SILVA FENNICA Silva Fennica 46(4) research articles www.metla.fi/silvafennica · ISSN 0037-5330 The Finnish Society of Forest Science · The Finnish Forest Research Institute Species-Habitat Associations in a Northern Temperate Forest in China Chunyu Zhang, Yazhou Zhao, Xiuhai Zhao and Klaus von Gadow Zhang, C., Zhao, Y., Zhao, X. & Gadow, K. v. 2012. Species-habitat associations in a northern temperate forest in China. Silva Fennica 46(4): 501–519. This contribution identiies species-habitat associations in a temperate forest in north-eastern China, based on the assumption that habitats are spatially autocorrelated and species are spatially aggregated due to limited seed dispersal. The empirical observations were obtained in a large permanent experimental area covering 660 × 320 m. The experimental area was subdivided into four habitat types using multivariate regression tree (MRT) analysis. According to an indicator species analysis, 38 of the 47 studied species were found to be signiicant indicators of the MRT habitat types. The relationships between species richness and topographic variables were found to be scale-dependent, while the great majority of the species shows distinct habitat-dependence. There are 188 potential species-habitat associations, and 114 of these were signiicantly positive or negative based on habitat randomization. We identiied 139 signiicant associations using a species randomization. A habitat is not a closed system it may be both, either a sink or a source. Therefore, additional to the randomization, the Poisson Cluster Model (PCM) was applied. PCM considers the spatial autocorrelation of species and habitats, and thus appears to be more realistic than the traditional randomization processes. It identiied only 37 associations that were signiicant. In conclusion, the deviation from the random process, i.e. the high degree of species spatial mingling may be explained by persistent immigration across habitats. Keywords dispersal limitations, indicator species, spatial autocorrelation, species richness, topographic differentiation Addresses Zhang and X. Zhao, Key Laboratory for Forest Resources & Ecosystem Processes of Beijing, Beijing Forestry University, Beijing 100083, China; Y. Zhao, Department of Landscape Architecture, School of Architecture, Tsinghua University, Beijing 100084, China; Gadow, Faculty of Forestry and Forest Ecology, Georg-August-University Göttingen, Büsgenweg 5, D-37077 Göttingen, Germany E-mail (Zhang) zcy_0520@163.com Received 1 March 2012 Revised 10 September 2012 Accepted 12 September 2012 Available at http://www.metla.i/silvafennica/full/sf46/sf464501.pdf 501 Silva Fennica 46(4), 2012 1 Introduction Spatial distributions of forest trees often exhibit patterns correlating with the variation of soil chemistry or topography in tropical forests (Harms et al. 2001, Itoh et al. 2003, Russo et al. 2005, Yamada et al. 2006, 2007, John et al. 2007) and in temperate forests (Zhang et al. 2009, 2010). This suggests that the ecological organization caused by niche differentiation may be important for maintaining species diversity and species coexistence. If environmentally biased spatial distributions principally result from niche differentiation, plant species should show particular habitat preferences. They would preferably occur in localities where they have competitive advantages, although spatial autocorrelation cannot be ignored when considering species-habitat associations (Legendre and Legendre 1998). A common assumption of most traditional statistical methods for species-habitat associations is that individuals are independently distributed with respect to conspeciics (Condit 1996, Clark et al. 1998, Plotkin et al. 2000). But the independence assumption is often violated by the patterns produced by short-distance dispersal and recruitment processes. The limited dispersal of seeds and short-distance recruitments would contribute to the spatial autocorrelation of species distributions (Condit 1996, Clark et al. 1998, Plotkin et al. 2000). Thus, the assumptions of independence of sample units are often violated by the pattern caused by the dispersal limitations and dependent recruitment processes of trees and shrubs. To test the contribution of habitat specialization to species coexistence, the relationships between the species spatial distribution and environmental factors need to be studied. In the northern temperate forests of China, the distribution patterns of individuals within a plant population generally tend to be more aggregated than random (Zhang et al. 2009). Furthermore, signiicant correlations between species and soil nutrients were found in these forests (Zhang et al. 2010). This suggests that habitat preferences are potentially important in explaining the spatial variation in tree communities. Nakashizuka (2001) maintained that habitat specialization remains a prominent hypothesis to explain the species coexistence in a temperate forest community. 502 research articles The objective of this study is to analyse some of the mechanisms generating differences in species abundance across habitat types. The fully mapped experimental area of 21 ha is located in a multi-species forest ecosystem in North-Eastern China. We assume that habitats are spatially autocorrelated and that the range of seed dispersal is limited. Based on previous ield observations, we expect substantial species-habitat associations in the experimental area. Speciic objectives of this study are (1) to determine possible scale-dependent associations between species richness and topographic variables; (2) to identify indicator species for a particular habitat and (3) to examine possible associations of trees and shrubs with distinct habitats. We will also discuss the effect of habitat differentiation in maintaining a high species diversity in the Jiaohe temperate forest. 