Climate and anthropogenic drivers of changes in abundance of C4 annuals and perennials in grasslands on the Mongolian Plateau

Abstract

Background

C4 plants have increased substantially during the past several decades in the grasslands of the Mongolian Plateau due to regional warming. Here, we explore how the patterns of abundances of C4 annuals and C4 perennials change over space and time.

Methods

A total of 280 sites with C4 plants were surveyed in four types of grasslands in 9 years. The relative biomasses of C4 plants (P_C4), C4 annuals (P_A4), and C4 perennials (P_P4) were calculated. Structural equation modeling was used to analyze the drivers of changes in P_A4 and P_P4.

Results

At the regional scale, P_A4 on average was 11% (±19%, SD) and P_P4 was 13% (±19%, SD). Spatially, C4 annuals dominated the C4 communities within an east–west belt region along 44° N and tended to spread toward northern latitudes (about 0.5°) and higher altitudes in the east mountainous areas. The abundance of C4 annuals decreased, while that of C4 perennials increased. The patterns of C4 annuals and C4 perennials were mainly controlled by temperature, growing season precipitation, and dynamics between the two life forms.

Conclusions

C4 annuals exhibited competitive advantages in normal and wet years, while C4 perennials had competitive advantages in dry years. Grazing as a main human disturbance increased C4 annuals, but had no significant effect on C4 perennials.

INTRODUCTION

The rising atmospheric CO₂ concentration and climatic warming have substantially altered the distribution and abundances of C3 and C4 plants over several decades recently in many ecosystems, particularly in grasslands, and this further influences the biogeochemical and functioning properties of these ecosystems (Cerling et al., 1993; Cowles et al., 2016; Sage & Kubien, 2003; Still et al., 2003). The differential responses of C3 and C4 plants to global change drivers are mainly associated with the essential physiological differences between C3 and C4 photosynthesis. The CO₂ concentrating mechanism in C4 photosynthesis enables a much higher CO₂ partial pressure at the site of Rubisco and consequently a very low (close to nil) photorespiration rate (Hatch, 1987), increasing the photosynthetic efficiency of C4 plants compared with C3 plants at high temperatures and/or low intercellular CO₂ (Ehleringer & Björkman, 1977; Gong et al., 2017; Sage & McKown, 2006). Ehleringer (1978) proposes that temperature is the major factor driving C3/C4 distribution since C4 plants have a higher quantum yield (the ratio of assimilated CO₂ to absorbed photosynthetically active radiation) than C3 plants under high temperatures (Ehleringer & Björkman, 1977). This mechanism has been termed the quantum yield hypothesis and has been used for predicting distributions of C3 and C4 species at regional to global scales (Collatz et al., 1998; Edwards et al., 2010; Still et al., 2003). In mixed C3/C4 grasslands of the Mongolian Plateau, growing season temperatures increased rather rapidly by almost 2°C between 1990 and 2000 (NOAA NCDC Climate Data Online, 2015) and stocking rates doubled between the 1980 and 2009 (Wittmer et al., 2010). The relative abundance of C4 plants in this area has increased by 10% following the increase of growing season temperatures in the last decades, based on regional comparison of carbon isotope ratios of recent wool and soil organic matter (Auerswald, Wittmer, Bai, et al., 2012; Wittmer et al., 2010). Wittmer et al. (2010) documented that the improvement in C4 habitat quality led to a spread of C4 plants toward northern latitudes and higher altitudes on the Mongolian Plateau that were previously too cold for C4 plants. However, it remains unclear how patterns of abundances of C4 annuals and C4 perennials change over space and time, which is critical for projecting responses of C3/C4 composition and functioning and services of mixed C3/C4 grasslands to climate change.

