 2011). When testing these hypotheses, it should be noted that the importance of environmental heterogeneity might vary between life stages (de Toledo et al. 2011; Metz 2012). Comita et al. (2007) suggested that individuals in the adult tree stage more often have significant distributional correlations with specific soil moisture conditions. Hu et al. (2012) further argued that the distribution of small trees is likely to be explained by dispersal limitation, whereas the distribution of adult trees is more likely to be determined by the environment. These studies postulate that the associations of mature trees with specific habitat conditions can be formed during the early life stages via differentiated growth and mortality between species (Metz 2012).

However, there are some concerns about these previous studies. First, the soil conditions were categorised in most cases on the basis of apparent topography or soil classification types (Gunatilleke et al. 2006; Comita et al. 2007; Chuyong et al. 2011; de Toledo et al. 2011; Metz 2012); however, soil conditions vary continuously rather than discretely in a spatial context (Baldeck et al. 2013). Thus, simplifying soil conditions by categorisation would reduce information related to soil environmental variation, and this may lower the robustness of any statistical analysis (Itoh et al. 2010). Second, analyses of tree distributions have often used inventory data from the well-established stages. The environmental regulation of specific distributions could change ontogenetically (as reviewed above); therefore, studies of the spatial configurations of trees should also consider younger life stages (Comita et al. 2007; Metz 2012; Ota et al. 2012; Oki et al. 2013). Third, previous studies were mostly conducted in tropical forests and rarely in temperate forests. Temperate forests comprise less species than tropical forests; therefore, the former forests should provide good samples for validating the generality of theories developed for tropical forests. Some previous studies suggest that the species assemblies of temperate forests may be niche-determined rather than chance-determined (Nakashizuka 2001; Myers et al. 2013). More recently, however, it has been suggested that niche partitioning in terms of habitat associations among species is not apparent, even in temperate forests (Shibata et al. 2010). Hence, dispersal limitation may also play an important role in temperate forests in shaping the diversity and spatial patterns of tree species.

The present study analysed the species distribution pattern in a temperate forest through various life stages, from seedlings to mature adults, to assess the importance of habitat associations and dispersal limitations. First, tree species distribution was analysed with respect to soil and light conditions during the ontogenetic life stages. The number of stems in each species during each life stage was modelled using an equation that included environmental factors as continuous explanatory variables and that considered spatial autocorrelation. The model parameters were estimated using a Bayesian framework, which tested whether each species had a specific niche space along environmental gradients, possibly resulting in the avoidance of interspecific competition. Second, intraspecific spatial correlations among life stages and interspecific spatial correlations in each life stage were analysed to assess the effect of dispersal limitation on mitigating the overall interspecific competition. Based on the results of these analyses, the following questions were addressed: (i) do tree species have different optimal distributions relative to soil and light conditions; (ii) do specific tree distributions change relative to environmental variables through ontogenetic life stages; (iii) are seedlings distributed more frequently in the vicinity of conspecific adult trees; (iv) are tree species distributed exclusively with each other in any life stage; and (v) what are the roles of niche assembly and dispersal limitation in the context of coexistence of tree species within a temperate forest?

Material and Methods

Study site

The present study was conducted in the Ogawa Forest Reserve (OFR), which is an old-growth deciduous hilly forest. OFR is located in the southern part of the Abukuma Mountains, Ibaraki Prefecture, central Japan (36°56′ N, 140°35′ E, 610 m a.s.l.). The mean monthly temperature is 10.7 °C, with highest monthly mean of 22.6 °C in August and lowest −0.9 °C in February. Annual precipitation is approximately 1910 mm, and highest in August and September and lowest in December and January, while the maximum snow depth often reaches 50 cm (Mizoguchi et al. 2002).

Within the 6-ha permanent plot (300 × 200 m; Fig. 1), which was established in 1987 in OFR (Masaki et al. 1992), there were 61 tree species with >15-cm DBH, i.e. 49 tall tree species and 12 shrub species. Fagaceae, Betulaceae and Acer were the dominant taxa in this forest. More detailed information on the tree species composition in OFR is available in Masaki et al. (1992) and Masaki (2002).

Details are in the caption following the image — **Figure 1**
Open in figure viewer PowerPoint

Topographic map of the 6-ha plot. The altitudinal interval of contours is 2 m. Soil moisture content (SMC; cm³ cm⁻³) is shown using 10-m grids on the basis of shade density.

The elevational difference within the plot was 50 m. A small stream ran from north to south through the middle of the plot. The typical soil types were Andic Haplumbrepts and other types (e.g. Andic Dystrochrepts) (Yoshinaga et al. 2002); further information on topography and soil types can be found in Yoshinaga et al. (2002). In a preliminary study (M. Takahashi et al., unpublished research), where soil samples were obtained at depths of 5–15 cm at 17 locations representing these soil types in the plot, soil nitrogen content did not vary significantly among soil types (anova P = 0.1), while the liquid phase varied significantly among soil types (anova P = 0.04). Liquid phase values showed a significant correlation with depth of 50 cm (r = 0.75, P < 0.01). Based on these preliminary results, we assumed that the volumetric soil water content at a depth of 15 cm could be used as a representative variable for soil conditions potentially affecting the tree species distribution (Kozlowski et al. 1991; Palmiotto et al. 2004; Engelbrecht et al. 2007).

Field methods

Tree inventory

We categorised four tree life stages: mature adult, immature juvenile, sapling and seedling. For stems >15 cm GBH (girth at breast height), reproductive trees were defined as adults and non-reproductive trees were defined as juveniles. The specific threshold between adults and juveniles was set at 0.3 × specific maximum girth within the plot, based on results presented in Shibata & Tanaka (2002) and their unpublished results. For smaller stems, those taller than 30 cm were defined as saplings, while those <30-cm high and >1 year old were defined as seedlings. The inventory was conducted for each of the stages with the sampling designs described below.

The 6-ha plot was demarcated into 600 cells using 10-m grids and further divided visually into sub-cells of 5 × 5 m (2400 sub-cells in total). Within each sub-cell, stems of juvenile and adult trees were identified in May 2005. Saplings of tree species were identified in a quadrat of 2 × 2 m placed at each of the grids (651 quadrats and 2604 m² in total) in May 2004. In 2006, a sub-quadrat (1 × 1 m) was placed within each of the 2 × 2-m quadrats (651 m² in total) and seedlings of the tree species were identified. During these inventories, approximately 7400 stems were identified in total across the life stages (1141 adults, 3180 juveniles, 1761 saplings and 1324 seedlings).

