Abstract:
Objectives To explore the characteristics of the spatial distribution of plant communities in the Qitai Desert Grassland Nature Reserve (QNR) in Xinjiang.
Methods Using high-resolution remote sensing imagery, a preliminary delineation of vegetation patches was conducted. To obtain relevant plant community parameters, field quadrat surveys were conducted in conjunction with the TWINSPAN, which was employed for the quantitative classification of plant quadrats. Subsequently, the classification results were used to construct a Random Forest model, which was then validated for accuracy, aiming to establish the spatial distribution patterns of dominant plant communities in the QNR.
Results The plant community in the QNR was composed of dominant species such as Reaumuria soongorica, Haloxylon ammodendron, Ephedra przewalskii, and Anabasis brevifolia. The TWINSPAN quantitative classification resulted in the division of 43 plant quadrats into 9 community types. Among these, the H. ammodendron + R. soongorica (G6) covered the largest area, whereas the Leymus angustus + Halocnemum strobilaceum + Phragmites australis + Poacynum pictum (G9) occupied a smaller area but exhibited the highest diversity. The E. przewalskii + A. brevifolia (G1) exhibited the lowest diversity index, dominance index, and evenness index. The accuracy of the Random Forest model reached 86.05%, with a Kappa coefficient of 0.8155.
Conclusions The plant communities in the QNR can be divided into 9 communities, forming dominant plant communities primarily consisting of shrubs and subshrubs. Influenced by factors such as soil type and water resource conditions within the reserve, these groups exhibit spatial discontinuity, appearing in a patchy distribution. This study combined high-resolution remote sensing imagery with field surveys, ultimately achieving a high-precision visualization of the spatial distribution patterns of plant communities in the QNR. The findings of this study have significant potential for broad application in vegetation classification.
下载: