Drought resistance of ten ground cover seedling species during roof greening

Roof greening is an important national policy for maintaining the hydrological balance in China; however, plant growth is limited by drought stress. This study aims to identify strong drought resistant plant species for roof greening from ten common species: Paeonia lactiflora, Hemerocallis dumortieri, Meehania urticifolia, Iris lactea var. chinensis, Hylotelephium erythrostictum, Sedum lineare, Iris germanica, Cosmos bipinnata, Hosta plantaginea, and Dianthus barbatus. By controlling the soil relative water content (RWC), we designed three treatments: moderate drought stress (40±2% < RWC < 45±2%), severe drought stress (RWC < 30±2%) and well-watered control (RWC > 75±2%). After the seedlings were provided different levels of water, their membrane permeability (MP), chlorophyll concentration (Chl), and superoxide dismutase (SOD), peroxidase (POD) and ascorbate peroxidase (APX) activity were measured. Finally, the membership function method was used to assess the drought resistance of these species. The results showed that C. bipinnata and M. urticifolia were not suitable for moderate or severe drought stress and did not survive. The other species presented variations in physiological and biochemical parameters. The MP of He. dumortieri, I. lactea and Ho. plantaginea showed minor changes between the well-watered control and drought stress. Most of the species showed reduced SOD activity under moderate drought stress but increased activity under severe stress. All of the plant species showed decreases in the protective enzymes POD and APX with increasing drought stress. The membership function method was applied to calculate the plant species' drought resistance, and the following order of priority of the roof-greening plant species was suggested: He. dumortieri > I. germanica > I. lactea > D. barbatus > Hy. erythrostictum > S. lineare > Ho. plantaginea > P. lactiflora.