Environmental variables

The volumetric water content was measured as an index of the soil moisture content (SMC) in autumn 2006 using an electronic sensor (HydroSense; Campbell Scientific Inc., UT, USA) 3 days after the last rainfall. The values at five points randomly selected within each quadrat at a depth of 15 cm from the soil surface were measured and averaged (Fig. 1). The number of days during May–September (153 days) in recent years (i.e. 2011–2013) when rainfall did not occur in the previous 3 days ranged from 94–107 days, suggesting that SMC data of this study represented 60–70% of the overall SMC in a growing season (based on data collected by the Japan Meteorological Agency at a site 4 km apart).

The light conditions were measured in summer 2006. We took hemispherical photographs using a digital camera (COOLPIX 5000, Nikon, Tokyo, Japan) with a fish-eye lens (FC-E8, Nikon) at 1 m above the ground surface at the centre of each quadrat. The relative photon flux density in each quadrat was estimated from the images using CanopOn (version 2.01), which was developed by A. Takenaka (http://takenaka-akio.org/etc/canopon2/).

In the following analyses, four subcells around a grid were merged (and redefined as a subplot) and the environmental variables obtained at the grid were assigned to stems in the corresponding subplot. A variable measured on the border of the 6-ha plot was assigned to one and two sub-cells (also redefined as a subplot of smaller area) at the corner and edge, respectively (i.e. areas of these subplots were smaller).

Analyses

Species with more than five adult trees were used in the present study (35 species; Table 1). We use a hierarchical Bayesian framework to model the relationship between the number of stems and an environmental variable.

Table 1. Number of stems sampled from each species in each life stage. The sampling areas (ha) were 6, 6, 0.24 and 0.06 for the adult, juvenile, sapling and seedling stages, respectively. Species with less than five seedlings and saplings are denoted as low-recruitment species. Nomenclature follows Satake et al. (1999). A hyphen indicates that number of stems was less than five.

taxon	adult	juvenile	sapling	seedling	recruitment
Betulaceae
Betula grossa	22	11	–	–	low
Ostrya japonica	30	31	–	6
Carpinus cordata	47	498	42	9
Carpinus japonica	15	30	–	–	low
Carpinus tschonoskii	38	54	5	10
Carpinus laxiflora	57	411	13	49
Fagaceae
Fagus crenata	39	110	474	24
Fagus japonica	150	475	126	69
Quercus crispula	14	23	–	34
Quercus serrata	132	149	–	239
Castanea crenata	41	13	–	–	low
Magnoliaceae
Magnolia obovata	5	10	–	–	low
Hamamelidaceae
Hamamelis japonica	18	14	28	–
Rosaceae
Prunus buergeriana	5	–	–	–	low
Prunus grayana	5	21	112	19
Prunus verecunda	26	36	–	8
Sorbus japonica	15	25	12	14
Malus tschonoskii	5	–	–	–	low
Sapindaceae
Acer amoenum	26	536	166	145
Acer sieboldianum	10	72	33	8
Acer tenuifolium	13	104	89	65
Acer rufinerve	33	22	31	170
Acer mono	23	143	92	105
Acer nikoense	8	23	–	–	low
Acer cissifolium	8	14	5	–
Sabiaceae
Meliosma myriantha	12	105	58	–
Aquifoliaceae
Ilex macropoda	11	40	45	10
Cornaceae
Swida controversa	59	63	18	18
Benthamidia japonica	10	55	15	–
Araliaceae
Acanthopanax sciadophylloides	6	21	72	51
Kalopanax pictus	11	5	–	18
Clethraceae
Clethra barbinervis	38	87	40	61
Styracaceae
Styrax japonica	15	20	75	8
Styrax obassia	148	343	21	28
Oleaceae
Fraxinus lanuginosa	20	49	90	77

Specific distribution throughout the environment

We assumed that the density of stems of the i-th species in the j-th subplot [or (sub)quadrat] (n_i,j) follows a Poisson distribution and varies around the unknown parameter E_i,j multiplied by the relative area (A_j) of the subplot (0.25 at the corner, 0.5 at the edge, and 1.0 elsewhere) and the (sub)quadrat (1.0 for all of them):

$urn:x-wiley:14358603:media:plb12253:plb12253-math-0001$ (1)

We used the following function to model E_i,j based on an environmental variable, x:

$urn:x-wiley:14358603:media:plb12253:plb12253-math-0002$ (2)

where a_i, b_i and c_i represent species-specific parameters; x_j is the value of an environmental variable (SMC or light level) that ranges from 0 to 1; and ρ_i,j is the spatial random effect on the j-th subplot [or (sub)quadrat]. The variable x_j was logit-transformed in the model to ensure that it ranged from −∞ to ∞. The parameter a_i is the degree of specialisation of the i-th species on the focal environmental variable. If a_i is 0, this function is flat, suggesting that this species is distributed independently of the values of the variable for this environment (i.e. generalist). However, if a_i is positive, this function is unimodal, suggesting that the number of stems of this species is higher at a certain range of this environmental gradient (i.e. specialist). If a_i is negative, the shape of the curve is concave and the number of stems of this species tends to be distributed in extreme environmental conditions. When a_i is positive, the parameter b_i represents the location of the peak of the convex curve, which is the optimal environment for the i-th species. The parameter c_i is the height of the curve, which roughly reflects the specific population size.

We then describe the variation in a_i, b_i and c_i using normal distributions with hyper-parameters:

$urn:x-wiley:14358603:media:plb12253:plb12253-math-0003$ (3)

$urn:x-wiley:14358603:media:plb12253:plb12253-math-0004$ (4)

$urn:x-wiley:14358603:media:plb12253:plb12253-math-0005$ (5)

We had no prior knowledge of these hyper-parameters, therefore their non-informative priors have a normal distribution,with mean = 0 and variance = 1000 for μ_a, μ_b, and μ_c and a gamma distribution with shape parameter = 0.001 and scale parameter = 0.001 for 1/ $urn:x-wiley:14358603:media:plb12253:plb12253-math-0006$ , 1/ $urn:x-wiley:14358603:media:plb12253:plb12253-math-0007$ and 1/ $urn:x-wiley:14358603:media:plb12253:plb12253-math-0008$ . Spatial random effects were modelled as an intrinsic Gaussian conditional autoregressive (CAR) effect (Latimer et al. 2006). For the j-th subplot [or (sub)quadrat], the CAR effect has a normal distribution with a mean ρ of its neighbours and variance:

$urn:x-wiley:14358603:media:plb12253:plb12253-math-0009$ (6)

where j+ denotes the set of neighbours and Nj+ is the number of neighbouring subplots [or (sub)quadrats] (with smaller values at the edge or the corner of the 6-ha plot). The number of adjacent subplots [or (sub)quadrats] that needed to be encompassed as the neighbourhood was defined on the basis of the spatial scale of distributional aggregation in each life stage (see below and Results). The conditional variance $urn:x-wiley:14358603:media:plb12253:plb12253-math-0010$ is a hyper-parameter. A gamma distribution with shape parameter = 0.001 and scale parameter = 0.001 was also used as a non-informative prior of 1/ $urn:x-wiley:14358603:media:plb12253:plb12253-math-0011$ .

