Seasonal Differences in Water-Use Sources of Impatiens hainanensis (Balsaminaceae), a Limestone-Endemic Plant Based on “Fissure-Soil” Habitat Function

: The southwestern mountains of Hainan Island are the southernmost region with tropical karst landform in China. The frequent alternation of dry and wet seasons leads to the loss of the mineral nutrients of limestone, creating karst ﬁssure habitats. Plants living in karst ﬁssure habitats for long periods of time have developed local adaptation mechanisms correspondingly. In the paper, hydrogen–oxygen stable isotope technology was applied to determine the water-use sources of Impatiens hainanensis in the dry and wet seasons, hoping to expound the adaptation mechanism of I. hainanensis in karst ﬁssure habitats to the moisture dynamics in the wet and dry seasons. In the wet season (May to October, 2018), the air humidity is relatively high in the I. hainanensis habitat; in the dry season (November 2018 to April 2019), there is a degree of evaporation. In the wet season, ﬁne-root biomass increases with soil depths, while coarse-root biomass decreases with soil depths; in the dry season, ﬁne-root biomass is lower and coarse-root biomass is higher compared with the wet season. It was found that the average rainfall reached 1523 mm and the main water-use sources were shallow (0–5 cm) and middle (5–10 cm) soil water, epikarst water, and shallow karst ﬁssure water during the wet season; the average rainfall reached 528 mm, and the deep (10–15 cm) soil water and shallow karst ﬁssure water were the main water-use sources during the dry season. Fog water has a partial complementary effect in the dry season. The differences in the distribution of root biomass and each source of water in the wet and dry seasons of I. hainanensis also reﬂect the different water-use strategies of I. hainanensis in the wet and dry seasons. In both dry and wet seasons, I. hainanensis formed a water-use pattern dominated by soil water and shallow ﬁssure water (0–15 cm) under the inﬂuence of the “ﬁssure-soil-plant” system in the karst region.


Introduction
The environment governs the geographical distribution, growth, and developmental status of plants, but the environment is influenced by plants at all times [1]. In the soilplant-atmosphere continuum, plant water is influenced by environmental physical factors such as soil, air moisture limitation, and the regulation of organisms themselves [2]. Water is the most important limiting factor for ecosystem processes and functions in deserts and ecologically fragile areas [3], and it plays a critical role in plant growth and development, survival, and distribution [4]. The intensity and frequency of precipitation determine the survival, composition, structure, and functional assemblage of plant species in ecosystems [5]. In the context of global climate change, changes in precipitation patterns, in Impatiens hainanensis Y. L. Chen is a perennial herb endemic to Hainan Island's limestone regions and is only distributed in the karst tropical mountain rainforests at altitudes of 190-1300 m, in karst fissures [31] (Figure 1). I. hainanensis communities are dominated by low scrub and sparsely distributed with few species, often form dense rocky microhabitats, and limit their own regeneration and settlement due to the environment of rocky microhabitats [32]. Soil water in shallow karst fissures is an important water source for plants, and stone gap microhabitats indirectly affect vegetation growth by influencing soil water and nutrients [33,34]. Therefore, we hypothesized that I. hainanensis has a well-developed and dense root system that climbs rocks, penetrates fissures, and obtains water replenishment in fissured soils and karst crevices. During the rainy season (May to October, average rainfall is 1523 mm), due to increased precipitation, excess rainwater is stored in shallow karst fissures, and I. hainanensis may primarily rely on shallow water. During the dry season (November to April, average rainfall is 528 mm), as reduced precipitation reduces shallow soil water, I. hainanensis may shift their use of water from shallow soil water to deep soil water. Here, we investigate the water use of I. hainanensis in shallow rift habitats during the dry and wet seasons based on hydrogen and oxygen stable isotope techniques and ask the following questions: (1) What are the differences in water δD (δ 18 O) of I. hainanensis during the dry and wet seasons? (2) What is the impact of karst geological drought on water use in I. hainanensis? (3) What are the water-use strategies of I. hainanensis in a "soil-fissure" environment?

