Abstract:Based on tree-ring samples of Picea crassifolia at four different elevations along a north-facing slope in the Anymaqin Mountains, northeastern Tibetan Plateau, we developed four ring-width chronologies. Statistical results showed that characteristics of chronologies are different with the increasing elevation, and the correlations of chronologies in the common interval decreased with the increasing elevation. Meanwhile, correlations between the chronologies of different elevation and the two climatic parameters (temperature and precipitaton) exhibited distinct difference: tree-rings were significantly and positively correlated with the October mean monthly temperature of precious year, and also were significantly but negatively correlated with prior August and current May and June mean monthly temperatures, respectively. All the correlations were fluctuating with the increasing elevation. Tree-rings were significantly and positively correlated with the September and October precipitation of precious years, but were decreasing along the increasing elevation. Tree-rings were also significantly and positively correlated with the May precipitation of current years, and the correlations were increasing along the increase of the elevation. Furthermore, tree-rings were also negatively correlated with temperatures of different time interval and the warming index, proving that the optimal growing season of Picea crassifolia is May-September. The results of response function analyses showed that the influence of temperature and precipitation explained most of the growth variance at the low-elevation site, but much less at the high-elevation site. The results were greatly different from previous studies that tree-rings were significantly and positively correlated with temperature at the upper forest limit, whereas were significantly correlated with precipitation at the lower forest limit. Principal component (PC) analysis was used to indicate regional variations in radial growth patterns. Of the four chronologies，the first PC explains 81.071% of their total variance. According to the response function analyses, the first PC of the four chronologies showed that moisture and temperature of the last stage of prior year growing season and the initial stage of current year growing season are the most important limiting factors of tree growth. This conclusion was further proved by regional model simulation results. Overall, our results indicated the similarities and discrepancies of Picea crassifolia ring-width chronologies on the same slope, and provided some background information for future dendrochronological studies in this area.