2 Materials and Methods 2.1 Study Area This study is based on a dataset obtained in a large permanent ield plot. The experimental site is located at (43°57.897´ ~ 43°58.263´N, 127°42.789´ ~ 127°43.310´E) in the Jiaohe Management Bureau of the Forest Experimental Zone in Jilin province, in Northeastern China. The research plot measures 320 m × 660 m and covers an area of 21.12 hectares. The altitude in the experimental area ranges from 425.3 m to 525.8 m above sea level. In the study area, the average annual temperature is 3.8 °C. And the hottest month is July with an average day temperature of 21.7 °C. The coldest month is January with an average day temperature of –18.6 °C. The average annual precipitation is 695.9 mm. The soil is a brown forest soil with a rootable depth ranging between 20 and 100 cm. The last recorded tree felling activities took place 50 years ago. The vegetation type represents a mixed broadleaf-conifer forest with 63 species (including three climber species). Altogether 53 916 individual trees with a breast height diameter (dbh) exceeding 1cm were tagged and mapped, and their species was identiied. The dbh value was measured at 1.3 m above ground level. Among the 63 woody species in Zhang et al. the research plot there are 47 abundant species, comprising at least one individual/ha. The species were identiied according to the records in the Chinese Virtual Herbarium (see http://www. cvh.org.cn/cms/). The dominant tree species are Ulmus davidiana var. japonica (Rehder) Nakai, Pinus koraiensis Siebold & Zucc., Juglans mandshurica Maxim., Tilia mandschurica Rupr.er Maxim., Carpinus cordata Bl., Acer mono Maxim., Fraxinus mandshurica Rupr., Tilia amurense Rupr. and Ulmus laciniata (Trautv.) Mayr. The top ive species in stem density are Acer mandshuricum Maxim., Syringa reticulata var. amurensis (Rupr.), Ulmus davidiana var. japonica, Carpinus cordata and Acer mono, respectively. The total basal area of dominant tree species and stem density of the top ive species are shown in Appendix 1 and 2. 2.2 Relationships between Species Richness and Topography The relative heights at the four corner nodes of each 20 m × 20 m cell were used to develop a variogram model of the entire research area. To examine the association between species richness and topographic variables at different scales, the altitude values were estimated for different cell sizes (5 m × 5 m, 10 m × 10 m, 30 m × 30 m, 40 m × 40 m and 50 m × 50 m) using block kriging (Legendre and Legendre 1998). Species richness in each cell was counted at each of these six different scales. Spearman rank correlation coeficients were calculated to test the relationships between species richness and the topographic variables at each of the six spatial scales. When the plot was subdivided into equally dimensioned cells, the intersections of grid lines were called “nodes”. The relative height differences among the nodes and the elevation of the starting node were measured. Thus it was possible to calculate the elevation of other nodes according to the height difference among nodes and the measured elevation of the starting node. The present study mainly focuses on the results of the 20 m × 20 m cell analysis. The elevation of a particular cell was calculated as the mean of the elevations of its four corner nodes. The cell slope for each of the ive cell sizes was estimated as the mean angular deviation from Species-Habitat Associations in a Northern Temperate Forest in China the horizontal plane of each of the four triangular planes which were formed by connecting three of its adjacent corners (Harms et al. 2001). The convexity of a cell was calculated as the elevation of the focal cell minus the mean elevation of the eight surrounding cells (cf. Yamakura et al. 1995). For the edge cells, convexity was taken as the elevation of the center point minus the mean of the four corners. Positive and negative convexity values respectively indicate convex (ridge) and concave (valley) land surfaces. The aspect of a cell can be obtained from the average angle of the four triangular planes that deviate from the north direction. Four maps show the spatial pattern of the four topographic variables using 20 m × 20 m cells (Fig. 1). Each cell shows the altitude (ranging from 425.3 m to 525.8 m above sea level with 100.5 m difference in altitude between the highest and lowest cells), the convexity (ranging from –6.6 m to 4.7 m), the slope (ranging from 1.4° to 39.2°) and the aspect (ranging from 41.9° to 329.7°). 2.3 Habitat Classiication and Indicator Species Multivariate regression tree (MRT) analysis was performed, following De’ath (2002), to classify habitat types according to topographic conditions and species composition. Distance-based MRT is a relatively new statistical technique that can be used to describe relationships between multispecies data and environmental characteristics. The dissimilarities used in distance-based MRT are calculated by Euclidean distances. Thus, one obtains clusters of sites by repeated splitting of the data, which are chosen to minimize the dissimilarity of sites within clusters. Habitats were delineated using threshold values for the topographic variables, while the species data was used to ind the best thresholds. Indicator values and associated probabilities were computed to identify the statistically signiicant indicator species in a speciic habitat type. Indicator species analysis combines a species relative abundance with its relative frequency of occurrence in the various groups of sites. Indicator value (di,c) of species was calculated as the product of the relative frequency (fi,c) and relative average abundance (ai,c) in clusters. 503 Silva Fennica 46(4), 2012 research articles a) Altitude b) Slope c) Convexity d) Aspect Fig. 1. Maps depicting four topographic variables at the scale of 20 × 20 m cells. a) Altitude from 425.3 m (white) to 525.8 m (black) above sea level; b) Convexity from –6.6 m (white) to 4.7 m (black); c) Slope from 1.4° (white) to 39.2° (black); d) Aspect from 41.9° (white) to 329.7° (black). For cluster c in set K: fi ,c = ai ,c = ∑ j ∈c pi , j nc (∑ j ∈cxi , j ) / nc K ∑ k =1((∑ j ∈ k xi , j ) / nk ) (1) di ,c = fi ,c × ai ,c where pi,j = presence/absence (1/0) of species i in sample j xi,j = abundance of species i in sample j nc = number of samples in cluster c The species indicator value is a maximum when all individuals of a species are found in a single group of sites and when the species occurs in all sites of that group. The statistical signiicance of the species indicator values is evaluated using a randomization procedure (Dufrêne and Legendre 1997). All calculations were done using the R statistical software (R Development Core Team 504 2010). MRT analysis was implemented using the “mvpart” library of R (De’ath 2010). Indicator species analysis was performed using R’s “labdsv” library (Roberts 2010). Rare species with less than 50 individuals within the 21.12 ha study area were excluded from the species–habitat association analysis. Trees at different life stages may have different ecological habitat preferences, as reported by Webb and Peart (2000), Comita et al. (2007) and Lai et al. (2009). In this study, we only focus on the relationship between species types and habitat types. Thus we assume that all individuals of a given species respond similarly to a speciic habitat type, regardless of their stage of development. 2.4 Testing Species-Habitat Associations Most methods for testing species-habitat associations assume that trees and shrubs are independently distributed with regard to conspeciic individuals. However, the assumption of inde- Zhang et al. pendence is often violated because of the limited range of seed dispersal and recruitment (Harms et al. 2001). In this study, we are using three methods to test this assumption. This study mainly focuses on the results based on a Poisson Cluster Model (PCM) analysis. 