Dynamic Changes in Plant Resource Use Efficiencies and Their Primary Influence Mechanisms in a Typical Desert Shrub Community

: Understanding plant resource use efficiencies ( RUE s) and their tradeoffs in a desert shrub community, particularly as it concerns the usage of water, light, and nitrogen, remains an ecological imperative. Plant RUE s have been widely used as indicators to understand plant acclimation processes to unfavorable environmental conditions. This study aimed to examine seasonal dynamics in RUE s in two widely distributed plant species in a typical desert shrub community (i.e., Artemisia ordosica and Leymus secalinus ) based on in-situ measurements of leaf photosynthesis, specific leaf area ( SLA ), leaf nitrogen concentration (i.e., N mass + N area ), and several site-related abiotic factors. Both species exhibited significant seasonal variation in RUE s, with a coefficient of variation (CV) > 30% and seasonal divergence among the various RUE s. Seasonal divergence was largely controlled by variation in stomatal conductance ( G s), which was in turn influenced by variation in soil water content ( SWC ) and water vapor pressure deficit ( VPD ). RUE s between species converged, being positively correlated, yielding: (i) r 2 = 0.40 and p < 0.01 for WUE; (ii) r 2 = 0.18 and p < 0.01 for LUE ； and (iii) r 2 = 0.25 and p < 0.01 for NUE . RUE s for A. ordosica were mostly larger than those for L. secalinus , but less reactive to drought. This suggests A. ordosica was more conservative in its usage of available resources and was, therefore, better able to adapt to arid conditions. Resource use strategies between species differed in response to drought. Desert shrubs are projected to eventually replace grasses, as drought severity and duration increase with sustained regional climate change.


