Water Use Characteristics of the Common Tree Species in Rock-dominated and Thin-soil Environments in Subtropical Monsoon Climate Region


 Variations in precipitation pattern under climate changes influence water availability that have important implications for plants water use and vegetation sustainability. However, the water use characteristic of the main tree species under different temporal-spatial of water availability remain poorly understood, especially in high temporal-spatial heterogeneity area, such as subtropical monsoon climate region of China. We investigated water use characteristics of the most widely and common natural trees, Mallotus philippensis and Celtis biondii , in edaphic and rocky habitats. We measured the δD and δ 18 O values of xylem and soil water and water potential of plant leaves during the wet season in 2020. The results showed that the two species mainly absorbed soil water from shallow layers and switched for deeper layers during the late of the wet season in both habitats. But the plant water sources were different in edaphic and rocky habitats when the antecedent precipitation was much high, deep layers soil water in the former and still shallow layers in the latter. The two species had no significant differences in water uptake depth, but notably distinction in the diurnal water potential ranges. M. philippensis maintained less negative predawn and midday water potential, whereas C. biondii showed higher diurnal water potential ranges. Besides, the water potential of C. biondii were negatively associated with antecedent precipitation amount. These results indicate that there is significant eco-physiological niche segregation but no ecohydrological segregation co-existing species in communities. Besides, antecedent precipitation amount and habitat differences were the main factors influencing the plant water uptake depth. While the relationship between leaf physiological traits and water availability was affected by the species types, rather than the habitats. Furthermore, during the long drought in growing season, there are probable divergent responses of M. philippensis and C. biondii , such as growth restriction and hydraulic failure. But when the precipitation is heavy and long, these natural species could increase the ecohydrological linkages between ecosystem and the deep-layer system in edaphic habitat.

poorly understood, especially in high temporal-spatial heterogeneity area, such as subtropical 23 monsoon climate region of China. We investigated water use characteristics of the most widely 24 and common natural trees, Mallotus philippensis and Celtis biondii, in edaphic and rocky habitats. 25 We measured the δD and δ 18 O values of xylem and soil water and water potential of plant leaves 26 during the wet season in 2020. The results showed that the two species mainly absorbed soil water 27 from shallow layers and switched for deeper layers during the late of the wet season in both 28 habitats. But the plant water sources were different in edaphic and rocky habitats when the 29 antecedent precipitation was much high, deep layers soil water in the former and still shallow 30 layers in the latter. The two species had no significant differences in water uptake depth, but 31 notably distinction in the diurnal water potential ranges. M. philippensis maintained less negative 32 predawn and midday water potential, whereas C. biondii showed higher diurnal water potential 33 ranges. Besides, the water potential of C. biondii were negatively associated with antecedent 34 precipitation amount. These results indicate that there is significant eco-physiological niche 35 segregation but no ecohydrological segregation co-existing species in communities. Besides, 36 antecedent precipitation amount and habitat differences were the main factors influencing the 37 plant water uptake depth. While the relationship between leaf physiological traits and water 38 availability was affected by the species types, rather than the habitats. Furthermore, during the 39 long drought in growing season, there are probable divergent responses of M. philippensis and C. dense and shallow fine root system absorbed more water from the soil surface layers and 70 habitat with seasonal sampling during the growing season on the subtropical monsoon climate 114 region of China. The primary objectives of the study were to (ⅰ) evaluate the water uptake pattern 115 of species in two habitats for the temporal-spatial heterogeneity of water availability; (ⅱ) 116 investigate how the leaf water potential and water use efficiency of species responds to season 117 variation with different precipitation; (ⅲ) explore the relationship between plant physiological 118 traits and water uptake pattern. The first hypothesis is that the plant water use pattern varies in the 119 seasonal changes with different precipitation in two habitats, and the second is that the plant water 120 uptake would coordinate with physiological characteristic of species, coupling with water 121 availability. 122 123 2 Materials and methods 124

