Nutrient variation induced by rodent disturbance in Haloxylon ammodendron as a target transfer strategy

Abstract Nutrients form a link between herbivores and plant. This study explored the physiological and ecological response mechanism of Haloxylon ammodendron population to rodent disturbance in Gurbantunggut Desert from the perspective of nutrient cycle. Through field investigation, we quantified rodent disturbance intensity (DI) to H. ammodendron and analyzed the ecological response mechanism of H. ammodendron population to rodent disturbance from the perspective of plant and soil nutrient cycling and changes. The results indicated that moderate rodent DI (number of effective burrows = 3–6) was the maximum limit that can be tolerated by H. ammodendron; the threshold for optimal H. ammodendron response to rodent disturbance was mild (number of burrows = 1–3). Meanwhile, the rodent disturbance caused significant nutrient enrichment (e.g., organic carbon, available phosphorus, and available potassium) in the deeper soil (at 20–40 and 40–60 cm depth) and significantly reduced the soil total salt content (p < .05). Furthermore, as the DI increased, the branches of H. ammodendron showed significantly increased soluble total sugar, crude fiber, and total nitrogen contents (p < .05) but significantly decreased crude fat and crude protein contents (p < .05); these results are related to the nutritional target transfer strategy evolved by H. ammodendron for long‐term resistance to rodent disturbance. The current study clarified the optimal disturbance model for mutually beneficial H. ammodendron–great gerbil relationship, on the basis of which the ecological response mechanism of H. ammodendron population to rodent disturbance in deserts was illustrated. The current study provides a scientific basis for the protection mechanisms of desert plants to rodent disturbance.


| INTRODUC TI ON
Plant-herbivore relationships are complex and involve interactions such as complex molecular, signaling, and strategy networks aimed at overcoming each other's defenses. Herbivores use various feeding strategies to obtain nutrients from plants. Plants, in turn, protect themselves by triggering defense mechanisms that suppress, block, or modify the metabolism of herbivores (Santamaria et al., 2013;Xiao et al., 2017). In response to the constant threat of herbivorous animals, plants have evolved a range of defense strategies. These include mechanical and chemical barriers that reduce the performance of herbivores; for instance, after a herbivore attacks a plant, it usually releases a wide variety of herbivore-induced plant volatiles (HIPVs) that can attract the herbivore's natural enemies and warn the neighboring plants of an imminent threat. Exposure to HIPVs increases the levels of defense signal hormones in plants, changes their defense status, and makes the undamaged branches more resistant to herbivores; moreover, surrounding plants may respond to HIPV signals by assessing the risks and producing plasticity response based on the actual environmental conditions (Pezzola et al., 2017;Rodriguez-Saona & Frost, 2010).
Of the many defense strategies, the mobile target strategy may be a robust response in an unpredictable and information-free environment (Kessler, 2015). Plants using the defense strategy of moving targets respond to attacks by changing their phenotypes, or even their genotypes. Moreover, considering the plastic performance of the defense strategies, these plants invest resources in defense only when the risk of damage increases (representing a cost-benefit trade-off), thereby avoiding the waste of resources (Potts, 2021).
The cost of complementary utilization of resources due to changes in plant genotypes is often manifested in the nutritional level; this increases resistance to herbivores (Mcart & Thaler, 2013;Velzen & Etienne, 2015), indicating a certain relationship of plant nutrient content and community structure with the number of herbivores. This relationship has been a concern of ecologists for a long time.
For instance, Awmack and Leather (2001) found that the relationship of plant nutrition with the growth and survival of herbivore population is always concave, and the difference in nutrient composition greatly affects the fecundity of the herbivore population. Moreover, Wetzel et al. (2016) found that plants inhibit herbivorous population via changes in plant nutrient levels; this is a key herbivore inhibition mechanism in natural systems. Joern et al. (2012) confirmed that nitrogen and phosphorus content is positively correlated with the number of herbivores and that other nutrients may have a role here.
Herbivores require various nutrients; however, most researchers have focused only on nitrogen (i.e., plant protein) and phosphorus contents and paid little attention to other nutrients. Moreover, most relevant reports have focused on insect-plant relationships and limited empirical studies on the relationship between plant nutrient content and community structure of herbivorous in desert ecosystems (Awmack & Leather, 2001;Scogings, 2011;Vieira-Neto et al., 2017;Wetzel et al., 2016;Wimp et al., 2010).
In desert ecosystems, rodents always live alongside desert plants. These rodents obtain the nutrients they need for their growth and development by feeding on the plants. Therefore, the nutrient types and contents in the plant tissues have a considerable effect on the rodents' growth and development. Haloxylon ammodendron, representing the most widely distributed vegetation in desert areas, plays an important ecological role in maintaining the stability of sand dunes, providing essential nutrient and water conditions for other vegetation demonstrating undergrowth, and maintaining arid ecosystem function and structure (Mares et al., 1997;Zhao et al., 2019).
The great gerbil (Rhombomys opimus)-a main, highly social rodent in the desert and semidesert areas of Northwest China-is widely distributed in H. ammodendron forests (Wen et al., 2020). It feeds on desert plants including not only H. ammodendron but also Kalidium foliatum, Tamarix taklamakanensis, Caragana sinica, and other psammophytes. Among them, the nutritious branches of H. ammodendron account for 80% of its food source (Liu et al., 2012). Rhombomys opimus burrows and nests under H. ammodendron shrubs and gnaws on their root phloem ; this activity has affected the growth of H. ammodendron forests in the Gurbantunggut Desert of Xinjiang, China (Gao et al., 2014).
In deserts with a relatively homogeneous population structure, R. opimus and H. ammodendron demonstrate long-term coexistence.
Rhombomys opimus mainly harm the branches in 2-3 years H. ammodendron, and only high intensity disturbance can affect the growth of H. ammodendron, while light feeding can increase branches number and crown width, which was beneficial to the growth of H. ammodendron. As a consequence of long-term feeding and disturbance by rodents, H. ammodendron has gradually evolved a unique evolutionary strategy to ensure its own survival. Simultaneously, to keep its food source, R. opimus has gradually evolved a corresponding survival mechanism to maintain a balanced relationship. So far, however, no reports have been published on this topic.