2.4.1 Randomized Habitat Maps This section presents the case where randomized habitat processes were modeled with dispersal limitations of species but no spatially autocorrelated habitat features. Random habitat processes were used to simulate habitat maps that were not autocorrelated. The true species map was used to indicate the dispersal limitations of species. Spatial dependency within species was evaluated by generating a series of random habitat maps. In these simulated maps the non-overlapping areas are identical in extent to the four habitat types of the true maps. Each simulated map included exactly 248 cells (20 m × 20 m) of habitat type 1, 85 cells of habitat type 2, 52 of habitat type 3 and 143 of habitat type 4. The habitat types were randomly permuted among the 528 cells. To assess the species associations, each simulated habitat map was matched with the true tree distribution map. Then the relative stem density of the focal species in each habitat type was counted. This procedure was repeated 10 000 times. Thus, the frequency distributions of the stem density estimates for each species in each habitat type were obtained from 10 000 simulated habitat maps. We then compared the relative stem density of a particular species calculated from the true habitat map with that from the simulated habitat maps. If the proportion of instances where {stem density of simulated habitat maps < stem density of true habitat map} was greater than 0.975, we assumed that the given species was positively associated with a particular habitat at the 0.05 level in a two-tailed test. Alternatively, if the proportion of instances where {stem density of simulated habitat maps > stem density of true habitat map} was greater than 0.975, we assumed that the given species was negatively associated with a particular habitat at the 0.05 level in the two-tailed test. Species-Habitat Associations in a Northern Temperate Forest in China 2.4.2 Randomized Species Maps This section presents the case where randomized species processes were modeled with no dispersal limitations of species but spatially autocorrelated habitat features. A complete spatial randomness (CSR) process was used to simulate no dispersal limitations of species. The true habitat map was used to indicate spatially autocorrelated habitats. Spatial dependency within habitat types was tested by generating a series of maps with random locations for each species, using the CSR process in the simulated species maps. Matching the true habitat map with the simulated species maps, we calculate the stem density of each species in each habitat type. This procedure was repeated 10,000 times to establish the frequency distribution of the estimated stem density. Signiicant deviations of the observed density values from the expected ones were assessed at 0.05 levels using a twotailed test. 2.4.3 Poisson Cluster Process Using poisson cluster processes, dispersal limitations of species and spatial autocorrelation of habitat maps were modeled. The Poisson Cluster Model (PCM) was used to simulate the dispersal limitations of species. The true habitat map was used to indicate spatially autocorrelated habitats. In a subsequent analysis, using Ripley’s K function, it was found that 47 species (Appendix 3) were not randomly distributed, but signiicantly aggregated. By choosing the PCM approach, it was implied that clusters arise from local propagation. The PCM models aggregation caused by local seed dispersal or gap recruitment. Thus, it was decided to simulate the spatial distribution of each of these species using the Poisson cluster process. The observed spatial aggregation is then used to develop expected species–habitat associations. Subsequently, we model the spatial aggregation of the species distributions using the PCM with Ripley’s K value as guiding parameter, as recommend by Diggle (1983), Plotkin et al. (2000) and John et al. (2007). The PCM model can be used to capture small-scale spatial aggregation in species distributions that are due to aggregated dispersal (Potts et al., 2004). The deinition and parameter 505 Silva Fennica 46(4), 2012 research articles estimation of the Poisson cluster process follows Plotkin et al. (2000). 3 Results aspect (east has low values, west has high values) is negative and highly signiicant at ine scales. This means that the number of species is increasing when the aspect changes at close range from west to east. 3.1 Species Richness and Topography 3.2 Habitat Types and Indicator Species The correlation between species richness and topography shows a distinct scale-dependence (Table 1). Highly signiicant or signiicant negative associations between species richness and altitude were found at iner scales, ranging from 5 m to 30 m, but vanished at higher scales. This result may not be very surprising, considering that fewer species will be able to establish themselves on the rather exposed higher altitudes. The correlations are very weak although signiicant. It means that there was almost no linear relationship between the studied habitat characteristics and species richness. Thus, some other factors than the studied ones may explain better the species richness or the species richness is affected by many factors each having a low impact alone. Convexity, which expresses the relative altitudinal difference between the focal cell and its surrounding neighbors, does not seem to affect species richness. Signiicantly positive associations between species richness and slope were found at ine and coarse scales. The number of species increases with increasing terrain steepness, but only at very close range, which means that the increasing richness is found at the transitions from the hill bottom (or the plateau) to the adjacent slope. This is plausible, but we are unable to provide an explanation for the signiicant association at the 50 m scale. The association between species richness and The experimental area was subdivided into four habitat types using a MRT method. The tree size was selected using a cross-validation procedure, with the four-leaf tree clearly identiied as having the smallest cross-validated relative error (CV error = 0.635; see Appendix 4). The geographical proile of the four habitat types is shown (Fig. 2). Each 20 m × 20 m cell is assigned to one speciic habitat, as indicated by the numbers 1, 2, 3 and 4. The topographic attributes of the four habitat types are presented in Table 2. Habitat type 1, which occurs in 248 cells, occupies the lowaltitudes. It is separated from habitat type 2, 3 and 4 by the lower altitudinal boundary of 453.6 m. Habitat type 2 (n = 85) occupies the east-facing cells with aspects less than 187°, in the