Study area and sampling sites characteristics 125
The study site is located in a small catchment (area = 1.14 km 2 ) in the Lutou Observation and 126 Research Station for north Luoxiao national forest park (28°31′7″-28°38′N, 127 113°51′52″-113°58′24″E), which is situated in the northeast of Hunan Province, China. Along the 128 part of the slope, the soil is mainly thin with a depth of 10-30 cm. At the foot of the slope and in 129 the depression, the soil is relatively thick of 70-90 cm with amounts of rock fragments. The other 130 part of the slope has a high exposed dolomite ratio, while the soil occurs discontinuous only in 131 carbonate rock gaps. Thus, the habitats were variable with the different outcrop ratio, such as 132 edaphic habitat with low outcrop ratio, continuous broken rock habitat with patches of soil, 133 isolated outcrops habitat, and so on. Springs sometimes appear at the bottom of hillslopes during 134 the rainy season or after rains in the drought season. The region has a subtropical mountainous monsoon climate, with mean annual precipitation of 1450.8 mm and an annual temperature of 136 18.5℃. The wet season lasts from late April to the end of September and provides >60% of total 137 annual rainfall, while the dry season extends from December to February (Nie et al., 2012). The 138 growing season spans from April to October. 139 The area was undergone dramatic deforestation caused by farming, grazing, and burning, then According to the distribution of these typical habitats, thick soil with rock fragments habitat 148 (edaphic habitat for short) and one continuous stone outcrops with soil fragments (rocky habitat 149 for short) were chosen at the foot of the Northwest-facing hillslope in two 20×20m sample plots 150 ( Fig. 1). The two habitats are 50 meters apart, while the elevation difference was about 5 meters. 151 In the edaphic habitat, the soil is relatively thick (about 90cm deep), horizontally interrupted by 152 smally outcrops, and vertically interrupted by small rocks. At the soil profile, the upper layer soil 153 (0-30 cm) is well-drained, while the lower layers (30-70 cm) are sticky with a low soil saturated 154 hydraulic conductivity (Ks) (Fu et al., 2015). Underneath the soil is a high-weathered dolomite 155 bedrock zone (70-90cm). The outcrop ratio is about 20% in this habitat. This habitat is covered by 156 dense vegetation, with Mallotus philippensis, Rhus chinensis, and Celtis biondii dominating the tree layer. Vitex negundo and Pyracantha fortuneana dominate the shrub layer. In the rocky habitat, 158 the outcrop ratio is more than 80%, and the range of height from the top of the outcrop to the soil 159 in the rock gaps is from 0.3m to 3m. The soil inlaid in the rock in a spotty pattern and is 160 discontinuous (average 30cm deep). Similarly, high-weathered dolomite bedrock zone is under the 161 soil. The vegetation is sparse in this habitat. The tree layer is dominated by M. philippensis, Ficus 162 tinctoria, and C. biondii, and the shrub layer is dominated by V. negundo. There is a intermittent 163 spring outflow near the two habitats at the bottom of the hillslope. 164

Plant and soil sampling 165
Plant and soil sampling were conducted simultaneously at the two habitats bimonthly on Jun 12 166 (middle wet season with high antecedent precipitation), August 5 (middle wet season with low 167 antecedent precipitation) and October 18 (early dry season) 2020. Besides, we also sampled on 168 May 18 in early wet season with 20-day drought. Two common species, adult M. philippensis 169 (DBH of from 5 to 11cm) and C. biondii (DBH of from 6 to 12cm) at each of the habitats, were 170 selected for the study. We selected four individuals per species for analysis. The leaf and plant 171 xylem samples from every selected plant were collected in each habitat. every selected plant was 172 collected in each stand-age tree per month. The fully sun-exposed, mature and healthy leaves in 173 the upper canopy from each selected plant were collected in different directions on each sampling 174 date. The leaves were mixed and packed into craft paper bags and brought them back to the 175 laboratory for measuring the plant leaves δ 13 C. Shoots ranging from 0.3 to 0.5 cm in diameter and 176 50-70, 70-90 cm) with an auger (sampling only at 70cm deep in the rocky habitat); and, five 180 replicates were collected at each layer. Among them, the high-weathered bedrock samples were 181 collected between 70-90cm in the edaphic habitat and 50-70cm in the rocky habitat. A subsample 182 of the soil samples was stored at -20°C for isotopic analysis, whereas the remainder of the 183 samples were sealed for measurement of gravimetric soil water content, obtained by oven drying 184 for one day. The volumetric water content (VWC) was converted according to gravimetric water 185 content and bulk density of each layer. 186