Study Area and Locations
Our study area was located in the southwestern mountainous region of Hainan Island (in the southernmost part of mainland China). The Bawangling Emperor Cave of Changjiang County (Sample Plot A), the Exianling of Wangxia Township (Sample Plot B), and the Mihouling of Dongfang City (Sample Plot C) are typical tropical rainforest karst landscapes and belong to the tropical monsoon climate ( Figure 2). As altitude increases, rainfall gradually increases, and relative humidity increases (mist and dew are often seen on the mountain). The geology of the study area primarily consists of limestone with intervals of metamorphic and sedimentary rocks [35]. Soil is mainly black or brown limestone. Soil cover is unevenly distributed, with the bedrock extensively exposed, steep slopes (63% of the slopes at ≥28 • ), thin soil layers, severe vegetation degradation, and low forest cover [35]. There was continuous outcrop of rocky habitat, including stone faces, caves, ditches, and crevices in each sample plot. The vegetation was mostly tropical evergreen monsoon rainforest, deciduous monsoon rainforest, and hilltop scrub. The understory was rich in dead branches, while shrubs and groundcovers were growing vigorously. There was epikarst water in the vicinity of the sample sites and no surface streams passing through; thus we did not consider the effect of surface water on plant water-use sources.

Sample Collection
There were three I. hainanensis sample plots at the study site (Sample Plot A: Bawangling of Emperor's Cave, Sample Plot B: Exianling, Sample Plot C: Mihouling), and each sample plot was set up with nine repeated samples (5 m × 5 m). We collected samples from these sites, three times per month between 6 May 2018 and 6 April 2019. We collected stem bases from three mature I. hainanensis plants from each sample square (three replicates). We collected soil samples using a vertical soil profile by digging within each sample square at a distance from the sampling plants, and soil samples were collected at different depths (0-5 cm, 5-10 cm, 10-15 cm), with three replicates per layer. Standard rain gauges were set up at each sample site to collect rainwater after each rainfall. Nearby epikarst water was collected to represent the stable isotopic composition of the upper stagnant zone and the shallow groundwater in the area. Fog water was collected in the morning between 07:00 and 10:00 when the fog is thickest using collection tanks in the dry season. Sampling was divided between the wet season (May to October, 2018) and dry season (November 2018 to April 2019). Precipitation samples, epikarst water samples, stem water samples, soil water, and fog water collected in the field were quickly packed into sampling bottles, sealed with a sealing film to prevent evaporation, and stored in an incubator at −4 • C to be taken back to the laboratory.

Soil Water Content Determination
One 15 cm deep vertical soil profile was dug around each of the selected I. hainanensis in each sample square. Soil samples were collected at 0-5 cm, 5-10 cm, and 10-15 cm depths in aluminum boxes, with 3 replicates from each layer, and weighed in the laboratory. We dried the soil in an oven at 105 • C to a constant weight (measured every 24 h until the weight remained constant) and weighed. We then calculated the water content of each layer of soil.

Investigation of Root Biomass
We selected nine well-grown I. hainanensis plants from each sample plot and collected the roots in layers by digging a soil profile around each plant. The root system was divided into three layers: 0-5 cm, 5-10 cm, and 10-15 cm. We then collected a volume of 4 cm × 4 cm × 4 cm from each soil layer, brought it back to the laboratory, crushed and sieved it, and then measured the root diameters with vernier calipers and graded according to fine (<1 mm), medium (1-2 mm), and coarse (>2 mm) roots. We weighted the fresh weight of the roots from each layer with an electronic balance, dried it at 75 • C in a drying oven to a constant weight (measured every 24 h until the weight remained constant), and then calculated the dry weight of the roots.