Introduction
Roof greening, which is regarded as the "fifth surface greening", is one of the fundamental measures for sponge cities and represents an important national policy for improving the relationship between city development and nature protection to maintain the hydrological balance in China [1,2]. As an important supplement of urban landscaping, roof greening can help a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 mitigate the urban heat island effect [3], improve air quality [4] and enrich the biodiversity of cities [5]; hence, this landscaping style has expanded throughout all of China. Green roofs can be categorized roughly into two types: those that consist of diverse types plants (shrubs, trees, grasses, and flowers), namely, intensive green roofs (IGRs), and those that consist of simple herbaceous plant species, namely, extensive green roofs (EGRs) [6]. To grow on roofs, plants face many challenges. Taking Beijing as an example, plants that compose green roofs suffer from restricted rainfall in winter, spring, and autumn and evaporation always increases with high summer temperatures. Drought is considered as one of the most common environmental stresses that currently affects plant growth [7,8]. When plants experience drought stress, reactive oxygen species (ROS) are produced [9] including singlet oxygen ( 1 O 2 ), superoxide radical (O 2 -), hydroxyl free radical (•OH), and hydrogen peroxide (H 2 O 2 ) [10]. ROS can reduce crop productivity and plant viability because they can cause oxidative damage to proteins, DNA, and lipids [11]. Accordingly, drought stress can not only disrupt leaf membrane permeability (MP) [12] but also reduce the chlorophyll concentration (Chl) [13] and the activity of superoxide dismutase (SOD) [14], peroxidase (POD) [15] and ascorbate peroxidase (APX) [16]. These indicators are used for measuring the degree of plant drought stress, and they are usually analyzed as a whole due to their close associations, such as the ability of antioxidative enzymes SOD, POD and APX [17] to quench ROS and protect the cell from damage.
Biological membranes are crucial aspects of living systems that control the organization and distribution of different chemical components [18], and maintain sufficient water in plant tissue to protect the organism from dehydration and carboxylation and prevent enzymes from inactivating [19]. Liposomes are colloid vesicles composed of a lipid bilayer membrane and a watery internal compartment [20], and they serve as transport carriers for the efflux of secreted proteins. Low temperature [21], drought [22], salt [23] and heavy metal [24] stress break the stability of the plant cell membrane system and proteins, thereby increasing biofilm fluidity, altering the conformation of proteins, and then leading to physiological, biochemical and metabolic imbalance and abnormalities [25]. MP is determined by electrolyte leakage [26] and could be estimated by measuring electrolytes seeping from the plant cells under environmental stress. Generally, the greater the value of MP, the more cell damage there is. Under the same water condition, a lower MP value implies a stronger adaptation of the plant species to the environment.
Chlorophyll, a green pigment, is widely distributed in plant leaves and stems [27]. It helps convert absorbed solar radiation into stored chemical energy [28] and binds to proteins within chloroplasts and affects the light-harvesting capability and photosynthesis of plants [29,30]. Upon drought stress, plant Chl is mainly affected by the physical destruction of chloroplasts and the inhibition of Chl a and Chl b functionality. Drought stress also causes the chloroplast matrix lamella to bend and swell [31], thereby impeding Chl synthesis and reducing its production [32]. In addition, reactive oxygen species (ROS, 1 [34]. SOD is an important enzyme that is ubiquitously expressed in aerobic organisms and catalyzes the dismutation of superoxide anions to hydrogen and molecular oxygen, which constitutes the first line of defense against ROS at the cellular level [35,36]. Based on the prosthetic metal at the active site, SODs are classified into three groups, namely, CuZn-SODs, Mn-SODs, or Fe-SODs [37], of which Mn-SODs are closely related to mitochondria [38] and CuZn-SODs are mainly located in the cytoplasm and chloroplasts of plant cells [39]. McCord and Fridovich [40] described the principle chemical reaction under the elimination of ROS by SODs as O 2 Existing in peroxisomes, glyoxysomes, vacuoles, the nucleus, and the extracellular matrix, SODs play a critical role in drought tolerance [41]. The SOD activity reflects the ability of plant species to adapt to environmental stress. Higher SOD activity values represent a stronger adaptation ability [42].
The antioxidant enzyme POD can scavenge and breakdown ROS [43] via the reaction RH 2 +H 2 O 2 !2H 2 O+R, in which H 2 O 2 is thoroughly converted into H 2 O [44,45]. Chen et al [46] and Wu et al [47] found that increased POD activity helped cucumber and Dendrobium moniliform alleviate oxidative damage under drought stress. POD can further scavenge peroxides induced by SOD, and the synergistic action between these enzymes constitutes the protective enzyme system of the organism. Changes in the activity of these enzymes under stress may reflect the plant resistance ability in adverse environments.
APX is a member of the class I heme peroxidases and an important enzyme in plant antioxidant defense systems, and APX with several isoenzymes has a strong ability to scavenge ROS [48][49][50][51]. APX has been found in most eukaryotes, including higher plants [52], where it plays a key role in the metabolism of H 2 O 2 . Stronger APX activity would more quickly remove H 2 O 2 , thus preventing oxidative damage [53]. Different kinds of POD isoenzymes have obvious tissue and organ specialization. Similarly, APX is distributed in chloroplasts and the cytoplasm. POD and APX differ in their composition, structure, substrate specialization affinity, and stability during the purification process [54].
The membership function method is widely used to assess plant stress resistance. For example, it is has been used to assess the drought tolerance of Malus According to "Beijing local standards, roof greening specification (DB11/T 281-2005)" [62], more than 20 species can be used for ground cover and roof greening. Studies have focused on most of these species for their drought resistance, although these studies were limited to one species or one family. Because of the lack of studies comparing the drought resistance between these species, this study aims to screen plant species with strong resistance under drought stress to provide government policymakers with scientific plant species choices to improve plant survival rates and save maintenance costs during roof greening. carry out the experiments here. Also, the 10 plant species used for experiments were all market-purchased, common ground cover plants. These plants are not endangered rare plants, not be protected.

