Effects of plant species richness on 13C assimilate partitioning in artificial grasslands of different established ages

Artificial grasslands play a role in carbon storage on the Qinghai–Tibetan Plateau. The artificial grasslands exhibit decreased proportions of graminate and increased species richness with age. However, the effect of the graminate proportions and species richness on ecosystem C stocks in artificial grasslands have not been elucidated. We conducted an in situ13C pulse-labeling experiment in August 2012 using artificial grasslands that had been established for two years (2Y), five years (5Y), and twelve years (12Y). Each region was plowed fallow from severely degraded alpine meadow in the Qinghai-Tibetan Plateau. The 12Y grassland had moderate proportions of graminate and the highest species richness. This region showed more recovered 13C in soil and a longer mean residence time, which suggests species richness controls the ecosystem C stock. The loss rate of leaf-assimilated C of the graminate-dominant plant species Elymus nutans in artificial grasslands of different ages was lowest in the 12Y grassland, which also had the highest species richness. Thus the lower loss rate of leaf-assimilated C can be partially responsible for the larger ecosystem carbon stocks in the 12Y grassland. This finding is a novel mechanism for the effects of species richness on the increase in ecosystem functioning.

grasslands established for different lengths of time are poorly understood. Previous research has found ecosystem carbon storage was strongly influenced by plant diversity 20,21 . Plant diversity increases root biomass or root carbon concentration [22][23][24] and microbial community activity [25][26][27] . Additionally, carbon storage increases with higher inputs of root carbon residues 28 or with more plant-derived recalcitrant compounds 29 and is directly influenced by the microbial community activity 21,[30][31][32][33][34] .
Prior studies in the alpine meadows of the Qinghai-Tibetan Plateau have indicated that community structure determines the ecosystem carbon storage capacity 18,[35][36][37] and plant photo-assimilated carbon cycling 38 . Another study found the species richness and root-shoot ratio were the major controlling factors of soil C stocks when comparing the native alpine meadow with winter grazing, cultivated perennial and annual pastures 18 . The variations in community structures in these studies are all associated with changes in species richness and proportions of graminate. In artificial grassland on the Qinghai-Tibetan Plateau, graminate is typically the cultivated grass species, some of which could be replaced by forbs with increasing planting years and different anthropogenic management strategies. Graminate are fast-growing grasses that typically have strong nutrient-absorbing capacity and low below-ground allocation compared with other forbs in alpine meadow environments. However, the effects of the different graminate and of the species richness on the ecosystem C store of artificial grasslands of varying ages remains unclear. These differences may have significant effects on the restoration and management of degraded grassland.
In this study, we hypothesize that a low proportion of graminate grasses or high species richness would have positive effects on the C store in high-altitude artificial grasslands. We used an in situ 13 C pulse-labeling experiment (plant diversity has pronounced effects on the enhanced accumulation of recently fixed carbon but not existing carbon 21 ) to test these hypotheses on three general grasslands that were ploughed from "black beach" for two years (2Y, young), five years (5Y, middle-aged) and twelve years (12Y, old). The purposes of this study were the following: (1) to quantify the partitioning of recently fixed carbon among shoot, root, and soil and fluxes in the plant-soil system and (2) to estimate the effects of graminate grasses and species richness of artificial grasslands on carbon cycling in the plant-soil system.

Results
Total carbon stocks and biomass. Biomass, root-shoot ratio, species richness and above and belowground total carbon stocks for different established ages of artificial grassland are shown in Table 1. Artificial grassland after 5 years had the least above-ground biomass, and the 12-year grassland had the most. The total carbon showed no significant differences in either above-or below-ground biomass for the three types of grassland. The species richness is 9.4, 2.6, and 15.5 for 2Y, 5Y and 12Y grasslands, respectively. 13 C dynamics and partitioning in the plant-soil system. After the labeling, 13 C in shoots of the three general types of artificial grasslands followed an exponential decrease during the chase period (Fig. 1). The decline reflects the newly assimilated 13 C allocated to below-ground stocks and the C loss by respiration. The dynamics of 13 C allocation in shoots were similar for three general types of artificial grasslands, but the recovery of 13 C in shoots differs significantly between the 2Y and 12Y artificial grasslands after the 1st day of labeling.
The newly assimilated 13 C allocated from shoots to below-ground C pools and to respiration between 3 h and 22 days after labeling amounted to 44.9, 45.5, and 26.0% of recovered 13 C in the grasslands aged 2Y, 5Y, and 12Y, respectively (Fig. 1). Different sections of the sites were used for below-ground C allocation (Fig. 2) and for respiration (Fig. 3). The amount of 13 C transferred to below-ground pools was greater in the 12Y and 5Y sites than in the 2Y site, and differences were significant among 12Y, 5Y and 2Y grasslands at all sampling times (Fig. 2). The maximum recovery of 13 C in below-ground pools occurred on the 1st day after the labeling in the 12Y site. Carbon loss by respiration for the 5Y site was greater on the 1st day and the 8th day of the chase period (Fig. 3), and the maximum loss of 13 C by respiration was not significant at 22 days, the end of the labeling time.
On the 1st day after labeling, assimilated 13 C in roots for the 2Y, 5Y, and 12Y plots was 11.8, 15.3, and 25.7%, respectively (Fig. 4a). The recovery of 13 C in roots showed significant differences for all sampling dates during the  Table 1. Biomass, root-shoot ratio, total carbon, species richness and ratio (graminate biomass/forb biomass) for the three types of grasslands. 2Y, planting of artificial grass for two years; 5Y, planting of artificial grass for five years; 12Y, planting of artificial grass for twelve years. Different letters indicate significant differences among the three general types of lands (n = 3, P < 0.05); ns represents non-significant differences.   labeling period. During the following days, the percentage of 13 C for 5Y grassland gradually increased to 18.5% at 22 days, whereas the percentage for 12Y grassland decreased to 13.2%, and in the 2Y plots, the percentage remained constant at 11.8% (Fig. 4a). At 3 hours after labeling, 1.7%, 3.5%, and 6.9% of the recovered 13 C was found in the soil of the 2Y, 5Y, and 12Y artificial grassland, respectively. The 13 C recovery in the 12Y site was much higher than in the other two grasslands (Fig. 4b), with a larger percentage of recovered 13 C located in the topsoil at a depth of 0-10 cm for the 12Y plots (Fig. 4c). The maximum recovery of 13 C in soil was 8 days after the labeling in the 2Y (4.9%) and 5Y (5.8%) plots (Fig. 4b), and significant differences were observed among the three grasslands at all sampling times ( Fig. 4c).
At the end of the chase period, 15.6%, 21.7% and 20.9% of recovered 13 C in the 2Y, 5Y, and 12Y grasslands, respectively, had been transferred to below-ground C pools ( Fig. 5). At 22 days after the labeling, 44.3% and 42.5% of the recovered 13 C was used for respiration in the 2Y and 5Y grasslands, respectively, more than in the 12Y site (34.9% of recovered 13 C) (Fig. 5). Within the below-ground C pools, the smallest proportion of 13 C was incorporated into the soil in all three types of grasslands ( Fig. 5). At 22 days after labeling, 18.5% of recovered 13 C was stored in roots in the 5Y plot, more than in the other two ages of grasslands (11.8% in 2Y and 13.2% in 12Y).

Mean residence time.
Mean residence time (MRT) of net assimilated C was determined to evaluate its lifetime in the plant-soil system. The longest MRT detected was 51.3 days in 12Y, followed by 39.8 days in 5Y and 37.9 days in 2Y.

Relationships between ratio (graminate biomass/forb biomass) and recovered 13 C in soil and
MRT. Relationships between ratio (graminate biomass/forb biomass) and recovered 13 C in soil and MRT were not significant in linear fittings. The quadratic curve fittings (% Soil -rec 13 C = − 0.99 × ratio 2 + 9.44 × ratio-12.61, and MRT = − 2.71 × ratio 2 + 25.34 × ratio-2.85) showed that the peak value of recovered 13 C in soil of artificial grassland was 9.89% when the ratio value reached 4.77, and the peak value of MRT was 56.46 when the ratio value was 4.68.
Species foliage 13 C. The 13 C values decreased in the foliage of three species with increases in the labeling time ( Fig. 6). For E. nutans and Poa annua, the 13 C values were greater in 5Y than in the other two grasslands at all labeling times except for the first day (P < 0.05) (Fig. 6a,c). The largest slope value of 13 C for E. nutans was observed in the 12Y plot from 3 h to one day after labeling (Fig. 6b), and the smallest slope value was in the 5Y plot (Fig. 6b), with similar results for P. annua (Fig. 6d). The relationship between slope values of 13 C from 3 h to one day for E. nutans and species richnesses of the three artificial grasslands was significant in linear fittings (Fig. 6b). However, the largest slope value of 13 C for forbs was observed in the 5Y plot (Fig. 6f) and the relationship between slope values of 13 C from 3 h to one day and species richnesses of the three artificial grasslands was not significant in linear fitting (Fig. 6f).

Discussion
Different land uses have significant effects on the exchange and sequestration of ecosystem C and the percentages of C allocated to root pools and soil pools differed significantly. We also observed the different C residence time for different age artificial grasslands.
Assimilated 13 C was recovered in shoots (70-85%), roots (13.3-22.9%), and soil (1.7-6.9%) at 3 h after labeling in all land types (Figs 1 and 4). Loss or export of recently fixed C of shoots was 20.4%, 31.3%, and 38.8% in the first 24 h for 2Y, 5Y, and 12Y, respectively, all of which are much lower than the value of 77% reported after two sequential pulse-labelings 39 . The maximum decrease rate in leaf occurred between 0 h and 24 h after labeling (Fig. 1), and the maximum translocation rate to below-ground pools occurred within 24 h after labeling (Fig. 2). This result is similar to those of studies reported earlier 18,40,41 . Loss through respiration, which increased between 1d and 22d in the 2Y and 12Y sites (Fig. 2) but remain unchanged in the 5Y site (Fig. 2), is consistent with the grazed plot studied by Hafner et al. 19 . However, this parameter showed no significant changes for the three plots during the labeling period.
Roots are considered the major C sink within the below-ground pools 40 . The results of the experiment indicated that more of the below-ground carbon was contained in roots than soil pools (Figs 4 and 5). The artificial grasslands in our study contained perennial species such as herbage plants and forbs and had developed a  relatively good rooting system that can be used as carbon storage. More than 70% of the allocated 13 C in soil was in the top 10 cm of soil (Fig. 4c), due to the superficial root biomass allocation in artificial grassland compared with natural alpine meadow.
Significant differences were found in the partitioning of 13 C into below-ground C pools among the 2Y, 5Y, and 12Y grasslands. For 5Y and 12Y, there was a significantly greater percentage of assimilated 13 C in below-ground C pools (12Y, 21.7%; 5Y, 20.9%), than for 2Y (15.6%) (Figs 2 and 5). The partitioning percentage was lower than in previous pulse-labeling studies, 30-50% reported by Kuzyakov et al. 42 , 58.7% reported by Wu et al. 40 , 40% reported by Hafner et al. 19 , 61% by Zou et al. 38 and 22-43% by Zhao et al. 18 . This may be because of the lower species richness (Table 1) and simpler community structures of artificial grasslands than in the aforementioned studies 18,38,40 . The ecological structure of artificial grasslands is not natural, and it is potentially unstable and fragile.
Artificial grasslands of different established ages showed different capacities for recovery of soil 13 C stock. The proportion of carbon remaining in the soil was higher in the 12Y plot (7.65%) than in the 2Y (3.76%) and 5Y (3.12%) plots (Figs 4 and 5); significantly, the values for the 2Y and 5Y plots were similar to the 3% reported in perennial E. nutans grassland studied by Zhao et al. 18 , but all three values were lower than reports from native meadow 18,19,38 . This was not controlled by changes in the proportion of graminate because our result showed that a moderate proportion of graminate would have the highest value of recovered 13 C. The 12Y plot had the highest species richness (15.5 ± 1.1) among the three artificial grasslands, which may explain this condition. According to the complementary niche hypothesis, higher species richness will benefit ecosystem function 43 . One of the reasons might be the high below-ground biomass for the 12Y site (Table 1). Many studies have shown that carbon allocation within the subsurface pool changed in response to species richness 14,44 . Moreover, coexisting species under conditions of high species richness may occupy different niches in the community, which improves the utilization efficiency and the competition for limited resources in soils 14 , so the plant roots of communities with higher species richness may produce more exudates, increase rhizodeposition and decompose faster, enhancing soil organic matter turnover [45][46][47] . In addition, our results revealed that the artificial grassland established for 12 years was progressing slowly toward the succession of natural grassland and had a more stable community structure than grasslands established for 2 or 5 years.
The MRT of ecosystem C in the 12Y grassland was 1.4 times longer than the 2Y. This implies that the photosynthetic C stocked in the 12Y has a lower rate of loss than that do other artificial grasslands. This finding was not caused by the proportion of graminate because the longest MRT was associated with moderate graminate proportions and not low graminate proportions. We found that the C transfer and loss rates in the 12Y grassland with higher species richness were lower (shallow slopes in Fig. 6) in the leaves of the graminate plants E. nutans and P. annua by labeling from 3 h to one day than those in the 2Y and 5Y with lower species richnesses. The lower loss rate of leaf-assimilated C can partially explain the larger ecosystem carbon stocks in the 12Y grassland. This result indicates the effect of species richness on system C stock is derived not only from diversified functions of increased species and from intrinsic changes in the functions of the dominant original species. Therefore, the transfer and loss of assimilated C in the plant aboveground material was also affected by biodiversity, though the majority of existing studies focused on the aboveground litter decomposition 48,49 and the below-ground microbial biomass and activity [50][51][52] .
One of the possible explanations that account for this lower loss rate of leaf-assimilated C is associated with decreased aboveground respiration in the 12Y. Previous studies reported that higher plant diversity mitigated C losses 53,54 , and a recent study demonstrated lower aboveground respiration for the perennial artificial grass (with higher biodiversity) than for the annual artificial grass (with lower biodiversity) on the Qinghai-Tibetan Plateau 18 . Alternatively, in all three artificial grasslands, E. nutans and P. annua were two of the fast-growing plant species, and these grasses were shown to have small root-shoot ratios. The reduced rates of transfer and loss of leaf assimilated C might derive from an adaptive resource allocation response to increased aboveground competition in the 12Y compared with the other artificial grasslands.

Materials and Methods
Site description. The study was conducted at a ranch in Maqin County, of the Golok Tibetan Autonomous Prefecture of Qinghai Province, China (34° 20′ -22′ N, 100° 30′ -29′ E) at an altitude of 4,120 m. The station has a plateau continental climate characterized by strong solar radiation with long, cold winters and short, cool summers. The annual average temperature is − 0.6 °C, ranging from − 10 °C in January to 11.7 °C in July. The mean annual precipitation is 513 mm, occurring mainly in the short summer from May to September. There is no entirely frost-free period. The natural vegetation type of the research area is alpine meadow, most of which has been degraded by human disturbance (overgrazing and fertilizer use) and climate change. The artificial grasslands were revegetated in the most severely degraded alpine meadows (often termed "black beach" or "black-soil land"), with mixed sowing of native perennial grasses (mixed E. nutans. and P. annua with the same proportion) in recent years.
In this study, we selected three 100 × 100 m sites that were artificially planted and restored from "black-soil land" and that have been growing for different lengths of time: two years (2Y), five years (5Y) and twelve years (12Y). All of the sites have been used as winter pasture by local herdsmen. All of the plant species were perennials. The species composition of 2Y was E. nutans, P. annua, Festuca sinensis, and the forbs mainly included Pedicularis kansuensis, Potentilla nivea, Gueldenstaedtia diversifolia, Viola yedoensis, Parnassia trinervis, Leontopodium nanum, and Oxytropis ochrocephala. The species composition of 5Y was E. nutans, P. annua, and the forbs mainly included P. nivea, P. trinervis. The species composition of 12Y included E. nutans, P. annua, F. sinensis, Stipa aliena. The forbs mainly included the following: P. kansuensis, Saussurea superb, P. nivea, G. diversifolia, V. yedoensis, P. trinervis, L. nanum, Gentiana straminea, Lancea tibetica, Ajania tenuifolia, Polygonum sibiricum and O. ochrocephala. The ratios of graminate to forbs were 7.3, 2.2, and 3.3 for the 2Y, 5Y and 12Y plots, respectively. The proportions of the different forbs varied. During the experimental periods, most plant species were at the Scientific RepoRts | 7:40307 | DOI: 10.1038/srep40307 developmental stages of flowering or fruiting. The rainfall was plentiful, and plants in all three types of artificial grasslands grew well without drought stress during our experiment. More details are provided in Table 1.
Pulse labeling. We randomly established three replicate plots (1 × 1 m, 50 cm height) in each artificial grassland. The closest distance between two neighboring plots was approximately 5 m. We installed a wire skeleton in each plot in the late afternoon of the day before isotope labeling and covered it with a transparent polyethylene film just before labeling (film with more than 96% transmittance of photosynthetically active radiation). Each chamber was composed of wire skeleton and polyethylene film. We inserted the chambers into the soil to a depth of approximately 5 cm and packed them with a nylon mesh with pore size of 45 mm. These meshes were extended to 10 cm depth to cut off roots from plants outside of the labeling chambers, while allowing nutrient and water exchange between the soil inside and outside of the chambers. Extra-fine earth was packed firmly around the base of the chamber to reduce gas leakage. The inner surface of the chamber was smeared with anti-fog agent to reduce water vapor condensation during labeling, which can increase light intensity and reduce the 13 CO 2 dissolved into water drops on the chamber's inner surfaces.
Pulse labeling was carried out at 11:00 in the morning on August 1, 2011, which was a clear day. Labeling began immediately after the polyethylene film was closed and tightly sealed. A beaker containing a solution of 40 mL distilled water and 8.0 g Na 2 13 CO 3 (with 13 C to 99 at.%) was placed in the center of the chamber in each plot, the 13 CO 2 was released by careful injection of 28.4 mL of 5 mol/L H 2 SO 4 into the beaker, and the chamber was then immediately sealed. The volume of 13 CO 2 released from the chemical reaction was 1.66 L at least. It would decrease quickly to a low level according to reported photosynthetic rate at both leaf and community levels 55 , so the total volume of released 13 CO 2 did small change in air pressure inside the chambers. The air in each chamber was circulated by an electric fan (12 V, 0.21 A) in the center of each chamber to guarantee a uniform air environment. The polyethylene film of the chambers was beaten gently by hand to remove the condensed water vapor, and the polyethylene film was removed after 3 hours. The flow rate and labeling time were determined according to photosynthetic rates and below-ground productivity as previously measured 18,35,38,40 , to ensure sufficiently higher 13 C abundances in both shoot and root samples, compared with unlabeled control samples, during the whole chase period. The CO 2 concentration in the chambers was not measured during labeling. It can be expected to decrease quickly to a low level according to reported photosynthetic rates at both leaf and community levels 6,55,56 . Sample collection. Shoot samples were collected 3 h and 1, 8 and 22 days after labeling. Above-ground plant parts of all species were harvested and pooled as shoot samples by clipping at the soil surface before the soil was sampled. Foliage samples of different species were collected separately. Only green shoots were used for further 13 C analysis. Soil cores of 5 cm in diameter were taken to 20 cm depth immediately and at 3 h, 1, 8, and 22 days after labeling. Soil and root samples were taken from two layers: 0-10 cm and 10-20 cm. All roots and soil in the cores were carefully extracted and sieved with a 2 mm screen. The soil samples that passed through the sieve were air-dried and stored at 4 °C before being analyzed for total carbon and 13 C. The sampled roots were carefully washed by wet sieving though a 0.5 mm screen to remove attached soil and dark-brown/black debris. The roots were further separated into living and dead components based on their color. The shoot and root samples were oven-dried at 105 °C for 48 h. Only data from living roots are mentioned in this work 40 .

