The deposition of reactive nitrogen (N) has substantially increased in subtropical regions due to human activities. However, the effects of long-term N addition on the water-use efficiency of subtropical forests are poorly understood. Here, we conducted an 11-year experiment in a subtropical Cunninghamia lanceolate plantation with four N-addition levels: N0, N1, N2, and N3 (equivalent to 0, 6, 12, and 24 g of N m-2 yr-1, respectively). A thermal dissipation probe system was used to calculate sap flow and plant biomass carbon was assessed by field investigation. The whole-plant water use and water-use efficiency were estimated. In addition, the δ13C of tree rings was used to indicate the plant intrinsic water-use efficiency. The results showed that N3 significantly increased the annual sap flow velocity, especially in summer and winter. Annual water use, plant growth, and water-use efficiency did not significantly differ among the N treatments, but water use tended to be higher in N3 than in N0. Furthermore, the reduction of δ13C between the pre-N treatment period and the post-N treatment period was 3.02%, 3.26%, 3.58%, and 5.28% for N0, N1, N2 and N3, respectively, which supported the inference that N addition could enhance water use. We conclude that long-term addition of high levels (but not of low levels) of N increased whole-plant water use in C. lanceolate plantations. Our results indicate that N deposition accompanied by high temperature and drought events may negatively affect water balance in subtropical forests.

Sap flow measurement and water use

In 2014, a thermal dissipation probe system (TDP) for sap flow measurement was installed in one sample tree per plot, so that there were three replicate sample trees for each treatment. The DBH of the sample trees were similar to the average (about 22.5 cm) in the plot. We assumed that that the sap flow of a sample tree with average DBH represented average sap flow for the plot. Two 2-cm-long probes separated by 10 cm in the vertical direction were inserted into the sapwood of the sample trees. The top probe was equipped with a heating component, which was heated continually at 0.2 W. The temperature difference between the two probes was influenced by the sap flow density near the heated probe. Therefore, the recorded difference in the temperature (△T) between the two probes was converted to sap flow velocity V (g cm^-2 s^-1) according to equations 1 and 2 (see next paragraph) (Granier 1987; Zhu et al. 2012).

Probes were uniformly installed on the same side of sample trees at a height of 1.3 m and were covered with aluminum foil to minimize the effects of solar heating on sap flow density. Hourly mean values were automatically recorded and stored on a data logger (CR10X, Campbell Scientific, Inc., Logan, USA). Every 3 months, the probes were removed and reinserted on a previously unused side of each tree in order to ensure the well-condition of probes. In this study, data collected from 1 January 2015 to 31 December 2015 were used. To better show the variation in sap flow velocity among N treatments, we multiplied each value for sap flow velocity by 10⁶, but we used original values to estimate the water use by the sample trees. Sap flow through the trunk can be approximately equivalent to the transpiration from the canopy (Kozlowski and Pallardy 1997). We cut down 12 trees (without probes) in each plot to estimate sapwood area (A_S; Table 1) as calculated with equation 1:

A_S =3.577DBH^1.328                         (1)

We then calculated the annual water use in each plot based on estimates of A_S per plot, tree number per plot, and the following equations:

      K = (dT_M – dT) / dT                (2)

       V = 0.0119 × K^1.231                 (3)

   W = As × V×3600                  (4)

where K is a dimensionless value, dT_M is the maximum temperature difference between two probes, dT is the instantaneous temperature difference, V is sap flow velocity (g cm^-2 s^-1), and W is water use (g h^-1). The whole-tree water use per year (kg year^-1 tree^-1) was calculated by summing water use in the year. We assumed that no sap flow occurred at night when the temperature difference reached its minimum.