| Experimental site
The field experiment was conducted at the southern edge of the Gurbantunggut Desert (44°26′-45°12′N, 86.2°06′-87°54′E) at an altitude of 314-436.8 m above sea level. This region has a typical temperate arid desert climate with an annual average temperature of 5-6.1°C; the maximum temperature can reach 43°C in the summer, and the minimum temperature can be less than −42°C in the winter. The annual cumulative temperature is as high as 3000-3600°C, and the mean annual evaporation is >2000 mm. This region is extremely dry, with an average relative humidity of 50-60%; and <45% during May and August (Huang et al., 2015). The annual precipitation is 100-160 mm, mainly occurring in the winter and during May-September and accounting for 70-80% of the annual precipitation. The stable snow cover, with a thickness of >13 cm in the winter, is an important source of deep water for vegetation growth and development in the spring (Xing et al., 2013). The landform in the region is mainly composed of fixed and semifixed dunes, sand ridges, composite dunes, and sandy interdune lowlands. This region has very few vegetation species, with H. ammodendron and Haloxylon persicum being the main species providing an approximately 20-30% of vegetation cover (Huang et al., 2015). This region has other plant species that are sporadically distributed-ranging from shrubs such as Tamarix spp., Calligonum sp., Ephedra distachya, Artemisia desertorum, and Nitraria sibirica to herbs species such as Alhagi sparsifolia, Salsola spp., Aristida peuuata, and Suaeda spp. (Kotler et al., 2002).
Human activity and global climate change have caused the natural vegetation in the desert to degrade seriously, leading to significant decrease in vegetation cover. In some areas, the vegetation cover has decreased to <5% (Ding et al., 2017). Moreover, the decline in the vegetation cover has severely restricted the food sources of desert rodents. Therefore, H. ammodendron-a high-quality sandfixing plant that is the main food source for R. opimus-has inevitably become seriously disturbed, especially in the middle and lower parts of sand dunes and the interdune areas, where there are clusters of R. opimus burrows. As a result, collapsed areas are widespread here.
In the densely R. opimus burrow areas, H. ammodendron has shown an irreversible decline and desertification. In some areas, although there are many R. opimus burrows, H. ammodendron is flourishing, especially with the dark green and highly vital collected branches.

| Sample plots set
In June 2019, four sample plots (1000 × 200 m 2 ) were set up at intervals of >1000 m. Three 50 × 50 m 2 quadrats were set up in each sample plot for different disturbance levels along the direction of a longitudinal dune. The morphological characteristics (i.e., plant height, crown width, basal stem, and health status) of all H. ammodendron in every quadrat were recorded.
Each subquadrat from the 50 × 50 m 2 quadrats in each sample plot was small and circular, measuring (2 × 2 × 3.14=) 12.56 m 2 , with H. ammodendron in the center were used to assess the levels of R. opimus disturbance (Figure 1), and 16 subquadrats were collected in each quadrat (four repetitions for each of the four different disturbance levels). In total, 192 (16 subquadrats per quadrats × 3 quadrats × 4 sample plots) subquadrats with different levels of R. opimus disturbance were created. At the same time, H. ammodendron plant areas not disturbed by R. opimus in the same quadrats were selected, and 108 plants (with somewhat consistent morphological characteristics) were selected according to the plant health status.

| Evaluation of the R. opimus effective burrow number
The R. opimus burrows under H. ammodendron shrub were randomly selected, and circular quadrats with a radius of 4 m and H. ammodendron plants in the center were created. Three replicates were set in each quadrat, and these quadrats are only used to evaluate the number of effective burrows. The stealing opening burrow method was adopted (Zhao et al., 2007): All R. opimus burrows in the sample were slightly blocked with sand, and after 24 h, the numbers of R. opimus burrows (i.e., the number of the reopened burrows) were recorded and considered the effective burrow numbers.

| Evaluation of R. opimus Dis
According to the relationship between the health status of H. ammodendron (percentage of healthy branches) and the R. opimus burrows under the bush (Figure 1), the R. opimus disturbance intensity (DI) was rated at four levels.

| Plant sample collection
Five 2-to 3-m-high H. ammodendron plants were randomly selected from each quadrat with or without R. opimus disturbance, and their branches at the same height in the four cardinal directions (i.e., south, west, and north) were collected and mixed thoroughly. They were placed in Kraft paper bags and transported to the laboratory.
The branch samples were dried to a constant weight at 60°C, ground into powder using a mortar and pestle, passed through a sieve with 40 mm mesh size, and then subjected to nutritional component analysis. As described by Hansen and Koroleff (2007), crude fat (CFT) and crude fiber (CF) contents were determined using Soxhlet extraction and the acid-base decooking method (Feng et al., 2015), respectively. As described by Xiang et al. (2020), inverted and reduced sugar contents (RS), with glucose as the standard, both were determined using GB/5009.8-2016 and GB/5009.7-2016, respectively. Total nitrogen (TNC) content was determined using the perchloric acid-sulfuric acid digestion method. Total phosphorus (TPC) and total calcium (Ca) contents were determined through atomic absorption spectrophotometry (Type 932GBC; Scientific Equipment).
As described by Li and Li (2000), soluble sugar (STS) content was determined using the anthranone colorimetric method. Soluble protein content was determined using the Coomassie bright blue method.
Finally, the methods of Li were used to determine proline (Pro) content (Li & Li, 2000).

| Soil sample collection
Next, H. ammodendron leaves were collected, and exfoliation and separation methods were employed to collect the rhizosphere soil of H. ammodendron. We collected all the soil attached to the H. ammodendron roots. In brief, the soil loosely attached to the root system was removed by shaking several times and the remaining soil more closely attached to the root system (about 2 mm in thickness), termed rhizosphere (R) soil (Riley & Barber, 1970;Zhang et al., 2019), was then collected. At the same time, the soil in which H. ammodendron was not grown was selected from the same quadrats and termed nonrhizosphere (S) soil.
Both R and S soils were obtained from three soil depths: 0-20, 20-40, and 40-60 cm. All soil samples were air-dried and sieved through a 2-mm mesh for physicochemical analysis according to Bao (2008). Soil total organic carbon (TOC), TN, TP, total potassium (TK), soil water, available nitrogen (AN), available phosphorus (AP), and available potassium (AK) contents were determined using the KCr 2 O 7 method, HClO 4 -H 2 SO 4 digestion method, Mo-Sb colorimetric method, atomic absorption spectrometry method, weighing method, alkaline hydrolysis diffusion method, atomic absorption spectrometry-based ammonium acetate extraction method, and 0.5 mol/L sodium bicarbonate-molybdenum blue colorimetry method, respectively. Soil pH in a 1:5 soil-deionized F I G U R E 1 Field test design sketch. Description: Ordinate is the health condition of Haloxylon ammodendron; abscissa is the number of effective burrows. N, G, M, and B are, respectively, no disturbance (the number of effective burrows was 0), mild disturbance (the number of effective burrows was 1-3), moderate disturbance (3-6 effective burrows), and severe disturbance (the effective number of burrows was >6) water mixture was measured on a pH meter (Seven Easy, Mettler-Toledo). Rhizosphere nutrient enrichment rate (E) was calculated using the following formula (2015):

| Data statistics and analysis
The

| Effects of R. opimus disturbance on soil nutrients in R and S soil
As shown in Figure 2, R. opimus disturbance had significant effects on R and S soil nutrients, but these effects differed at different soil depths (p < .05). In Figure 2A, soil TOC under G was higher in R than S soil at 0-20 cm and 40-60 cm but not 20-40 cm. In addition, soil TOC under G and M was significantly higher than under N and B in R soil at 20-40 cm and 40-60 cm but not 0-20 cm (p < .05) (Figure 2A). At 0-40 cm, TN content was significantly higher under M disturbance than under other DIs (p < .05; Figure 2B).
Under N, G, and M, TK content differed significantly between R and S soils (p < .05; Figure 2C). In R soil, TP content was not affected by R. opimus disturbance at 0-20 cm (p > .05) but was significantly affected by R. opimus disturbance at 20-40 cm (p < .05). In S soil, TP content at 20-60 cm was significantly higher under B than under in other samples (p < .05; Figure 2D).
Moreover, AN content significantly differed under N and G in both R and S soils (p < .05). AN content was significantly higher E = (R content − S content) ∕S content × 100 % F I G U R E 2 Effects of different disturbance intensity on TOC, TN, TK, and TP in the rhizosphere and nonrhizosphere soil of Haloxylon ammodendron. Description: Line chart (mean with standard error) with different lowercase letters indicated Rhizosphere soil (R) nutrients significant difference (p < .05); line chart (mean with standard error) with different capital letters indicated non-rhizosphere soil (S) nutrients significant difference (p < .05); * represents significant differences between rhizosphere and nonrhizosphere nutrients at same disturbance level (p < .05). Ordinate is the content of soil organic carbon (TOC) (A), soil total nitrogen (TN) (B), total potassium (TK) (C), and soil total phosphorus (TP) (D); abscissa is group name (N, G, M, and B are, respectively, no disturbance (the number of effective burrows was 0), mild disturbance (the number of effective burrows was 1-3), moderate disturbance (3-6 effective burrows), and severe disturbance (the effective number of burrows was >6)). R and S were rhizosphere and nonrhizosphere soil nutrients of H. ammodendron, respectively at 0-20 and 40-60 cm in R soil under G than in other samples (p < .05). AN and AP contents in S soil under B were significantly higher (p < .05; Figure 3A,B). AK contents in R soil were significantly higher under G (p < .05), whereas it was significantly higher in S soil under B (p < .05; Figure 3C). As R. opimus DI increased, total sugar (TS) content decreased significantly (p < .05; Figure 3D).
Finally, rhizosphere E demonstrated that under G, TN, TP, AN, AP, and AK were enriched but TS was depleted. Moreover, under M, TOC, TN, and AN were enriched, whereas TN, AP, and AK were depleted under B (Table 1).

| Correlation analysis between R. opimus disturbance and soil nutrients
The Pearson correlation analysis results showed that soil nutrient contents were significantly correlated with R. opimus DIs (p < .05; Table 2). Specifically, DIs were significantly and positively correlated with soil TP and AN contents (p < .05). Moreover, it demonstrated a highly significant and negative correlation with soil AP and TS contents in R soil (p < .01) but a significant and positive correlation with soil TN, TK, AK, TP, and AN contents in S soil (p < .01). In R and S soils, TS content was significantly and negatively correlated with TOC, TP, TN, and AN contents (p < .01); AK content was significantly and positively correlated with TP, AN, and AP contents (p < .01); AN content was significantly and positively correlated with TP contents (p < .05); and TN content was significantly and positively correlated with TOC content (p < .01).

| Effects of R. opimus disturbance on H. ammodendron nutrient components
The results showed that R. opimus disturbance significantly affected the nutrient content in the collected H. ammodendron branches ( Figure 4). Specifically, as R. opimus DI increased, the contents of STS ( Figure 4A), CF ( Figure 4C), and TNC ( Figure 4E) in the collected H. ammodendron branches all showed a significant increasing trend, whereas the contents of crude fat (CFT) ( Figure 4D) and crude protein (CP; Figure 4H) showed a significant decreasing trend (p < .05).
Additionally, the content of reducing sugar (RS) was higher in N, while all other treatments recorded similar reducing sugar content ( Figure 4B). The content of TPC was higher in N, and the least was recorded in G ( Figure 4F). In particular, TPC content under G was significantly lower than that under other DIs (p < .05). Moreover, Ca content was significantly higher in G and M and lower in N and B ( Figure 4G).

F I G U R E 3
Effects of different disturbance intensity on AN, AP, AK, and TS in the rhizosphere and nonrhizosphere soil of Haloxylon ammodendron. Description: Line chart (mean with standard error) with different lowercase letters indicated Rhizosphere soil (R) nutrients significant difference (p < .05); line chart (mean with standard error) with different capital letters indicated nonrhizosphere soil (S) nutrients significant difference (p < .05); * represents significant differences between rhizosphere and nonrhizosphere nutrients at same disturbance level (p < .05). Ordinate is the content of available nitrogen (AN) (A), available phosphorus (AP) (B), available potassium (AK) (C), and total salt (TS) (D); abscissa is the group name that has identical meanings as described in Figure 2. R and S were rhizosphere and nonrhizosphere soil nutrients of H. ammodendron, respectively

| Effects of R. opimus disturbance on the N/P ratio in the collected branches and R soils
As shown in Figure 5, R. opimus disturbance affected N/P ratio in R soils (p < .05). Specifically, the mean N/P ratios under N, G, M, and B were, respectively, 19.16, 31.95, 28.89, and 29.78 in the collected H. ammodendron branches and 0.14, 0.19, 0.26, and 0.15 in R soils.

| Relationship between the nutrients in the H. ammodendron branches and R soils
Nutrient contents in the collected H. ammodendron were significantly affected by R. opimus disturbance. As shown in Table 3, R. opimus DIs were significantly correlated with all the nutrients, except TPC and Ca, in the collected H. ammodendron branches (p < .05). In the terms of relationship between nutrient composition of collected H. ammodendron branches and R soils, STS content was significantly and positively correlated with TOC and AN content but significantly and negatively correlated with AP and TS (p < .05).
In R soils, RS content was significantly and positively or negatively correlated with the contents of all nutrients, except TP and AK (p < .05). In particular, CF content was significantly and negatively correlated with AP and TS contents (p < .05). TNC content was significantly and positively correlated with TOC and TN contents but significantly and negatively correlated with AP and TS contents (p < .05). TPC content was significantly and negatively correlated with TOC, TP, AN, and AK contents but significantly and positively correlated with TS content (p < .05). Ca content was significantly and positively correlated with TOC, TN, and AN contents (p < .05). Description: disturbance intensity (DI) (N, G, M, and B are, respectively, no disturbance (the number of effective burrows was 0), mild disturbance (the number of effective burrows was 1-3), moderate disturbance (3-6 effective burrows), and severe disturbance (the effective number of burrows was >6)). The enrichment ratio (E%) with a value >0 indicates nutrient enrichment and a value <0 imply nutrient depletion. TOC: soil organic carbon, TN: soil total nitrogen, TP: soil total phosphorus, TK: total potassium, AN: nitrogen, AP: available phosphorus, AK: available potassium and, TS: total salt. The values in bold in Table 1 have no special definition. The values are the Pearson correlation coefficients. The correlation coefficient r of Pearson is between −1 and 1, r < 0 is negative correlation, and r > 0 is positive correlation. * represent significances at p < .05; ** represents significances at p < .01.

Variables TOC (g/kg) TN (g/kg) TP (g/kg) TK (g/kg) AN (mg/kg) AP (mg/kg) AK (mg/kg) TS (mg/g) DI
CP content was significantly and negatively correlated with TOC and AN contents but significantly and positively correlated with AP and TS contents (p < .05).
Furthermore, nutrient composition in the collected branches of H. ammodendron significantly correlated (p < .05; Table 4). Specifically, STS content was significantly and negatively correlated with RS, CFT, and CP contents but significantly and positively correlated with CF and TNC contents and RNP (N/P ratio) (p < .05). RS content was significantly and positively correlated with TK, AP, AK, and TS but significantly and negatively correlated with TOC, TN, TP, AN, and DI (p < .05). CF content was significantly and negatively correlated with CFT and CP but significantly and positively correlated with TNC content (p < .05). CFT content was significantly and negatively correlated with TNC content and RNP (N/P ratio) but significantly and positively correlated with CP content (p < .05). TNC content was significantly and negatively correlated with CP content but significantly and positively correlated with RNP (N/P ratio) (p < .05). CP content was significantly and negatively correlated with RNP (N/P ratio) (p < .05).

| Effects of environmental factors on nutrient composition in the H. ammodendron branches
The RDA results showed that environmental factors significantly affected the nutrient composition in the H. ammodendron branches a F I G U R E 5 Effects of different disturbance intensity on N:P ratio in Haloxylon ammodendron and rhizosphere soil. Description: Ordinate is N:P ratio in H. ammodendron and rhizosphere soil; abscissa is the group name that has identical meanings as described in Figure 2 ( Figure 6), the total variance between environmental factors and nutrient composition in the collected branches was 5.72, and the explanatory variables accounted for 95.2% of the total variance. The first two principal component axes accounted for 86.07% of the total variance in the nutrient composition-environmental factor relationship; of them, the first and second ordination axes were responsible for 58.71% and 27.65% of the total variance, respectively.
As shown in Figure 6, R soil TS contents and DIs were the major factor contributing to the differences in the functional traits of the branches of H. ammodendron (p = .002)-accounting for 86.36% of the total variance (Table 5)-followed by TOC and AN contentsaccounting for 46.2% (p = .002) and 43.6% (p = .004) of the total variance, respectively (Table 5)

| R. opimus disturbance significantly affected soil nutrient contents in R and S soils
Low nutrient content is a main feature of soil in desert areas. The "fertile island" effect, under the shrub of desert vegetation, is a crucial accumulation phenomenon observed under limited soil F I G U R E 6 Redundancy analysis (RDA) for environmental factors. Description: Environmental factors are generally represented by arrows. The length of the arrow line represents the degree of correlation between a certain environmental factor and community and species distribution, and the longer the arrow, the greater the correlation. When the angle between the environmental factors is acute, it means that there is a positive correlation between the factors, an obtuse angle implies a negative correlation. Red and blue arrows represented rhizosphere soil factors and nutrient composition of Haloxylon ammodendron assimilative branches, respectively TA B L E 3 Correlation analysis of rhizosphere soil and branches nutrients of Haloxylon ammodendron in different disturbance intensities The values are the Pearson correlation coefficients. The correlation coefficient r of Pearson is between −1 and 1, r < 0 is negative correlation, r > 0 is positive correlation. * represent significances at p < .05; ** represent significances at p < .01. resources and the main mechanism in shrubs that facilitates their nutrient utilization and adaptation to barren environment in desert areas Tu et al., 2011;Yuefei et al., 2018).