Seedlings
One-year-old seedlings of ten species, i.e., Paeonia lactiflora, Hemerocallis dumortieri, Meehania urticifolia, Iris lactea var. chinensis, Hylotelephium erythrostictum, Sedum lineare, Iris germanica, Cosmos bipinnata, Hosta plantaginea, and Dianthus barbatus, were provided by the Yu-Quanying flower market, a large and popular wholesale market in Fengtai District, Beijing that supplies most ornamental plants for Beijing City. The plant species present various propagation modes and other characteristics ( Table 1).

Field soil collection and preparation
The field soil was collected from a wild wetland area in Milu Park, and it was then air-dried and ground to powder for the transplantation experiment. The soil was moderately saline (pH = 7.89) and presented available nitrogen, available phosphorus, and available potassium contents of 24.7 mg�kg -1 , 18.9 mg�kg -1 , and 322 mg�kg -1 , respectively [63].

Plant transplanting
The transplantation program was as follows: first, approximately 400 g powdered soil was placed in a plastic pot that was 20 cm tall and 13 cm in diameter and had 3 small holes at the bottom for drainage. Second, after removing the plastic wrap surrounding the roots, the seedlings were carefully planted at the pot's center. Third, another approximately 300 g of powdered soil was placed into the pot to cover the roots and then compressed tightly by hand. The seedlings were watered every 10 min, three times in total, to ensure that enough water was available to support plant growth. The transplantation was a success if new leaves and fresh stems were developed. The seedling survival rate reached 99% one week after replanting.

Drought stress treatment design
Three treatments were designed for each of the ten species: two drought stress level treatments, which included moderate drought stress (MDS or moderate; the water content in the soil varied from 40±2%~45±2%) and severe drought stress (SDS or severe; the water content in the soil was less than 30±2%), and one control group (CG or well-watered), which was under sufficient soil water conditions (the water content in the soil was over 75±2%) [64,65]. For each treatment, three replicates were performed. Drought stress was dependent on natural evaporation. During the drought stress period, a WET-2™ sensor made by Delta-T Devices, Ltd., Cambridge, UK, was applied to measure the water content. Once the relative water content (RWC) of the soil met the requirements of the experiment, the plant seedlings were maintained under those conditions for approximately two days to ensure that changes in plant physiology and biochemistry had occurred. For the well-watered treatment, the seedlings were watered every four days.

Leaf sampling
All plants grew new leaves ten days after transplanting, which indicated the plants' roots had developed by the time leaves could react to the plant's physiological status. Referring to the sampling method in VDI-Guideline 3975 Part 11 [66], at least 15 g of healthy leaves was collected for each replicate. The leaf samples were placed into sealed plastic bags under a portable ice-box at 0~4˚C before being transferred to the lab for further physiological and chemical analysis.