lower right and upper left of the plot. Habitat type 3 (n = 52) and habitat type 4 (n = 143) are found in the westward-facing cells with aspects exceeding 187°, in the upper right and upper left of the plot. Habitat type 3 occupies altitudes below, habitat type 4 altitudes above 465.7 m. According to the MRT analysis, 38 of the 47 species occurring in the experimental area, were found to be signiicant indicators of the habitat types. Fifteen of these, listed in Appendix 5, are signiicant indicators of habitat type 1, ive of habitat type 2, seven of habitat type 3, and eleven of habitat type 4. Table 1. Correlation coeficients showing the degree of correlation between species richness and four topographic variables at six different spatial scales Scales 5 m×5 m 10 m × 10 m 20 m × 20 m 30 m × 30 m 40 m × 40 m 50 m × 50 m 506 Altitude –0.1263*** –0.2537*** –0.2896*** –0.1767** –0.1341 0.0565 Convexity Slope 0.0056 –0.0163 –0.0934* –0.0596 –0.2208* –0.1869 0.0332** –0.0084 –0.0055 0.0290 0.1966* 0.3020** Aspect –0.0453*** –0.0690* –0.0542 0.0715 0.1483 0.2518* Zhang et al. Species-Habitat Associations in a Northern Temperate Forest in China 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 3 3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 3 3 3 3 1 1 1 1 1 1 1 1 1 1 1 2 3 3 3 3 3 3 1 1 1 1 1 1 1 1 1 3 3 3 3 3 3 3 3 3 1 1 1 1 1 1 1 3 3 3 3 3 3 3 3 3 3 1 1 1 1 1 1 4 4 4 4 4 4 3 3 3 3 1 1 1 1 1 1 4 4 4 4 4 4 4 4 4 3 3 1 1 1 1 1 4 4 4 4 4 4 4 4 4 4 3 1 1 1 1 1 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 44 4 3753 1 3 1 1 1 1 1 1 4 5410 4 4 4 4 4 4 50 40 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 3 4 4 490 4 14730 3 4 1 1 1 4615 1 1 1 1 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 45025 2 2 2 485 2 2 2 2 1 1 1 5 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 1 51 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 5 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 49 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1 0 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1 48 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 0 4 4 4840 2 4 2 2 4 4 2 2 475 2 2 2 4 4 2 2 2 2 2 470 4 4 2 2 2 2 2 3 465 2 2 2 2 2 2 3 2 2 2 2 2 2 3 460 2 2 2 2 2 1 3 455 2 2 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 445 1 1 10 1 1 1 1 1 1 414 1 1 14301 1 1 1 1 1 1 1 435 1 1 1 1 1 1 1 45 2 4 4 4 3 3 3 3 1 1 1 1 1 1 1 1 4 4 4 4 4 4 4 4 4 4 2 2 2 2 1 1 Fig. 2. Map showing the distribution of the four habitat types at the 20 m × 20 m scale. The lines show the elevation contours at 5 m intervals. Fig. 3. Map showing the spatial distribution of two species with distinct habitat preference: Ulmus davidiana var. japonica (red dots) and Carpinus cordata (black dots). Background colors: green = habitat type 1; yellow = habitat type 2; red = habitat type 3; blue = habitat type 4. Table 2. Topographic attributes of the four habitat types. Habitat types Min Habitat 1 Habitat 2 Habitat 3 Habitat 4 426.3 453.7 454.0 465.7 Elevation Mean 439.9 466.6 459.6 489.4 Max 453.5 501.2 465.7 519.7 Min Convexity Mean Max Min Slope Mean Max Min Aspect Mean Max –6.6 –2.9 –1.4 –1.1 –0.5 0.5 0.0 0.6 1.5 4.3 2.9 4.7 1.4 5.3 5.5 3.9 10.3 15.7 13.5 15.9 38.3 39.2 33.1 37.3 41.9 140.5 188.9 187.6 187.5 164.1 231.2 239.9 329.7 185.9 275.7 286.3 Some species show distinct habitat-dependence, which becomes apparent when their spatial distributions are mapped. An example involving two species Ulmus davidiana var. japonica and Carpinus cordata, each of which has a distinct habitat preference, is presented (Fig. 3). Ulmus davidiana var. japonica correlates negatively with the elevation, the slope, and the convexity. Carpinus cordata, on the other hand, correlates positively with the elevation, the slope, the convexity 507 Silva Fennica 46(4), 2012 research articles Table 3. Signiicant habitat associations, based on three methods: randomized habitats, randomized species and PCM. The symbol “+” represents signiicant positive, “−”signiicant negative associations. Habitat association Habitat 1 + Habitat 2 + Habitat 3 + Habitat 4 + Total positive Habitat 1 − Habitat 2 − Habitat 3 − Habitat 4 − Total negative Grand Total Randomized habitats 16 2 0 24 42 18 24 25 5 72 114 Randomized species 21 10 13 20 64 21 27 7 20 75 139 PCM 16 4 5 10 35 1 1 0 0 2 37 and the aspect. Both correlations are signiicant. This explains the different species-habitat associations of the two species and their distribution in the experimental area. 3.3 Species-Habitat Associations Altogether 47 tree and shrub species, accounting for 99.7% of all individuals, were available to identify possible species-habitat associations within the research plot. Out of 188 potential species-habitat associations (four habitat types × 47 species), 114 were signiicantly positive or negative based on the habitat randomization procedure. The randomized-species model reveals 139 signiicant associations, whereas the PCM only shows 37 associations which are signiicant. The PCM process identiies comparatively few associations, but is much more realistic than the randomized-species and habitat randomization processes which do not consider the spatial autocorrelation of species and habitat, respectively. Based on the PCM model, 34 out of 47 species were signiicantly associated with one or more habitat types, only two of these were negative. The habitat randomization produced 39, the species randomization all 47 species with signiicant associations (Table 3; see also Appendix 6, 7 and 8). These associations were positive as well as negative. Again, habitat randomization and the randomized-species approach produced an inlated number of signiicant associations. The PCM model appears to give more realistic results. Table 4 shows four cross tabulations with the numbers of common species-habitat associations, based on the PCM model. Few species are signiicantly associated with more than one habitat. Most habitat pairs have zero common specieshabitat associations which are either signiicantly positive or negative. Signiicant associations across several habitat types are very rare. Table 4. Cross tabulations of common species-habitat associations based on the PCM. The symbol “+” indicates signiicant positive, “−” signiicant negative associations; N indicates no signiicant associations. Habitat 1+ Habitat 1N Habitat 1− Habitat 1+ Habitat 1N Habitat 1− Habitat 1+ Habitat 1N Habitat 1− 508 Habitat 2+ Habitat 2N Habitat 2− 0 4 1 16 26 0 0 1 0 Habitat 3+ Habitat 3N Habitat 3− 0 5 0 16 25 1 0 0 0 Habitat 4+ Habitat 4N Habitat 4− 0 10 0 16 20 1 0 0 0 Habitat 3+ Habitat 2+ Habitat 2N Habitat 2− 1 4 1 Habitat 4+ Habitat 2+ Habitat 2N Habitat 2− 1 9 0 Habitat 4+ Habitat 3+ Habitat 3N Habitat 3− 0 10 0 Habitat 3N 4 38 0 Habitat 4N 3 33 1 Habitat 4N 5 32 0 Habitat 3− 0 0 0 Habitat 4− 0 0 0 Habitat 4− 0 0 0 Zhang et al. 4 Discussion Habitat specialization is a prominent topic used to explain particular patterns of species coexistence in a forest community. Several studies in tropical forests could identify some species-habitat associations, but similar investigations seem to be lacking for temperate forest ecosystems. Accordingly, the purpose of this study is to broaden our understanding of the structuring forces in a temperate multi-species forest ecosystem with particular reference to species-habitat associations based on topographic features. We found that spatial autocorrelation of species and topographic variables may confuse the contribution of topographic variation to plant spatial patterns. The spatial distributions of most of the tree and shrub species in our experimental plot were distinctly aggregated. Species autocorrelation may be the result of seed-dispersal limitations of most tree species. For this reason, it was necessary to reduce the effects of autocorrelation, regarding species as well as habitat. 