Sample plot design can affect the efficiency and accuracy of shrub coverage measurements in shrub-encroached grasslands

Shrub encroachment has become a global concern. The fraction of shrub coverage (FSC) is an important indicator that reflects the distribution of shrubs and the degree of shrub encroachment in grasslands. The line-point intercept (LPI) method is commonly used for FSC measurement in field surveys, but it's often associated with issues of poor accuracy and low measurement efficiency. Here, we focus on Caragana microphylla shrubencroached grassland as a case study. We used normalized difference vegetation index (NDVI) data obtained from unmanned aerial vehicle (UAV) to derive the actual values of FSC (FSCT) and to simulate measurements using the LPI method. We compared the results for measurements based on different sample plot designs, including variation in plot shape and line distribution, the number of sample lines, and the sample point spacings. The aim was to investigate the influence of these main parameters of sample plot design upon measurement results, in order to provide a reference for FSC measurements in field experiments. We first calculated FSCT and stratified the plots according to coverage levels. Based on this, we further evaluated the measurement accuracy of the LPI method under different parameter settings. The results revealed significant systematic errors in the measurement from circular plots. For square plots, the minimum number of lines required for the "high," "medium," and "low" coverage groups are 48, 16, and 8 (for 80% accuracy), and likewise, 108, 24, and 16 (for 90% accuracy), respectively. Futuremore, point spacings of 0.1 m and 0.5 m can achieve the same accuracy as the original spacing (0.02 m). We conclude that the systematic errors in circular plots are caused by the radial distribution of lines, whereas in square plots, achieving an efficient measurement requires considering different coverage levels, ensuring an adequate number of lines, and setting a relatively smaller point spacing.