Measurement and calculations.
To remove all carbonates, soil samples were prepared by washing in 0.1 mol/L HCl until no air bubbles appeared. The acid-treated samples were oven-dried at 105 °C for 24 h. They were then ground manually with a pestle and mortar to homogeneously fine powders. Shoot and root samples were also ground using a MM 200 steel ball mill (Retsch GmbH, Haan, Germany). Carbon contents and the 13 C/ 12 C ratio in the samples were measured with a MAT 253 stable isotope ratio mass spectrometer system (Thermo Fisher Scientific Inc., Bremen, Germany).
The natural abundance of 13 C in samples is expressed as 13 C % units relative to Pee Dee Belemnite. To facilitate comparisons with other studies, we also calculated the enrichment values as 13 C (at%) excess, the increase in 13 C atoms due to pulse labeling expressed as the percentage of total carbon atoms in the sample, using the following equations:  13 13 where R sample is the isotope ratio ( 13 C/ 12 C) of the sample, and 0.011237 is the ratio of 13 C/ 12 C in Pee Dee Belemnite. 13 C(at%) represents the percentage of 13 C atoms among the total carbon atoms. To estimate the amount of 13 C incorporated into various plant and soil pools, the following equation was used: where carbon pool size is the carbon content in shoots, roots and soil and was assumed to be constant during the whole chase period. Atmospheric background was corrected using 13 C(at%) excess instead of 13 C as in the above equation.
= × % of C C C 100 (5)   recovered  13  13  t,amount  13 0,amount Here, 13 C 0, represents the weight (mg m −2 ) of 13 C in the pool 3 hours after the labeling. Ecosystem respiration was calculated according to the following equation: where 13 C respiration is the % of recovered 13 C of ecosystem respiration, and 13 Cshoot is the % of recovered 13 C in shoots 13 C belowground is the % of recovered 13 C in the below-ground C pool.
Mean residence time is the average time a C atom remains in a compartment and is defined as the ratio of the holding capacity (pool size) and (net) C flux through the pool 35 . MRT was determined by Eqs 7 and 8 57 : 13 t 13 max kt where m( 13 C) t is the mass (in mg) of C present in the plant-soil compartment at t time, m( 13 C) max is the amount of 13 C at the peak, and t is the time after labeling: Statistical analysis. Normality of aboveground and below-ground plant biomass and SOC stocks was tested using the Kolmogorov-Smirnov test. The significance of differences among the three treatments considering the aboveground and below-ground plant biomass and 13 C stocks was tested by ANOVA, which was calculated separately for each layer; P < 0.05 was considered statistically significant for treatment means. We used nonlinear least squares (function Bnls^) to fit Eq. 5. Statistical analysis was performed using SAS 9.2 for Windows.