Quantifying correlations among leaf traits and trait differences between plant functional types (PFTs) not only gives insight into selective pressures that have shaped vegetation evolution, but also plays a crucial role in predicting vegetation productivity shift with climate and land use change. The photosynthetic characteristics, leaf nitrogen content (N_mass), leaf mass per area (LMA), as well as their relationships in trees, shrubs and grasses growing in three sites, Ningshan County located in Qinling Mountain, Fuxian County and Shenmu County both located in the Loess Plateau were studied. The results showed that there were significant differences (p<0.05) in the saturated photosynthetic rate (P_max), photosynthetic nitrogen use efficiency (PNUE), quantum yield of PSⅡelectron transport (Ф_PSⅡ), leaf nitrogen content (N_mass) and leaf mass per area (LMA) of three PFTs among Ningshan County, Fuxian County and Shenmu County. Trees, shrubs and grasses in Shenmu County had significantly higher mean values of P_max than those in Ningshan County and Fuxian County. The P_max, PNUE, Ф_PSⅡ, potential efficiency of primary conversion of light energy of PSⅡ(Fv/Fm), Nmass and LMA differed significantly (p<0.05) among trees, shrubs and grasses in three sites, and grasses had higher mean values of Pmax and PNUE than shrubs and trees did. The LMAs of trees, shrubs and grasses increased as climatic drought increased gradually from Ningshan County through Fuxian County to Shenmu County. The mean values of LMA for the three PFTs ranked in the order of trees>shrubs>grasses, and the LMA in trees was about twice times that in grasses.
In 60 plant species including trees, shrubs and grasses growing in Ningshan County, Fuxian County and Shenmu County, the LMA was significantly negatively correlated with the N_mass and PNUE (p<0.01), while the P_max was positively correlated with the N_mass (p<0.05). The correlation analysis among photosynthetic parameters showed that the P_max was significantly positively correlated with the PNUE (p<0.01), but negatively correlated with the Ф_PSⅡ (p<0.01).