The posterior distributions of the parameters were separately obtained for each life stage and for each of light (only for seedling and sapling stages) and SMC level using a Markov chain Monte Carlo (MCMC) method with OpenBUGS (Lunn et al. 2009). Three chains were initialised with different values of these parameters. The convergence of MCMC was assessed visually, with reference to Gelman and Rubin's convergence diagnostics ( $urn:x-wiley:14358603:media:plb12253:plb12253-math-0012$ ) smaller than 1.1 (Gelman et al. 2004). The number of iterations required for convergence varied from 80,000 to 220,000, depending on the life stages and environmental variables. We used 20,000 additional iterations and thinned the chains every two steps to obtain a posterior sample size of 10,000, which yielded the posterior distributions. When the credible interval of parameter a_i encompassed 0, the i-th species was assumed to be generalist in that specific life stage. For species with less stems, the estimated values (a, b, c) might converge on values of the hyper-parameters (μ_a, μ_b, μ_c). To confirm this, the estimated values of the parameters were plotted against the specific sample sizes.

Based on the estimated function, the distribution probability of the i-th species in environment x could be obtained using the following equation:

$urn:x-wiley:14358603:media:plb12253:plb12253-math-0013$ (7)

We solved this equation numerically. A random number, k, was uniformly generated between 0 and 1, the expected number of stems (E) in environment k was calculated, and the number of stems (N) was randomly obtained using a Poisson distribution, where E is the mean. If the value of N was 3, for example, we obtained the vector {k, k, k}. This procedure was repeated until the total length of the vectors reached 100,000 and f(x) was assessed as the probability density.

A niche overlap index was calculated for each pair of species at each life stage as percentage of similarity of species distribution (C) (Whittaker & Fairbanks 1958):

$urn:x-wiley:14358603:media:plb12253:plb12253-math-0014$ (8)

where, p_i,x represents the number of species i relative to the conspecific total at environment x. To calculate C, the logit-transformed environmental gradient was divided into classes with a width of 0.1 (i.e. breaks at −10, −9.9, −9.8, ··· 9.8, 9.9 and 10), and the number of each species relative to its total number in each class was calculated using f(x) obtained above.

To test if niche overlap at the community level was larger (or smaller) than expected by assuming a random assembly of species, parameters a and b were randomly assigned to 35 species in the following way (value of parameter c was set to 1). For parameter a, 35 random values were generated based on a uniform distribution (ranging from minimum to maximum values estimated in the analyses). Parameter b was also generated in the same way. This community was named the Random community. Another extreme pattern was also assumed: the maximum of the estimated values for parameter a was assigned to every species [i.e. the narrowest breadth of a species distribution curve f(x)], the values of parameter b were evenly distributed into 35 classes between maximum and minimum values estimated and assigned to 35 species. This pattern of species assembly assumes a situation in which the niche is most effectively partitioned among species. This community was named as Uniform community. The mean (μ_m) and SD ( $urn:x-wiley:14358603:media:plb12253:plb12253-math-0015$ ) of logit(C) for a community m (e.g. 595 values of C for 35 species) were compared among adult, juvenile, sapling and seedling stages and Random and Uniform communities, which were estimated using a Bayesian framework as follows:

$urn:x-wiley:14358603:media:plb12253:plb12253-math-0016$ (9)

$urn:x-wiley:14358603:media:plb12253:plb12253-math-0017$ (10)

$urn:x-wiley:14358603:media:plb12253:plb12253-math-0018$ (11)

where, μ_h, μ_τ, $urn:x-wiley:14358603:media:plb12253:plb12253-math-0019$ and $urn:x-wiley:14358603:media:plb12253:plb12253-math-0020$ are hyper-parameters. Non-informative priors for the first two parameters were a normal distribution with mean = 0 and variance = 1000, and the inverses of the last two parameters was a gamma distribution with shape parameter = 0.001 and scale parameter = 0.001. Estimation with MCMC was undertaken in the same way as described earlier, with 9000 iterations, discarding the first 1000 steps as a burn-in.

Spatial analyses

The spatial scale of aggregation of each species in each life stage was evaluated by calculating L(r) transformed from Ripley's K-function K(r) (Lan et al. 2009):

$urn:x-wiley:14358603:media:plb12253:plb12253-math-0021$ (12)

A negative value of L(r) indicated regular distribution and a positive value indicated aggregated pattern at scale r. First, adult and juvenile trees were randomly distributed within the sub-cells to which they belonged. For saplings and seedlings, the trees with the same number as the (sub)quadrat to which they belonged were randomly distributed within the corresponding subplot. These distributions were used as the observed distribution. Second, a random distribution of trees within the 6-ha that had the same number as each life stage was generated. This random distribution was generated 20,000 times and used to estimate the expected range (i.e. 95% confidence envelope) of L(r), and the spatial scale of aggregation was evaluated on the basis of a comparison with the observed L(r).

To evaluate specific seed dispersal limitation, the univariate L(r) was expanded to a bivariate function L₁₂(r) (Lan et al. 2009), and the spatial correlations between the adult and seedling (or sapling) stages were tested by comparing observed and expected values of L₁₂(r). In this test, a random distribution was generated 20,000 times by torus translation. Similarly, the interspecific spatial correlations were tested in each stage for all of the species pairs. The number of pairs was 595, 528, 253 and 276 for the adult, juvenile, sapling and seedling stages, respectively (Table 2).