Precipitation and spring sampling 187
Rainwater samples were routinely collected for each rain event above 5mm from May 2020 to 188 December 2020. The isotopic values of precipitation were not collected from January to April due 189 to the COVID-19 pandemic impacting. The collection equipment was designed based on the new 190 device for monthly rainfall sampling for the Global Network of Isotopes in Precipitation (Agency, 191 2002). The rainwater samples were stored in cap vials, wrapped in parafilm and stored in a freezer 192 until the analysis of stable isotopes. Temporal distribution of rainfall data and other meteorological 193 data were collected at a meteorological station located in the middle of the same small catchment. 194 Spring water discharged from June 1 to November 29, but cutoff between July 25 to August 29. 195 The spring was sampled regularly during the outflow period. Both rainwater and spring water 196 were stored in cap vials, wrapped in parafilm, and frozen until stable isotope analysis. 197

Isotopic analyses 198
The water was extracted from xylem and soil using automatic cryogenic vacuum distillation 199 Sciences. The  13 C in the plant leaves were analyzed using an isotope ratio mass spectrometer 204 (IRMS, MAT253, Thermo Fisher Scientific, Bremen, Germany). 205 The isotope composition is reported in  notation relative to V-SMOW as 206 (1) 207 Where X represents D, 18

Leaf water potential 215
Predawn and midday water potentials (pd and md, respectively) of leaves were measured in 216 the wet seasons (simultaneously with isotope sampling) with a pressure chamber (PMS 217 Instruments Co., Corvallis, OR, USA). Samples (n= 5 per species) were collected from branches 218 that were fully exposed to the sun, 2/3 of the way up of the canopy, at least 2 m above ground and 219 for predawn water potential between 4:00 to 6:00 h and midday measurements were subsequently 220 conducted between 12:00 and 14:00 h on the same day. and spring as the potential deep water sources. Plant water source partitioning was determined by 224 the Bayesian mixing model MixSIAR (version 3.1.7) (Stock and Semmens, 2013). The raw 225 isotopic ratios of the xylem water were input into MixSIAR as the mixture data. The averages and 226 standard deviations of the soil water isotopes in the different soil layers were the source data. The 227 discrimination was set to zero for both δD and δ 18 O because there is generally no isotopic 228 discrimination of water during plant water uptake by roots (Ehleringer and Dawson, 2010). 229 For the subsequent analysis and comparison, the plant water sources were divided into shallow 230 Independent-samples T test and One-way ANOVA were used to detect the differences in plant 241 water sources and water potential among the species, habitats and their seasonal differences. Post 242 hoc comparisons were based on Tukey's HSD. Moreover, Pearson correlation was used to conduct 243 the correlation analysis, and the figures were plotted with Origin software version 9.0. 244

