Methods

Site description

This study was conducted at the Xiaoliang Research Station for Tropical Coastal Ecosystems of the Chinese Academy of Sciences (Xiaoliang Station, 21°27′N, 110°54′E). Xiaoliang Station is located in the southwest of Guangdong Province, China, and has a tropical monsoon climate. In the studied region, the mean annual temperature is 23 ºC, and the mean annual precipitation ranges from 1 400 to 1 700 mm. There is a clear seasonal variation, with the wet season lasting from April to October, and the dry season lasting from November to March. The soil is classified as a latosol developed from granite (Wang et al. 2014). It is estimated that the annual wet N deposition was approximately 40 kg N ha-1 in 2012 in this region (Mo et al. 2015), and the forest is generally regarded as a P-limited ecosystem (Mo et al. 2019).
The study site was located approximately 5 km from the coast in a secondary broad-leaf mixed forest on coastal land with a very small slope, which was. The forest was restored from a Queensland peppermint (Eucalyptus exserta ) plantation by introducing 312 plant species between 1964 and 1975. Thereafter, natural colonization during succession displaced most of the planted tree species, and the area developed as a relatively typical secondary evergreen broad-leaf mixed forest, with a high biodiversity and a complex community that was similar to a natural forest (Chen et al. 2016). The most common tree species in the study site are as follows:Castanopsis fissa , Cinnamomum camphora , Carallia brachiata , Aphanamixis polystachya , Ternstroemia pseudoverticillata , Acacia auriculaiformis , Cassia siamea , Albizia procera , A. odoratissima , Leucaena leucocephala , Aquilaria sinensis , Chakrasia tabularis ,Syzygium levinei , Schefflera heptaphylla , S. hancei , S. bullockii , Uvaria macrophylla ,Schefflera octophylla , Psychotria rubra , and Aporusa dioica (Li et al. 2015, Wang et al. 2017a).

Experimental design

A randomized block design experiment with N and/or P fertilization was established in the secondary tropical forest in September 2009 (Zhao et al. 2014, Chen et al. 2016). In each block, four plots (10 m × 10 m) were included, with five replicate blocks in total, and the adjacent blocks were separated by a 50-m buffer region. The four treatments, including N addition (+N), P addition (+P), N and P addition (+NP), and a control (CK, no addition of mineral nutrients), were assigned randomly to the four plots within each block. The edges of each plot were trenched to a depth of 20-cm, put by a PVC broad and surrounded by a 2 m wide buffer. Because a large number of fine roots were distributed in surface soils, the trenches largely inhibited the transfer of nutrients among different treatments, as evidenced by clear differences in extractable soil P between fertilized and unfertilized treatments after six years of fertilization in 2015 (Mo et al. 2019). In this study, the soil available P concentrations were significantly different between unfertilized and fertilized treatments, as described in our previous studies (Li et al. 2015, Mo et al. 2019).
Fertilizers were regularly applied bi-monthly from 2009-2017 to achieve the total amounts of N and P equivalent to 100 kg ha-1yr-1. Every fertilizer application consisted of quantitative fertilizer (NH4NO3 and/or Na2HPO4) were dissolved in 30 L of groundwater, which was then applied to the corresponding plots uniformly using a backpack sprayer near the soil surface. Thirty liters of groundwater was also applied to control plots (Wang et al. 2014, Li et al. 2015). It is estimated that the amount of added water in each plot was equivalent to 0.08% and 0.35% of rainfall inputs in the wet and dry seasons, respectively, so that the effects of added water on soil moisture could be ignored (Mo et al. 2015).

Sampling and measurements

Based on previous vegetation surveys at the study site, we selected four native woody tree species that occurred sufficiently frequently in plots in the four treatments, including S. bullockii , U. macrophylla , S. octophylla , and P. rubra (Table 1), all of which are perennial shrubs or small trees with shade tolerance and C3 carbon fixation, or evergreen tree species. Leaf samples were collected in June or July of 2012, 2015, and 2017 (continuous yearly sampling would be harmful to the sampled shrub individuals because a relatively large leaf amount was sampled in every sampling event). Briefly, we randomly selected three to five individuals of each species in each treatment, and then sampled fully expanded and healthy mature leaves. After labeling the samples in Ziploc bags, all leaf samples were immediately transferred back to the laboratory. Each leaf sample was washed and air-dried, then placed in an oven at 105 ℃ for 30 min and maintained at 65 ℃ for 72 h to a constant weight. Subsequently, leaf samples were ground and sieved using 0.15 mm sieves, after removing petioles and main veins, and leaf P concentrations were measured spectrophotometrically after digestion with sulfuric acid (H2SO4). Leaf N concentrations were determined using the Kjeldahl method (Wang et al. 2013).
The leaf mass per area (LMA) was measured in 2015 using a portable leaf area meter (LI-3000A, LI-COR Biosciences). Photosynthetic nutrient use efficiency for N (PNUE) and P (PPUE) was defined as the rate of net photosynthesis per unit N or P expressed on a leaf dry mass basis. Leaf P was partitioned into four fractions: structural P, metabolic P (including Pi), nucleic acid P, and residual P, using sequential extraction (following Kedrowski (1983) with modifications by Hidaka and Kitayama (2011)). Details on how these measurements were conducted can be found in Mo et al. (2019).
The standard anthrone colorimetric method was employed to coarsely measure the concentrations of leaf NSCs, for the rapid analysis of large samples (Dubois et al. 1956, Hewitt 1958, Buysse and Merckx 1993, Li et al. 2016a). For the determination of soluble sugar concentration (total sugar, sucrose, and fructose), 0.1 g of ground leaf sample was placed into a 50 ml tube and mixed with 10 ml 80% (v/v) alcohol, extracted in 90℃ water for 10 min, three times. All the extracted solutions were transferred into 50 ml flasks, and the final volume was adjusted to 50 ml for the measurement of soluble sugar via the standard anthrone colorimetric method. The residue after three extractions in the 50 ml tube was treated with 30% (V/V) perchloric acid (HClO4) for 12 h, then extracted in 80℃ water in 10 min. After that, the extracted residue was cooled down and filtered, with the final volume adjusted to 50 ml in a flask for the determination of starch. Finally, the sugar and starch concentrations in all the extracted solutions were measured from the absorbance of anthrone colorimetric at 620 nm using a UV-vis spectrophotometer. The soluble sugars and starch concentrations were calculated per dry matter of leaves (mg g-1) (Li et al. 2016a, Xie et al. 2018). In this study, the NSC concentrations can be defined as the sum of the soluble sugar concentrations and starch concentrations (Hoch et al. 2003).

Data analysis

As our experiment used a randomized block design, for each species, we used species-specific linear mixed model analysis to examine the effects of N, P, and N×P on the leaf soluble sugar and starch concentrations, and N, P, and N:P ratios for each species in 2012, 2015 and 2017. Given the variation among species, the linear mixed models were used to test the difference on the soluble sugar and starch, N, P concentrations and N:P ratios. In these models, species (S), N-addition (N), and P addition (P) were considered as the fixed effects, and block, nested by sampling year, was included in the models as a random factor. Relative effects (RE) were quantified by the ratio of the variable in the experimental group (+N, +P, +NP) to the control group (CK), minus one (In Fig. 1). All data analyses were performed in Excel 2013 and IBM SPSS Statistics 19.0. Results are reported as significant when p <0.05.