Table 2. Spatial correlations evaluated on the basis of L(r) for the species pairs in each life stage.

stage	number of pairs	interspecific spatial correlation
stage	number of pairs	positive	not significant	negative
adult	595	86 (14.5%)	405 (68.1%)	104 (17.5%)
juvenile	528	149 (28.2%)	236 (44.7%)	143 (27.1%)
sapling	253	65 (25.7%)	121 (47.8%)	67 (26.5%)
seedling	276	102 (37.0%)	127 (46.0%)	47 (17.0%)

All the calculations for L and L₁₂ were implemented for the range of r from 0 to 30 m at every 0.5-m interval. If multiple r-values existed, whereby the observed Ls were more than the 95% confidence envelope and these values were not continuous, we adopted the minimum value as the spatial scale of aggregation.

Results

The SMC values ranged from 0.05 to 0.8 (cm³ cm⁻³), which corresponded to −2.9 to 1.4 (mean = −1.16) after logit transformation (Fig. 2, upper). The light level ranged 0.005 to 0.25, which corresponded to −5.3 to −1.1 (mean = −3.18; Fig. 2, lower). Among the 35 species tested, two, 12 and 11 species had less than five stems in the juvenile, sapling and seedling stages, respectively (Table 1). Seven species were low-recruitment species, with less than five saplings and seedlings according to the sampling method used in the present study.

Analysis based on L(r) showed that over 60% of the species had aggregated distributions during any life stage, and the scales of spatial aggregation for the aggregated species were estimated as 10.7 ± 6.3 m, 7.6 ± 6.5 m, 5.0 ± 3.5 m and 6.2 ± 7.7 m for the adult, juvenile, sapling and seedling stages, respectively. Based on these results, the nearest neighbouring subplots [or (sub)quadrats] (i.e. a distance of 10 m) were used to simplify evaluation of the spatial autocorrelation in the following analyses of any species.

Analysis based on the light level as an environmental variable did not detect any significant bias of the parameter a from 0 in either the seedling or sapling stage (data not shown). Thus, the distributions of seedlings and saplings of the species tested were not significantly affected by the spatial heterogeneity of light conditions in this forest; therefore only the SMC results were considered in the following analyses. The estimated values of parameters a, b and c with respect to SMC are shown in Fig. 3 and summarised in Appendices S1–4. For parameter a (Fig. 3, top), all the species had significant positive values in the adult stage (ca. 0.3–0.5), with the exception of Acer mono (ca. 0.2), which had a credible interval that encompassed 0. In the juvenile stage, all species other than Kalopanax pictus (<0.2) had significant positive values (ca. 0.2–0.3). Saplings and juveniles had similar values, and seedlings tended to have the highest values in comparison with other stages. The values of species with smaller sample sizes did not tend to converge.

For parameter b (Fig. 3, middle), most species had values around −1.5 (which corresponded to SMC ≈ 0.2 cm³ cm⁻³) in the seedling and sapling stages. In the juvenile stage, some species tended to have smaller values that ranged from −2 to −1 (SMC 0.1–0.3), and values in the adult stage tended to have a wider range than those in the preceding stages, i.e. from −2.5 to 0.7 (SMC 0.1–0.7). The values of species with smaller sample sizes did not tend to converge.

For parameter c, the estimated values tended to be higher with increasing sample sizes, irrespective of the life stage, as expected (Fig. 3, bottom).

The probability densities in terms of SMC are shown in Fig. 4. In the seedling stage, most species had similar functions and a narrower distribution range than at other stages in terms of SMC. The range distribution in the sapling stage was larger than that in the seedling stage. The differentiation of optimal values between species appeared to begin in the juvenile stage, and this tendency was more conspicuous in the adult stage. Low-recruitment species did not show a consistent tendency of a biased distribution toward wetter or drier conditions.

The niche overlap index for the adult stage was significantly smaller than for younger stages, but significantly larger than the Random and the Uniform communities in terms of credible intervals of the posterior of the mean (Fig. 5). The niche overlap index at the juvenile stage was intermediate between the adult and other stages, but was relatively closer to the sapling and seedling stages. Among the six communities tested, the Uniform community showed minimum values of niche overlap.

The intraspecific spatial correlations between adults and seedlings were not significant for 24 species at any spatial scale and were significant only for three species (Fagus crenata, Quercus crispula and Quercus serrata) beyond a spatial scale of 16.0 ± 9.5 m. In the sapling stage, correlations were not significant for 23 species and significant for Fagus crenata and Acer tenuifolium beyond the spatial scale of 20.0 m and 22.5 m, respectively. For adult–juvenile relationships, about half of the species (16 species) showed significantly positive spatial correlation beyond the spatial scale of 8.3 ± 5.5 m, and the other 17 species did not show significant correlations (data not shown).

Among the pairs of different species, no more than 30% had significant negative spatial correlations at specific spatial scales in any of the life history stages (Table 2).

Discussion

The results of the present study in a temperate forest are in accordance with previous findings from tropical forests in some respects. Comita et al. (2007) suggested that individuals in the adult tree stage have more frequent significant distributional correlations with soil moisture conditions. Hu et al. (2012) also argued that the distributions of adult trees are more likely to be determined by the environment than those of younger trees. This study also postulated that the association of mature trees with specific habitat conditions could be formed ontogenetically, probably via differentiated growth and mortality between life stages, as in tropical forests (Metz 2012). However, whether niche partitioning along environmental gradients contributes to coexistence of species in this forest is unclear. Furthermore, the role of dispersal limitation also appeared to be less significant.

Does habitat association maintain species diversity?

Based on results of the present study, the answer to the question “Do tree species have different optimal values for their distributions relative to the soil and light conditions?” is yes with respect to SMC but no with respect to light. With the exception of Acer mono in the adult stage and Kalopanax pictus in the juvenile stage, all the species had significantly positive values for parameter a with respect to SMC, suggesting that their distribution patterns had unimodal curves along the SMC gradient and that each species had an optimal value on this gradient.

Based on estimated values of b, the optimal SMC tended to be more variable among species in later life stages, while interspecific competition was not completely relaxed at every stage, as shown by mean values of the niche overlap indices, which were significantly larger than expected for a random assembly. In a previous study (Masaki et al. 1992), habitat segregation among species was suggested on the basis of visual assessments of specific distribution within the plot. However, the quantitative soil condition data collected in this study showed that habitat was partitioned less effectively than could be randomly assumed. Furthermore, the forest studied did not include an extremely wet or dry habitat (e.g. swampy habitat; Harms et al. 2001) and did not experience dry summer each year (as in seasonal evergreen tropical forests; Baltzer et al. 2008). Considering that SMC mostly ranged between 0.1 and 0.5 in the plot (Fig. 2, upper), and that most of the species distribution overlapped within this range (Fig. 4), efficient niche segregation along the SMC gradient was difficult. Although Palmiotto et al. (2004) reported specific different responses to soil conditions on the basis of an experimental study, their experiment used extreme soil conditions to elucidate the specific differences more clearly. Heterogeneity in soil conditions at the landscape scale, where extreme soil conditions are found, may be sufficient to determine tree distributions (Laurance et al. 2010). However, at the local scale, spatial variations in soil conditions may not fully explain the species assemblies.