The quantum yield hypothesis explains well the observed global distributions of C4 vegetation (Collatz et al., 1998; Edwards et al., 2010; Still et al., 2003). However, it provides little insight into the successional process of C4 vegetation, although the relative abundance of C3 and C4 species is the competitive outcome of different plant functional types. For instance, C4 annuals and C4 perennials may respond differently to the improvement of habitat quality due to their different velocities of occupying new habitats according to Grime's C–S–R plant strategy concept (Craine, 2005; Grime, 1977; Hodgson et al., 1999). The C, S, and R indicate competitive, stress-tolerant, and ruderal strategies, respectively (Grime, 1977). In the cool-temperate grassland of the Mongolian Plateau, annuals have been adapted to a ruderal strategy, opportunistically and rapidly using the improved conditions in temperature and colonizing new open habitats. Dominant perennials, in contrast, are long-lived and allocate only a small proportion of annual production to seeds, and thereby use a stress-tolerant strategy. Some perennials in undisturbed habitats, which use a competitor strategy, utilize resources better than others (Grime, 1988). Grime's C–S–R plant strategy concept also suggests that ruderal plants dominate the initial phase, while stress-tolerant species dominate the terminal phase of vegetation succession (Caccianiga et al., 2006; Grime, 1977). Namely, annuals that can produce a large quantity of seeds to ensure rapid rehabilitation after disturbance should be the first to take advantage of the improved habitat quality of increasing growth temperature and open space. In contrast, perennials that conservatively use resources should increase their abundance in the long term, and the advantage of the conservative strategy is especially prominent when resources are limited. Hence, C4 annuals and C4 perennials may not have equally contributed to the substantial increase in C4 plants in the grassland of Mongolian Plateau. It may be expected that C4 annuals profited first from the change in environmental conditions favoring C4 plants, but would have been gradually replaced by C4 perennials. However, this prediction has not been tested yet.

In semiarid grasslands on the Mongolian Plateau, water is a key limiting resource to plant growth (Bai et al., 2004, 2008). The C4 plants on the Mongolian Plateau in general develop relatively shallow root systems (S. H. Chen et al., 2001; Jigjidsuren & Johnson, 2003; Yang, Auerswald, Bai, & Han, 2011) because their growth is restricted to summer months when the temperature peaks synchronically with rainfall and the subsoil moisture from winter has already been exploited by C3 plants (Nippert & Knapp, 2007; Yang, Auerswald, Bai, & Han, 2011). C4 annuals depend even more on summer rainfall because their germination is triggered by a critical minimal amount of rainfall in warm months (Ehleringer, 1985). The proportion of C4 annuals and C4 perennials may partly depend on the amount and distribution of rainfall within the growing season (May–August), which are also highly variable over years (Bai et al., 2004; Wittmer et al., 2008). Therefore, the potential effect of growing season precipitation on C4 annuals and C4 perennials needs to be assessed when exploring the dynamic changes in both functional types after the improvement of C4 habitat quality.

Grazing is a major driver for changes in species composition, plant community structure, and ecosystem functioning in grasslands on the Mongolian Plateau (Bai et al., 2007, 2012; W. H. Li et al., 2017). This is particularly true in the Inner Mongolian grassland, where the numbers of livestock have increased by 11-fold during the past 60 years, and as a result, 55%–80% of the total grassland area in the Inner Mongolia Autonomous Region has been degraded (Briske et al., 2015). Long-term grazing leads to decreases in species richness and aboveground biomass in this region (Bai et al., 2007). A recent study reported that cattle grazing for seven decades shifted plant species composition from codominance by a C3 midgrass (Pascopyrum smithii) and a C4 shortgrass (Bouteloua gracilis) to dominance by B. gracilis in the western North American Great Plains (Derner et al., 2019). Annuals may profit and colonize the new bare ground since they have large soil seed banks (Coffin & Lauenroth, 1989) and show opportunistic growth, especially after grazing (Kinucan, 1992). Hence, the proportion of C4 annuals and C4 perennials in a gradually warming climate may be modified additionally by grazing.

