Does the intensive grazing and aridity change the relations between the dominant shrub Artemisia kopetdaghensis and plants under its canopies?

Abstract The interspecific plant interactions along grazing and aridity stress gradients represent a major research issue in plant ecology. However, the combined effects of these two factors on plant–plant interactions have been poorly studied in the northeast of Iran. To fill this knowledge gap, 144 plots were established in 12 study sites with different grazing intensities (high vs. low) and climatic characteristics (arid vs. semiarid) in northeastern Iran. A dominant shrub, Artemisia kopetdaghensis, was selected as the model species. Further, we studied changes in plant life strategies along the combined grazing and aridity stress gradients. In this study, we used relative interaction indices calculated for species richness, Shannon diversity, and species cover to determine plant–plant interactions using linear mixed‐effect models (LMM). The indicator species analysis was used to identify the indicator species for the undercanopy of shrub and for the adjacent open areas. The combined effects of grazing and aridity affected the plant–plant interactions and plant life strategies (CSR) of indicator species. A. kopetdaghensis showed the highest facilitation effect under high stress conditions (high grazing, high aridity), which turned into competition under the low stress conditions (low grazing, low aridity). In the arid region, the canopy of the shrub protected ruderals, annual forbs, and grasses in both high and low grazing intensities. In the semiarid region and high grazing intensity (low aridity/high grazing), the shrubs protected mostly perennial forbs with C‐strategy. Our findings highlight the importance of context‐dependent shrub management to restore the vegetation damaged by the intensive grazing.

Grazing is a key biotic stress in dry rangelands due to its extensive application and its potential to change the community structure and species composition, and to degrade the ecosystem services (Diaz et al., 2007;Jankju, 2016, Kéfi et al., 2007Li et al., 2013, Rahmanian et al., 2019. Previous studies examining the herbivores as main drivers have reported that unattractive, toxic, or thorny plants may induce positive indirect (i.e., grazer-mediated) effects on palatable herbs, shrubs, or trees (Bakker et al., 2004;Callaway et al., 2005;Smit et al., 2006). Grazing may affect the plant-plant interactions as well (Soliveres et al., 2011). The result of these interactions will be highly dependent on the ability of the nurse plant to moderate the effects of herbivores and on the tolerance of the facilitated species to grazing (Baraza et al., 2006;Vandenberghe et al., 2009). Further, nurse plants may protect the neighbors against herbivory and enhance their survival by increasing resource availability (Acuña-Rodríguez et al., 2006;Rand, 2004).
The CRS strategy, distinguishing between the competitors, ruderals, and stress tolerators (CSR- Grime, 1979), provides a detailed view on the characteristics of indicator species for different types of vegetation, that is, across plant growth forms and differences in the intensity of environmental stress and disturbance and/or grazing (Grime, 1977;Hodgson et al., 1999).
However, previous studies have reported varying effects of grazing on plant-plant interactions. For instance, Soliveres et al. (2011) showed that rabbit grazing caused positive interactions between the bunch grass (Stipa tenacissima) and saplings of the shrub (Retama sphaerocarpa) during winter and autumn. However, because of higher grazing intensity in the summer, the interspecific interactions shifted to neutral. Similarly, Holthuijzen and Veblen (2015) found that positive interactions between Artemisia tridentata ssp. wyomingensis and Poa secunda decreased with increasing grazing intensity in the arid region because grazing reduced the productivity during the drought periods more intensively.
This may result in the absence of positive interactions between plants due to different stressors Verwijmeren et al., 2014;Michalet et al., 2014). On the contrary, Noumi et al. (2016) showed that suppressive effects of shrubs on Acacia tortilis seedlings shifted to positive with increasing grazing stress.
Therefore, an increase in facilitation due to the combination of these two stressors can be expected. This research aims at investigating plant-plant interactions, accounting for the combined effects of grazing and aridity.
The selected dominant species, Artemisia kopetdaghensis, is an aromatic shrub that is widely distributed, ranging from warm and arid to cold and semiarid steppes of northeast Iran (180-400 mm) and parts of Turkmenistan (Memariani, 2016). We used A. kopetdaghensis and its understory plants as a model system to study the combined effects of grazing and climate (arid region: high/low grazing, semiarid region: high/low grazing). Our aim was to answer the following questions: (a) What are the dominant interactions between A. kopetdaghensis as a target species and its surrounding understory herbs? (b) Are the relations between A. kopetdaghensis and plants in its surroundings affected by aridity and intensive grazing? And (c) do the combined effects of grazing and aridity differ between various plant life strategies?

| Study area
We selected two sites along the precipitation gradient in northeastern Iran in the Khorassan-Kopet Dagh floristic province of the Irano-Turanian region, located between 35°43′-36°44′N and 58°40′−60°27′E. Based on meteorological data, Khajeh Kalat has an arid climate and Baharkish has a semiarid climate, expressed by De Martonne aridity index (see Table 1 for more details). The mean annual precipitation (20-year mean) is 255 mm in Khajeh Kalat and 385 in Baharkish. Artemisia kopetdaghensis Krasch.M.

| Sampling design
The two studied regions were 1,600 ha and 1,035 ha in size for the arid and semiarid regions, respectively. The HG and LG sites were of similar size in both climatic regions. The distance between individual sampling areas within each climatic region was less than one kilometer. The HG and pairwise LG sites were relatively homogenous in terms of topography, land use, and vegetation, and the only substantial difference between the paired HG and LG sites was in their grazing intensities. The LG sites were located within fences that have prevented grazing for around 35 years, whereas HG sites were open and therefore have suffered from long-term overgrazing. Each plot was characterized by geographic coordinates and altitude. In 2017, the number of individuals and percentage cover of all vascular plant species were recorded between April and June, when the growing season peaks in this region ( Figure 1).
The decision about the grazing status of the sites (high grazing intensity vs. occasional/low grazing intensity) was based on the median number of dung droppings (Marques et al., 2001) (see more information on the grazing history in Table 1).
The sampling design was arranged in a hierarchical way: In each of the two climatic regions (arid and semiarid), we selected six sampling areas, with a high-grazed and a low-grazed site in each sampling area, arranged in a pairwise way (hereafter referred to as HG and LG sites). Then, we sampled three plots under the A. kopetdaghensis

| Statistical analyses
Relative interaction intensity (RII) was used to assess the effect of shrubs on undercanopy vegetation (Armas et al., 2004) and was calculated based on the cover, richness, and diversity (expressed as The differences in RII indices for species richness, cover, and diversity between the HG and LG sites and between the arid and semiarid regions were tested using linear mixed-effect models, with "sampling areas" as a random effect, "climatic region" and "grazing" as fixed effects, and RII-based richness (RII-Richness), cover (RII-Cover), and Shannon H (RII-Shannon diversity) as response variables.
All univariate analyses were performed in the R software, using the NLME package. The script for the model testing the interaction between "climate" and "grazing" was "lme (Relative interaction in-tensity~climatic region * grazing, random = ~1|sampling area and LDMC were inserted into the "StrateFy" spreadsheet 3 to calculate C, S, and R percentages for each species (Pierce et al., 2017).

| The effect of climate and grazing interaction on plant-plant relations
We found significant effects of both grazing and aridity on plantplant interactions, expressed by the RII indices. In particular, the RII indices for species richness, cover, and Shannon diversity were all positive in high aridity/high grazing conditions. The RII values were negative for species richness, Shannon diversity, and cover in the low aridity/low grazing conditions and also for Shannon diversity in the low aridity/high grazing (Table 2).

| Interaction intensity along the stress gradient
For all three indices (cover, richness, and Shannon diversity), RII was higher in the arid than in the semiarid climatic region (Figure 3). The RII indices for species' cover, species' richness, and Shannon diversity were all positive on both the LG and HG sites in the arid region, indicating a facilitative effect of the target shrub, Artemisia kopetdaghensis (Figure 3; Appendix A). However, the response of RII to the grazing intensity varied with climatic conditions. A significant facilitation (expressed by the positive RII values) was recorded in the semiarid region for species' cover, richness, and Shannon diversity, but only on the HG sites. The RII values for species' covers, richness, and Shannon diversity were negative for the LG sites in the semiarid region, indicating competition rather than facilitation by the dominant shrub (Figure 3; Appendix A).  (Table 3).

| Plant strategies and indicator species
There were not many indicator species in the open plots, all of them annual/perennial stress-tolerant species (e.g., Poa bulbosa in HG sites of the arid region; Aegilops triuncialis and Achillea biberesteinii in LG sites of the semiarid region; Table 3).

| Shrub canopy-mediated abiotic and biotic stresses
Changes in relative interaction intensity (RII) indicate changes in interaction type along a combined gradient of grazing and aridity. For all of the three RII indices (cover, richness, and Shannon diversity), there was a decreasing trend from the highest toward the lowest stress levels (Figure 3). The shrub (A. kopetdaghensis) showed facilitative effects, preserving species diversity and richness and the total cover of species under its canopy. However, the facilitative effect was significantly stronger in the drier climatic region. Previous researchers (Bertness & Callaway, 1994;Brooker & Callaghan, 1998;Butterfield et al., 2016) have documented increases in the facilitation effect of the shrub by moderating the aridity stress. In arid environments, facilitation usually involves increasing the water and nutrient availability (Holzapfel & Mahall, 1999). Besides that, the shade from the shrub reduces extreme temperatures and decreases evaporation from the soil, which may further facilitate the germination of seeds and growth of seedlings. Therefore, this may explain why the shrubs show higher facilitation in the arid than in the semiarid regions (Farzam & Ejtehadi, 2017;Smit et al., 2007;Tirado et al., 2015).
The effect of A. kopetdaghensis canopy was consistently facilitative under intensive grazing. As A. kopetdaghensis is unpalatable, it is not usually grazed by livestock during the growing season.

| Relative interaction index along the stress gradients
Various results have been reported, and some researchers indicated that the amelioration of abiotic stress was more important than protection from grazing (Arroyo et al., 2015;Howard et al., 2012).
On the contrary, other studies demonstrated that grazing was a more important driver of the plant-plant interactions than abiotic stress in the African savanna (Filazzola et al., 2017;Louthan et al., 2014).
In the arid region, strong facilitation was observed in both grazing intensities, suggesting that the protection from aridity is more important than protection from intensive grazing Soliveres et al., 2011). Accordingly, a theory by Smit et al. (2009) predicts relatively low importance of protection from grazing in water-limited environments. In arid climates, herbivores are sparsely distributed, and the availability of water or nutrients is more critical for vegetation than for protection from grazing (Ellis & Swift, 1988). In water-limited environments, the shrubs usually improve soil fertility and microclimate under their canopies (Cortina & Maestre, 2005, Maestre et al., 2009. Also, shade from shrubs' and trees' canopy can retain soil moisture at the soil surface and facilitate neighbors with shallower roots (Maestre et al., 2003). Therefore, the dominant shrub may promote species richness and productivity by providing safe microsites for species growing in harsh conditions (Bruno et al., 2003).
On the other hand, in the semiarid region, where plants presumably grow in higher water availability, livestock grazing played a critical role in determining the type and relative intensity of the shrub's interaction with undercanopy species. The effect of the shrub's canopy (RII) was positive on the HG sites, but changed to negative with lower livestock grazing intensity. In harsh conditions such as high grazing intensity, the positive RII means the shrub can directly enhance survival rate, growth, and reproduction of other species by providing a more suitable environment under its canopy.
However in low grazing intensity, a negative RII means that herbs prefer to grow in the open areas rather than under the canopy of shrubs, where they need to compete for light, nutrients, and water (Graff et al., 2007;Le Bagousse-Pinguet et al., 2012).  (Caccianiga et al., 2006). However, on the LG sites in the arid region, the canopy protects the surrounding plants from aridity only, so the stress is less intensive than on the HG site. Therefore, most of the indicator species under the shrub's canopy were annual forbs and grasses with R strategy on the LG site. Stress-intolerant species were better candidates for facilitation than stress-tolerant species (Graff & Aguiar, 2011). For instance, in the arid conditions of Mediterranean shrublands, the stress-avoidant species, with high specific leaf area and rapid growth, coexist with species featured by very low specific leaf area (Gross et al., 2013).

| Indicator species response to plant interactions in the condition of stress
In the semiarid region, A. kopetdaghensis canopy showed a facilitative effect on the HG sites, supporting the establishment of species with C-selected strategy, such as Elymus hispidus or Lactuca orientalis, which have larger leaves and are generally more palatable to livestock (Tajali, 2012). This is mainly because A. kopetdaghensis is an unpalatable, stress-tolerant shrub, and its canopy creates microsites, protecting other species against grazing by large herbivores. In  (Table 3). As suggested by Grime (1977), when the disturbance is relatively low, species with S-strategy can maintain their dominance in a community by occupying aboveground and belowground space rather than by competing for resources.

| CON CLUS IONS
Our results document that local-scale biotic processes, such as facilitation by the shrubs, are important determinants of diversity patterns.
In general, shrubs are known to provide refugee for species in harsh conditions, such as high aridity or overgrazing. Furthermore, we argue that the discrepancy in the literature on changes in plant-plant interactions may be partially explained by differences in plant life strategies of species in the conditions of the combined effect of biotic (grazing) and abiotic (aridity) stress. Therefore, in the arid region, drought-escaping species such as ephemerals and ruderals (R-selected) and species tolerating stress (S-selected) but avoiding herbivory are highly dependent on the facilitation under the canopy of shrubs. However, in sites without severe aridity, canopy of the target shrub protected competitive species (C-selected) in the conditions of high grazing intensity (low aridity/high grazing). On the contrary, on sites without intensive grazing and severe aridity (low aridity/low grazing), facilitative effects of the shrub turned to competitive. Restoration approaches are urgently needed, especially for dry rangelands, degraded by intensive grazing in countries that have limited resources. Understanding the role of plantplant interactions can significantly contribute to designing a sustainable management of both arid and intensively grazed areas.

ACK N OWLED G EM ENTS
This research is based on the results of a Ph.D. thesis by S. Rahmanian.
The financial support was provided by the Ferdowsi University

CO N FLI C T O F I NTE R E S T
None declared.  LG −0.14 −0.38 −0.11