Introduction
Drylands (arid and semiarid areas) cover 39% of the earth's land surface and are home to about 20% of the world's human population [1]. Ecosystems in drylands are highly vulnerable to global environmental change and desertification [2]. Given the speed and intensity of climate change and socioeconomic development that risk aggravating environmental and socioeconomic problems on various spatial scales (e.g., land degradation, poverty, declining food and water security), research on both social and ecological system processes, as well as their interactions, is urgently needed [3,4]. Recent studies highlighting the importance of semiarid ecosystems in their contribution to terrestrial net primary production are also key to forming meaningful sustainable development goals (SDGs) for drylands [1,[4][5][6].
Plants in drylands generally experience more environmental stress, especially as severity and duration of droughts continue to increase under a warming climatic regime [7][8][9]. Consequently, it is expected that there will be changes in the supply of available plant resources, such as water (W), light (L), and nitrogen (N) for photosynthetic assimilation, as well as for the maintenance of functional plant structures. This may lead to changes in plant resource use efficiencies (i.e., RUEs, or individually through WUE, LUE, and NUE). Plant RUE (and its components) is defined as the ratio of photosynthetic assimilation to the consumption and subsequent usage of resources [10,11]. Photosynthesis is generally limited by W, L, and N, although other abiotic factors, such as temperature (T), can also play a crucial part, particularly at seasonal timescales [12]. The survival and growth of plants could be influenced by RUEs, which could ultimately cause fluctuations in ecological functioning. Plant RUEs have been widely used to understand plant acclimation processes to unfavorable environmental conditions expected to accompany climate change and associated extreme weather events.
Plant resources may change in unison. Changes in one resource will likely induce changes in other resource uses at the leaf, species, and community level, conditional on the temporal scale [13]. Changes in the availability of one resource may result in changes, not only in the use efficiency of that resource, but also in the use efficiencies of other resources [10,14]. For instance, water supply increases NUE, but decreases WUE, whereas the addition of N slightly increases WUE at the expense of NUE at the leaf level. This suggests that the shift in the availability of one resource could engender disproportionate constraints on the use efficiencies of other plant resources [15]. Large transpiration rates and high solar irradiation proceed to maximize instantaneous photosynthesis, which correlates positively with NUE and negatively with both WUE and LUE [16]. Consequently, the individual response in WUE, LUE, and NUE to changes in environmental conditions are not self-governing. Therefore, plants may exhibit divergent tradeoffs between the various RUEs during acclimatization to prevailing site conditions. To better model plant responses to climate change in arid and semiarid areas, relative changes (tradeoffs) among WUE, LUE, and NUE and their causes need to be quantified.
RUEs and their response to environmental change have received considerable interest (e.g., [17][18][19][20]). Changing environmental factors constrain the variation in plant RUEs, as the factors influence the supply and demand of resources [20,21]. RUEs integrated over monthly timescales (i.e., RUEm) vary seasonally in response to variation in resource availability and weather conditions [16]. Crops and native vegetation that are adapted to water-limited conditions, achieve adaptation mainly by dehydration avoidance and escape, rather than having an innate ability to function in a dehydrated state or be desiccation tolerant [22]. Fischer and Maurer have shown that agricultural cultivars yield biomass under drought as a function of yield potential (yield without drought), drought susceptibility index, and severity of drought [23]. Improved WUE is usually expressed in an improved yield under water-limited conditions, only when there is a need to balance crop water use against a limited and known soil water supply [22]. Plant breeders and physiologists have long been concerned with drought resistance in plants. Just how dryland plants adjust their RUEs to resist drought remains uncertain.
Species is by far the most influential factor explaining the variance in leaf photosynthetic assimilation [24]. RUEs, key plant function, acclimate to environmental fluctuations through coordination and tradeoffs among plant morphological, physiological, and biochemical traits in achieving optimal resource usage and adaptation to prevailing environmental conditions [18,25]. Plant production in water-limited environments is very often affected by constitutive plant traits that allow maintenance of a high plant water status [22]. Owing to the likely diversity of factors and factor interactions to RUEs, dynamic changes in RUEs and their primary influence pathways or mechanisms are not entirely understood, and thus further studies are needed, especially for dryland ecosystems.
Plant ecosystem health in arid and semiarid areas is rapidly deteriorating as water shortages become more frequent [26]. Responses of plant communities to environmental change are typically regulated by the dominant species in the community [27,28]. It has been reported that the neighborhood of a tree can have a significant impact on functional traits involved in resource use [29]. Shrub-dominated plant communities in drylands of northwest China are commonly associated with the presence of Artemisia ordosica Krasch. (shrub) and Leymus secalinus (Georgi) Tzvel. (desert grass) [19]. These communities are widespread throughout the semiarid regions of the Mu Us Desert [30]. Their sustainability and presence, however, are being greatly affected by severe limitations in available soil water emergent in many parts of the desert complex [26,31].
This study aims to understand the characteristics and influence mechanisms and pathways responsible for the observed variation in RUEs in both A. ordosica and L. secalinus in response to dry conditions. The specific objectives of the study are to: (i) examine seasonal dynamics in the individual components of RUEs (i.e., WUE, LUE, and NUE); (ii) determine whether a level of convergence exists in RUEs between the two plant species; and (iii) clarify the role of environmental factors in the control of seasonal dynamics in RUEs. The study will have important implications for clarifying the acclimatization capacity of dryland shrub-grass associations and understanding how drylands respond to ongoing environmental change. It is a key theme of SDGs of drylands, regarding their social-ecological system dynamics and drivers [4,32]. The study has the potential to deliver both novel scientific insights and development impact consistent with the aspirations of United Nations' SDGs for 2030.

Site Description
The study site is located at the Yanchi Research Station of Beijing Forest University (37°42′31″ N, 107°13′47″ E, 1530 m above mean sea level), Ningxia, northwest China (Figure 1). The site is representative of a transitional zone between arid and semiarid conditions at the southern edge of the Mu Us Desert. The prevailing regional climate is temperate arid and semiarid, characterized with scarce rainfall, irregularly distributed, and variable from year to year. The mean annual temperature (based on 1954-2020 data) is 8.4 °C, and the mean annual precipitation is 293 mm. Most precipitation (>70%) occurs during June-September (data source: Yanchi Meteorological Station, Yanchi Research Station). The soil at the site involves sand, with a bulk density of 1.54 ± 0.08 g cm −3 , a total porosity of 35.70 ± 3.83%, a field capacity of 20.31 ± 3.33% (g g −1 × 100), and a permanent wilting point of 3.64 ± 0.37% in the upper 10 cm of the soil profile (g g −1 × 100; mean ± standard deviation, n = 16). The landscape of this region is a typical inland dune ecosystem with very distinct habitat types [33]. The shrub cover at the site consists mostly of A. ordosica, and smaller amounts of Hedysarum mongolicum Turez.and Salix psammophila C. The most abundant herbaceous species include L. secalinus, Stipa glareosa P. Smirn., Pennisetum centrasiaticum Tzvel., and Setaria viridis (L.) Beauv.. The main canopy of the community is about 1 to 1.5 m tall. Shrub roots are distributed mainly in the upper 10-50 cm of the soil.