In this study, we examine the spatiotemporal patterns and drivers of changes in the relative abundance of C4 annuals and C4 perennials in mixed C3/C4 grasslands on the Mongolian Plateau based on data collected systematically from 280 field sites with C4 plants between 2001 and 2009. Specifically, we address three questions: First, how do the patterns of abundances of C4 annuals and C4 perennials change over space and time in grasslands on the Mongolian Plateau? Second, what are the relative roles of temperature, precipitation, and dynamics between C4 annuals and C4 perennials in determining the spread pathways of the two life forms of C4 plants? Third, how do variations in growing season precipitation (i.e., wet, normal, and dry years) and grazing affect the relative abundance of C4 annuals and C4 perennials? To address these questions, we hypothesize that (1) temperature and growing season precipitation are key climatic drivers controlling the patterns of C4 plants at a regional scale; (2) C4 annuals will be replaced by C4 perennials in much drier habitats (Supporting Information: Figure S1); and (3) grazing favors C4 annuals over C4 perennials.

MATERIALS AND METHODS

Study area

This study was conducted in grasslands on the Mongolian Plateau located in the Autonomous Region of Inner Mongolia of China and the Republic of Mongolia (Supporting Information: Figure S2). The study area is located at 40°–48° N in latitude (approximately 800 km) and 106°–119° E in longitude (approximately 1200 km), with elevation ranging from 800 to 1700 m above sea level (a.s.l.). The mean annual precipitation (MAP) increases from 100 mm year⁻¹ in the west part to 400 mm year⁻¹ in the east and north parts of the study area. Most of the precipitation (approximately 75%) falls during the growing season (May–August). The mean annual temperature (MAT) and the mean temperature of July (MJulT) range from −2 to 8°C and from 16 to 24°C, respectively. The spatial variation of temperature and precipitation follows altitudinal and latitudinal trends. Almost the entire study area is grazed by sheep and goats with little cropland. C4 perennials usually start to grow earlier than C4 annuals. For instance, Cleistogenes squarrosa, a dominant perennial, starts to grow at the beginning of June. Salsola collina, a widely distributed annual, starts to grow in July or August depending on precipitation (S. H. Chen et al., 2001; Jigjidsuren & Johnson, 2003).

Field survey

The field surveys were conducted in late July or in August of 2001–2009. The entire study area could not be surveyed every year because of its large spatial extent and the low road density and quality. Among these sites, only 280 sites had C4 plants, which were used in the following analysis. Four community types, including typical steppe, desert steppe, sand dune, and mountain meadow, were investigated. All C4 species—and for comparison also all C3 species—were recorded on an approximately 100 m × 100 m area at each site (Table 1). Positions and altitudes of sampling sites were recorded with a mobile GPS. For all individual C4/C3 species and the whole C4/C3 community in total, the relative contribution to above-ground biomass was measured by clipping all plants with three to five 1 m × 1 m quadrats, which were randomly selected within a 100 m × 100 m area. C4 species were assigned to annuals and perennials according to the literature (Jigjidsuren & Johnson, 2003; Wang, 2003). Land use was classified as recently grazed (248 sites) or ungrazed (32 sites) based on the presence of fresh feces or sheep and goats.

Table 1 Frequency of occurrence and life forms of C4 species across the 280 sampling sites

Species	Frequency	Life form
Cleistogenes squarrosa	0.6	Perennial	Grass
Salsola collina	0.5	Annual	Forb
Kochia prostrata	0.1	Perennial	Semi-shrub
Setaria viridis	0.1	Annual	Grass
Amaranthus retroflexus	<0.1	Annual	Forb
Aristida heymannii	<0.1	Annual	Grass
Artemisia sieversiana	<0.1	Annual	Forb
Bassia dasyphylla	<0.1	Annual	Forb
Chloris virgata	<0.1	Annual	Grass
Digitaria ischaemum	<0.1	Annual	Grass
Eragrostis minor	<0.1	Annual	Grass
Euphorbia humifusa	<0.1	Annual	Forb
Micropeplis arachnoidea	<0.1	Annual	Forb
Tribulus terrestris	<0.1	Annual	Forb
Anabasis brevifolia	<0.1	Perennial	Semi-shrub
Artemisia adamsii	<0.1	Perennial	Forb
Cleistogenes songorica	<0.1	Perennial	Grass
Kochia krylovii	<0.1	Annual	Forb
Atriplex sibirica	<0.1	Annual	Forb
Salsola passerina	<0.1	Perennial	Semi-shrub