Stable Isotope Measurement
Soil water and stem water were extracted using evaporative cooling through a moisture vacuum extraction system (LI-2000). We used a Finnigan Delta V Advantage Isotope Ratio Mass Spectrometer (Thermo Fisher Scientific, Inc., USA) and Flash 2000 HT elemental analyzer measurements (±<1 ‰ for δD and ±<0.2 ‰ for δ 18 O) at the Flash, Stable Isotope Laboratory, Tsinghua University, to measure all water samples including atmospheric precipitation, soil water, epikarst water, and stem water samples. Hydrogen isotope ratios can be expressed in terms of the difference in thousands (‰) relative to the Vienna Standard Mean Ocean Water (V-SMOW) as follows:

. Utilization of Precipitation by Plants
Based on the δD value of the potential water source, we used the atmospheric precipitation and deep soil water (10-15 cm) as the two water-source ends of the plant to calculate the plant's uptake utilization at this precipitation using binary linear mixing model with the following equation: P p is the proportion of plant water that is at the precipitation end of the scale, δD R is the δD value of the water source at the precipitation, δD SW is the δD value of the water source at the deep soil, and δD p is the δD value of plant stem water.

Contribution of Precipitation to Soil Water
Using atmospheric precipitation and epikarst water as the two source ends of soil water, the contribution of precipitation to soil water in the layer was calculated using a binary linear mixing model with the following equation: P SW represents the proportion of the water source at the precipitation in each layer of soil moisture, δD R is the δD value of precipitation water sources, δD R is the δD g value of epikarst water sources, and δD SW represents the δD value of soil water in each layer.

Plant Utilization of Soil Water at Different Depths
Isotopic values of stem water and potential water sources for each layer were substituted into the IsoSource model to calculate the proportion of soil water utilized by plants for each potential water source.
δX p is the δD and δ 18 O values of stem water. δX i is the δD and δ 18 O values of the potential water source, and fi is the plant's utilization of the potential water sources.

Statistical Analysis
We used the statistical analysis software SPSS 26.0 for all data analysis. We conducted generalized linear model analyses to test the relationship between δD and δ 18 O of atmospheric precipitation, plant stem water, and fog water. We used a mean comparison and ANOVA to compare the differences in δD and δ 18 O between precipitation, plant, fog water, soil water in each layer, and epikarst water after different precipitation events in the wet and dry seasons. We used a multifactor ANOVA to test the effects of precipitation events, soil layers, and the contribution of precipitation to δD and δ 18 O of soil water in each layer.
All statistically significant differences were tested at 0.05 level.

Dynamics of Atmospheric Precipitation δD (δ 18 O) and Precipitation with Wet and Dry Season
The average annual rainfall in I. hainanensis habitat in 2018-2019 was 984 mm, with precipitation in 2018-2019 mainly concentrated in May to October (wet season), accounting for 82% of the annual precipitation, and the average annual evaporation was 1433 mm, 1.5 times the annual precipitation, with little variation in temperature from previous years ( Figure 3). The hydrogen-oxygen isotope ratios for the atmospheric precipitation showed obvious seasonal differences in δD (δ 18 O) values between dry and wet seasons at all three sample plots. During the wet season, the variation of δD (δ 18 O) in the atmospheric precipitation is between −29.1 and −2.1‰, which is a relatively large variation. In the dry season, the variation of δD (δ 18 O) in the atmospheric precipitation is between −11.8 and −4.9‰, which is a relatively small variation ( Figure 4).

Atmospheric Precipitation δD (δ 18 O) Characteristics
According to the relationship diagram of δD (δ 18 O) for all three sample plots ( Figure 5), the local meteoric water line (LMWL) deviates from the global meteoric water line (GMWL) to different degrees due to seasonal differences in moisture sources and moisture factors. The one-variable linear regression analysis of δD on δ 18 O for atmospheric precipitation showed that the slope in the equation for the atmospheric precipitation line at the study sample sites was small and the intercept was large during the wet season compared with the global atmospheric precipitation line. It also showed that the slope and intercept in the equation for the atmospheric precipitation line were both small during the dry season. The slope and intercept of the atmospheric precipitation line equation are smaller in the dry season than when compared with the wet season. This indicates that the air humidity in the I. hainanensis sample plots is relatively low due to local precipitation and that there is a certain degree of evaporation.