4.1 Habitat Types and Indicator Species The experimental plot was subdivided into four habitat types sharing topographic characteristics. Habitat type 1 has the largest area and occupies the low altitudes. Almost half of all the signiicant species-habitat associations were identiied in habitat type 1, when considering both habitat and species autocorrelation. The lower plateau is characterized by moist and wet soil conditions. The cells on the lower plateau and lower gentle slopes (habitat types 1 and 3) have two types of indicators: light and moisture-demanding pioneer and shade tolerant understorey species, including Betula spp., Euonymus spp., Acer spp., Malus baccata (L.) Borkh. and Syringa reticulata var. amurensis. The high lying areas and steep slopes (habitat types 2 and 4) are characterized by well-drained soil conditions. These habitat types have more climax indicator species, such as Pinus koraiensis, Acer mono, Tilia amurensis, and Quercus mongolicus Fisch. Some harvesting has taken place in the experimental area during the early 1960’s, but details are not available. Species-Habitat Associations in a Northern Temperate Forest in China Based on the results obtained with the PCM model, the percentage of species showing signiicant associations with habitat type 1 (45.9%) and habitat type 4 (27.0%) is consistent with the relative habitat areas in the two dominant habitat types. Habitat type 1 occupies 47 per cent of the total experimental area, habitat type 4 only 27 per cent. This almost exact match is likely to be accidental. However, the bigger areas naturally support a greater number of species which are candidates for signiicant species-habitat associations. Habitat types 2 and 3 represent the transitional zones between the low and high plateau. Again, the total percentage of habitat-associated species in both types together (27.0%) closely matches that of their total area (25.9%). Thus, the number of habitat-associated species in a speciic habitat may depend on the available habitat area. This is a species-area phenomenon. Due to the increasing number of trees, more habitat specialists can be found in habitat types that occupy larger areas. This is simply an effect of the species-area relationship. However, this effect does not contradict the occurrence of species specialization. The majority of the species in our experimental area showed signiicant positive or negative associations with speciic habitat types. The complex topography in the study area includes habitat types which are preferred by different tree and shrub species, resulting in a high proportion of distinct species-habitat associations. Our indicator species analysis showed that 38 out of 47 of the more common species were indicative of speciic habitat types. The indicative power was rather variable among the different indicator species. For example, in habitat type 1, Ulmus davidiana var. japonica showed the highest indicator value (0.62), while Betula platyphylla Suk. had the lowest indicator value (0.10). 4.2 Species-Habitat Associations Almost all the studies about species-habitat associations, based on data from large experimental areas, were conducted in tropical or subtropical forest ecosystems. Harms et al. (2001) identiied six habitat types in the tropical forest in Barro Colorado Island, where many species had strong species-habitat associations. Similar results were 509 Silva Fennica 46(4), 2012 obtained in other tropical forests (Harms et al. 2001, Itoh et al. 2003, Russo et al. 2005, Yamada et al. 2006, 2007, John et al. 2007). Our study appears to be among the irst ones based on data from a large experimental area in a temperate forest ecosystem. Altogether 34 species (72.3% of the total number) are associated with at least one habitat type, based on the conservative PCM model. Interestingly, these 34 species only produce 35 signiicantly positive species-habitat associations (out of 136 potential associations). Almost each of these species shows a unique positive association with a particular habitat. Ulmus davidiana var. japonica is an exception, being signiicantly associated with two habitat types simultaneously. The patterns of different species specializing in the different habitat types can be interpreted by the life history strategies (for example, shade tolerance, growth and mortality rates, etc). According to Nakashizuka et al. (1992) and Masaki et al. (1999), shade-tolerant species have recruitment rates that are almost equivalent to, or signiicantly larger than mortality rates. This means that their populations can be maintained or may even increase under natural conditions. Understory shade-tolerant species, such as Lonicera ruprechtiana Regel., Rhamnus davurica Pall., Euonymus macropterus Rupr., Viburnum sargenti Koehne, are neutrally associated with all four habitat types in the research plot. Some dominant canopy species, such as Tilia amurensis, Pinus koraiensis, Acer mono and Ulmus spp. show a strong adaptability toward habitat variations. These species did not indicate any particular habitat preference and had neutral associations with all habitat types. Species may differ from each other in their habitat preference. They are presumed to be capable of locating themselves in different positions along habitat gradients. For example, Fraxinus mandshurica specializes on habitat type 1, while Quercus mongolica clearly prefers habitat type 4. This corresponds to the ecological characteristics of the two species. Fraxinus mandshurica prefers fertile, moist and well drained sites, and usually occurs on the gentle slopes. The cells of habitat type 1 are located on these sites. Quercus mongolica prefers dry sites, and is usually found on the hilltop and lower ridges. That is where the cells of habitat type 4 are found. 