Association classification in natural and artificial Larix gmelinii forests
There were totally 423 species in all the plots, belonging to 212 genera, 79 families. The species belonging to Compositae and Ranunculaceae of the angiosperms held a dominant position in the species community composition. Natural forests were classified into 5 plant associations by TWINSPAN, including Ass.I (Larix gmelinii-Betula fruticosa-Carex meyeriana), Ass.II (Larix gmelinii-Spiraea chamaedryfolia+Corylus heterophylla-Eriophorum vaginatum), Ass. III (Larix gmelinii-Ledum palustre+Vaccinium vitis-idaea-Deyeuxia angustifolia), Ass.IV (Larix gmelinii-Rhododendron dauricum-Pyrola dahurica), Ass.V (Larix gmelinii-Carex) (Table 1). Artificial forests were classified into 4 plant associations by TWINSPAN, including Ass.I (Larix gmelinii-Betula fruticosa-Carex meyeriana), Ass.II (Larix gmelinii-Spiraea chamaedryfolia+Corylus heterophylla-Eriophorum vaginatum), Ass.V (Larix gmelinii-Carex), Ass.VI (Larix gmelinii-Carex Larix gmelinii-Sorbaria sorbifolia-Geranium wilfordii) (Table 2).
The response of different associations to environmental gradients
In the CCA analysis, the Monte Carlo permutation test indicated significant eigenvalues for all canonical axes (P<0.01). The species-environment correlations of CCA axes respectively were 0.944, 0.777, 0.806, 0.708. The cumulative percentage variance of the species-environment relationship for CCA axes was 83.5% in natural forests (Table 3). The species-environment correlations of CCA axes respectively were 0.950, 0.922, 0.862, 0.861. The cumulative percentage variance of the species-environment relationship for four CCA axes was 80.4% in artificial forests (Table 4). CCA ordination performed well in plant association distribution along environmental gradients. In addition, natural forests had better fitting effect than artificial forests.
In two-dimensional CCA ordination diagram of natural forests, Asso.I and Asso.III usually distributed in higher elevation and slope, shady slope, lower annual mean temperature. Asso.II was immensely limited by annual precipitation, and mainly distributed in moist sunny slope with higher canopy density. Asso.IV usually distributed in cold shady slope with lower annual precipitation. Asso.V usually distributed in lower elevation and more flat areas (Fig.2). In artificial forests, Asso.I distribution trends was similar to the trends in natural forests. Asso.II and Asso.V were usually distributed in lower elevation with higher canopy density. Asso.VI was immensely limited by annual mean temperature and annual precipitation in lower elevation (Fig.3).
The main limiting environmental factors and species diversity patterns in different associations
In natural forests, environmental variable ranking of marginal effects in order was annual mean temperature, annual precipitation, elevation, slope aspect, canopy density, slope. Five environmental factors had significant marginal effects on community distribution (P < 0.05). Removing the collinearity of leading factors, only 5 environmental factors past through significance test. The conditional effect order of environmental factors didn’t change (Table 5).
In artificial forests, environmental variables ranking of marginal effects in order was elevation, annual mean temperature, annual precipitation, slope aspect, canopy density, slope. Five environmental factors had significant marginal effects on community distribution (P < 0.05). Removing the collinearity of leading factors, only 2 environmental factors past through significance test. The conditional effect order of environmental factors changed, and ranking of environmental variables in order was elevation and annual precipitation (Table 6).
Species richness ranged from 16.07 to 28.18, the Shannon-Wiener index ranged from 1.98 to 2.44, and evenness ranged from 0.69 to 0.81 in natural forests (Fig.4). Species richness ranged from 14.25 to 26.18, the Shannon-Wiener index ranged from 1.56 to and 2.46 evenness ranged from 0.58 to 0.71 in artificial forests (Fig.5). Ass.III and Ass.IV only appeared in natural forests and Ass.VI only appeared in artificial forests. Species diversity in Ass.I, Ass.II, Ass.V of natural forests were higher than that of artificial forests.