Changes of Soil Water Content
During the dry season, the variation in soil water content ranged from 14.1 to 19.8% in Sample Plot A, from 18.6 to 26.5% in Sample Plot B, and from 16.3 to 19.4% in Sample Plot C ( Figure 6). Soil water content of the 0-5 cm surface layer is affected by evaporation less than the soil water content at the 5-10 cm and 10-15 cm depths, and the soil water content gradually increases with increasing soil depth. During the wet season, the variation of soil water content in Sample Plot A ranges from 45.2 to 33.1%, the variation of soil water content in Sample Plot B ranges from 42.2 to 27.3%, and the soil water content in Sample Plot C ranges from 45.1 to 30.2%. The soil moisture content in the 0-5 cm surface layer is higher than at the 5-10 cm and 10-15 cm depths due to the influence of precipitation in the wet season, and the soil moisture content gradually decreases with increasing soil depth. Overall, soil moisture content is significantly higher in the wet season than in the dry season.

Contribution of Precipitation to Soil Water at All Levels
During the wet season, the highest contribution of precipitation to soil water in the 0-5 cm surface layer was between 58.1 and 78.5%, and the higher the amount of precipitation, the higher the contribution of precipitation to soil water at the 0-5 cm surface layer. The contribution of precipitation to soil water in each layer decreased gradually with the increasing soil depth. In the dry season, the contribution of precipitation to soil water in the deep layer of 10-15 cm was between 55.2 and 77.4% at the highest point, the contribution to soil water in the surface layer at the 0-5 cm depth was only 26-46%, and the contribution of precipitation to soil water in each layer gradually increased as the soil depth increased (Figure 8).

Vertical Distribution of Roots
The vertical distribution of root biomass (i.e., the percentages of coarse, medium, and fine roots in each layer of soil in the total root biomass) of I. hainanensis for the three sample sites is shown in Figure 9. The biomass of fine roots accounted for 65.2% of the total root biomass, and the biomass of coarse roots (>2 mm) was mainly distributed at the 0-5 cm depth of surface soil and accounted for 64.8% of the total root biomass. The proportion of fine root biomass increased with the increasing of soil depth, and the proportion of coarse root biomass decreased with the increasing soil depth. During the wet season, the fine roots (≤1 mm) of I. hainanensis were mainly distributed at 10-15 cm deep soil and accounted for 55.6% of the total root biomass, while the coarse roots (>2 mm) were mainly distributed at 0-5 cm surface soil and accounted for 60.3% of the total root biomass. During the dry season, coarse root biomass accounted for 23.7% more of the total root biomass than coarse root biomass in the wet season. During the wet season, the proportion of fine root biomass from the total root biomass of I. hainanensis was 20.3% less than that of the dry season.

Utilization of Dry and Wet Season Precipitation by I. hainanensis
The utilization rate of I. hainanensis to different precipitation conditions during the wet and dry seasons is shown in Figure 12. In the wet season, the utilization rate of I. hainanensis for precipitation in sample plots A, B, and C ranged from 60.1 to 86.3%, 62.1 to 76.6%, and 65.2 to 75.8%. In the dry season, the utilization rate of I. hainanensis for precipitation in sample plots A, B, and C ranged from 50.3 to 68.2%, 22.2 to 38.6%, and 33.4 to 58.5%. Compared to the wet season, I. hainanensis use more precipitation than the dry season. The response of I. hainanensis to wet season precipitation is more rapid and highly utilized.