510 research articles The distribution of some species relects regeneration following local disturbance. Betula platyphylla, for example, is a light-demanding and shade-intolerant tree species. The species is positively associated with habitat type 1 where heavy cutting disturbance occurred about 50 years ago. Viburnum burejaeticum Regel et Herd which is a distinctly light-demanding species, showed a particular preference for habitat type 3 and a negative association with habitat type 2. These habitat types differ with regard to aspect. The cells in habitat type 3 are located on sunny south-facing slopes, while most cells in habitat type 2 are found on rather more shaded south-westerly slopes. Habitat differentiation and spatial limitations due to species dispersal are two key factors that contribute to structuring forest communities. The habitat differentiation theory is based on the idea that there is a trade-off between growth/survival rates and resource availability (Kitajima 1994, Wright et al. 2003), producing environmentally dependent species preferences. This theory seems to represent the existing mainstream view (Clark et al. 1999, Svenning 2004, Harms et al. 2001, Wright 2002). Several studies have shown that forest community structure is governed by seed dispersal limitation and demographic stochasticity (Hubbell et al. 1999, Bell 2001, Hubbell 2001). These indings are supported by evidence of seed dispersal limitation (Hubbell et al. 1999, Dalling et al. 2002) and nonenvironmental spatial dependency in species distributions (Svenning 2001, Tuomisto et al. 2003). The understanding of species-habitat associations depends largely on the ability to reduce the effects of autocorrelation, regarding species as well as habitat. Spatial autocorrelation cannot be ignored when considering species-habitat associations (Legendre and Legendre 1998). Restricted seed dispersal and autocorrelated habitat types may create an artiicially inlated effect to specieshabitat associations. This study has shown that randomizing species (under the CSR model) and randomizing habitat types seriously overestimate the number of distinct habitat associations. For this reason, the more realistic and conservative PCM model was also used. The PCM approach identiied much less signiicant species-habitat associations (only 37, see Table 3) than the species and habitat randomization methods (114 and Zhang et al. 139 signiicant associations, respectively). Our results indicate that habitat differentiation and dispersal limitation are not mutually exclusive. Both contributed simultaneously to the maintenance of the particular distribution of the tree and shrub species in our experimental area. A species which is positively associated with a particular habitat can be expected to have a greater competitive advantage than other species that are neutrally or negatively associated with the same habitat. According to an earlier result presented by Harms et al. (2001) in a tropical forest, if negative associations were used to identify disappearing (“sink”) subpopulations within the research area, then the list of species neutrally or positively associated with a particular habitat type would be equal to the number capable of sustaining populations if the plot was composed of only that habitat type. According to our PCM tests, 32 out of 47 species were neutrally or positively associated with the four habitat types. Habitat types 1 and 2 were negatively associated (avoided) by Acer tegmentosum Maxim. and Viburnum burejaeticum respectively, based on the PCM model. This leaves 46 species which are neutrally or positively associated with either habitat type 1 or 2. No negative species-habitat associations were found in habitat type 3 and 4. Habitat types are not closed systems, and there are immigrations between them via seed dispersals on short- and/or long-distances. This suggests that species assemblages will be inluenced signiicantly by species input from the surrounding habitat types. Thus, the high proportion of species in these assemblages might be maintained by persistent immigration, not by habitat differentiation among coexisting species. Sink areas are subsidized by sources and even strong habitat preference of trees and shrubs does not seem to provide suficient support for the hypothesis that habitat differentiation maintains high species diversity. Future studies of species-habitat associations should therefore examine whether habitat, if not subsidized by immigration, would be able to support non-negative tree population growth. Species-Habitat Associations in a Northern Temperate Forest in China Acknowledgements This research is supported by the National Basic Research Program of China (973 Program; 2011CB403205), the Fundamental Research Funds for the Central Universities (HJ201019), the National Natural Science Foundation of China (31200315), and the 12th ive-year National Science and Technology plan of China (2012BAC01B03). References Bell, G. 2001. Neutral macroecology. Science 293: 2413–2418. Clark, D.B., Clark, D.A. & Read, J.M. 1998. Edaphic variation and the mesoscale distribution of tree species in a neotropical rain forest. Journal of Ecology 86: 101–112. — , Palmer, M.W. & Clark, D.A. 1999. Edaphic factors and the landscape-scale distributions of tropical rain forest trees. Ecology 80, 2662–2675. Comita, L.S., Condit, R. & Hubbell, S.P. 2007. Developmental changes in habitat associations of tropical trees. Journal of Ecology 95: 482–492. Condit, R., Hubbell, S.P. & Foster, R.B. 1996. Changes in tree species abundance in a neotropical forest: impact of climate change. Journal of Tropical Ecology 12: 231–256. Dalling, J.W., Muller-Landau H.C., Wright S.J. & Hubbell S.P. 2002. Role of dispersal in the recruitment limitation of neotropical pioneer species. Journal of Ecology 90: 714–727. De’ath, G. 2002. Multivariate regression trees: a new technique for modeling species–environment relationships. Ecology 83: 1105–1117. — 2010. mvpart: multivariate partitioning. R package version 1.3-1. Available at: hhttp://cran.r-project. org/. Diggle, P. 1983. Statistical analysis of spatial point patterns. Academic, London. Dufrêne, M. & Legendre P. 1997. Species assemblages and indicator species: the need for a lexible asymmetrical approach. Ecological Monographs 67: 345–366. 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 511 Silva Fennica 46(4), 2012 89: 947–959. Hubbell, S.P. 2001. The uniied neutral theory of biodiversity and biogeography. Monographs in population biology. Princeton University Press, Princeton, New Jersey, USA. — , Foster, R.B., O’Brien, S.T., Harms, K.E., Condit, R., Wechsler, B., Wright, S.J. & Loo & de Lao, S. 1999. Light-gap disturbances, recruitment limitation, and tree diversity in a Neotropical forest. Science 283: 554–557. Itoh, A., Yamakura, T., Ohkubo, T., Kanzaki, M., Palmiotto, P.A., LaFrankie, J.V., Ashton, P.S. & Lee, H.S. 2003. Importance of topography and soil texture in the spatial distribution of two sympatric dipterocarp trees in a Bornean rainforest. Ecological Research 18: 307–320. 