Woody Plant Community Diversity And The Scale Effects In A Subtropical Forest

Forest biodiversity loss is becoming a serious ecological crisis to human beings,which attract extensive attention from ecologists.For further monitoring and studying forest biodiversity,many forest biodiversity permanent monitoring plots have been established all over the world,providing substantially long-term and stable information.However,most of the studies were carried out based on one single scale and different or even contrasting results of measuring and assessing forest biodiversity were observed on different scales.Therefore,multi-scale studies should be taken into account to fully understand forest biodiversity more comprehensively and objectively.Growth,development and distribution of plant are controlled by environmental factors.Forest classification according to environmental gradient and sequently detecting indicator species of every specific forest classification are helpful to understand responses of forest community to environmental variation and can provide references to forest management and biodiversity conservation.Thus,we established a 10 hm~2 permanent monitoring plot in Kanghe Provincial Nature Reserve in Dongyuan County,Guangdong Province,China and detected the scale effect onαdiversity,dominant species composition and standing wood structural characteristics.We also used different sample scales and plot constructing methods to simulate species-area and species rank-abundance relationships,and conducted indicator species analysis based on different forest community type classification methods to test the responses of forest community to environmental gradients.Major findings were as follows:(1)A total of 50070 individuals were recorded,belonging to 2 class 34 orders 52families 96 genera 153 species,respectively.Fagales,Theales,Laurales,Myrsinales and Rubiales have the many individuals.Fagales contributed most individuals and basal area(BA)to forest community,and Theales had similar contribution to individuals as Fagales did.The forest community mainly consisted of tropical and temperate floristic components.Respectively,numbers of genera and individuals of tropical floristic components were 2.9and 1.9 times as many as those of temperate floristic components,while BA of temperate floristic components contributed more to the forest community,which is 2.2 times as many as that of tropical floristic components.(2)On the scale of 25,100,400 m~2,respectively,highly significant differences of community structural diversity characteristics were observed among plots(P<0.01),including 7 indices ofαdiversity(species richness,Shannon-Wiener diversity index,Pielou’s evenness index,abundance,N:S ratio,kurtosis and skewness),statistics of abundance based on plots,tree height and total basal area.According to multiple comparisons,no significant difference in tree height on 3 sampling scales and in kurtosis and skewness on the scale of 100 and 400 m~2 was found respectively,and significant difference in other indices on 3 sampling scales was observed among plots(P<0.05).(3)Icreasing number of dominant species and decreasing important value were found as the sampling scale increased.However,rank of specific dominant species stay releatively consistent on 3 sampling scales.Saplings and seedlings contributed most individuals to forest community,accounting for 51.1%and 30.2%of total individuals,respectively.Highly abundant species and abundant species accounted for 63.5%and 22.8%of total individuals,78.4%and 12.1%of total basal area,but 5.9%and 10.5%of total species,respectively.Rare species and highly rare species accounted for less than 0.4%and 0.5%of total basal area,respectively,but together accounted for 37.9%of total species.(4)Species-area relationships constructed by different methods showed different scale effect.Different models used to simulate species-area curves constructed by one specific method showed difference in goodness of fit,but a specific model used to simulate species-area curves constructed by different methods showed no significant difference in goodness of fit.According to goodness of model fitting and confidence intervals of parameters,Lomolino function was the best model for simulating species-area curves constructed by contiguous pure random sampling,Cumulative Weibull function was the second and Exponential function was the third.Power function performed well in simulating species-area curves constructed by nested sampling and contiguous adjacent random sampling.(5)Common species had a few species while occasional species had a large number of species which had great contribution to species diversity.Species rank-abundance curves of the forest community were close to lognormal distribution,indicating a large mature community and stable environment.Differences in performance of 4 models used to simulate species rank-abundance curves constructed by different sampling methods on different scales were observed.Four models used to simulate species rank-abundance curves constructed by contiguous pure random sampling and contiguous adjacent random sampling performed consistently on different scales.Lognormal distribution model performed best when sampling area was no less than 0.1225 hm~2,and broken stick model was the best model when sampling area was 0.0025 hm~2,and Zipf model performed best under other sampling area conditions.With respect to nested sampling,Zipf model performed best when sampling area was no more than 0.0225 hm~2,and lognormal distribution model performed best when sampling area was larger than 0.0225 hm~2.(6)Forest community was classified by CCA with northness,eastness,elevation and slope as environmental factors.Only elevation and slope had large contribution to ordination axes.Elevation was highly significantly and correlated with the first ordination axis(P<0.01,rintra=0.992),while slope was highly significantly and correlated with the second ordination axis(P<0.01,rintra=-0.961).Plots were classified into 3 forest community types by PAM cluster analysis based on LC scores of the first two ordination axes.Standard errors of max dissimilarity,average dissimilarity and diameter were relatively small within cluster,which indicated similar homogeneity was found among plots within the same community type.Significantly higher silhouette width was also found when the forest was classified into 3 communty types.Both results came to the same conclusion that the most reasonable classification method is whern the forest was classified into 3 community types.88 plots belonging to community type 1 were located at middle and low elevation.51 plots belonging to community type 2 were mainly located at middle and high elevation but slow slope.111 plots belonging to community type 3 were located at middle and high elevation but steep slope.Indicator species analysis showed the preference of indicative plant species to environmental factors and different environmental gradients of3 community types.(7)250 plots(each 400 m~2)were classified into 5 forest community types by multiple regression tree(MRT),using elevation and convexity as grouping nodes,based on Hellinger distance.Community type I was located at the peak and on the ridge with high elevation.Community type V was located on the ridge with low elevation.Community type II,III and IV were located in the valley with high,middle and low elevation,respectively.Three indices of multiple-site beta diversity based on binary data showed significant variation in species composition among plots of the same community type,but similar complexity of species composition among 5 community types.The total multiple-site beta diversity of community type IV at low elevation was the highest,while that of community type V was lowest and those of community type I,II and III at middle and high elevation were similar.Species turnover among plots was the key reason why species composition of5 community types showed variation among plots.Indicator species analysis showed that indicative plants of community type I and II at high elevation mainly consisted of highly abundant,abundant and common species and indicative plants of community type IV and V at low elevation mainly consisted of abundant,common and occasional species.