Photosynthesis Gas-Exchange Measurements
The measurements were taken from plants enclosed within three 5 × 5 m 2 plots. Three individuals of both A. ordosica and L. secalinus were randomly selected in each plot (n = 9) assigned for in situ measurements [26,34,35]. Photosynthesis was measured on fully developed leaves on the south-facing side of each assigned plant individuals, every 10 days from May to September 2019. The measurements were taken with a portable LI-6400 photosynthesis system, equipped with 2×3 cm 2 transparent light source chamber (Li-Cor Inc., Lincoln, NE, USA). Five measurements were taken from each individual plant, during each measurement period. All leaf gas-exchange characteristics were measured six times every ~2-h from 8:00-18:00 Local Beijing Time (Greenwich Mean Time + 8 h). Rates of photosynthesis and transpiration (Pn and E, in μmol m −2 s −1 and mmol m −2 s −1 , respectively) and stomatal conductance (Gs, mol m −2 s −1 ) were measured, which were expressed per unit leaf area. The leaf area of fresh leaves was measured at the same time using an LI-3100C leaf area meter (LI-COR Environmental, Lincoln, NE, USA).

Measurement of Biotic Factors
Two samples of 10 leaves were collected immediately after each gas-exchange measurement on nearby plants with similar characteristics to the target plants, for both leaf area measurement and N content. After each leaf area measurement, fresh weight of sampled foliage was measured with an electronic balance. The fresh leaves were then ovendried at 75 °C for 48-h for dry weight determination. The specific leaf area (SLA, cm 2 g −1 ) was then calculated as the ratio of fresh leaf area to dry weight. The leaf N concentration was determined both on a dry weight and leaf area basis (i.e., Nmass and Narea, g kg −1 and g m −2 , respectively) using the Kjeldahl method [10].

Measurement of Abiotic Factors
All meteorological variables were measured with sensors installed on 6-m tall eddycovariance tower, assembled next to the sampling area. Tower-based measurements included: (i) air temperature (T, °C) and relative humidity (RH, %) taken with a thermohygrometer (HMP155A, Vaisala, Finland); (ii) net radiation (Rn, W m −2 ) with a CNR-4 sensor (Kipp & Zonen, Netherlands); and (iii) incident photosynthetically active radiation (PAR, μmol m −2 s −1 ) with a quantum sensor (PAR-LITE, Kipp & Zonen, The Netherlands). Half-hourly soil water content (SWC, m 3 m −3 ) was monitored at a 30-cm depth within a 10-m radius around the tower, with three replicate sensors (ECH2O-5TE, Decagon Devices, Pullman, WA, USA). Rainfall (PPT, mm) was measured using a tipping bucket raingauge (TE525WS, Campbell Scientific Inc., Logan, UT, USA) set in an opening approximately 50 m from the tower. Vapor pressure deficit (VPD, kPa) was calculated from tower-based measurements of RH and T.

Data Processing and Analysis
Water, light, and nitrogen use efficiencies (i.e., WUE in μmol mmol −1 , LUE in mol mol −1 , and NUE in μmol g −1 s −1 , respectively) were calculated individually with the equations (1)-(3): where Pn is the net photosynthesis, E the transpiration, PAR the leaf surface photosynthetically active radiation, and Narea the amount of N per unit leaf area. Monthly water, light, and nitrogen use efficiencies were subsequently calculated as averages of daily WUE, LUE, and NUE. Major terms and their acronyms appear in Table A1.