Determining the MP, Chl, SOD activity, POD activity, and APX activity
The MP (%) of the leaves was calculated as MP ¼ where L t is the relative electrical conductivity of the plant material in the drought stress treatments and L Cg is the relative electrical conductivity of the material in the control group. The relative electrical conductivity where S 1 is the original conductivity of the deionized water with fractured fresh leaves, S 2 is the conductivity of the boiled deionized water with fractured leaves, and S 0 is the conductivity of deionized water [67]. The leaf MP was determined using a Thermo Scientific™ Orion 3-star inductivity-measuring device. Before the test, all sample leaves were flushed with deionized water 3 times and residual water on the leaf surface was removed by absorbent paper.
Chl was estimated according to the method described by Arnon [68] and Zhang et al [69] in detail. Three grams of fresh leaf material was crushed with a mortar and extracted with 10 mL of 80% acetone for 15 min. The extracted solution was then centrifuged at 2500 rpm (F = 34.9 g) for 3 min and measured at wavelengths of 643 nm, 645 nm, and 663 nm via a spectrophotometer (Metash™ UV-6100A). Calculations were performed via the formulas below.
Chlorophyll a ðChla; mg � L À 1 Þ ¼ 12:7A 643 À 2:69A 645 The total chlorophyll of the solution where Chl a and Chl b refers to the concentration of chlorophyll a and chlorophyll b of the extracted solution; A 643 , A 645 and A 663 refer to the absorbance of the measured solution at wavelengths of 643 nm, 645 nm and 663 nm, respectively; C T (mg/L) is the total chlorophyll of the solution; V represents the total volume of the extracted solution (mL); and W is the weight of the extracted leaf (g). In the final result, Chl (mg�g -1 ) refers to the chlorophyll content contained within each gram of leaf sample.
Crude enzyme extracts from the leaves were used to measure the SOD, POD, and APX activity. Approximately 0.5 g fresh leaves was added with a slight amount of CaCO 3 , highpurity quartz sand and 5 mL of phosphate buffer (0.05 mol�L -1 ) and then crushed into a powder in a mortar under freezing conditions. The mixture was subsequently transferred to a 10 mL centrifuge tube and then diluted with deionized water to 10 mL. The samples were then centrifuged at high speed (F = 13000 g) for 20 min at 0~4˚C [67].
The SOD and POD reaction systems were established as described by Zhang et al [67], and the APX reaction system was described by Tang et al [70] as shown below: Two copies of the reaction system solution for each leaf sample were configured following Table 2-SOD mentioned above, one of which was put into a test tube and illuminated with 4000 lx for approximately 20~30 min at room temperature, the other one was put into the check tube wrapped in aluminum foil to avoid illumination. Once the color of the solution transition started, the reaction was immediately stopped. The final solution absorbance value was determined with a Metash™ UV-6100A spectrophotometer at a wavelength of 560 nm.
where A 0 is the absorbance of the check tube solution; As is the absorbance of the test tube solution; V T (mL) is the total volume of the samples; V 1 (mL) is the volume of the reaction system; and W (g) is the weight of the fresh leaves. All the components were put into a test tube, and then the components of the POD reaction system were added (Table 2-POD). The solution's light absorption value was recorded for each tube (the wavelength was maintained at 470 nm). The solution was read every 1 min, and each solution was recorded 5 times in 5 min. where ΔA 470 is the change in absorbance during the reaction period, W (g) is the weight of the sample, t (min) is the reaction time, Vs (mL) is the volume of the reaction system, and V T (mL) is the total volume of the sample. With respect to the APX reaction system (Table 2-APX), the mixture was put into a test tube, after which the light absorption values were recorded at 290 nm every minute; this step was repeated 5 times. The formula to calculate the APX activity was as follows: where ΔA 290 is the change in absorbance during 5 min, V 1 (mL) is the volume of the crude enzyme, V 2 (mL) is the volume of the crude enzyme involved the reaction (0.1 mL in this test), t (min) is the reaction time (5 min in this test) and W (g) is the weight of the fresh leaves. FW is short for fresh weight. SPSS 17.0 and Excel 2010 for Windows were used to calculate the mean, SD, etc. Multiple comparisons of the means by the least significant difference (Tukey's honestly significant difference [HSD]) test were performed on the 5 parameters (MP, Chl, SOD activity, POD activity, and APX activity) under the two drought stress treatments and the control group. ANOVA was used to determine significant differences between 10 species and between three treatments (P<0.05).

Data analysis
Following a drought resistance assessment method [71] for plant species based on the membership function value in fuzzy mathematics was used, the MP, Chl, SOD activity, POD activity, and APX activity results can be integrated into a single value for each species. The membership function value was calculated as follows: where the lowercase "i" and "j" represent the plant species and the parameter type, respectively; " b X ij " is the mean value of the parameter "j" of the species "i"; "X imax " and "X imin " represent the maximum and minimum of the parameter "j" of the "i" species; and " b X ij " is the membership function value and represents the drought resistance of the seedlings. The average of the membership function value was then applied to estimate the adaptive capability of the plants under drought stress. After calculation according to formula (1) or (2), only positive " b X ij " value was chosen as the result. The formula for the average was as follows (where "n" represents the number of parameters, and "X i " represents the average " b X ij "): Results and discussion

Soil relative water content
The soil relative water content between the three drought stress levels are significantly different, with the average of 74.8~85.3% for the well-watered control group, 38.3~45.2% for the moderate drought level, and 15.4~24.3% for the severe drought level (Table 3). These findings were consistent with the designed levels.