Meteorological factors and isotopic compositions of precipitation 246
The total precipitation was approximately 2121 mm in 2020 (Fig. 1), 52.69% higher than the 247 multiyear mean (1961-2017) precipitation (1450.8 mm) (Ding et al., 2020). While the distribution 248 of rainfall was temporally uneven (Fig. 1), 79.32% of the rainfall occurred during the wet season. 249 It was noted that there are two extreme precipitation events in Sep. 7 (282.2mm) and Jun. 7 250 (115.2mm). On the other hand, no effective rainfall records were collected in the 20 days from Apr. 251 to May in the wet season. The first sampling took place after the 20-day drought. The other three 252 samplings were conducted in the sunny day after 1-2 days of rainfalls. The accumulated 253 precipitation amount ten days before the last three samplings were 283.6mm, 49.4mm, and 254 55.4mm, respectively. 255 The isotopic compositions of the precipitation showed a large fluctuation during the study 256 period (Fig. 1). The mean  D of the precipitation was -48.69 ‰, the mean  18 O of the 257 precipitation -7.88 ‰. The relatively depleted isotopic values of precipitation occurred when it 258 rains continuously for a long time with high precipitation. The D of ten days precipitation before 259 three samplings in middle and late wet season were ranging from -23.55 to -57.52 ‰, -34.54 to 260 -68.36 ‰, -40.76 to -51.02 ‰, respectively. The  18 O of precipitation before three samplings were 261 ranging from-5.27 ‰ to-8.15 ‰, -7.68 ‰ to-9.65 ‰, -6.54 ‰ to-7.4 ‰, respectively. 262

Variation in isotopic composition of soil water and spring 268
The δD and δ 18 O values of soil water in the different habitat varied with soil depth and season 269 (Fig. 2, Fig. 3). In edaphic habitat, the average δD value of soil water was -45.56 ± 16.05 ‰ 270 (mean ± S.D.), and average δ 18 O value was -6.55 ± 1.73 ‰. The average δD and δ 18 O values of 271 soil water in rocky habitat were −44.6 ± 16.58 ‰ and −6.7 ± 1.96 ‰, respectively. There were no 272 significant differences (p=0.84 for δD, p= 0.79 for δ 18 O) in the soil isotopic compositions in the 273 different habitats. In the early wet season with 20-day drought, the soil water isotopes displayed 274 depleted with soil depth (Fig. 2a, Fig. 3a). In the middle wet season with high precipitation before sampling, D and  18 O values of water at soil profile were consistent with recent rainfall values 276 (Fig. 2b, Fig. 3b). In late two sampling, the soil water isotope composition converged at the top 277 and bottom layers, which were similar to recent rainfall values (Fig. 2c, d)

Variations in soil water content and water uptake patterns 290
The VWCs of the two habitats displayed clear vertical and seasonal variations (Fig. 4). The 291 average VWCs were 43.42 ± 7.68% in edaphic habitat and 38.24 ± 8.42% in rocky habitat during 292 the study periods. The VWCs of shallow soil layers in the two habitats differed significantly 293 (p<0.001). However, the middle and deep soil moisture showed no significant differences in the 294 two habitats. In the early wet season with 20-day drought, the VWC of the shallow layer was 295 lowest in the two habitats and the soil moisture increased with depth (Fig. 4a). In the middle and 296 late wet season, the VWC exhibited slightly increasing tendency along the soil profile in the 297 edaphic habitat but decreasing tendency in the rocky habitat. Among them, the soil moisture in 298 edaphic habitat had the highest values in middle wet season (Fig. 4c).

303
The two tree species mainly took up soil moisture throughout the wet season in two habitats 304 (Fig. 5). While the plants used different layers soil water in seasonal variation with no significant species differences (p>0.05). In the early wet season with 20-day drought and middle wet season 306 with low precipitation before sampling, both M. philippensis and C. biondii in two habitats 307 utilized the largest proportion of shallow soil water (64.97%, 0-30cm). In the middle wet season 308 with high precipitation before sampling, two species in rocky habitat also absorbed more than 309 67.14% of its water from shallow soil layers. While the mean water uptake fractions of the two 310 tree species in edaphic habitat were 64.45% for middle and deep soil layers (50-90cm,). In the late 311 wet season, the M. philippensis and C. biondii in edaphic habitat obtained more than 74.82% of its 312 water from the shallow and deep soil layers. While in rocky habitat, the two species mainly 313 extracted soil water from shallow and middle layers (82.13%). 314