Statistical Analysis
The coefficient of variation (CV) was used to quantify the seasonal variation in RUEs (i.e., WUE, LUE, and NUE) and biotic factors. To analyze relationships among WUE, LUE, and NUE for a given species and to determine whether a level of convergence exists in RUEs between the two plant species, a standard major axis (SMA) operation was performed using sma, a procedure available in the smatr R-package. Paired student t-tests were performed in pairwise comparisons of WUE, LUE, and NUE for A. ordosica and L. secalinus. To clarify the role of factors in the control of seasonal dynamics in RUEs, stepwise regression was used to find the most critical biophysical factors responsible for controlling RUEs. In addition, structural equation models (SEM) were used to assess the direct and indirect contributions of biotic and abiotic factors to variations in RUEs. The significance level was set at p = 0.05.
Drought days in 2019 were defined as those days with daily mean SWC < 0.1 m 3 m −3 [9,26,36]. The average dates for the onset and end of the growing season were set at early May and end of September. Spring, summer, and autumn were defined as occurring in May, June-August, and September, respectively. All statistical analyses were performed in R-Studio ver. 3.6.3 (The R Development Core Team) and Origin2017 (OriginLab, Northampton, MA, USA).   Figure 3A-C). Values of photosynthetic parameters for A. ordosica were generally greater than those for L. secalinus. Plant parameters of SLA, Nmass, and Narea for A. ordosica varied less than those of L. secalinus over the same period, with CV of 16, 10, and 20% for A. ordosica, and 24, 12, and 34% for L. secalinus (Figure 4). Leaf N concentrations (with respect to both Nmass and Narea) were lower for A. ordosica than those for L. secalinus ( Figure 4B,C).
Monthly WUE lowered in summer, with minima of 0.90 and 0.94 μmol mmol −1 in July for A. ordosica and L. secalinus ( Figure 6A), respectively. Nitrogen use efficiency peaked in summer for both species, with maxima of 6.97 and 2.65 μmol g −1 s −1 in August ( Figure 6C). Light use efficiency peaked at different months of the growing season for the two species, i.e., LUE being maximum in spring for A. ordosica, with a value of 0.033 mol mol −1 , and in summer for L. secalinus, with a value of 0.016 mol mol −1 ( Figure 6B).

Controlling Factors on Variations in RUEs
Resource use efficiencies were mainly affected by Gs, SWC, and VPD for both species (Table 1). Specific leaf area only affected RUEs in L. secalinus. Control of biophysical factors on RUEs was through their direct and indirect effects on Pn, E, and Narea (Supplementary Material Figures S1A-C and S2A-C). Moreover, RUEs were essentially more strongly regulated by Pn than by E, PAR, or Narea (Supplementary Material Table S1). Net photosynthesis was more directly affected by Gs and VPD, while E was more directly affected by

Variations in RUEs and Their Controlling Factors
Variation in plant RUEs can indicate plant growth strategies in different environmental conditions [37,38]. Previous studies have reported that environmental fluctuations can cause changes in RUEs [16,17,26]. Our finding that RUEs is significantly affected by SWC and VPD over the growing season (Table 1), is supported by previous findings in aridshrub species, such as S. psammophila and H. mongolicum [26,39]. Soil water content and VPD determined seasonal variation in RUEs by controlling Pn and E, but more predominantly Pn (Supplementary Material Figures S1A-C and S2A-C). High VPD induced stomatal closure, preventing excessive water loss, resulting in decreased Gs and, thus, a disproportionately larger decrease in Pn, than in E [40][41][42]. This led to a lowering of WUE, as shown by its negative relationship with VPD and Gs (Table 1). Soil water boosted by intermittent rain pulses, normally resulted in an elevated E, more than observed in Pn, leading to a reduction in WUE during the summer (Supplementary Material Figures S1D and  S2D) [43][44][45].
The result that LUE was positively controlled by Gs and SWC (Table 1), is commonly observed in desert plants [46][47][48], for which light is not a limiting factor. Soil water supply in dry areas is often improved by intermittent rain pulses, leading to larger photosynthetic capacities [36]. When soil water was enough to meet atmospheric demand, Pn increased along with Gs (Supplementary Material Figures S1A-C and S2A-C) [9,45], causing LUE to be greater, and vice versa. Offset of water restrictions increased the efficiency by which PAR was converted to photosynthates. A reduction in stomatal constraint probably also played a positive role.
Since Narea changed very little over the growing season, the observed seasonality in NUE was mainly due to changes in Pn (Figures 3A and 4C; Supplementary Material Table  S1), establishing a positive relationship between NUE and Gs through its positive effect on Pn (Table 1). Higher stomatal conductance had caused Pn to be higher [11], which may have caused NUE to be greater during the summer.
Overall, variations in RUEs were largely controlled by Gs. Stomatal conductance was itself mostly controlled by SWC and VPD. Seasonal patterns in RUEs were due to the control applied by intermittent rain pulses and associated soil water supplies. This result confirmed that the presence of water was responsible for the large changes observed in resource use and efficiencies in arid and semiarid grasslands [10,13,49,50]. It submits that drier conditions at our site more likely decreased SWC in situ, which limited plant physiological activity and growth. It may be important for vegetation-climate models of net primary productivity to address these fundamentals in predicting plant responses in dryland ecosystems.
A. ordosica and L. secalinus, as indicator species, can be selected on the basis of their trait values' responsiveness to environmental factors and their importance both locally and regionally [9,19,39], for monitoring trends in ecosystem-level properties across environmental gradients (e.g., pollution, drought, fertility) [51]. These field measurements always involve a balance between the number of replicates and precision. The number of replicates selected should depend on species variability in the trait of interest, as well as on the number of species sampled [34]. Pérez-Harguindeguy et al. showed that the minimum and preferred number of replicates for different traits is five and ten, respectively, based on common practice [34]. Prior studies in semiarid shrublands usually involved three to seven replicates [26,35,39,45]. Given constraints of time, we selected nine replicates for each species for an improved assessment, and thus could guarantee validity of our study results.

