Measuring Plant Spatial Point Pattern:A Novel Photograph Orientation Method Based On GigaPan System

Plant spatial pattern analysis is an important method for investigating the population structure, intra- and inter-species competition, and the interactions between plant and environment. As the first step of point pattern analysis, the coordinates of plant individuals have to be determined in two-dimensional space, which is critical for further data analysis. But this is often a labor-consuming and time-cost work. Photography orientation method has been proposed to collect the spatial point pattern data, which divides the main plot into many small sub-plots and takes photos for each sub-plot. This method is inefficient due to the complex procedure in handling lots of sub-plots. Gigapan is a new image capture system, which can produce panoramic photos with high resolution (giga pixels), recoding details and characters in variable spatial resolution from individual to landscape scale. In this study, we aim to establish a new photography orientation method with higher efficiency, by combining the advantages of data collection of GigaPan system and data analysis of geographic information system (GIS). In field work, for the same set of equipments, the efficiency is affected by the plot size and number of reference points. Specifically, the efficiency is higher with larger plot size and fewer reference points, which may decrease the accuracy of data.I selected three communities with different complexity (i.e. artificial lawn, low shrub tussock, and high shrub tussock) to validate the new method and to examine the effect of plot size and number of reference points on the data accuracy. The procedure of this new method is as follow:Firstly, the gradients of the plot size and number of reference points were designed, and the coordinates of reference points and test points were measured using laser range finder. Secondly, the GigaPan system was used to obtain the pictures of the plot with different plot size. These pictures were corrected in arcGIS. Finally, the coordinates of test points were determined in arcGIS, which were compared to the results of laser range finder. The Root Mean Square error (RMSE) and the distance error rate between test points were calculated to test the effect of plot size and number of reference points on the data accuracy.The main conclusions of this study are as follows:1) The performance of new method was satisfying for simple community (e.g. artificial lawn and low shrub tussock). Data accuracy was affected by plot size and number of reference points, and the measurement error was less than 5% in a 100 m2 plot with more than 20 reference points.2) The performance of new method decreased significantly in complex community (i.e. high shrub tussock), where the accuracy and efficiency was low and some plant individuals were difficult to be mapped.3) The field efficiency of our new method is greater than ten times and at least four times, relatively, than the method of laser range finder and previous photograph orientation method (i.e. sub-plot shooting).4) As a case study, the new method was used to determine the spatial point patterns of Arenaria densissima and Rhododendron impeditum. Arenaria densissima was random at 0-50cm, and clumped at 50-250cm, which was similar to Alex’s study in 2008. It further validate the power of this new method.In summary, we established an efficient method based on Gigapan to collect the spatial pattern data with high efficiency in grassland or low shrub community. The new method is also useful to monitor the long-term change of species composition and abundance within plant community, providing insights into multi-scale community succession and community assembly processes.