Abstract
We selected four Populus euphratica Oliv. forest plots (100 m × 100 m) in the upper reaches of the Tarim River in the Xinjiang Uygur Autonomous Region of China. Each of the four forest plots was chosen to represent a different growth and death stage of P. euphratica forest: juvenile forest, mature forest, dying forest, and dead forest. In each plot, we measured the coordinates, DBH, height, and status of all P. euphratica individuals. We used (1) spatial pattern analysis to explore spatial distribution patterns and associations of live trees and dead trees, (2) a random mortality model to test whether the tree death was random or non-random, and (3) a generalized linear mixed-effect model (GLMM) to analyse factors related to tree survival (or death). In the juvenile plot, live trees were significantly aggregated at all scales (p < 0.05); while in the mature and dying plots, live trees were more aggregated at small scales and randomly distributed at larger scales. Live trees and dead trees showed a significantly positive association at all scales in the juvenile plot (p < 0.05). While in the mature and dying plots, live trees and dead trees only showed a significantly positive association at scales of 0-3 m (p < 0.05). There was significant density-dependent mortality in the juvenile plot; while mortality was spatially random at all scales in the mature and dying plots. The distance from the river showed significantly negative correlations with tree survival (p < 0.01). DBH and height had significantly positive associations with tree survival in the juvenile, mature, and dying plots (p < 0.05). In extreme drought, dying trees appeared to be shape-shifting into more shrub-like forms with clumps of root sprouts replacing the high canopies. The shift under extreme drought stress to more shrub-like forms of P. euphratica may extend their time to wait for a favourable change.