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 inluence spatial distributions of tropical tree species. PNAS 104: 864–869. Kitajima, K. 1994. Relative importance of photosynthetic traits and allocation patterns as correlates of seedling shade tolerance of 13 tropical trees. Oecologia 98: 419–428. Lai, J.S., Mi, X.C., Ren, H.B. & Ma, K.P. 2009. Species-habitat associations change in a subtropical forest of China. Journal of Vegetation Science 20: 415–423. Legendre, P. & Legendre, L. 1998. Numerical ecology. 2nd English ed. Elsevier Science BV, Amsterdam. Masaki, T., Tanaka, H., Tanouchi, H., Sakai, T. & Nakashizuka, T. 1999. Structure, dynamics and disturbance regime of temperate broad-leaved forests in Japan. Journal of Vegetation Science 10: 805–814. Nakashizuka, T. 2001. Species coexistence in temperate, mixed deciduous forests. Trends in Ecology & Evolution 16: 205–210. Plotkin, J.B., Potts, M.D., Leslie, N., Manokaran, N., LaFrankie, J. & Ashton, P.S. 2000. Species-area curves, spatial aggregation, and habitat specialization in tropical forests. Journal of Theoretical Biology 207: 81–99. Potts, M.D., Davies, S.J., Bossert, W.H., Tan, S. & Supardi, M.N.N. 2004. Habitat heterogeneity and niche structure of trees in two tropical rain forests. Oecologia 139: 446–453. R Development Core Team. 2011. R: a language and environment for statistical computing. R Founda- 512 research articles tion for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0. Available at: http://www.Rproject.org/. Roberts, D.W. 2010. labdsv: Ordination and multivariate analysis for ecology. R package version 1.4-1. Available at: hhttp://cran.r-project.org/. Russo, S.E., Davies, S.J., King, D.A. & Tan, S. 2005. Soil-related performance variation and distributions of tree species in a Bornean rain forest. Journal of Ecology 93: 879–889. Svenning, J.-C., Kinner, D.A., Stallard, R.F., Engelbrecht, B.M.J. & Wright, S.J. 2004. Ecological determinism in plant community structure across a tropical forest landscape. Ecology 85: 2526–2538. — 2001. Environmental heterogeneity, recruitment limitation, and the mesoscale distribution of palms in a tropical montane rain forest (Maquipucuna, Ecuador). Journal of Tropical Ecology 17: 97–113. Tuomisto, H., Ruokolainen, K. & Yli-Halla, M. 2003. Dispersal, environment, and loristic variation of western Amazonian forests. Science 299: 241–244. Webb, C.O. & Peart, D.R. 2000. Habitat associations of trees and seedlings in a Bornean rain forest. Journal of Ecology 88: 464–478. Wright, S.J. 2002. Plant diversity in tropical forests: a review of mechanisms of species coexistence. Oecologia 130: 1–14. — , Muller-Landau, H.C., Condit, R. & Hubbell, S.P. 2003. Gap-dependent recruitment, realized vital rates, and size distributions of tropical trees. Ecology 84: 3174–3185. Yamada, T., Tomita, A., Itoh, A., Yamakura, T., Ohkubo, T., Kanzaki, M., Tan, S. & Ashton, P.S. 2006. Habitat associations of Sterculiaceae trees in a Bornean rain forest plot. Journal of Vegetation Science 17: 559–566. — , Zuidema, P.A., Itoh, A., Yamakura, T., Ohkubo, T., Kanzaki, M., Tan, S. & Ashton P.S. 2007. Strong habitat preference of a tropical rain forest tree does not imply large differences in population dynamics across habitats. Journal of Ecology 95: 332–342. Yamakura, T., Kanzaki, M., Ohkubo, T., Ogino, K., Chai, E.O.K., Lee, H.S. & Ashton, P.S. 1995. Topography of a large-scale research plot established within a tropical rain forest at Lambir, Sarawak. Tropics 5: 41–56. Zhang, C., Gao, L. & Zhao, X. 2009. Spatial structures in a secondary forest in Changbai Mountains Northeast China. Allgemeine Forst- und Jagdzeitung (AFJZ) 180: 45–55. Zhang et al. Species-Habitat Associations in a Northern Temperate Forest in China — , Zhao, X. & Gadow, K. v. 2010. Partitioning temperate plant community structure at different scales. Acta Oecologica 36: 306–313. Total of 36 references Appendix 1. The total basal area of dominant tree species in research plot. Tree species Ulmus davidiana var. japonica Pinus koraiensis Juglans mandshurica Tilia mandschurica Carpinus cordata Acer mono Fraxinus mandshurica Tilia amurense Ulmus laciniata Total basal area (m2/ha) 0.4309 0.6833 1.8705 0.6098 0.3813 1.2182 0.6023 0.8771 0.9912 Appendix 2. Stem density of the top ive species in research plot. Tree species Acer mandshuricum Syringa reticulata var. amurensis Ulmus davidiana var. japonica Carpinus cordata Acer mono Stem density (stems/ha) 150 412 141 393 240 513 Silva Fennica 46(4), 2012 research articles Appendix 3. List of the 47 species used in the analysis Species* Family Life forms Betula platyphylla Acer mandshuricum Syringa reticulata var. amurensis Acer ginnala Lonicera ruprechtiana Rhamnus davurica Euonymus macropterus Padus racemosa Ulmus davidiana var. japonica Acanthopanax senticosus Acer barbinerve Ulmus macrocarpa Berberis amurensis Lonicera maackii Philadelphus schrenkii Rhamnus schneideri var. manshurica Betula costata Pinus koraiensis Juglans mandshurica Fraxinus rhynchophylla Maackia amurensis Phellodendron amurense Viburnum sargenti Tilia mandshurica Ulmus laciniata Euonymus verrucosus Aralia elata Corylus mandshurica Quercus mongolica Acer trilorum Viburnum burejaeticum Carpinus cordata Acer tegmentosum Actinidia arguta Acer mono Abies holophylla Malus baccata Pyrus ussuriensis Populus davidiana Fraxinus mandshurica sorbus alnifolia Euonymus alatus Rhamnus ussuriensis Rhamnus parvifolia Rhamnus diamantiaca Lonicera praelorens Tilia amurensis Betulaceae Aceraceae Oleaceae Aceraceae Caprifoliaceae Rhamnaceae Celastraceae Rosaceae Ulmaceae Araliaceae Aceraceae Ulmaceae Berberidaceae Caprifoliaceae Saxifragaceae Rhamnaceae Betulaceae Pinaceae Juglandaceae Oleaceae Leguminosae Rutaceae Caprifoliaceae Tiliaceae Ulmaceae Celastraceae Araliaceae Betulaceae Fagaceae Aceraceae Caprifoliaceae Betulaceae Aceraceae Actinidiaceae Aceraceae Pinaceae Rosaceae Rosaceae Salicaceae Oleaceae Rosaceae Celastraceae Rhamnaceae Rhamnaceae Rhamnaceae Caprifoliaceae Tiliaceae Tree Tree Small tree/tree Shrub/small tree Shrub Shrub/small tree Shrub Tree Tree Shrub Small tree Shrub/tree Shrub Shrub Shrub Shrub Tree Tree Tree Tree Tree Tree Shrub Tree Tree Shrub Shrub Shrub Tree Tree Shrub Tree Tree Liana Tree Tree Tree Tree Tree Tree Tree Shrub Shrub Shrub Shrub Shrub Tree * The species were identiied using the records in the Chinese Virtual Herbarium (see http://www.cvh.org.cn/cms/) 514 Zhang et al. Species-Habitat Associations in a Northern Temperate Forest in China Appendix 4. Multivariate regression tree for the species composition data. Euclidean distance was used for splitting. Barplots show the multivariate species mean at each node, and the numbers of sites are shown in parentheses. 