Tradeoffs between RUEs
High NUE occurred predominantly when LUE was high in both species ( Figure  7C,F), such as those seen in boreal trees [52], suggesting that a level of convergence existed between LUE and NUE. It was previously reported that plants can reduce the constraints on carbon uptake by maximizing the use efficiency of the most limiting resource, while lowering the use efficiency of resources that are more abundant [53]. Some prior studies have shown a negative association between LUE and NUE, in contrast to our results [16,54]. One plausible explanation for this discrepancy may be associated with the fact that LUE in plants reflect processes of carbon fixation. This is not the case with NUE, where NUE is mostly the product of carbon fixation and protein synthesis, which are weakly coupled to carbon uptake. Temporal integration also tends to decrease the relative importance of resource availabilities and cause long-term RUE responses to differ from short-term observations [52,55]. As a key ecological function, long-term relative changes in RUE deserve further investigation for improved understanding in how plants may respond to climate change, especially with long-lasting effects of highly variable precipitation and extreme aridity in dryland ecosystems.
In A. ordosica, variation in NUE and LUE were mostly explained by variation in Pn, leading to a positive correlation between LUE and NUE (Supplementary Material Table  S1). Increased Pn was accompanied by an elevated Gs, increasing LUE and NUE when leaf N concentrations were relatively stable [56]. In L. secalinus, positive correlation between LUE and NUE were explained by variations in Pn and Nmass (Figures 3A and 4B; Supplementary Material Table S1). The plasticity of LUE to high insolation levels was largely due to Pn being limited by carboxylation capacity and associated Nmass [57]. High insolation acted to maximize instantaneous photosynthesis, which correlated positively with LUE due to reduced absorption of saturated light, and positively with NUE due to lower N investment that maximized Pn during the summer [58,59].
There was neither correlation between WUE and LUE, nor correlation between WUE and NUE in both species ( Figure 7A,B,D,E). The Mu Us Desert is occasionally affected by drought; SWC was shown to be a key factor in controlling plant RUEs and constraining plant growth (Table 1) [10,26,45,50]. A. ordosica has an ability to avoid functional damage by reducing transpiration losses through stomatal closure during periods of excessive dryness [9,39]. Consequently, its RUEs response to drought largely depended on the physiological control conveyed by Gs. Physiological and structural variations in Gs and SLA both regulated RUEs in L. secalinus (Table 1). Drought-coping variations in plant morphological and physiological traits could have led to differences in tradeoffs between RUEs in the two species, as observed in prior studies [18,29,60]. The results indicated that dominant species have adaptive differentiation of resource use-related traits to achieve local coexistence in dryland ecosystems.