Membrane permeability
Plant cells dehydrate when they suffer drought stress, which leads to mechanical damage to the membranes [72]. Greater MP values led to more cytosolic exosmosis and further damage to the plant cellular structure. However, it was hard to distinguish which species had stronger or weaker drought resistance when they were under the well-watered control because they had not been affected by drought yet. In this study, C. bipinnata and D. barbatus presented significantly higher MP values than the other species (Fig 1), followed by the MP value of M. urticifolia and Hy. Erythrostictum, for which the MP value was significantly different relative to the remaining species. On the contrary, Liliaceae and Iridaceae family species presented low MP values under the well-watered control. The change of MP values implies the different physiological characteristics of various plants.
With the drought stress treatment, the above four plant species with higher MP values expressed different tolerance features. C. bipinnata did not survive under severe drought stress, and M. urticifolia did not survive under moderate or severe drought stress. Both species are annual herbs, and their root growth is strongly inhibited by the lack of water [73,74]. Although leaf sampling occurred only ten days after transplanting, the plant roots were transplanted with the original moist rooting medium and the seedlings were shaded and fully watered, which was beneficial for root development. Previous studies identified a strong relationship between new leaf germination and plant survival rate [75]. In addition, leaf biomass was positively correlated with root biomass, implying that root length developed when the plants grew new leaves [76]. In this study, plants under the well-watered control have all grown new leaves and even buds at sampling, thus demonstrating that the root has developed. Leaf sampling could be conducted in 3 days, 5 days, 10 days, etc. once the soil RWC matched the designed drought levels [77][78][79]. Therefore, the withering of C. bipinnata and M. urticifolia was induced by drought stress instead of a short growth period. The MP values of D. barbatus did not change significantly.

PLOS ONE
Drought resistance of ten ground cover seedling species Under drought stress, most species showed significantly increased MP values, including P. lactiflora, He. dumortieri, I. lactea, S. lineare, I. germanica, and Ho. plantaginea, among of which P. lactiflora showed significantly increased MP values only under severe drought stress (Fig 1). The MP value of I. germanica increased the most by 36% and 56% under moderate and severe conditions, respectively, and that of S. lineare increased by 31% and 55%, respectively. These findings indicated that these species' membranes were damaged under drought stress.
Among Plant drought resistance is closely related to its cell membrane system stability [31]. Usually, the cell membrane is first affected by drought stress [84], and then the cell structure is damaged and MP increases, which leads to the extravasation of extracellular electrolytes, which is why MP values increase when plants are subjected to drought stress. The stability of the MP values of these two species indicated that their cell membrane was undamaged. Therefore, the osmotic adjustment ability of multicolored carnation leaves is strong enough to avoid damage to the protoplast membrane under drought stress treatment.