Meteorological data

The precipitation and temperature within the different years deviated considerably from the MAP and the MJulT, and this deviation differed among sites. Hence, growing season (May–August) precipitation (gP) and temperature were determined for each sampling site based on the year and time of sampling. This was done following the approach of Wittmer et al. (2008), which interpolates between meteorological stations based on 1.5 km × 1.5 km resolution maps of MAP, mean monthly precipitation, MJulT, and MAT (Climate Source Inc.). In brief, the difference between each climatic variable under consideration and its long-term average of the last normal period (1961–1990) was calculated for each meteorological station (N = 46). These differences were then geostatistically interpolated and added to the high-resolution maps of the means. This procedure incorporates the temporal details of the individual sampling years while retaining the spatial details of the long-term maps, which account for factors such as topography, rain shadows, lake effects, and temperature inversions.

To explore how variations in growing season precipitation (i.e., wet, normal, and dry years) affect the relative abundance of C4 annuals and C4 perennials, we used the standardized precipitation-evapotranspiration index (SPEI) to consistently identify and quantify wet and dry years (Isbell et al., 2015). SPEI, a standard normal variable for water balances aggregated over a given number of months at a particular location, allows comparison of drought severity through time and space (Begueria et al., 2010). For example, for a particular location, SPEI-4 = 0.67 for August in a sampling year corresponds to a level of seasonal wetness that had historically occurred once every 4 years during the months of May–August. We extracted SPEI values at 280 sampling sites from a global SPEI database provided on 0.5° × 0.5° grids (). The SPEI-4 for August in the sampling year were used to classify each experiment year as dry, normal, and wet years. Over the period 2001–2009, dry/wet years were defined as those that historically occurred once in 4 years, that is, SPEI-4 = 0.67, with SPEI-4 > 0.67 indicating wet years and SPEI-4 < −0.67 representing dry years. The other years were assigned to normal years.

Statistical analysis

The relative biomass of C4 plants (P_C4), C4 annuals (P_A4), and C4 perennials (P_P4) in the total plant aboveground biomass and the aboveground biomass of C4 annuals relative to the C4 community were calculated. To identify the relative role of succession in determining the spread pathways of C4 annuals and perennials, we used number of years since 2000 (an extremely dry year for the whole region since the 1960s) as a proxy for successional time. Geostatistical analysis was carried out using the GNU R 2.7.2 (R Development Core Team, 2008) with the auxiliary packages geoR (Ribeiro & Diggle, 2001) and gstat (Pebesma, 2004). This was done to estimate meteorological data at the sampling sites from the measured data at meteorological stations and to obtain a map of P_A4. Probability distributions of P_A4 and P_P4 were obtained from the measured data using the kernel density method (Silverman, 1986) with Gaussian kernels as implemented in GNU R 2.7.2. The relative biomass of C4 annuals in the C4 plant domain is bounded between 0 and 1. These boundaries were maintained during kernel density estimation using the reflection method according to Silverman (1986) at the boundaries. The bandwidth of density estimation was defined via Silverman's “rule of thumb.” Structural equation modeling (SEM) was performed to analyze different hypothetical pathways that may explain temperature and precipitation linkages to P_A4 and the effect of P_P4 on P_A4 (Figure S1). Fast growth of plants commonly occurs in July when the temperature peaks synchronically with rainfall in the Mongolian grassland (Bai et al., 2004). Thus, the mean July temperature (MJulT), current year July temperature (JulT), mean July precipitation (MJulP), and current year July precipitation (JulP) were used as temperature and precipitation variables in SEM. The effect of grazing was not included in SEM because quantification of grazing intensity at site level was unavailable. Thirteen sites from sand dunes and mountainous meadows were excluded in the SEM analysis due to their low sampling frequency. The data for P_A4 and P_P4 were log-transformed before SEM analysis to improve the normalization. Because zero cannot be log-transformed, the data were slightly adjusted as: $\mathit{Adjusted}{P}_4=({\mathit{P}}_4+0.001)\times 0.998,$where P₄ denotes either P_C4, P_A4, or P_P4. Statistical analyses on data with this adjustment did not differ from those on other data compressions that bring the extreme values closer to zero (Clary, 2008). Analysis of variance (ANOVA), followed by least significant difference (LSD) multiple range tests, was used to evaluate the effects of the annual wetness on P_C4, P_A4, and P_P4. Thirty-two pairs of the grazed and ungrazed (reference) sites were selected to examine the effect of grazing on the relative biomass of P_C4, P_A4, and P_P4. ANOVA and LSD multiple range tests were performed to assess the effect of grazing on P_C4, P_A4, and P_P4. All regression procedures were carried out using SPSS Version 16.0 (SPSS Inc.).