Relationships in RUEs between the Two Species
The finding that RUEs in A. ordosica was positively and linearly correlated with those in L. secalinus (Figure 8), provides some evidence of convergence. This result is consistent with trends seen in other studies [e.g., 26,29,38], such as those reported for piñon pine and juniper [11]. RUEs in dominant plant species growing in harsh environments, such as arid [61], semiarid Mediterranean [38], and semiarid shrubland (this study) are subject to the effects of various levels of drought intensity and duration. Clearly, convergence in RUEs in the dominant plant species can be largely explained by soil water limitations experienced at these sites. It further confirms that ecosystem functioning is more sensitive and vulnerable to highly variable precipitation, extreme water scarcity, and pronounced fluctuations in diurnal temperatures in drylands [2,4].
Although convergence existed in RUEs between the two species, there was a difference in their magnitudes. RUEs in A. ordosica were mostly greater than those in L. secalinus (Figures 5 and 6; p < 0.01, Figure 9), due to their elevated leaf photosynthetic capacity and Pn ( Figure 3A). Prominent SLA in A. ordosica means that it has a larger capacity to capture light and acquire nutrients more directly than in L. secalinus. These results suggest that shrubs may be better suited for arid conditions, as previously proposed by Zha et al. and Wu et al. [9,26,36]. Compared to RUEs in L. secalinus, variation in RUEs was more limited in A. ordosica ( Figure 5A,B), indicating that desert shrubs may be more resistant to drought. Suppression of SLA, Nmass, and Narea was greatest in A. ordosica, compared to that in L. secalinus (Figure 4). Soil water content (SWC) and VPD had more influence on LUE and WUE in L. secalinus, suggesting that the species is more responsive to drought than A. ordosica. Past studies have reported similar findings in other desert plants (e.g., [9,19,62]).
There was divergence in the relationship between plant drought tolerance and its RUEs [11]. Desert plants either select resource acquisition or resource conservation in their response to drought [63]. In this study, A. ordosica demonstrated a resource-conservation strategy [10,26,39]. This explains the high RUEs and low sensitivity to environmental change observed in A. ordosica, compared with L. secalinus (Figures 5 and 6; Supplementary Material Figures S1D-F and S2D-F). A. ordosica had higher RUEs, and thus could accelerate phenological change, improve photosynthetic production, and complete its life cycle as quickly as possible to avoid dehydration and escape drought, which resulted in its better acclimatization to dry environments in comparison with L. secalinus [64]. With continued increases in drought severity and duration and associated lowering of groundwater reserves anticipated with future climate change, desert shrubs are projected to eventually replace grasses, impairing the function of natural ecosystems and, thus, diminishing the number and quality of ecological services they provide [2,65]. Our results suggest that the preservation of plant biodiversity in arid and semiarid drylands is crucial to buffer the undesirable effects of climate change and desertification.

Conclusions
Both species had similar seasonal patterns in RUEs (i.e., WUE, LUE, and NUE) and exhibited significant seasonal variation in RUEs from May-September 2019 (CV > 30%). Monthly WUE lowered in July and NUE peaked in August for both A. ordosica and L. secalinus. Light use efficiency peaked at different months for the two species. SWC and VPD were the environmental factors that affected variation in RUEs the most by regulating Gs. For a given RUE, there was a convergence in resource use efficiency between the two species, with A. ordosica exhibiting lower overall variation compared to L. secalinus. However, the tradeoffs between RUEs were divergent in the species. It was determined that A. ordosica was more adaptable to arid conditions than L. secalinus. Vegetation succession may lead to shifts in ecosystem composition in favor of more drought tolerant species, such as A. ordosica, in the near-to-immediate future. This may result in a reduction in biodiversity and ecosystem functioning, and enhanced vulnerability to global climate change and human disturbance in drylands. Dryland social-ecological systems are especially sensitive to rapid rates of physical and social change, such as those associated with climate change and urbanization. Our findings demonstrate the need to strengthen dryland ecosystem management methods for sustainable livelihoods and to advance progress towards the implementation of 2030 SDGs for drylands.

Data Availability Statement:
The data and materials that support the findings of this study are all available from the corresponding author upon reasonable request.

Acknowledgments:
We would like to thank X.W. Yang, S.J. Liu, G.P. Chen, C. Zhang, Y. Luo and H. Tian for their assistance with field measurements and instrument maintenance. The U.S.-China Carbon Consortium (USCCC) supported this work by way of helpful discussion and the exchange of scientific ideas.

Conflicts of Interest:
The authors declare no conflicts of interest.