No effective niche segregation with respect to the light environment was detected in the seedling and sapling stages. Another previous study in OFR suggested that saplings of some tree species were present at a higher density under canopy gaps than in other areas (Abe et al. 1995). However, this result depended on intensive sampling efforts under the gaps. In this forest, sites with abundant light (e.g. light level >0.1) were very rare (Fig. 2), which probably explained the failure to detect light effects on seedling and sapling distributions. Under the current conditions, the community of seedlings and saplings should be mainly composed of more shade-tolerant species. Several species that had few seedling and sapling populations, such as Betula grossa and Castanea crenata, are expected to decrease in the future (Masaki et al. 1992). The majority of the other species are considered to be less gap-dependent in the study forest (Abe et al. 1995); therefore, current spatial variation in the light environment is not considered to be critical for the coexistence of tree species. If an exceptionally large and intense disturbance occurs in future, we may be able to detect niche segregation in terms of the light environment in seedling and sapling stages.

Thus, we postulate that the coexistence of tree species within this old-growth forest might not be substantially facilitated by niche segregation in terms of both the soil and the light environments.

Do tree distributions change with the life stages?

Initially, we expected that seedlings of any species would be distributed as habitat generalists because the seeds of most species in this forest were widely dispersed, having a highly scattered distribution (Masaki et al. 1994; Nakashizuka et al. 1995; Shibata & Nakashizuka 1995; Tanaka et al. 1997). However, distribution of seedlings was concentrated within a specific range in the soil moisture environment, just as habitat specialists. Seedlings are often vulnerable to extreme soil water availability conditions (Comita & Engelbrecht 2009). This may explain why early stages of the trees in this forest had distribution modes at modest conditions in terms of SMC.

In contrast, the tree specific habitat diverged with the life stage: nine changed to wetter and 15 did to drier habitats from the seedling to adult stages (Appendix S1, S4). Changes of distribution to wetter habitats with life stage has been found in many previous studies (Gunatilleke et al. 2006; Comita et al. 2007; Chuyong et al. 2011), suggesting that a preference for wetter habitats in later stages may be common among tree species in both tropical and temperate forests. However, one should also consider that ontogenetic physiological development of desiccation tolerance over life stages (Ishida et al. 2005) could enable the majority of species to change habitat to drier areas in the adult stage.

Metz (2012) concluded that species become more niche-partitioned in later life stages because of specific differential mortality in different habitats. We consider the change of habitat with life stage in this forest could be explained by similar mechanisms, and this can be tested in future by monitoring the growth and survival of younger trees.

Are species exclusively distributed in space?

Hu et al. (2012) found that species distribution was determined by dispersal limitation in the younger stages. However, no evidence of effective dispersal limitation was found in the present study. The seedlings and saplings of most species had independent distributions relative to their conspecific adults, except for a few species with large seeds. Juveniles displayed a positive spatial correlation with conspecific adult trees for about half of the species studied, but these correlations between older-stage trees were likely to result not from dispersal limitation, but rather from changes in habitats preferred by conspecific adult trees. In addition, most species had spatially nonexclusive distributions compared with others (>80% of species pairs) in any life stage. Thus, dispersal limitation did not appear to facilitate the coexistence of multiple species within this forest community, according to statistical analysis.

There are several possible reasons for the absence of effects of dispersal limitation in this study. First, this may be a consequence of the seed dispersal patterns in this forest. Approximately half of the species in this forest were found to have winged seeds (i.e. Betulaceae, Acer and Fraxinus), which resulted in a highly scattered seed distribution (Nakashizuka et al. 1995; Shibata & Nakashizuka 1995; Tanaka et al. 1997). Bird-dispersed seeds also had a very scattered distribution in this forest (Masaki et al. 1994) compared with the patterns modelled in tropical forests (Schupp et al. 2002). In addition, even the acorns of Quercus have been reported to be well scattered in this forest via secondary dispersal through rodents (Iida 1996). Thus, dispersal limitation may be less likely in this forest than expected in tropical forests.

Second, density-dependent mortality is also causal. Previously, Masaki & Nakashizuka (2002) and Shibata et al. (2010) analysed the current-year seedling mortality and showed that most species experienced highest mortality during the year immediately after germination, which was positively and negatively correlated with the conspecific seedling density and distance from the nearest conspecific mature tree, respectively. This mechanism may result in no contagious distribution of >1-year-old seedlings around conspecific mature trees.

Thus, dispersal limitation may only play a restricted role in this temperate forest in regulating species assembly. Most species could potentially reach a vacant site after the death of a mature tree via effective seed dispersal. Furthermore, many species in this forest exhibit synchronised seed production over the years, irrespective of taxon (Shibata et al. 2002). This means that every species has an equivalent opportunity to be recruited to a site wherever and whenever it becomes vacant.

How is tree species diversity maintained?

Why was difference in habitats not clear in the present study? We offer three possible explanations to this question. First, interspecific niche differentiation may have been detected if other environmental variables had they been included in the analysis. The spatial variation in resources other than SMC, such as nitrogen (although it is spatially less variable than SMC in this forest) and phosphorus, could affect the species distribution patterns (John et al. 2007). Effects of these other factors need to be tested in future studies.

Second, the temporal oscillation in environmental conditions may have effects on habitat association. The adult trees of low-recruitment species may be the legacy of past disturbances in this forest (Masaki et al. 1992). At present, this old-growth forest has a low proportion of light gaps (Fig. 2, lower). The current light regime is unlikely to be beneficial for recruitment of light-demanding species. However, if an intensive and large-scale disturbance occurs in this forest, the spatial heterogeneity in light conditions will be sufficient to affect recruitment patterns of tree species. The roles of spatiotemporal dynamics in coexistence of tree species need to be tested on the basis of long-term monitoring studies.

