Welcome to visit Scientia Silvae Sinicae,Today is

Scientia Silvae Sinicae ›› 2023, Vol. 59 ›› Issue (10): 57-65.doi: 10.11707/j.1001-7488.LYKX20210872

• Research papers • Previous Articles     Next Articles

Scaling Effects of Edge Correction Methods on Spatial Structure Parameters

Shuai Yu1,Tijiu Cai1,Pide Zhang2,Minglei Ren2,Haiyu Zhang2,Cunyong Ju1,*   

  1. 1. Key Laboratory of Sustainable Forest Ecosystem Management of Ministry of Education School of Forestry, Northeast Forestry University Harbin 150040
    2. Naozhigou Forestry Farm of Dongning Forestry and Grassland Bureau Mudanjiang 157299
  • Received:2021-11-30 Online:2023-10-25 Published:2023-11-01
  • Contact: Cunyong Ju

Abstract:

Objective: Suitable edge correction methods were selected to eliminate the influence of edge effects on the spatial structure parameters of the sample plots and to provide a theoretical basis for forest spatial structure analysis. Method: The edge trees were removed according to different edge correction methods such as the Voronoi diagram-based method, the internal buffer method and the nearest-neighbor (NN) method in the sample plots of different sizes, then the structural parameters i.e. the uniform angle index, the size differentiation index, the mingling index and the tree species diversity mingling were calculated in terms of those remained trees within the different-size plots, and we further analyzed how these structural parameters varied against the changes of the sample plots' sizes so as to recognize the applicability of the three different correction methods on different plots. Result: For those plots with each side length of no more than 40 m, the internal buffer method removed the least amount of edge wood. As the scale of plots increased furthermore, the internal buffer method started to removed the most amount of edge wood while the NN method removed more edge trees than the Voronoi diagram-based method, but the differences were not significant. The spatial distribution patterns of small sample plots mostly appeared as clumped or random, and the spatial distribution pattern of large sample plots tends to be clumped but close to random. The mean value of size differentiation index of each sample plot gradually transitioned to intermediate wood status when the side length of the sample plot increased. The mean curves of the simple mingling index showed a trend of strong mixed degree, and the tree species diversity minglings were between medium and strong mixed degree. Conclusion: When the side length of the sample plot was no more than 40 m, the structural parameters varied greatly with the size of the sample plots changing. When the side length of the sample plot was greater than 60m, the edge correction or not had little effect on the calculation of the size differentiation index, the mingling index, and the tree species diversity mingling. The internal buffer method has applicability limitations in calculating the uniform angle index since it did not induce convergence of the index like the other two methods. In general, the NN method is less dependent on the plot scale than the internal buffer method, and is the best performing one among the three methods. The three methods used in this paper all belong to minus-sampling method, the calculation of the structural parameters only utilized the retained trees in the sample plots and wasted part of the survey information. As an alternative, the mirror replication or eight-neighborhood translation correction method may form a larger sample plot (i.e. plus-sampling method) to offset the edge effects, but whether they have scale effects is a question that needs to be further investigated in the future.

Key words: edge effect, internal buffer method, the nearest-neighbor method, spatial distribution, mingling index

CLC Number: