Belowground interactions differ between sympatric desert shrubs under water stress

Abstract Understanding the relationships among species is central to ecological research; however, many knowledge gaps remain regarding how desert plant species interact. In the present study, we assessed the effect of rainfall on the belowground interactions and root morphology of two desert shrubs, Reaumuria soongorica (Tamaricaceae) and Salsola passerina (Chenopodiaceae), from three communities with similar landforms and soil environments. The roots of both R. soongorica and S. passerina were deeper when grown together than grown singly. Interestingly, the belowground biomass of R. soongorica was higher, but the belowground biomass of S. passerina was lower when grown together than when grown alone. This suggests that S. passerina benefitted from the association with R. soongorica. When grown together under conditions of low rainfall, the roots of R. soongorica were deeper than those of S. passerina, which suggests that R. soongorica is more robust than S. passerina when subjected to periods of decreased rainfall. We concluded that the symbiotic relationship between these two shrub species can lead to deeper roots and that the plants are affected by rainfall availability. Combined with the output results of climate change models, we speculated that the distribution area of these two species will expand to the west, which has important implications on how the interactions of other desert species may change in response to climate variability.


| INTRODUC TI ON
Ecology is the study of the interactions among organisms and their environment. Most studies have focused on the relationship between organisms and the environment and, more specifically, the relationships between plants and animals. These studies have revealed many interesting phenomena, including coevolution (Baum, 2018;Ehrlich & Raven, 1964;Rausher, 2001), special signal transmission (Knauer & Schiestl, 2015), and parasitic relationships (Fritz, Moulia, & Newcombe, 1999;Schmidt et al., 2017;Su et al., 2017). Currently, there is a deficit in the information needed to classify the types of biological interactions. This may be because previous studies have mainly assessed the interactions between individuals and have neglected the effects of community-level interspecies relationships. Additionally, previous studies have mostly focused on the interspecies relationships of plants at the species level (Adler et al., 2018;Mangla, Sheley, & Radosevich, 2011), and relatively, few studies have assessed the individual-level relationships, especially in desert ecosystems.
Reaumuria soongorica and Salsola passerina are desert shrubs that are usually constructive or dominant species in the desert ecosystem of northwest of China (Chen, Zhu, Cao, Song, & Zhong, 2011;Zheng et al., 2008). Both shrubs exhibit two growth patterns, which is unusual among desert shrubs. In addition to growing independently, Reaumuria soongorica plants sometimes grow close (within a few centimeters or even less than a centimeter) to S. passerina plants, and their branches overlap each other and the roots become intertwined. Because the distribution of vegetation in deserts is generally sparse, this growth pattern does not usually occur between other desert species. In the present study, we referred to this growth pattern (i.e., of highly overlapping ecological niches) as "paired growth" (Figure 1a, b) and the widespread independent growth pattern of one plant species as "single growth." These two growth patterns provide an ideal experimental basis for studying the interspecific relationship of desert plants.
Considering predicted global climate change scenarios, many studies have focused on the impact of environmental factors on the life cycles or the adaptability of animal and plant species; however, few studies have focused on how climate change might affect the relationships between plant species. Once the impact of climate change on species was demonstrated, it became clear that the disruption of the original balance between species would be inevitable.
The morphology of a plant root system can reflect adaptations to its environment (Stanley et al., 2018); however, most studies on plant root system morphology have mainly explored how plants acquire resources or adapt to climate change (Pilon et al., 2013). Some studies have examined the distribution of plant root systems within a single plant community, that is, looking at the relationships among species or individuals (Tron et al., 2015). The main influence of an interspecies relationship within the same vegetation type (except vines) is the distribution of roots among neighbors. In this study, we explored the differences between the growth patterns of the root systems of two desert shrubs.
Three hypotheses guided our study: (a) The primary environmental factor that affects the growth and distribution of R. soongorica and S. passerina is the quantity of annual rainfall, that is, plants at sites receiving greater rainfall would have a greater biomass. (b) The relationship between R. soongorica and S. passerina is mutualistic, that is, plants of both species would grow better when grown together than when grown singly. (c) The mutualism between R. soongorica and S. passerina is more intense in the presence of environmental stress, that is, that plant competition would be weakened and the positive interactions intensified by an increase in environmental stress (Bertness & Callaway, 1994). To tests these three hypotheses, we compared the growth of R. soongorica, and S. passerina grown singly and grown together under conditions with an environmental gradient. This study demonstrated the changes in the interactions between these two desert shrub species along a rainfall gradient and thus presents important indications of how desert species interactions may change in response to changing climatic conditions.