Dynamics of Fog Water
A linear regression analysis of fog water δD and δ 18 O in the I. hainanensis showed that the linear correlation between soil water δD and δ 18 O was highly significant (Figure 13). This indicates that I. hainanensis were able to utilize both atmospheric precipitation and other water sources (e.g., fog water) in the area. During the dry season, fog water δD and plant water δD are closely related (Figure 14), indicating that I. hainanensis take up and use fog water. Fog water has a partial water recharge effect on I. hainanensis in the dry season.   During the wet season, the δD (δ 18 O) values of stem water, soil water, and epikarst water are very close to the regional and global precipitation line, which indicates that stem water, soil water, and epikarst water mainly came from atmospheric precipitation. The close relationship between the δD of epikarst water and the δD of local precipitation in the wet season indicates that atmospheric precipitation is the main source of epikarst water. The values of epikarst water δD (δ 18 O) were also close to the regional precipitation line and the values of stem water and soil water. This also indicates that I. hainanensis absorbs epikarst water as well as atmospheric precipitation and soil water in this region. During the dry season, atmospheric precipitation has less influence on the δD of epikarst water. Additionally, the δD (δ 18 O) values for plant water, soil water, and fog water were close to the regional and global precipitation lines when compared with epikarst water. This indicates that plant and soil water of I. hainanensis is mainly derived from atmospheric precipitation and fog water. During the wet season, the increase in precipitation leads to the epikarst water level rising, and the deep soil receives rising recharge from epikarst water in addition to recharge from precipitation. In the dry season, there was little recharge from epikarst water to the soil. In both the wet and dry seasons, soil filling makes the shallow karst fissures a major channel for subsurface leakage and the main soil-supporting space for karst areas. Soil filling in shallow karst fissures slows down the rate of water infiltration and leakage to a certain extent, and I. hainanensis take advantage of this unique microhabitat to absorb sufficient water to sustain themselves in karst fissures.

Differences in Moisture Sources of I. hainanensis during the Wet and Dry Seasons
The δD (δ 18 O) values for the atmospheric precipitation in the I. hainanensis sample plots were smaller when compared with global atmospheric precipitation δD (δ 18 O) values. This indicates that because of the influence of the tropical monsoon climate and high precipitation, the fractionation of hydrogen and oxygen isotopes in precipitation is smaller in these sample plots when compared with other regions. The seasonal precipitation variation makes the hydrogen-oxygen isotope fractionation during precipitation less obvious, and there is a close correlation between the hydrogen-oxygen isotope composition of precipitation and various environmental factors such as season [38][39][40][41]. The results from this study show that the isotopic composition of atmospheric precipitation in the I. hainanensis sample plots was correlated with precipitation (i.e., there was a significant precipitation effect), which may be caused by differences in the main water vapor sources between the wet and dry seasons in the I. hainanensis sample plot. In the wet season, the δD values of plant water of I. hainanensis ranged between the δD values of shallow and middle soil water and the δD values of epikarst water, indicating that plant water of I. hainanensis mainly came from soil water and epikarst water. During the dry season, the δD values of I. hainanensis plant water were between the δD values of atmospheric precipitation and deep soil water, indicating that I. hainanensis plant water mainly came from precipitation and deep soil water. In this study, the utilization of shallow and middle soil water by I. hainanensis reached 59.6% and 36.1% in the wet season, while in the dry season, the utilization of middle and deep soil water by I. hainanensis reached 15.9% and 77.5%. This difference in water use reflects the functional characteristics of the soil type and the community structure of I. hainanensis in shallow karst rift habitats, as well as the adaptation of I. hainanensis to different water conditions between the dry and wet seasons. The temporal and spatial dynamics of soil water δD are influenced by differences in precipitation δD and precipitation magnitudes, which are dependent on the intensity of precipitation and evaporation of soil water δD values and infiltration [42][43][44][45]. In the wet season, soil water δD values were in between precipitation values and epikarst water values. This indicates that the soil water in the I. hainanensis sample plots was mainly derived from atmospheric precipitation and epikarst water. As the intensity of precipitation increases during the wet season, the rate of water displacement in the soil becomes faster, leading to faster infiltration of precipitation into the soil, and influencing the precipitation on the soil to become more pronounced. In the wet season, the rise in the epikarst karst water level causes the deep soil water and epikarst water to be recharged by atmospheric precipitation. In the dry season, the epikarst water level decreases, the recharge of soil water mainly comes from atmospheric precipitation, the surface soil water is recharged of soil water, and surface soil water mainly comes from atmospheric precipitation. Surface soil water is affected by evaporation and precipitation, resulting in an increased evaporation rate, which is often subject to drought stress [46][47][48][49][50], and is consistent with the findings of Xu et al. [51]. As precipitation increases, the rate of infiltration of precipitation from the soil surface to the deeper soil layers accelerates, the effect on deep soil water δD becomes more obvious, and the contribution of precipitation to soil water gradually increases.

