Introduction

Due to a combination of human impacts and climatic changes, grasslands have suffered substantial degradation during past decades in many areas of the world, including in arid and semi-arid regions of China1,2,3. To keep track of such changes and predict further degradation, conventional indicators such as plant height, cover, density and biomass and physical and chemical characteristics of soil are used to monitor rangeland degradation4,5. Here, δ13C values are evaluated as a microcosmic index for monitoring processes of grassland degradation.

Grazing disturbance not only reduced the δ13Cleaf value of palatable species such as annuals, perennial grasses and legume (Table S1, Fig. 1a,b), but also converted the photosynthetic pathways of three native species from C3 to C4 (Table S2). The underlying mechanism is most likely that the degraded sites experienced decrease of both leaf area index and the ratio of leaf blades to stem6, leading to a decrease of the net photosynthetic rate3 and an increase of the intercellular carbon dioxide concentration (Ci) (Table S3). As adjacent NG and OG plots in one site were assumed to be equal in ambient carbon dioxide concentration (Ca), observed changes in δ13Cleaf can be described by the follow equation7,

Figure 1
figure 1

The δ13Cleaf values in non-grazed and over-grazed sites in the seven types of grassland.

Images show latitudinal and longitudinal trends in δ13Cleaf variation. Panel (a) represents δ13Cleaf on non-grazed (NG) grasslands, while panel (b) shows δ13Cleaf of the same species on over-grazed grasslands (OG).

Full size image

In addition, the δ13Cleaf values were positive correlation with elevation and latitude, while negative correlation with longitude respectively (Fig. S1(a~c)), which may be due to atmospheric variations in δ13C constituting only a minor portion of the observed differences in plant tissue composition and the elevational differences in carbon isotope composition appearing to reflect real differences in discrimination by plants8. Moreover plants δ13C content at high altitude depends largely on hours with high radiation9,10.

Values of δ13Croot were significantly increased in most degraded grasslands and decreased in others (Fig. S2). In addition, δ13Croot had a similar value to δ13Cleaf in NG sites, but showed an increase and then a decrease along degradation gradient (Fig. S3). The δ13Croot was significantly increased in most grasslands types, this was possibly because increasing root/shoot ratio and biomass allocation to the roots was an important adaptive response of plants to grazing, reflecting that a high proportion of root biomass in total biomass can enhance the capacity of plant to tolerate environmental stresses and buffer external disturbances5,11. Plant transport dynamics, such as temporal changes in C allocation and metabolic processes along the transport pathways in the phloem of stems, determined the coupling of the isotope signals above and below ground9. Since the metabolites were enriched in 13C with reference to photosynthetic products, the large amount of stem reduction and diminution of individual plant size in OG plots shortened the transport and metabolite pathway, leading to the relative 13C depletion of metabolites transported to roots9. Mechanisms for δ13Croot enrichment in lightly grazed plots (LG) and depletion in OG plots may be that root in LG plot (Fig. S3) were still able to buffer animal disturbances (intermediate disturbance hypothesis). However, heavy grazing significantly removed leaf and stem biomass, resulting in a reduction in δ13Croot4. The differences in isotopic signatures between plant organs are consistent with the notion that the 13C enrichment of particular compounds occurs in a basipetal direction12. The δ13Croot values showed different correlation with geographic factors at high and all altitude (Fig. S1(d)~(f,m)~(o)), especially decreased significantly with an increase of elevation and increased with an increase of latitude and longitude at low elevation (Fig. S1(j,k)), which generated a well performed multiple regression model (Fig. 2a,b).

Figure 2
figure 2

The multiple regression of δ13Croot (at low elevation) and δ13Csoil (at all elevations) with latitude and elevation.

Panel (a) shows the regression of δ13Croot (at low elevation <2000m) in NG grasslands (R2 = 0.972 p < 0.0001 n = 45), while panel (b) shows the δ13Croot regression in OG grasslands (R2 = 0.993 p < 0.0001 n = 45). Panel (c) shows the δ13Csoil regression at all elevations (0~5000m) regression in NG grasslands (R2 = 0.810 p < 0.0001 n = 188), while panel (d) shows the δ13Csoil regression in OG grasslands at all elevations (R2 = 0.576 p < 0.0001 n = 188).