Third, spatial scales should be considered (Wang et al. 2012). Most previous studies in tropical or subtropical forests were conducted in larger-scale inventory plots (e.g. 50 ha), and these studies often categorised habitats into several types (each ca. 10–20 ha) on the basis of analyses (Gunatilleke et al. 2006; Comita et al. 2007; Chuyong et al. 2011). However, these studies did not consider the species assembly rules within each category, which is similar to the present study in terms of spatial scale. We consider it likely that species assembly could be regulated through a neutral process at the local scale (<10–20 ha), whereas partitioning of the environment would be detected among guilds of species at larger scales (Pearse et al. 2013).

To conclude, we propose two hypotheses for further studies. First, as also suggested in tropical forests, species assemblies could be controlled through a neutral process in temperate forests where the habitats of species appear to be identical rather than well partitioned in terms of SMC. Second, seed dispersal limitation might not play a substantial role in shaping local species assemblies in temperate forests because of more scattered seed distribution than in tropical forests. More comparative studies of temperate forests are required to establish a general and convincing theory regarding the mechanism that maintains tree species diversity in forest communities worldwide (Shen et al. 2013).

Acknowledgements

We thank Dr. T. Nakashizuka, Dr. T. Nagaike, and Dr. T. Tanaka for helpful support throughout this study. We also thank Dr. N. Miyashita for critical suggestions on statistical modelling, and Dr. A. Itoh for useful comments on an earlier version of the manuscript. We would like to thank Enago (www.enago.jp) for the English language review. This study was partly funded by the Technology Development Fund (S-9) of the Ministry of the Environment, Japan. Part of the data used in the present study was derived from the Forest Dynamics Database provided by the Forestry and Forest Products Research Institute and from the Monitoring Sites 1000 Project launched by the Ministry of the Environment, Japan. The meteorological data of the Japan Meteorological Agency was obtained from the Agriculture, Forestry and Fisheries Research Council, Japan.

