N. reynaudiana seedlings were used as experimental material to investigate their foraging strategy to spatially heterogeneous P distribution. Seeds of N. reynaudiana were purchased from a company in Kunming, Yunnan Province (Yunnan Jinye Eco-construction Group Co., Ltd ) and sown in climate chamber set at 25°C, 75% relative humidity, and a photoperiod of 12 h light (photon flux density 4000 lx) and 12 h dark. After 2 weeks, the germinants were transferred to nutrient-rich humus substrate and left to grow in a greenhouse set at 29.3°C/23°C (day/night); photon flux density of 21 mol quanta m-2 d-1, and ca. 42.7% and 67.7% relative humidity during the light and dark periods of the experiment, for four weeks. Seedlings with relatively uniform size (12 cm ± 1 cm in height, 4 cm ± 1 cm in root length, and 10 mg ± 1.5 mg in fresh weight) were selected for this experiment.
Experimental design and Growth conditions
To investigate the foraging strategy of N. reynaudiana seedlings to spatially heterogeneous P supply, a factorial experiment, which involved three levels of patch strength (0 + 0/30, 0 + 7.5/30 and 7.5 + 0/30 mg P. kg-1) and three levels of patch size (small, medium and large) were established (Figure 5). The treatment 0 + 0/30 involved no P addition in the central interior patch where the seedlings were initially planted and intended to simulate the effect of high patch strength on root proliferation in the exterior P-deficient and P-replete patches. Similarly, the treatment 0 + 7.5/30 were intended to simulate the effects of moderate patch strength on root proliferation in the exterior P-deficient and P-replete patches. The treatment 7.5 + 0/30 mg P. kg-1 involved addition of small amount of P in the central interior patch and intended to see if root proliferation between P-deficient and P-replete patches was influenced by initial P encounter. In addition, a homogeneous P supply with three levels (15/15, 18.75/18.75 and 15/15 + 7.5 mg P. kg-1 per patch), corresponding to the total P concentration in heterogeneous P supply, was included to compare total biomass production between heterogeneous and homogeneous P supply. The P concentrations used in this study were determined based on P availability in the soil in southern China. A total of 12 treatments were applied in the experiment: 9 heterogeneous and 3 homogenous P treatments.
The experiment was conducted using pots (diameter: 40 cm, height: 32 cm) that were divided into ½, ¼ and 1/6 baffle less compartments using plastic separator to simulate large, medium and small patches, respectively. Each pot compartment was filled with a mixture of sand and sodium polyacrylate prior to application of different P concentrations as the sodium polyacrylate (0.5–1 mm in diameter) strongly adsorbed the applied P and hence prevented its movement between compartments but capable of releasing nutrients slowly and evenly and could be freely penetrated by roots. According to the weight of each pot, the sodium polyacrylate and washed river sand, the different concentrations of KH2PO4 solution (as P source) were mixed and packed in a volume ratio of 1:3 in each pot. In each treatment pot, one N. reynaudiana seedling was planted in a cylindrical tube (30 cm in length and 8 cm in diameter) filled with sand or sand mixed with 7.5 mg KH2PO4·kg-1 at the center of the pot. Once the plants were placed correctly at the center of the treatment pots, the cylindrical tube and patch separators were removed carefully (Figure 5). Each treatment had three replicates of independent plants.
The experiment was carried out in the greenhouse of the Forestry College, Fujian Agriculture and Forestry University under the following environmental conditions: 29.3°C/23°C (day/night); photon flux density of 21 mol quanta m-2 d-1, and ca. 42.7% and 67.7% relative humidity during the light and dark periods of the experiment, respectively. Seedlings were left to grow under this condition for three months. To meet the growth requirements for other nutrients, 50 mL of nutrient solution adjusted to a pH of 5.5 was supplied to each pot compartment every 5 days. Macro-nutrients were supplied according to a modified Hoagland solution as 0.51 gL-1 KNO3，0.82 gL-1 Ca(NO3)2，0.49 gL-1 MgSO4•7H2O and 0.136 gL-1 KCl (Chen et al. 1992). Micro-nutrients were also supplied according to Amon formula as 2.86 gL-1 H3BO3, 0.08 gL-1 CuSO4•5H2O, 0.22 gL-1 ZnSO4•7H2O, 1.81 gL-1 MnCl2•4H2O, 0.09 gL-1 H2MoO4•H2O and 20 gL-1 Fe2+EDTA. The seedlings were watered every two to three days depending on the moisture content of the substrate.