Effect of Karst Drought on Water Use of I. hainanensis
During the dry season when precipitation is low and there is an insufficient water supply at the shallow layer, I. hainanensis mainly use stable deep water, water from shallow fissures, and fog water. This results in I. hainanensis distributing more coarse and fine roots when compared with the wet season when plants accelerate water uptake and use. Plants such as Cyclobalanopsis glaucoides and Pistacia weinmannifolia in karst areas also use more deep soil water in the dry season and switch to medium soil water in the wet season [52]. Plants will preferentially use water stored in the surface soil and shallow fissures when rainfall is relatively abundant and recharge the aquifers below the surface, which reduces the energy used to absorb water and promotes plant growth [53][54][55][56][57]. In addition, highaltitude karst areas are foggy in the dry season, with strong transpiration and high plant water demand, soil water cannot be supplied to the plants in time, and thus I. hainanensis use some of the fog water to increase their own water-use efficiency and reduce plant transpiration water loss.
During the wet season, precipitation is adequate, soil water is effectively recharged, water-use layers are close together, soil water is mainly concentrated in the shallow and middle layers, and I. hainanensis turns to the short-term available water in the shallow and middle layers. Related studies have shown that the source of water for perennial desert succulents in arid desert environments is deep soil water or groundwater, with other annuals with shallow roots using shallow groundwater recharged by precipitation [57]. This is similar to the fact that I. hainanensis mainly use deep soil water in the dry season, and some plants in arid areas shift from mainly absorbing deep soil water or groundwater to mainly absorbing shallow soil water recharged by rainfall with the onset of the rainy season and in areas with abundant precipitation may also develop upper and lower double root systems, using both deep and shallow soil water [58]. However, considering the specific geology of karst and desert areas, local plants use water not only from soil water but also deep groundwater in the rainy season for desert plants, while karst plants also use shallow fissure water and surface karst water, and even fog water due to climatic factors [59]. Plants in different arid ecosystems also develop different water-use strategies to adapt to the different local habitats. The different sources of water use in the wet and dry seasons of I. hainanensis reflect both similar water-use patterns to those of plants in arid regions, while at the same time showing local adaptation mechanisms to tropical karst seasonal precipitation and moisture environments.

