Plant growth and productivity in terrestrial ecosystems are limited by N and P (Aerts and Chapin, 2000; Güsewell, 2004; Vitousek et al., 2010; Dijkstra et al., 2016). Plant N demand is largely determined by photosynthetic N use efficiency, which varies among species because of leaf morphology or photosynthesis differences (Lambers and Poorter, 1992; Garnier et al., 1995). The variation in the requirements of P is related to internal phosphate recycling rates or related metabolic pathways differences among plants (Vance et al., 2003; Chen et al., 2013). N and P contents in plants are jointly affected by plant intrinsic properties, carbon synthesis characteristics and meteorological and pedologic factors (Li et al., 2007). In terrestrial ecosystems, N and P availability is important for community function and vegetation composition, and N:P ratios changing in plants may influence the competition balance among species in their communities (Güsewell et al., 2002; Agren, 2004; van der Hoek et al., 2004; Han et al., 2005; Fujita et al., 2010). The growth and productivity performance of species is affected by N and P fertilization, because plants need to maintain a certain N:P stoichiometric balance to function properly at the specific conditions (Fujita et al., 2010). Consequently, it is critical to determine the N and P requirements of each species in order to minimize interspecific competition and increase yield in their mixtures (Isaac et al., 2012; Neugschwandtner and Kaul, 2015, 2016).

High exploration of the root system for water and nutrient acquisition is required for improving N and P uptake efficiency when soil nutrients are scarce (Hinsinger et al., 2011a; Isaac et al., 2012). In limited nutrient environments, two or more species growing under intercropping or mixture plantations, may increase N and P acquisition and use efficiency, because such systems are supposed to enhance mutual positive interactions between plants (Zhang and Li, 2003; Li et al., 2007; Hinsinger et al., 2011a,b). In that respect, the cereal-legume intercropping system have advantages in growth, yield and resource use efficiency compared to respective sole cropping, because of their complementarity and facilitation in using light, water, and nutrition (Zhang and Li, 2003; Li et al., 2007; Hinsinger et al., 2011a; Betencourt et al., 2012; Xia et al., 2013; Tang et al., 2014; Neugschwandtner and Kaul, 2014, 2015). Although the improved acquisition of N under the cereal-legume intercropping is well known, only few recent studies indicated that similar effects may occur for P (Hinsinger et al., 2011a; Betencourt et al., 2012; Xia et al., 2013). In some dry environments, like the semi-arid Loess Plateau of China, the soil environment is hostile for plant growth as water deficits often occur and the soil erosion causes losses of N and P (Meng et al., 2008; Lü and Han, 2010; Liu et al., 2013; Niu et al., 2016). To understand the acquisition and allocation of limiting N and P in plants is useful to determine their adaptation strategy to the environment, and find appropriate measures to increase their productivity and improve their function in the region.

Native dominant species have great eco-adaptation to regional climatic and soil environment, and play an important role in ecological conservation and services in the region. Understanding the N and P relations in these species under different water and nutrient (i.e., N and P) availability is critical to clarify their co-existence mechanisms, and to choose appropriate measures to improve their productivity in natural grassland. In the natural grassland community of the semi-arid Loess Plateau of China, there are two co-dominant perennial species, the C₄ grass Bothriochloa ischaemum and the C₃ legume subshrub Lespedeza davurica. When they are grown in mixtures under different water and fertilizer levels, biomass production and water use efficiency are higher than when grown alone (Xu et al., 2013). The different mixtures and levels of water and fertilizer supply, particularly N and P combined, affect the N and P concentration in different organs of the grass B. ischaemum (Xu et al., 2016). Here, we report a study aimed to examine the N:P stoichiometry of the matched legume species L. davurica, from both the organ and whole plant level in response to soil water, mixture proportions and fertility changing. Our aims were to clarify the N and P accumulation and N:P ratio in L. davurica at organ and plant level, and examine the effects of soil water and N and P fertilizers on N:P stoichiometry in L. davurica intercropped with B. ischaemum in various mixture proportions.