| Study area and site description
The main codistribution areas of R. soongorica and S. passerina occur in the northern Qinghai-Tibet Plateau, usually in the long and narrow area of the southern Badain Jaran and Tengger deserts ( Figure 1c). This area has an arid or semi-arid climate characterized by drought, low precipitation, high evaporation, and extreme temperature differences in summer (Li et al., 2017). We selected three sampling locations in a piedmont alluvial plain with thick soil layers that have low human impact ( Figure 1c). Within each location, we selected sampling sites with similar landforms and soil environments that presented a gradient in climate and altitude (Table 1). The western part of S3 was sparsely distributed, and the eastern part of S1 was gradually replaced by other plant communities. The plant species structure of the sites was simple. In all sampling sites, R. soongorica and S. passerina were the dominant species; however, a few other species were present, for example, Nitraria tangutorum, Lycium barbarum, and Kalidium foliatum.

| Growth status of existing vegetation
We surveyed the growth status of vegetation in each sampling site in October and September 2016. We randomly laid out three quadrats (100 × 100 m) that were situated at least 50 m away from each other.
We further divided up each of the quadrats into plots (5 × 5 m). In each plot, we measured and recorded the plant height and crown breadth of all R. soongorica and S. passerina. We also counted how many of each species exhibited "paired growth" and "single growth" patterns.
Density was measured as the total number of plants divided by the total area of the plot. The paired ratio was calculated as the number of paired growth patterns divided by the total number of plants.

| Sample-plant selection
In each study site, we sampled three representatives of single R. soongorica, single S. passerina, and paired R. soongorica and S. passerina.
We specifically chose sample plants with normal growth forms that lacked signs of physical damage. To ensure the comparability and representativeness, the sampled plants were chosen between the highest distribution area of the plant height and crown breadth, based on the survey results.

| Sampling and processing
Height and canopy width were measured for single growing R. soongorica and S. passerina and for plants growing in close proximity. The aboveground parts were cut, placed in labeled ziplock bags, and refrigerated (0°C) to reduce the influence of respiration. The root systems were excavated (with a radius of 0.8 m from the base of each plant) with soil 10 cm belowground forming a layer until all of the root systems were dug out. Soils were screened with a 0.1 cm soil sieve, and stones and grit were removed. Roots were placed in labeled ziplock bags and refrigerated (0°C). All samples were transported to and stored in a laboratory. Roots of other plants were differentiated according to their color-morphological texture flexibility, and R. soongorica and S. passerina roots were separated from each other. Soil adhering to the seedlings was brushed off, and the morphological index (e.g., root  surface area) of the root systems was analyzed using Epson scanners and WinRHIZO Basic 2008a. Roots and stems were separately dried in an oven at 105°C for 2 hr and then maintained at 80°C for 48 hr, after which the dry biomass (aboveground biomass and each layer of belowground biomass) was determined using an analytical balance.

| Calculation
The root extinction coefficients of R. soongorica and S. passerina under conditions of single growth and paired growth were calculated using Gale's nonlinear model on plant root vertical distribution (Gale & Grigal, 1987), as follows: where Y represents the cumulative percentage of root biomass from the soil surface to the depth d (cm) of a certain soil layer, and β is the root extinction coefficient. The larger the value of β (closer to 1), the greater the proportion of plant roots distributed in deeper soil layers.
The root biomass paired-single ratio represents the mean paired root biomass of a species to the mean single root biomass and was calculated as follows: where a ratio larger than 1 indicates that the root system was more developed in paired growth than single growth, and a ratio smaller than 1 indicates that the root system was more developed in single growth than paired growth.

| Data use and statistical analysis
We Other data were tested using a nonparametric analysis and Wilcoxon rank-sum test. Significant differences were determined based on multiple comparisons using Kruskal-Wallis tests. The changes in the paired growth of the two plant species in response to varying environmental gradients were then compared. The results of ANOVA revealed that the underground biomass of both R. soongorica and S. passerina was significantly (p < .05) smaller in sites with lower annual rainfall (Table 3). When grown alone, R. soongorica commonly formed a tap root with a deeper root system. In contrast, S. passerina grown alone had fibrous root systems and roots that were mainly distributed in shallow layers.