Water Utilization of I. hainanensis under "Fissure-Soil" Habitat Characteristics
Plants can respond to changes in their environment with changes to morphology, physiology, and other characteristics that allow them to exhibit certain functional strategies [60][61][62]. Plants in karst regions have strong root systems, lithogenic, penetrating fissures, climbing rocks, and even roots encasing rocks to find water and nutrients in rock fissures and surface karst zones [63,64]. In this study, the coarse roots (>2 mm) of I. hainanensis were mainly distributed at the 0-5 cm surface layer of soil and accounted for 64.8% of the total root biomass in the dry season, while the fine roots (<1 mm) were concentrated at the 10-15 cm deep layer of soil and accounted for 65.2% of the total fine root biomass. Compared with the dry season, the coarse roots (>2 mm) and fine roots (<1 mm) of I. hainanensis in the wet season were less distributed in both surface and deep layer soil. The distribution of coarse roots (>2 mm) and fine roots (<1 mm) was lower in both topsoil and deep soil layers when compared with the dry season. Differences in plant strategies between the wet and dry seasons lead to different distributions of fine and coarse roots. Differences in root biomass distribution determined the ability of these plants to absorb and utilize soil water and the extent they can take up nutrients [25,65]. The depth of root distribution of desert plants is related to the location of the soil water used by plans [66,67]. Shallow-rooted plants are sensitive to precipitation and have a high capacity to use shallow soil water, but rainfall tends to evaporate, and shallow soil moisture is very variable; deep-rooted plants can penetrate several meters below the soil [68]. Many perennial plants in semi-arid and seasonally dry areas have dimorphic root systems that can use upper soil water in the wet season and take up deep soil water in the dry season [69]. In contrast, the root distribution depth of some plants in the karst area is related to the location of soil water, different seasonal root distribution strategies for water uptake, and utilization under the influence of alternating wet and dry seasons. In the dry season, I. hainanensis mainly utilized part of the precipitation and deep soil water. This indicates that soil water stored in shallow karst fissure water was the main source of plant stem water during this time. Due to the bare rocks in the I. hainanensis habitat, the surface soil moisture decreased and leaked severely, increase of plant coarse root distribution was more conducive to I. hainanensis climbing rocks and absorbing water deep into the shallow fissured soil, and the water absorption was more relatively concentrated. Shallow karst fissure water became a secondary water source for plant water utilization [70][71][72]. In the wet season, the increase in the distribution of fine roots also leads to an increase in the capacity for I. hainanensis to absorb and utilize soil water and the plant's range of water uptake. I. hainanensis mainly utilized part of the precipitation, shallow, and middle soil water. This indicates that I. hainanensis has reduced the use of shallow karst fissure water in favor of surface soil; however, shallow karst fissure water is still the main source of water for plants. The water content of surface soils on karst slopes declines rapidly after rainfall, while precipitation infiltrates into shallow fissures soils and remains stored in the shallow fissures for a long time. Shallow fissures are an important part of the surface karst zone and the main way of subsurface leakage for karst slope soils. Therefore, shallow karst fissures serve as the "container" for soil storage in special karst habitats [73,74]. The "fissure-soil" system makes the shallow karst fissures not only a pathway for subsurface soil leakage but also provides roots the space, water, and nutrients needed for growth. The "fissure-soil" system changes the water process in karst slope, with fissured soil slowing the infiltration rate of fissured water, and stores some water in the soil [26,27,75]. The "fissure-soil" system plays an important ecological function in supporting karst ecosystems and regulating hydrological processes.

Conclusions
In this study, we used stable isotope techniques to analyze the seasonal variation of water sources in the dry and wet seasons of I. hainanensis in a karst "fissure-soil" habitat. In the wet season, the main water sources of I. hainanensis were shallow and middle soil water, shallow karst fissure water, and epikarst water. Among them, the I. hainanensis in the Emperor Cave area was relatively more affected by the precipitation in the wet season when compared with the other two sample plots. Response was faster, and the utilization rate was higher, which may be related to the fact that this sample plot is located at the foot of the mountain, the slope is gentle, the soil thickness is larger, and the gravel content is lower. In the dry season, I. hainanensis uses deep soil water and shallow karst fissure water for its main water sources. As precipitation decreases and evaporation increases, most of the precipitation is lost through soil evaporation, surface run-off, and leakage, and precipitation becomes invalid. Most of the water in the shallow fissures is consumed, and only a small part is stored in the shallow fissures filled with litter and organic matter. The effective water in the soil layer decreases rapidly, and the effective water in the epikarst zone also slowly weakens. In the contiguous rocky habitats in the karst area, shallow karst fissures provide space for soil and growth space for plant roots, forming a "fissuresoil-plant" system. I. hainanensis has formed a water-use mode dominated by shallow karst fissure water. This shallow karst fissure water can help the I. hainanensis community maintain water consumption and can allow us to forecast the water-use strategies for this plant in this habitat under changes between the wet and dry seasons.