Legume and grass growing together is a common phenomenon in natural community. However, less attention has been given to nutrient behavior in different organs in their mixtures. Together with the sister paper, the studies provides the evidences of the N:P stoichiometric changes with mixture proportions in grass-forage mixture systems of native species in semiarid Loess Plateau (Xu et al., 2016). N and P concentration in different organs of the legume L. davurica varied with the SWC and fertilization when mixed with B. ischaemum, but their changing patterns were different. Leaf and stem N concentration of L. davurica did not vary significantly in response to mixture proportions, which may be associated to the low N requirements of this leguminous specie or it can maintain relatively stable N concentration in both organs under different conditions (Fujita et al., 2010; Sardans et al., 2012). In the grass-legume mixtures, it is likely that the grass absorbed N that was released by the legume root and nodule root system, resulting in a more efficient use of nutrients (Hauggaard-Nielsen and Jensen, 2005; Chen et al., 2013; Neugschwandtner and Kaul, 2014, 2015). Here N concentration in the roots of L. davurica under sole cropping was higher than under intercropping, indicating that mixture plantation decreased N enrichment in roots of L. davurica (Figure 1). It also may be caused by strong N acquisition of the neighboring shallow-root grass species B. ischaemum, whose fibrous root system intermingled with the legume roots and acquired the legume-derived N (Xu et al., 2016; Dhamala et al., 2017).

P concentrations in different organs of L. davurica under intercropping decreased compared with its monoculture, especially under +P or +NP treatments. There were two likely processes that may influence P concentration in L. davurica, as its proportion decreased in the mixtures: the first was that P fertilization tended to increase P availability, and the second was that limited P resources from fertilizer and soil was equally used by each single plant of L. davurica. Therefore, the increased numbers in the mixtures would decrease the P reallocation, limiting plant P uptake and P concentration in organs of L. davurica (Mi et al., 2015). Similar results were reported for the soybean-citrus intercropping, where P uptake and distribution amongst the organs of soybean decreased compared with their respective monoculture (Zhou et al., 2006). This contrasts with what was expected that P acquisition would be improved in both legumes and cereals under their mixtures because soil rhizosphere pH decreased, and P activities in the soils and roots (e.g., rhizosphere soil acid phosphatase activities) and rhizosphere soil P concentration increased (Inal et al., 2007; Betencourt et al., 2012). Unfortunately, we did not measure soil rhizosphere pH and phosphatase activity in the soils and roots in this study. P concentration in different organs or whole plant level of L. davurica decreased as its numbers decreased in the mixtures, implying that intraspecific competition intensification may increase its P uptake.

The higher N or P concentrations in the roots than in other organs and the higher N or P concentrations in the leaves than in the stems of L. davurica, could be explained by competition between the plant organs for N and P demand for growth, and the advantage that an organ may have for being close to the nutrient source. This is because nutrients are allocated to distant plant organs, after the demand of the adjacent organs to the nutrient source are met (Yang et al., 2014). However, this contradicts the findings from shrubs, where nutrient concentration in the roots increased in parallel with their neighbors, the stem and the leaves (Yang et al., 2014). Nutrient allocation is also driven by the plant and organ demand for N-rich proteins and P-rich RNAs (Vrede et al., 2004). Additional N and P allocation to the leaves are important to maintain leaf photosynthesis and photosynthate remobilization (Li et al., 2010). In dry environments (or dry years), a high N concentration in the leaves may provide adaptation to drought and competition by exploiting greater light availability while also increasing water use efficiency (Wright et al., 2003; Li et al., 2010; Yang et al., 2014).

Leaf N:P concentration ratio was higher than 16.0 under -NP and +N treatments, indicating a P limitation (Koerselman and Meuleman, 1996; Tessier and Raynal, 2003). Across most treatments (e.g., soil water regime, mixture proportion and fertilization), N:P ratio in leaf and root was significantly lower than 14.0, indicating a N limitation (Craine et al., 2008; Li et al., 2010). At whole plant level, the N:P ratio values were between 14.0 and 16.0 under -NP and +N treatments, while significantly smaller than 14.0 under +P and +NP treatments. These suggest that the species may have low P requirements, and mixture and fertilization changed the P accumulation in L. davurica (Zhou et al., 2006; Hinsinger et al., 2011a). Normally, N:P ratio is higher in structural organs than in metabolically active organs, as indicated here root had significantly lower N:P ratio (Xiao et al., 2004; Kerkhoff et al., 2006; Xu et al., 2013).

