Based on two tree-ring maximum latewood density （MXD） chronologies of Picea schrenkiana from the Manas River Basin, Xinjiang, the response characteristics of MXD to climate variation was discussed. Correlation analysis between MXD chronologies and instrumental records from Shihezi meteorological station showed that each chronology was significantly and positively correlated with the maximum monthly average temperature in July-August, and especially, the regional chronology （RC） was the most highly correlated variable （r=0.54, P〈0.001）. Afterwards, the maximum average temperature in July-August was reconstructed using RC. Comparison among reconstructed temperature, observed values, and the drought index （Is） confirmed that precipitation would affect MXD when the absolute value of Is is greater than 1.5σ （｜Is｜ 〉 2.5 during 1953-2008） or near to 1.5a over a 2-3 year period. The response characteristics are related to the semiarid climate of the study area. In dry years, lack of precipitation would limit the thickening of latewood cell walls and, as a result, impact MXD. Therefore, compared with relatively humid regions, the response of tree-ring MXD to air temperature similarly would be influenced by extreme moisture conditions in semiarid areas, and MXD, as a temperature proxy, should be used prudently on a limited scale.
The influence of temperature and precipitation on maximum latewood density （MXD） was mainly discussed in this paper, based on the samples of Picea schrenkiana from the Manas River Basin, Xinjiang, China. The correlation analysis between MXD and instrumental records from the Shihezi Meteorological Station showed that the MXD was positively related to the mean maximum temperature throughout the growing season at high elevations. Comparatively, the ring-width at low altitudes was limited by the precipitation in May-June. The composite chronology by MXD sequences was highly correlated with the mean maximum temperature in July- August （r = 0.54,p 〈 0.001）, which was then reconstructed by the composite chronology. The comparative analysis on the reconstructed temperatures, observed values, and drought indices （Is） revealed that precipitation would affect MXD when the absolute value of Is was greater than 1.5a （i.e., ｜Is｜ 〉 2.5） in the period of 1953-2008 A.D. or close to 1.5σ for 2-3 consecutive years. The response characteristics are linked with the semiarid climate in the study area. In a single year or consecutive years of extreme dryness, the lack of precipitation would limit the thickening oflatewood cell walls and thus impact the MXD. All in all, if a MXD chronology is aimed to reconstruct temperature history, the moisture conditions at the sampling site should be considered prudently.