515 Silva Fennica 46(4), 2012 research articles Appendix 5. Results of indicator species analysis. 516 Species Habitat type Indicator value Ulmus davidiana var. japonica Acer trilorum Syringa reticulata var. mandshurica Lonicera maackii Malus baccata Fraxinus mandshurica Rhamnus davurica Padus racemosa Juglans mandshurica Euonymus macropterus Euonymus alatus Phellodendron amurense Acer ginnala Maackia amurensis Betula platyphylla Carpinus cordata Pinus koraiensis Actinidia arguta Acer tegmentosum Aralia elata Acer mandshuricum Philadelphus schrenkii Acer barbinerve Abies holophylla Rhamnus schneideri var. manshurica Euonymus macropterus Betula costata Acer mono Sorbus alnifolia Ulmus laciniata Tilia mandshurica Tilia amurensis Corylus mandshurica Fraxinus rhynchophylla Quercus mongolica Rhamnus parvifolia Lonicera praelorens Acanthopanax senticosus Habitat 1 Habitat 1 Habitat 1 Habitat 1 Habitat 1 Habitat 1 Habitat 1 Habitat 1 Habitat 1 Habitat 1 Habitat 1 Habitat 1 Habitat 1 Habitat 1 Habitat 1 Habitat 2 Habitat 2 Habitat 2 Habitat 2 Habitat 2 Habitat 3 Habitat 3 Habitat 3 Habitat 3 Habitat 3 Habitat 3 Habitat 3 Habitat 4 Habitat 4 Habitat 4 Habitat 4 Habitat 4 Habitat 4 Habitat 4 Habitat 4 Habitat 4 Habitat 4 Habitat 4 0.6178 0.6054 0.5329 0.4643 0.4508 0.3961 0.3574 0.353 0.3134 0.2823 0.2693 0.2429 0.1157 0.1147 0.1007 0.5743 0.2395 0.2172 0.1769 0.0403 0.2847 0.2149 0.1892 0.1184 0.117 0.1117 0.0961 0.3386 0.3122 0.31 0.2752 0.2721 0.2457 0.2106 0.1983 0.0663 0.0641 0.0453 Probability 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.006 0.001 0.001 0.023 0.001 0.01 0.001 0.001 0.026 0.016 0.002 0.033 0.005 0.014 0.047 0.002 0.003 0.001 0.001 0.001 0.025 0.008 0.001 0.026 0.004 0.032 0.05 Zhang et al. Species-Habitat Associations in a Northern Temperate Forest in China Appendix 6. Species-habitat associations using the PCM model. “p” and “n” denote to “positive” and “negative”, respectively. *, ** and *** indicate signiicance at p < 0.05, 0.01 and 0.001, respectively. Species Habitat 1 Betula platyphylla Acer mandshuricum Syringa reticulata var. amurensis Acer ginnala Lonicera ruprechtiana Rhamnus davurica Euonymus macropterus Padus racemosa Ulmus davidiana var. japonica Acanthopanax senticosus Acer barbinerve Ulmus macrocarpa Berberis amurensis Lonicera maackii Philadelphus schrenkii Rhamnus schneideri var. manshurica Betula costata Pinus koraiensis Juglans mandshurica Fraxinus rhynchophylla Maackia amurensis Phellodendron amurense Viburnum sargenti Tilia mandshurica Ulmus laciniata Euonymus verrucosus Aralia elata Corylus mandshurica Quercus mongolica Acer trilorum Viburnum burejaeticum Carpinus cordata Acer tegmentosum Actinidia arguta Acer mono Abies holophylla Malus baccata Pyrus ussuriensis Populus davidiana Fraxinus mandshurica Sorbus alnifolia Euonymus alatus Rhamnus ussuriensis Rhamnus parvifolia Rhamnus diamantiaca Lonicera praelorens Tilia amurensis p* Habitat 2 Habitat 3 Habitat 4 p*** p*** p*** p*** p** p*** p*** p* p* p*** p*** p*** p*** p* p** p* p* p** p* p*** n** n*** p*** p*** p*** p* p* p*** p* p*** p** p*** p* p*** p*** p*** 517 Silva Fennica 46(4), 2012 research articles Appendix 7. Species-habitat associations using the CRS model. “p” and “n” denote to “positive” and “negative”, respectively. *, ** and *** indicate signiicance at p < 0.05, 0.01 and 0.001, respectively. Species Habitat 1 Habitat 2 Betula platyphylla Acer mandshuricum Syringa reticulata var. amurensis Acer ginnala Lonicera ruprechtiana Rhamnus davurica Euonymus macropterus Padus racemosa Ulmus davidiana var. japonica Acanthopanax senticosus Acer barbinerve Ulmus macrocarpa Berberis amurensis Lonicera maackii Philadelphus schrenkii Rhamnus schneideri var. manshurica Betula costata Pinus koraiensis Juglans mandshurica Fraxinus rhynchophylla Maackia amurensis Phellodendron amurense Viburnum sargenti Tilia mandshurica Ulmus laciniata Euonymus verrucosus Aralia elata Corylus mandshurica Quercus mongolica Acer trilorum Viburnum burejaeticum Carpinus cordata Acer tegmentosum Actinidia arguta Acer mono Abies holophylla Malus baccata Pyrus ussuriensis Populus davidiana Fraxinus mandshurica Sorbus alnifolia Euonymus alatus Rhamnus ussuriensis Rhamnus parvifolia Rhamnus diamantiaca Lonicera praelorens Tilia amurensis p*** n*** p*** p*** n** n*** n*** n*** n*** 518 p*** p*** p*** p*** n* n*** p* n*** p*** p*** n*** n*** p*** n*** p*** p*** n*** n*** n*** p*** n*** n*** n*** p*** p*** n*** n*** n*** n*** p*** p** n*** p*** n*** p*** n*** p*** n*** Habitat 3 p*** n*** n*** n*** n*** p*** n*** n* p*** n*** n*** n*** p*** p*** n*** p* n*** n*** p*** n* n* n*** p*** n*** p*** n*** n*** p*** p*** p*** p*** n* n*** p*** p*** n*** n*** p* p*** n*** n*** n*** n*** n*** p*** p*** n** p*** p** n*** p*** n*** n*** p*** p*** p*** n*** p* p*** p*** n*** n* p*** p*** p*** n*** n*** p*** n*** p*** n*** n*** n*** p*** n*** p*** n*** n*** n*** p** p*** p*** p*** n*** n*** n*** Habitat 4 p*** n* p*** n*** p*** p*** Zhang et al. Species-Habitat Associations in a Northern Temperate Forest in China Appendix 8. Species-habitat associations based on habitat randomization. “p” and “n” denote to “positive” and “negative”, respectively. *, ** and *** indicate signiicance at p < 0.05, 0.01 and 0.001, respectively. Species Habitat 1 Habitat 2 Habitat 3 Betula platyphylla Acer mandshuricum Syringa reticulata var. amurensis Acer ginnala Lonicera ruprechtiana Rhamnus davurica Euonymus macropterus Padus racemosa Ulmus davidiana var. japonica Acanthopanax senticosus Acer barbinerve Ulmus macrocarpa Berberis amurensis Lonicera maackii Philadelphus schrenkii Rhamnus schneideri var. manshurica Betula costata Pinus koraiensis Juglans mandshurica Fraxinus rhynchophylla Maackia amurensis Phellodendron amurense Viburnum sargenti Tilia mandshurica Ulmus laciniata Euonymus verrucosus Aralia elata Corylus mandshurica Quercus mongolica Acer trilorum Viburnum burejaeticum Carpinus cordata Acer tegmentosum Actinidia arguta Acer mono Abies holophylla Malus baccata Pyrus ussuriensis Populus davidiana Fraxinus mandshurica Sorbus alnifolia Euonymus alatus Rhamnus ussuriensis Rhamnus parvifolia Rhamnus diamantiaca Lonicera praelorens Tilia amurensis p*** n** n*** n*** n** n* n** n*** n*** n** n*** n*** n** n*** p*** p*** p*** p*** n*** n*** n*** n*** n*** n*** p*** p*** n*** n*** p*** n** n*** n*** p*** p* n*** n*** n*** n*** n*** n*** n*** n*** n*** p*** p*** p* n*** p* p*** n*** n*** n** n*** n*** p*** p*** p*** n** n*** n*** p*** p*** n*** n*** n*** p*** p** p*** p* n*** n* p*** p*** n** Habitat 4 n*** n*** n*** p** p*** p*** n*** n* n** n*** p*** p** p*** p*** p*** n*** n*** n*** n*** p*** n*** p*** n*** n*** n*** n*** n*** n*** p*** n** p*** n*** n*** n*** n*** p*** n** p*** p*** p*** 519