Our previous paper showed that the aggressivity values of B. ischaemum were bigger than 0 in their mixtures, which meant that B. ischaemum was the dominant while L. davurica the dominated species in their mixtures. In the study, the pot depth was about 30 cm, and the main soil depth for their root distribution was in 0–20 cm soil depth, which would strengthen the interaction between their root systems (Xu et al., 2015). This would explain why P facilitation may occur rather than competition between the two-species under intercropping. It is likely, that L. davurica was the preferred user of this pool when intercropped with B. ischaemum in comparison with respective monoculture (Li et al., 2008; Xu et al., 2013). Another reason could be that strong competition between them for accessing the same soil horizons because both are native species whose roots mainly distribute in the top soil layer and the small pots may restrict root distribution and proliferation (Li et al., 2006; Zhou et al., 2006; Hinsinger et al., 2011a).

Water is the main factor limiting plant growth and production in arid and semiarid region on the Loess Plateau of China. Nutrient uptake and concentrations in plants would be reduced by drought due to decreased transpiration flow and root growth, and lower soil nutrient mobility (Neugschwandtner and Kaul, 2015, 2016). In the specific water condition, rational fertilization is favorable for plant growth and production (Shan and Chen, 1993; Xu et al., 2016). Here, we found that single plant biomass production of L. davurica increased about 10% under N and/or P fertilization treatments, while the effects were mixture ratio-specific (Supplementary Figure S3). The N and P concentration in organ and whole plant level generally increased as N and/or P fertilizer were applied, while they were not exactly the same for all treatments. Along or with N, P fertilization could not only increased biomass production, but also nutrient concentration in each organ of L. davurica across mixture proportions and soil water regimes. For example, compared with 40%FC soil regime, single plant biomass production of L. davurica under 80%FC increased 35.8% at 10:2 of B. ischaemum to L. davurica after +NP treatment, while N concentration in leaf and whole plant level decreased 10 and 12%, respectively (Figures 1, 4, 5). In the same mixture ratio, compared with 80%FC, the single plant biomass production of L. davurica under 40% FC soil regime decreased about 27% after N fertilization (i.e., +N treatment), but N concentration in whole plant level increased about 10.7% (Figures 1, 5). All these suggested that P fertilization played an important function in improving biomass production and increasing nutrient concentrations in different organs of the legume, and also implied that the improvement of biomass production after fertilization may possess “dilution effect” on nutrient concentration in plant, besides “enrichment effect” (Craine et al., 2008; Neugschwandtner and Kaul, 2015; Li et al., 2016).

On the semiarid Loess Plateau of China, wild and native plant species experience the harsh environment with variable soil water and infertile soil nutrient conditions. Artificial fertilization was normally applied in order to increase their growth, and thus to improve vegetation coverage, accelerate successional process and mitigate desertification in the region. Results showed that P applied alone or combined with N, affected not only the biomass production, but also the N and P accumulation in each component of L. davurica in the mixtures, suggesting that P element should not be neglected when considering rehabilitating degraded grassland through fertilizer application. N and/or P fertilizers significantly affected N and P concentrations at both organ and whole plant levels, and the effects were closely related with mixture proportions, implying that exogenous fertilization may change the N:P stoichiometry of native dominant species in natural grasslands, and such effects may depend on their relative abundance in the community. Our results also provided clearly evidence of inconsistent changes of N:P stoichiometry in the organ and whole plant level under same treatment, suggesting that the organ differences should be considered when studying nutrient absorption and cycling between or among plant species. Because B. ischaemum and L. davurica are perennial species, and here we only reported the nutrient behaviors during the establishment period in their controlled mixtures, it is worthy and necessary to test their nutrient requirements as growth years extended, especially in the field natural conditions. Further research also needs to be taken on the nutrient cycling and balance at the specific mixture proportion in response to water and fertilization treatments to clarify the detailed mechanisms behind.

BX and YC conceived and designed the experiments. WX, ZW, and ZC performed the experiments. WX, ZW, ZC, and JP analyzed the data. BX and JP contributed reagents/materials/analysis tools. BX and YC wrote the paper.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.