Chlorophyll concentration
In the well-watered control, the studied plant species presented Chl concentrations from 6.06 to 47.69 mg�g -1 FW, and the fleshy Crassulaceae species H. erythrosticum and S. lineare presented the lowest Chl concentrations (Fig 2). The plants' Chl concentrations were affected by light intensity and environmental temperature, which affect the opening and closing of stomata and photosynthetic rates of plant leaves and then affect the accumulation of carbohydrates, which is consistent with the Chl concentration of plants [85]. In the study, all plants grew in a stable light intensity environment because the experimental area had a roof, which can prevent the effects of strong sunshine or rain. As for the ambient temperature, no extreme temperatures were encountered during the experiment. The leaves of these plants were collected at the same time after drought stress under the same environmental temperature. Thus, the changes in Chl concentration should be caused by drought stress instead of temperature or sunlight.
Upon drought stress, change trends of plant Chl concentration were various. Four species showed significantly (P<0.05) increased Chl concentrations under the different extents of drought stress, including P. lactiflora, Hy. erythrostictum, S. lineare and Ho. plantaginea (Fig  2). The increased Chl concentration under moderate or severe drought stress might be due to the increase of the stem cell mass and cell number of the leaves, thus forming a Chl condensation phenomenon, as in P. lactiflora [86]. Additionally, Liu et al [87] reported that the Chl a and b of Hy. erythrostictum would be increased during the day but decreased during the night under drought stress, which probably indicates more photosynthetic pigments were produced to promote photosynthesis of Hy. erythrostictum under drought stress. However, the photosynthetic pigment content was decreased at night for maintaining its normal physiological activities.
The species I. germanica and I. lacteal showed increased Chl concentrations under moderate stress, and then these values decreased under severe drought stress (Fig 2). Most Iridaceae plants are shade plants [88], some of which feature colorless leaves [89] and possess lower Chl concentrations than sun plants leaves [90]. Zhou [91] researched seven Iris species and also found that I. germanica had a higher Chl concentration in the early stage of drought stress than the control group.
D. barbatus did not change the Chl concentration under drought stress treatment (Fig 2), which indicated that the species had a strong self-repair and regulate ability during drought stress, and its leaves had a relatively good physiological and biochemical state, which could maintain normal photosynthesis and strong resistance during drought.
Only one species, He. Dumortieri, significantly decreased the Chl concentration with the increase of drought stress (Fig 2), which indicated that chlorophyll synthesis was interrupted and the chlorophyll decomposed under drought.

Superoxide dismutase activity
SOD activity is very sensitive to drought [92]. In this study, the SOD activity of six species, He. dumortieri, I. lactea, Hy. erythrostictum, I. germanica, Ho. plantaginea, and D. barbatus, initially significantly (P<0.05) decreased under moderate drought stress but increased under severe drought stress (Fig 3). The reduction of SOD activity under the moderate condition implied that a considerable amount of ROS was produced to damage plant cells and tissues, thus leading the plant cells to undergo oxidative damage. The change trends of SOD activity can be adopted to judge the species' drought resistance. Those species with higher SOD activity under drought can be considered to have strong drought resistance [96]. The SOD activity of P. lactiflora and S. lineare increased under either the moderate or severe drought stress. The difference of SOD activity might be caused by the expression of various isozymes, which induced the accumulation of the antioxidant substance in plants leaves that started up the antioxidant protection system when plants were under moderate and severe drought stress [97]. The degrees of SOD increase of S. lineare were higher than that of P. lactiflora, which is consistent with a previous study [98] in which S. lineare had stronger SOD activity. Overall, in this study, the SOD activity in most species increased after severe drought stress, which suggested that drought stress-induced SOD activity increases in these plant species to help them eliminate ROS. These plant species presenting increased SOD activity showed advantages in terms of their drought stress response, and proper drought exposure could significantly improve plant resistance to sustained drought stress [95].

Peroxidase activity
In the study, all plant species showed decreased POD activity under the moderate or severe drought stress, with most showing declining POD activity as the drought stress increased (Fig  4). Compared with the well-watered control, the POD activity of He. dumortieri, Hy. erythrostictum, S. lineare, I. germanica, Ho. plantaginea and D. barbatus was reduced significantly (P<0.05) during the moderate drought stress. The POD activity of these species was significantly (P<0.05) reduced under the severe drought stress (Fig 4). Such declining trends of POD activity with drought stress were contrary to the increased SOD activity trends.
Under drought stress, stronger POD activity might be attributed to the plant defense mechanisms against free radical formation resulting from water deficit [99]. According to Fig 4, the POD activity of I. lacteal was reduced under both the moderate and severe drought stress and was significantly (P<0.05) higher than that of the species that survived under severe drought stress except for He. dumortieri and Ho. plantaginea. This finding may indicate the species that have stronger resistance to drought.
Under severe drought stress, P. lactiflora, S. lineare, I. germanica and D. barbatu showed significantly lower POD activity values compared with the other species and other water conditions (Fig 4), which may be related to the POD enzyme reaching its tolerance limit and decreasing rapidly.
In general, the POD activity of all the species tended to decrease when the seedlings experienced drought stress. As a special enzyme to eliminate H 2 O 2 , the reduced POD activity in this study might have been caused by ROS elimination because different protective enzymes work together as a whole, with the elimination of O 2 -SOD increasing H 2 O 2 production.
However, a very high concentration of H 2 O 2 was beyond the reach of POD activity [100], which indicates that although SOD and POD are both antioxidases and can cooperate in scavenging ROS, drought stress might lead to a different enzyme system to resist adverse drought environments.