Supporting Information

References

Abe S., Masaki T., Nakashizuka T. (1995) Factors influencing sapling composition in canopy gaps of a temperate deciduous forest. Plant Ecology, 120, 21–32.
Web of Science® Google Scholar
Baldeck C.A., Harms K.E., Yavitt J.B., John R., Turner B.L., Valencia R., Navarrete H., Davies S.J., Chuyong G.B., Kenfack D., Thomas D.W., Madawala S., Gunatilleke N., Gunatilleke S., Bunyavejchewin S., Kiratiprayoon S., Yaacob A., Supardi M.N.N., Dalling J.W. (2013) Soil resources and topography shape local tree community structure in tropical forests. Proceedings of the Royal Society B, 280, 20122532.
10.1098/rspb.2012.2532
PubMed Web of Science® Google Scholar
Baltzer J.L., Davies S.J., Bunyavejchewin S., Noor N.S.M. (2008) The role of desiccation tolerance in determining tree species distributions along the Malay-Thai Peninsula. Functional Ecology, 22, 221–231.
10.1111/j.1365-2435.2007.01374.x
Web of Science® Google Scholar
Brown C., Burslem D.F.R.P., Illian J.B., Bao L., Brockelman W., Cao M., Chang L.W., Dattaraja H.S., Davies S., Gunatilleke C.V.S., Gunatilleke I.A.U.N., Huang J., Kassim A.R., Lafrankie J.V., Lian J., Lin L., Ma K., Mi X., Nathalang A., Noor S., Ong P., Sukumar R., Su S.H., Sun I.F., Suresh H.S., Tan S., Thompson J., Uriarte M., Valencia R., Yap S.L., Ye W., Law R. (2013) Multispecies coexistence of trees in tropical forests: spatial signals of topographic niche differentiation increase with environmental heterogeneity. Proceedings of the Royal Society B, 280, 20130502.
10.1098/rspb.2013.0502
CAS PubMed Web of Science® Google Scholar
Chuyong G.B., Kenfack D., Harms K.E., Thomas D.W., Condit R., Comita L.S. (2011) Habitat specificity and diversity of tree species in an African wet tropical forest. Plant Ecology, 212, 1363–1374.
10.1007/s11258-011-9912-4
Web of Science® Google Scholar
Comita L.S., Engelbrecht B.M.J. (2009) Seasonal and spatial variation in water availability drive habitat associations in a tropical forest. Ecology, 90, 2755–2765.
10.1890/08-1482.1
PubMed Web of Science® Google Scholar
Comita L.S., Condit R., Hubbell S.P. (2007) Developmental changes in habitat associations of tropical trees. Journal of Ecology, 95, 482–492.
10.1111/j.1365-2745.2007.01229.x
Web of Science® Google Scholar
Engelbrecht B.M., Comita L.S., Condit R., Kursar T.A., Tyree M.T., Turner B.L., Hubbell S.P. (2007) Drought sensitivity shapes species distribution patterns in tropical forests. Nature, 447, 80–82.
10.1038/nature05747
CAS PubMed Web of Science® Google Scholar
Gelman A., Carlin J.B., Stern H.S., Rubin D.B. (2004) Bayesian data analysis, 2nd edn. Chapman & Hall, New York, USA.
Google Scholar
Gunatilleke C.V.S., Gunatilleke I.A.U.N., Esufali S., Harms K.E., Ashton P.M.S., Burslem I.A.U.N., Ashton P.S. (2006) Species–habitat associations in a Sri Lankan dipterocarp forest. Journal of Tropical Ecology, 22, 371–384.
10.1017/S0266467406003282
Web of Science® Google Scholar
Harms K.E., Condit R., Hubbell S.P., Foster R.B. (2001) Habitat associations of trees and shrubs in a 50-ha Neotropical forest plot. Journal of Ecology, 89, 947–959.
10.1111/j.1365-2745.2001.00615.x
Web of Science® Google Scholar
Hu Y.H., Sha L.Q., Blanchet F.G., Zhang J.L., Tang Y., Lan G.Y., Cao M. (2012) Dominant species and dispersal limitation regulate tree species distributions in a 20-ha plot in Xishuangbanna, southwest China. Oikos, 121, 952–960.
10.1111/j.1600-0706.2011.19831.x
Web of Science® Google Scholar
Hubbell S.P. (2001) The unified neutral theory of biodiversity and biogeography. Princeton University Press, Princeton, NJ, USA.
PubMed Google Scholar
Iida S. (1996) Quantitative analysis of acorn transportation by rodents using magnetic locator. Plant Ecology, 124, 39–43.
10.1007/BF00045142
Web of Science® Google Scholar
Ishida A., Yazaki K., Hoe A.L. (2005) Ontogenetic transition of leaf physiology and anatomy from seedlings to mature trees of a rain forest pioneer tree, Macaranga gigantea. Tree Physiology, 25, 513–522.
10.1093/treephys/25.5.513
PubMed Web of Science® Google Scholar
Itoh A., Ohkubo T., Nanami S., Tan S., Yamakura T. (2010) Comparison of statistical tests for habitat associations in tropical forests: a case study of sympatric dipterocarp trees in a Bornean forest. Forest Ecology and Management, 259, 323–332.
10.1016/j.foreco.2009.10.022
Web of Science® Google Scholar
John R., Dalling J.W., Harms K.E., Yavitt J.B., Stallard R.F., Mirabello M., Hubbell S.P., Valencia R., Navarrete H., Vallejo M., Foster R.B. (2007) Soil nutrients influence spatial distributions of tropical tree species. Proceedings of the National Academy of Sciences USA, 104, 864–869.
10.1073/pnas.0604666104
CAS PubMed Web of Science® Google Scholar
Kozlowski T.T., Kramer P.J., Pallardy S.G. (1991) The physiological ecology of woody plants. Academic Press, San Diego, CA, USA.
10.1093/treephys/8.2.213
Google Scholar
Lan G., Zhu H., Cao M., Hu Y., Wang H., Deng X., Zhou S., Cui J., Huang J., He Y., Liu L., Xu H., Song J. (2009) Spatial dispersion patterns of trees in a tropical rainforest in Xishuangbanna, southwest China. Ecological Research, 24, 1117–1124.
10.1007/s11284-009-0590-9
Web of Science® Google Scholar
Latimer A.M., Wu S., Gelfand A.E., Silander J.A. Jr (2006) Building statistical models to analyze species distributions. Ecological Application, 16, 33–50.
10.1890/04-0609
PubMed Web of Science® Google Scholar
Laurance S.G.W., Laurance W.F., Andrade A., Fearnside P.M., Harms K.E., Vicentini A., Luizão R.C.C. (2010) Influence of soils and topography on Amazonian tree diversity: a landscape-scale study. Journal of Vegetation Science, 21, 96–106.
10.1111/j.1654-1103.2009.01122.x
Web of Science® Google Scholar
Li L., Huang Z., Ye W., Cao H., Wei S., Wang Z., Lian J., Sun I.F., Ma K., He F. (2009) Spatial distributions of tree species in a subtropical forest of China. Oikos, 118, 495–502.
10.1111/j.1600-0706.2009.16753.x
Web of Science® Google Scholar
Lin Y.C., Chang L.W., Yang K.C., Wang H.H., Sun I.F. (2011) Point patterns of tree distribution determined by habitat heterogeneity and dispersal limitation. Oecologia, 165, 175–184.
10.1007/s00442-010-1718-x
PubMed Web of Science® Google Scholar
Lunn D., Spiegelhalter D., Thomas A., Best N. (2009) The BUGS project: evolution, critique and future directions. Statistics in Medicine, 28, 3049–3067.
10.1002/sim.3680
CAS PubMed Web of Science® Google Scholar
Masaki T. (2002) Structure and dynamics. In: T. Nakashizuka, Y. Matsumoto (Eds), Diversity and interaction in a temperate forest community, Ogawa Forest Reserve of Japan. Springer, Tokyo, Japan, pp 53–65.
10.1007/978-4-431-67879-3_6
Google Scholar
Masaki T., Nakashizuka T. (2002) Seedling demography of Swida controversa: effect of light and distance to conspecifics. Ecology, 83, 3497–3507.
10.1890/0012-9658(2002)083[3497:SDOSCE]2.0.CO;2
Web of Science® Google Scholar
Masaki T., Suzuki W., Niiyama K., Iida S., Tanaka H., Nakashizuka T. (1992) Community structure of a species-rich temperate forest, Ogawa Forest Reserve, central Japan. Vegetatio, 98, 97–111.
10.1007/BF00045549
Web of Science® Google Scholar
Masaki T., Kominami Y., Nakashizuka T. (1994) Spatial and seasonal patterns of seed dissemination of Cornus controversa in a temperate forest. Ecology, 75, 1903–1910.
10.2307/1941595
Web of Science® Google Scholar