Full size image

Data from seven types of grassland show that after grazing, δ13Csoil rose in some and fell in other grassland types (Fig. 3). These contrasting trends in δ13Csoil after grazing may be related to the significant difference in the interaction between non-grazed, over-grazed treatments and region (Fig. 3, p < 0.0001), indicating that different grassland types have different tolerance to the same grazing intensity. The model for factors affecting δ13Csoil was developed as follows (Fig. 3):

Figure 3
figure 3

The δ13Csoil values at three depth intervals in non-grazed and over-grazed sites in seven grassland types.

Panel (a) shows the latitudinal and longitudinal trend in δ13Csoil variation in all layers on non-grazed (NG) grasslands and panel (e) shows the latitudinal and longitudinal trend in δ13Csoil variation in all layers on over-grazed (OG) grasslands. Panels (b–d) show δ13Csoil variation at sampling depths of 0–5 cm, 5–10 cm and 10–15 cm, respectively, on NG grasslands, while panels (f–h) show variation at the same depth in OG sites. The p-value of depth, treatment and region are 0.001, 0.035 and 0.000 respectively. The p-value of interaction between treatment and region is 0.0001. In the model developed, the interaction of treatment, depth and region is taken into consideration. Grazing lead to δ13Csoil increase at middle latitude and longitude sites and decrease at high and low latitudes and longitudes at 0–5 cm and 5–10 cm depth, while the opposite trend occurs at on 10–15 cm depth. Grazing disturbed δ13Csoil of the surface soil more than it disturbed of deeper soil layers.

Full size image

Heavy grazing may disrupt the structure of soil aggregates and surface crust because of livestock trampling, which increases the occurrence of medium and small particle-sized organic matter, induces soil organic matter to decompose more rapidly and leads to the soil being susceptible to water and wind erosion13. We considered that this situation would lead to a better moisture conditions in NG plots compared to OG ones, A much larger amount of soil moisture was evaporated in the grazing plots, probably because of less litter coverage on the soil surface3. It has been proved that better moisture conditions favor more open stomata and the preferential uptake of 12C over 13C14. Based on the G. D. Farquhar theory7, the developing application15 and the relative equal climate conditions at our adjacent NG and OG plots, we reasoned that δ13Cleaf was determined indirectly by the ratio of net photosynthetic rate to stomata1 conductance (A/g). The δ13C was enriched along the plant axis downward, leading to an increase of δ13Csoil9,16. Grazing induced the relative drought stress, favor more close stomata, leading to g decreasing and A/g variation. Consequently, δ13Csoil was relative depleted or enriched at OG plots (Fig. 3). Moreover, drought decreases the contribution of recently assimilated C to soil CO2 efflux and can increase the residence time of recently assimilated C in leaf biomass17. Up to 40% of photosynthates are exudated by roots and are rapidly respired or invested in biomass by rhizosphere microorganisms18. Since root carbon storage might supply autotrophic respiration and enable respiration rates temporarily, the respiration of microbes around the rhizosphere might decline more rapidly after root is removed by grazers, thus decreasing belowground biomass14,19. Values of δ13Csoil in the seven OG plots showed a higher correlation with elevation and latitude (R2 = 0.810, p < 0.0001) than δ13Csoil in the seven NG plots (R2 = 0.576, p < 0.0001). This possibly indicates that δ13Csoil was more related to geographic (Fig. S1(g)~(i),(o)) and even indirect climatic factors20, when grassland is degraded (Fig. 2c,d).