Variations in water potential of plant leaves and its linkage water uptake depth
The ψpd and ψmd of the two species exhibited profoundly seasonal variation during the sampling 320 period (p<0.01), which were less negative in the middle wet season than those in the early and late 321 wet season (Table 1)

Changes in plant uptake depth in response to precipitation amount in two habitats 351
The response of plant water source proportion in each soil layers to precipitation amount ten 352 days before sampling were distinct in two habitats (Fig. 7). In the edaphic habitat, tree species 353 absorbed less water from shallow layers and more deep soil water with the precipitation increases 354 (Fig. 7a, c). While the trees maintained high water uptake from shallow layers in the rocky habitat 355 whatever precipitation variations (Fig. 7d). Meanwhile, there were significant negative linear 356 relationships between the water source proportion of middle and deep soil layers and precipitation 357 (Fig. 7e, f).

Contrasting leaf water potential responses to precipitation amount between two species 367
The response of the diurnal ranges of water potential to precipitation amount ten days before 368 sampling were different in two tree species (Fig. 8). The Δψ of M. philippensis did not increase 369 from no rain to high rainfall with relatively low values in two habitats. However, the diurnal 370 ranges of water potential for C. biondii showed lower values with the precipitation increases in the 371 edaphic and rocky habitat. biondii also demonstrated that they were both in sufficient water supply in the two habitats (Fig.  432   6). 433

Water use characteristics and physiological changes in the different tree species 434
The two coexisting plants either in the edaphic or rocky habitat exhibited no significant 435 differences in water uptake pattern with seasonal changes, indicating that they had the same 436 eco-hydrological niche and no water source segregation. This result was inconsistent with a 437 previous study in the similar study area, which found that the tree and shrub had different water 438 use sources in the dry season (Nie et al., 2012). But other studies showed that the coexisting 439 species usually had water competition in mixed stand in non-karst regions (Liu et

498
In this study, stable isotope technique and pressure chamber were applied to detect the seasonal 499 water use characteristic of two common tree species in edaphic and rocky habitats on the 500 subtropical monsoon climate region. The results showed that the two species mainly absorbed soil 501 water from shallow layers and switched for deeper layers during the late of the wet season in both 502 habitats. But the plant water sources were different in edaphic and rocky habitats when the 503 antecedent precipitation was much high, deep layers soil water in the former and still shallow 504 layers in the latter. The two species had no significant differences in water uptake depth, but 505 notably distinction in the diurnal water potential ranges in the same habitat. These results indicate 506 that there is significant eco-physiological niche segregation but no ecohydrological segregation for 507 co-existing species in communities. Besides, antecedent precipitation amount and habitat 508 differences were the main factors influencing the plant water uptake depth. While the relationship 509 between leaf physiological traits and water availability was affected by the species types, rather than the habitats. Thus, during the long drought in growing season, there are probable divergent 511 responses of M. philippensis and C. biondii, such as growth restriction and hydraulic failure. But 512 when the precipitation is heavy and long, these natural species could increase the ecohydrological 513 linkages between ecosystem and the deep-layer system in edaphic habitat. 514

515
We thank the associate editor and anonymous reviewers for their helpful and valuable 516 comments that improved this manuscipt. This research was funded by the National Natural 517 Science Foundation of China (41807162, 31860236) and the Hunan Province Natural Science 518 Foundation (2019JJ50994). 519 520 We declare that we have no financial and personal relationships with other people or 521 organizations that can inappropriately influence our work, there is no professional or other 522 personal interest of any nature or kind in any product, service and/or company that could be 523 construed as influencing the position presented in, or the review of, the manuscript entitled "Water 524 use characteristics of the common tree species in rock-dominated and thin-soil environments in 525 subtropical monsoon climate region". 526