TA B L E 4 Correlation analysis between nutrients in the assimilation branches of Haloxylon ammodendron
In this study, we found that R. opimus disturbance significantly affected the processes of nutrient fixation, transfer, and redistribution and soil nutrient content in both R and S soils (Figures 2 and 3 Table 1). This was mainly due to the burrowing activities by R. opimus. We found that R. opimus burrows were mainly distributed in the soil layer at the depths of 20-40 cm depth. We found that R. opimus could store a considerable amount of food in the burrow area. This food is rich in nutrients and becomes the source of organic matter and other nutrients in deep soil layers (Herrera et al., 1999). In addition, the burrow structure can affect the soil ventilation and water conditions and shape the micro-topography (Russell & Maclean, 2008), which is conducive to microbial decomposition and thus enriches deep soil nutrient content (Kuznetsova et al., 2013).
Although the severe disturbance by R. opimus increased TOC content in the soil surface layer at 0-20 cm depth, the taproot and lateral root of H. ammodendron were inevitably damaged by the over-dense burrowing activities of R. opimus; this greatly weakened or even terminated the transport function of the H. ammodendron roots, thus greatly reducing their adsorption capacity, especially causing nutrient deficiency in the middle (20-40 cm) and deep (40-60 cm) soil layers ( Figure 3A). The rhizosphere E reflects not only the degree of nutrient enrichment but also the strength of plant rhizosphere effect (Yang & Liu, 2015). significantly and negatively correlated with the soil total salt content (Table 2), possibly because soil density decreased and moisture changed during rodent burrowing activity. In addition, urine and feces was found to stimulate microbial activity and promote nitrogen-fixing microorganism accumulation (Hawkins, 1996;Kuznetsova et al., 2013), thereby diluting soil salt concentration.
At this level, R. opimus disturbance could aid H. ammodendron in avoiding soil salinization risk to a certain extent.
In the desert ecosystem, the fertile island effect causes the redistribution of soil resources (such as soil moisture, nutrient, and salt) in the whole system, resulting in continuous occurrence and development of "plundered" and barren soil patches. This promotes the invasion of herbaceous plants by shrubs, resulting in continuous degradation of herbaceous plants and acceleration of ecosystem desertification (Liu et al., 2006 Figure 5)-possibly representing the "nutrition target transfer" defense strategy employed by plants against herbivores. In summary, the nutrient content changes in the collected H. ammodendron branches represent not only a physiological response to R. opimus disturbance but also a defense strategy against R. opimus.
The plant N/P ratio is an effective tool to indicate the nutrient limitation of plants in the study area (Ien & Moen, 2009). Zhang et al. (2004) proposed that the N/P ratio threshold can explain the N and P limitation in the study area: The growth of a species may be limited by N, P, and both N and P contents when the N/P ratio is <21, >23, and 21-23, respectively. Our current results showed that as R. opimus DIs increased, H. ammodendron growth in the study area gradually changed from N restriction (N/P = 19.16) to P restriction (N/P > 23; Figure 5). Moreover, soil TN content was significantly and positively correlated with R. opimus DIs, indicating that the nitrogen accumulation effect of H. ammodendron rhizosphere is strong under mild-to-moderate R. opimus DIs (E = 33-61%; Tables 1 and 2). Thus, moderate R. opimus disturbance somewhat alleviated the pressure of nitrogen deficiency in the desert ecosystem and supplemented a certain amount of nitrogen as a source.

ACK N OWLED G M ENTS
In this study, we would like to thank professor W.Z for her guidance, all the authors for their joint efforts. We also would like to thank many graduate students who are not listed as co-authors and to the staff involved in the collection of soil samples. This work was supported by the National Natural Science Foundation of China (32060374) and Talent Initiation Project of Shihezi University (R C Z K201952).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest. Hongwei Chen: Investigation (supporting); Methodology (supporting); Software (supporting). Wenjie Xiang: Investigation (supporting); Resources (supporting); Software (supporting).

O PEN R E S E A RCH BA D G E S
This article has earned an Preregistered Badge for making publicly available the digitally-shareable data necessary to reproduce the reported results. The data is available at https://doi.org/10.6084/ m9.figsh are.14844633.