Although the trends in δ13Croot and δ13Csoil were not in the same direction after grazing (Fig. 3 and S2), the difference interval of δ13Croot and δ13Csoil in OG plots Δr-sOG (Δr-sOG = Δroot-soilOG = δ13CrootOG − δ13CsoilOG) was little changed when compared to Δr-sNG (Δr-sNG = Δroot-soil NG = δ13CrootNG − δ13CsoilNG) (Fig. 4). This indicates that despite the complex processes of belowground carbon allocation, the C isotopic signature of soil and roots after grazing (Δr-sOG and Δr-sNG) is a promising approach to partitioning C sources of soil respiration, monitoring belowground biological activity and potentially identifying and quantifying the mechanisms of C stabilization and release9. Previous studies proved that carbon isotope indicators can provide information on diffuse air pollution21, reconstruct the past climate change using the δ13C of buried soils22 and reconstruct palaeoclimatic of lake throughflow using isotope data of plant macroremains and authigenic carbonates23 and reconstruct Phyllocladus using δ13C in a range of New Zealand proxies and macrofossils24 and so on. This study provided another possible idea of degraded grassland reconstructions, isotope data (Fig. 4) can provide complementary information to reconstruct and analyse the environmental perturbations of grazers. The isotope approach more or less contributes to construct more detailed global isotopic gradients in grassland when combined with δ13C, δ18O and δ15N of other species and other field (forest and ocean) worldwide24.

Figure 4
figure 4

The linear regression of the difference interval between δ13Croot and δ13Csoil at over-grazed and non-grazed sites.

Δr-sNG represents the difference interval of δ13Croot and δ13Csoil in non-grazed plots (Where Δr-sNG = Δroot-soil NG = δ13CrootNG − δ13CsoilNG (‰)), Δr-sOG represent the difference of δ13Croot and δ13Csoil in over-grazed plots (Δr-sNG = Δroot-soil NG = δ13CrootNG − δ13CsoilNG (‰)).

Full size image

Overall, the microcosmic index δ13C of leafs and roots allows an integrated understanding of the process of grassland degradation and can be a powerful tool permitting both tracing of C molecules and an integrated view of biological processes in the degradation of ecosystems across space and time.

Methods

Survey of representative grassland regions

The experiment selected seven types of grasslands in the Eurasian grassland zone: plain meadow, meadow grassland, typical grassland, desert grassland, temperate marsh grassland, alpine meadow and mountain desert grassland. The entire zone extends from northeast to southwest, stretching latitudinally over 16° 3′ and longitudinally over 36°11′, with an elevation range of 4200 m and a distance of nearly 3000 km across the zone. These grasslands have been over-grazed for nearly 40 years. More detailed information on the sites is provided in the Supplementary Information.

Experimental design and sampling of the surveyed grassland areas

Different land use types in each site were selected as study plots. These were fenced in the 1970s as non-grazed plots (NG) and severely degraded grassland that had been over-grazed (i.e. OG plots). Sites were sampled in August 2014 by randomly selecting three blocks in each plot (Fig. S4).

Method for δ 13C analysis

The δ13C analysis used a cavity ring down spectrometer (CRDS) (Picarro G2201-I Picarro, Inc. Santa Clara, CA, USA) provided by the Isotope Analysis Laboratory of Inner Mongolia Agriculture University. The leaves intercellular CO2 concentration (Ci) of leaves was measured using an LI-6400 Portable Photosynthetic System (Li-Cor, Lincoln, NE, USA).

Data calculation

Data were analyzed with SPSS version 20.0 (SPSS for Windows, Chicago, IL, USA) and figures were charted with SigmaPlot version 12.5 software. Correlation analysis was performed to investigate relationships between δ13C values and geographical factors. Univariate analysis using a general linear model was performed on δ13Csoil with depth, region and grazing treatments and on δ13Croot with region and grazing treatments. The same procedure was also used on the difference interval of δ13Croot and δ13Csoil. Univariate and multivariate linear regression analysis and nonlinear regression were performed on δ13C values and geographical factors. Means were tested with Tukey’s test when treatment effects were considered statistically significant (P < 0.05).

Additional Information

How to cite this article: Yao, H. et al. Stable carbon isotope as a signal index for monitoring grassland degradation. Sci. Rep. 6, 31399; doi: 10.1038/srep31399 (2016).