Measurement of root morphological traits
To examine temporal variation in root morphological traits, plants were harvested after 30, 60 and 90 days of treatment application by first draining down the sand from each pot with water, and then the roots were tied up in bundle from each compartment in both homogeneous and heterogeneous P treatments and thereafter carefully pulled out each seedling. All fine roots were collected from the growing media in each compartment. The roots from the different compartments were cleaned with distilled water separately, quickly dried with paper towels, and scanned with digital scanner (STD1600 Epson USA) with non-overlapping tiles. The root morphological traits (total root length, total root surface area, total root volume and average root diameter) from each compartment were determined by WinRhizo root analysis system (Version 4.0 B; Regent Instruments Inc, Canada).
Analysis of P contents
After 90 days of growth, plants were harvested and the P concentration in leaves, stems and roots were determined by digesting 1.0 g of dry plant material in 10 mL of acid mixture (H2SO4:HClO4 10:1). Phosphorus was determined by the molybdenum-blue colorimetric method. For each replicate, three independent samples were analyzed and the average value recorded. P content was computed by multiplying P concentration of the sample by dry mass of the respective organ; P translocation was computed as a ratio of shoot P to total P; and P use efficiency was calculated as the ratio of shoot dry mass to shoot P.
Analysis of biomass production
To determine dry matter of roots, shoots (stems and leaves) and whole-plant biomass, the shoots and roots of the harvested plants were oven-dried first at 105°C for 30 min and then at 79°C until constant mass. Then the dry mass of roots and shoots were determined using sensitive balance (with precision of 0.0001 g). Whole-plant biomass was calculated as sum of root and shoot dry masses.
To examine the foraging strategy of N. reynaudiana seedlings to spatially heterogeneous P supply, four-way between-groups multivariate analysis of variance (MANOVA) was performed using time, patch strength, patch size and high P versus low P compartments as fixed independent variables while taking root morphological traits as dependent variables. The inflation of Type 1 error was controlled by Bonferonni adjustment for multiple comparisons (Quinn and Keough, 2002). With four treatments in our study, time, patch strength, patch size and high versus low P compartments, six possible pair-wise comparisons could be made, thus the Bonferonni adjusted p value was 0.0083 (0.05/6). Results of the statistical analyses were considered significant if p < 0.0083 and to show tendencies if 0.0083 < p < 0.05. Three-Way ANOVA was performed to examine differences in root dry mass and root P contents with respect to patch strength, patch size and between P-replete and P-deficient patches while Two-ANOVA was performed to determine the effects of patch strength and patch size on leaf and stem P contents, P translocation and P use efficiency.
To examine the foraging behavior and its benefit in whole-plant biomass production, root morphological plasticity (Mplast), root physiological plasticity (Pplast) and sensitivity index (SI) were calculated for each level of patch strength and patch size. Mplast is increased root proliferation in nutrient-rich patch than in nutrient-poor patch while Pplast is increased rate of nutrient uptake in the nutrient-rich patch than nutrient-poor patch [11, 18, 41]. Thus, Mplast and Pplast were calculated following Mou et al.  and Zhang et al  as:
Where Rdm and RP were root dry mass and root P contents, respectively. For the homogeneous treatment, Mplast and Pplast were computed as the ratio of root dry mass/root P content in two opposite patches. Mplast and Pplast are expected to be one in the homogenous treatment, but greater than one in the heterogeneous treatment.
SI, as an index of how the total plant biomass responded to heterogeneous P supply compared to homogenous P supply , was computed as the ratio of plant total biomass (total root dry mass and shoot dry mass) in the heterogeneous treatment to that in the homogenous treatment for each level of patch strength and patch size. The following formula was used to calculate SI:
Where TotBm in hetero and homo stands for total biomass in heterogeneous and in homogenous P distribution environment, respectively.
Two-way ANOVA was performed to examine the effects of patch strength and patch size on shoot (leaf + stem) dry mass and total biomass (shoot + root), Mplast, Pplast and SI. Means that showed significant differences were compared by Tukey’s Post hoc test (p < 0.05). All statistical analyses were computed using SPSS Statistical Package (SPSS 20.0, SPSS Ins., Chicago, IL, U.S.A.).