Ascorbate peroxidase activity
Under well-watered conditions, P. lactiflora presented the highest APX activity value at 0.98 U�min -1 �g -1 FW, although the other species had APX activity from 0.09 to 0.26 U�min -1 �g -1 FW. Seven species showed increased APX activity, with P. lactiflora, He. dumortieri, Hy. erythrostictum, I. germanica, Ho. plantaginea, and D. barbatus presenting significantly higher APX values under moderate drought than under well-watered and severe drought stress conditions (P<0.05). However, the APX activity of S. lineare decreased significantly (P<0.05) under moderate and severe drought stress. At severe stress, most plant species reduced their APX value no more than 0.50 U�min -1 �g -1 FW, with the lowest at 0.07 U�min -1 �g -1 FW for Ho. plantaginea (Fig 5).
An interesting phenomenon in this study was that the SOD and APX activity of some plants seemed to complement each other. The SOD activity of He. dumortieri, I. lactea, Hy. erythrostictum, I. germanica, Ho. plantaginea, and D. barbatus decreased under moderate drought stress and increased under severe stress (Fig 3), whereas the APX activity displayed the opposite pattern. Moreover, the highest values of APX activity in these species were recorded in the moderate drought stress treatment, which may suggest that APX activity was first activated by the early or moderate drought stress to scavenge ROS. When the SOD activity increased under severe drought stress, the APX activity decreased at the same time, which suggests that SOD plays a dominant role in ROS scavenging during severe drought stress. The weakening of APX activity under severe drought stress indicates that its antioxidant capability is temporary and limited [101].

Assessment of plant drought resistance
We determined the drought resistance of ten ground cover species through six physiological indicators: MP, Chl, SOD activity, POD activity, and APX activity. However, it is difficult to judge which plant species has better drought resistance only based on individual parameters. Therefore, it is reasonable to use the membership function method that applies fuzzy mathematics to weigh these indicators and ultimately assess the drought resistance of the ten species. The calculated results following the formula of the membership function are shown in Table 4. The findings indicate that He. dumortieri was the most drought resistant species while C. bipinnata and M. urticifolia were not suitable for moderate or severe drought stress due to withering. In addition, P. lactiflora survived the weakest drought resistance. The order of plant resistance to drought stress was as follows: He. dumortieri > I. germanica > I. lactea > D. barbatus > Hy. erythrostictum > S. lineare > Ho. plantaginea > P. lactiflora > M. urticifolia > C. bipinnata (Table 4).

Conclusions
This study investigated how ten common plant species were tolerant to levels of drought stress and showed that drought stress disrupted plant growth because the same conditions were not observed under the well-watered treatment. Five parameters (MP, Chl, SOD, POD, and APX activity) changed under moderate and severe drought stress. The main results are as follows.
First, C. bipinnata and M. urticifolia failed to survive the drought stress and were not suitable for both moderate and severe drought stress.
Second, each plant species had quite different physiological and biochemical parameters. He. dumortieri, I. lactea, and Ho. plantaginea maintained a stable MP value after experiencing drought stress. Most species (except P. lactiflora and S. lineare) showed reduced SOD activity under moderate drought stress but increased activity under severe drought stress. However, the plant species showed decreased POD activity and APX activity when the drought stresses increased.
Third, complementary relationships might occur among SOD, POD and APX activity, and SOD may play a dominant role in scavenging ROS under severe drought stress while APX and POD are responsible under moderate drought stress.