Aims Previous studies have demonstrated there exist the inherent relationships between landscape pattern and road networks. New techniques for analyzing the distribution of nodes on a network had been developed, called the network K-function (for univariate analysis) and network cross K-function (for bivariate analysis). Using these methods, we analyzed the spatial pattern of plantations, focusing on the their spatial clustering in Xishuangbanna in Yunnan Province. In particular, we addressed explicitly: (1) the spatial distribution of plantations with road network from 1970 to 2000 to investigate the effects of road disturbance; (2) the effects of plantations spread on the coniferous forests and broad-leaved forests using the bivariate analysis.
Methods As a widely used method in other fields, K-function plays an important role in spatial data analysis in plant ecology recently. But it also forms a major challenge in present ecological researches. Like most other spatial statistical methods, the K-function assumes a homogenous environment to calculate the Euclidean distance between points (or straight-line distance as "the crow flies"), and thus becomes an inappropriate tool for analyzing point patterns confined along irregular road networks. A method to conduct K-function analysis of point patterns on a network is recently based on adaptation of Ripley's K-function (i.e. network K-function analysis). It is a second-order spatial point pattern analysis, using the variance of interpoint distances to describe two-dimensional point distribution patterns and will greatly benefit spatial analysis in road ecology and field boundary studies. And we further used the network cross K-function to analyze the effects of the plantations among the road network on the coniferous forests and broad-leaved forests. The network K-function will become a useful statistical tool to analyze ecological data along roads, field margins, streams and other networks, with the improvements in software and advancements in computer hardware technology in the future.
Important Findings Our study provided evidences that, plantations expanded proximity to road network. The results of the Kernel and the network K-function analyses showed that the number of plantations increased and tended to cluster with the road networks. In 1970, only three strong clusters of plantations existed in the northern part, but in 1990, there were plenty of strong clusters in Xishuangbanna. Univariate spatial pattern analysis using the network K-function showed that significant clustering of populations observed at all values of d in 1970 and 1990, but the observed curves is not smooth in 1970. Significant spatial clustering was found for the plantations up to the d value at 60 km, and then significant large-scale repulsion of clusters of populations varied from 80 to 200 km in 2000. At the same time, the network cross K-function analysis showed that the expanding plantations had remarkably impact on coniferous forests along the road networks. In 1970 and 1990 the curves had no significant change. In 1970, there was a negative relationship in 0-60km between the coniferous forests and plantations, then significant large-scale positive correlation. However, by 2000, K curve had a clear change. The distributions of these two forests were significantly repulsion at scales d > 70 km. And from 1990 to 2000, there was not obvious relationship at any significant scale between plantations and broadleaf forests, which suggested that the spread of plantations had no remarkable impact on broadleaf forests.