| RE SULTS
Compared with single growing plants, the root biomass of R. soongorica in all soil layers was higher when grown with S. passerina.
In contrast, S. passerina root biomass was significantly (p < .01) lower when grown with R. soongorica than when grown alone. In addition, roots of S. passerina grew deeper, and the biomass of its root system in the lower soil layers increased. In both species, the difference between the growth in the paired-and single growth treatments increased with an increase in the severity of the environments ( Figure 3).
In all sample areas, the root surface area of R. soongorica was significantly (p < .01) larger in the paired growth conditions than when grown alone, whereas the difference between paired and singly grown S. passerina was smaller. The vertical distribution of R.   (Figure 4).
The root extinction coefficients of R. soongorica were similar for all plants (paired and singly grown) and were not significantly (p > .05) affected by the environmental gradient. The root extinction coefficients of paired S. passerina plants were higher (i.e., roots were deeper) than singly grown plants, when grown in a suitable (i.e., S2) environment ( Figure 5a). The root biomass paired-single ratio of R.
soongorica was greater than 1. The root biomass paired-single ratio of S. passerina was <1 and decreased with the increase in the environmental gradient (Figure 5b).

| D ISCUSS I ON
Our findings supported the first hypothesis, that annual precipitation is the primary environmental factor that affects the growth and distribution of R. soongorica and S. passerina. By choosing sites with similar latitudes, topographies, and soils, we were able to test for a potential correlation between the growth and distribution of the two study species in response to altitude and rainfall. Only plant density and annual rainfall were significantly negatively correlated (p < .05), and no significant correlations were observed between the plant growth attributes and environmental factors tested. In arid regions, moisture is the primary limiting factor (Zhang, Zhao, Liu, Fang, & Feng, 2016) and rainfall is the main source of plant water (Nie, Chen, Wang, & Yang, 2012). Thus, according to the "most limiting factor" concept presented by Liebig (1840), moisture is the primary limiting factor for plants in arid ecosystems (Thomas, Hoon, & Dougill, 2011). The decreased distribution of the two plants in the eastern sites with larger annual rainfall may, therefore, be the re- Holub, 2009), and the effect of neighboring plants (Hendriks et al., 2015). Both the increase and decrease in rainfall may increase root biomass, whereby plants grow well and have larger plant parts under favorable conditions; however, root morphologies may not necessarily change substantially. However, under water stress, plants need to distribute more biomass underground to expand their roots to acquire more water. In this study, the aboveground parts of the same sampling sites were similar sized, and the significant increase in the underground parts represented a significant increase in the ratio of the root system. We speculated that the acquisition of water by S.

TA B L E 3
The root surface area and biomass of Reaumuria soongorica and Salsola passerina growing in moderate (S1), suitable (S2), and severe (S3) environments in the Hexi Corridor  Figure 3e); however, S. passerina received some benefit from the paired growth (i.e., had an extremely significantly lower root biomass, p < .001, Figure 3f). In the SGH proposed by Bertness and Callaway (1994), increased environmental stress leads to increased intensity and the importance of mutual benefits. The SGH has been confirmed by many studies (Mori, 2018) In this study, precipitation was the limiting factor. Our findings have important implications given that the global climate change model predicts that precipitation will continue to rise in the region for decades to come (IPCC, 2014). Additional observational studies may increase the accuracy of predictions regarding the future distribution of desert shrubs such as R. soongorica and S. passerina. For example, these drought-tolerant shrubs may move westwards as a result of increased precipitation. The distribution and density of each species may also change in response to climate variations, and in some cases, their ranges may even increase (Kolanowska et al., 2017). (17YF1WA161). We thank Wiley Editing Services for its linguistic assistance during the preparation of this manuscript. We thank Shuo Jiao for help regarding the data analysis and Tingting Xie for guidance regarding the use of technical terminology.

CO N FLI C T O F I NTE R E S T
The authors declare there are no conflicts of interest.

AUTH O R CO NTR I B UTI O N S
LS and YL conceived and designed the experiment; ZZ and YW performed the experiment and collected the data; ZZ analyzed the data and wrote the first draft of the manuscript; and all authors contributed substantially to the revisions.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data that support this article are accessible in the Dryad reposi-