Metz M.R. (2012) Does habitat specialization by seedlings contribute to the high diversity of a lowland rain forest? Journal of Ecology, 100, 969–979.
10.1111/j.1365-2745.2012.01972.x
Web of Science® Google Scholar
Mizoguchi Y., Morisawa T., Ohtani Y. (2002) Climate in Ogawa Forest Reserve. In: T. Nakashizuka, Y. Matsumoto (Eds), Diversity and interaction in a temperate forest community, Ogawa Forest Reserve. Springer, Tokyo, Japan, pp 11–18.
10.1007/978-4-431-67879-3_2
Web of Science® Google Scholar
Myers J.A., Chase J.M., Jiménez I., Jørgensen P.M., Araujo-Murakami A., Paniagua-Zambrana N., Seidel R. (2013) Beta-diversity in temperate and tropical forests reflects dissimilar mechanisms of community assembly. Ecology Letters, 16, 151–157.
10.1111/ele.12021
PubMed Web of Science® Google Scholar
Nakashizuka T. (2001) Species coexistence in temperate, mixed deciduous forests. Trends in Ecology & Evolution, 16, 205–210.
10.1016/S0169-5347(01)02117-6
CAS PubMed Web of Science® Google Scholar
Nakashizuka T., Iida S., Masaki T., Shibata M., Tanaka H. (1995) Evaluating increased fitness through dispersal: a comparative study on tree populations in a temperate forest, Japan. Ecoscience, 2, 245–251.
10.1080/11956860.1995.11682290
Web of Science® Google Scholar
Oki S., Akiyoshi T., Hoshino D., Shibata M., Matsushita M., Hoshizaki K. (2013) Interactive effect of canopy and fluvial disturbances on sapling community structure and species diversity in a montane riparian forest. Ecoscience, 20, 194–203.
10.2980/20-2-3609
Web of Science® Google Scholar
Ota T., Masaki T., Sugita H., Kanazashi T., Abe H. (2012) Properties of stumps that promote the growth and survival of Japanese cedar saplings in a natural old-growth forest. Canadian Journal of Forest Research, 42, 1976–1982.
10.1139/x2012-141
Web of Science® Google Scholar
Palmiotto P.A., Davies S.J., Vogt K.A., Ashton M.S., Vogt D.J., Ashton P.S. (2004) Soil-related habitat specialization in dipterocarp rain forest tree species in Borneo. Journal of Ecology, 92, 609–623.
10.1111/j.0022-0477.2004.00894.x
Web of Science® Google Scholar
Pearse W.D., Jones F.A., Purvis A. (2013) Barro Colorado Island's phylogenetic assemblage structure across fine spatial scales and among clades of different ages. Ecology, 94, 2861–2872.
10.1890/12-1676.1
PubMed Web of Science® Google Scholar
Rosindell J., Hubbell S.P., Etienne R.S. (2011) The unified neutral theory of biodiversity and biogeography at age ten. Science, 26, 340–348.
Web of Science® Google Scholar
Rüger N., Huth A., Hubbell S.P., Condit R. (2009) Response of recruitment to light availability across a tropical lowland rain forest community. Journal of Ecology, 97, 1360–1368.
10.1111/j.1365-2745.2009.01552.x
Web of Science® Google Scholar
Satake Y., Hara H., Watari S., Tominari T. (1999) Wild flowers of Japan: woody plants I & II. Heibonsha, Tokyo.
Google Scholar
Schupp E.W., Milleron T., Russo S.E. (2002) Dissemination limitation and the origin and maintenance of species-rich tropical forests. In: D.J. Levey, W.R. Silva, M. Galetti (Eds), Seed dispersal and frugivory: Ecology, evolution and conservation. CABI Publishing, New York, pp 19–33.
10.1079/9780851995250.0019
Google Scholar
Seidler T.G., Plotkin J.B. (2006) Seed dispersal and spatial pattern in tropical trees. PLoS Biology, 4, 2132–2137.
10.1371/journal.pbio.0040344
CAS Web of Science® Google Scholar
Shen G., He F., Waagepetersen R., Sun I.F., Hao Z., Chen Z.S., Yu M. (2013) Quantifying effects of habitat heterogeneity and other clustering processes on spatial distributions of tree species. Ecology, 94, 2436–2443.
10.1890/12-1983.1
CAS PubMed Web of Science® Google Scholar
Shibata M., Nakashizuka T. (1995) Seed and seedling demography of four co-occurring Carpinus species in a temperate deciduous forest. Ecology, 76, 1099–1108.
10.2307/1940918
Web of Science® Google Scholar
Shibata M., Tanaka H. (2002) Reproductive traits of trees in OFR. In: T. Nakashizuka, Y. Matsumoto (Eds), Diversity and interaction in a temperate forest community, Ogawa Forest Reserve of Japan. Springer, Tokyo, Japan, pp 95–108.
10.1007/978-4-431-67879-3_9
Google Scholar
Shibata M., Tanaka H., Iida S., Abe S., Masaki T., Niiyama K., Nakashizuka T. (2002) Synchronized annual seed production by 16 principal tree species in a temperate deciduous forest, Japan. Ecology, 83, 1727–1742.
10.1890/0012-9658(2002)083[1727:SASPBP]2.0.CO;2
Web of Science® Google Scholar
Shibata M., Masaki T., Tanaka H., Niiyama K., Iida S., Abe S., Nakashizuka T. (2010) Effects of abiotic and biotic factors and stochasticity on tree regeneration in a temperate forest community. Ecoscience, 17, 137–145.
10.2980/17-2-3163
Web of Science® Google Scholar
Tanaka H., Shibata M., Nakashizuka T. (1997) A mechanistic approach for evaluating the role of wind dispersal in tree population dynamics. Journal of Sustainable Forestry, 6, 155–174.
10.1300/J091v06n01_10
Google Scholar
de Toledo J.J., Magnusson W.E., Castilho C.V., Nascimento H.E.M. (2011) How much variation in tree mortality is predicted by soil and topography in Central Amazonia? Forest Ecology and Management, 262, 331–338.
10.1016/j.foreco.2011.03.039
Web of Science® Google Scholar
Vieilledent G., Courbaud B., Kunstler G., Dhôte J.F., Clark J.S. (2010) Individual variability in tree allometry determines light resource allocation in forest ecosystems: a hierarchical Bayesian approach. Oecologia, 163, 759–773.
10.1007/s00442-010-1581-9
PubMed Web of Science® Google Scholar
Wang S., Tang Z., Qiao X., Shen Z., Wang X., Zheng C., Fang J. (2012) The influence of species pools and local processes on the community structure: a test case with woody plant communities in China's mountains. Ecography, 35, 1168–1175.
10.1111/j.1600-0587.2012.00045.x
Web of Science® Google Scholar
Webb C.O., Peart D.R. (2000) Habitat associations of trees and seedlings in a Bornean rain forest. Journal of Ecology, 88, 464–478.
10.1046/j.1365-2745.2000.00462.x
CAS PubMed Web of Science® Google Scholar
Whittaker R.H., Fairbanks C.W. (1958) A Study of plankton copepod communities in the Columbia Basin, southeastern Washington. Ecology, 39, 46–65.
10.2307/1929966
Web of Science® Google Scholar
Yoshinaga S., Takahashi M., Aizawa S. (2002) Landforms and soil characteristics in Ogawa Forest Reserve. In: T. Nakashizuka, Y. Matsumoto (Eds), Diversity and interaction in a temperate forest community, Ogawa Forest Reserve. Springer, Tokyo, Japan, pp 19–26.
10.1007/978-4-431-67879-3_3
Google Scholar

Citing Literature

Volume17, Issue2

March 2015

Pages 449-458

Filename	Description
plb12253-sup-0001-AppendixS1.pdfapplication/PDF, 3.8 KB	Appendix S1. Model parameters (a, b, and c) estimated on the basis of the Bayesian framework for the adult stage.
plb12253-sup-0002-AppendixS2.pdfapplication/PDF, 3.8 KB	Appendix S2. Model parameters (a, b, and c) estimated on the basis of the Bayesian framework for the juvenile stage.
plb12253-sup-0003-AppendixS3.pdfapplication/PDF, 3.4 KB	Appendix S3. Model parameters (a, b, and c) estimated on the basis of the Bayesian framework for the sapling stage.
plb12253-sup-0004-AppendixS4.pdfapplication/PDF, 3.5 KB	Appendix S4. Model parameters (a, b, and c) estimated on the basis of the Bayesian framework for the seedling stage.

Heterogeneity in soil water and light environments and dispersal limitation: what facilitates tree species coexistence in a temperate forest?

Abstract

Introduction