Environmental factors affect carbon isotope discrimination in plants. Therefore, carbon isotope discrimination records a wealth of environmental information, such as temperature, precipitation etc., and the predominant factor's influence is significant. Based on this hypothesis, we determined the carbon isotope composition of Tamarix growing in the southern margin of the Taklimakan Desert where temperature and water are key growth-limiting factors. Mean annual temperature (MAT) showed a significant impact on variation in carbon isotope discrimination value (DS), but precipitation not. In multiple stepwise regression analysis, MAT was the first and the only variable selected into the prediction model of DS against MAT and precipitation, indicating that the effect of temperature on carbon isotope discrimination was predominant. The correlation between DS and the variation rate of precipitation in previous year shows that precipitation, as a major factor in arid regions, exerts a certain "lag effect" on the growth of Tamarix. The reconstructed MAT regression model was established according to the DS with the least squares method, the result bolster the argument that the planet's temperatures have shifted significantly in the last half-century.