RESULTS

Species composition and dominance of mixed C3/C4 communities

Altogether, six perennial C4 species and 14 annual C4 species were observed across 280 sampling sites. C. squarrosa and S. collina were dominant species in each group. All other C4 species were rare both across all sites and within a given site (Table 1). These species were mainly grasses and forbs in about equal proportions, while semi-shrubs were rare. In contrast, there were much more C3 species (in total 136 species) across the sampling sites (data not shown). Given a large number of C3 species, there were no species dominating in general, although the genera of Stipa, Artemisia, Leymus, Agropyron, and Caragana were frequently found among the perennials, while Chenopodium album was the most frequent annual C3 species (Table S1). The annuals (in total 15 species) contributed only 11% of all C3 species, while annuals were much more important among the C4 species and contributed 70% of these species. The C3 annuals were typically found in small patches of high disturbance like the outcasts of marmot burrows, while the C4 annuals were also found in other places. Also, in terms of contributions to biomass, the annuals were unimportant among the C3 species. Only in three cases did annuals contribute more than 10% of the total C3 biomass. Hence, there was no correlation between both photosynthetic groups that could indicate a common reason influencing the ratio of annuals and perennials.

Spatial patterns of abundances of C4 annuals against C4 plants

Spatially, P_A4 was above average in areas where C4 abundance, in general, was high. This was within an east–west belt region along 44° N (Figure 1). However, as indicated by the isolines of P_A4 across the study area, higher P_A4 occurred slightly toward northern latitudes (about 0.5° or 50 km). This was especially pronounced at the eastern mountainous part of the study area. Further, P_A4 was above the average C4 abundance at the southern edge of the study area, where the highest C4 abundance was found (between the towns of Sonid Youqi, Bayan Sum, and Xianghuang Qi). Hence, the annuals dominated within a belt around the centers of C4 abundance (Figure 1).

[IMAGE OMITTED. SEE PDF]

Temporal changes in C4 annuals and C4 perennials

On average, P_C4 was 24.0%, with a SD of 23.5%, and it did not show any temporal trend over the study period (Figure 2a). However, when C4 species were classified into C4 annuals and perennials, P_A4 significantly decreased (N = 8, R² = 0.92, p < 0.001) and P_P4 increased over time (Figure 2b,c). The increase in P_P4 was not significant when all years were included (N = 8, R² = 0.12, p = 0.392), but the trend of P_P4 became significant when 2002 was excluded (N = 7, R² = 0.91, p = 0.001). The outlier 2002 was mainly caused by two reasons: First, 2002 was a dry year, which favored C4 perennials (see below). Second, the large study area made it impossible to survey all the sampling sites each year, leaving the possibility that the outlier 2002 may have resulted in part from spatial variation-induced bias. Indeed, when the data were restricted to a small area (115°00′–118°00′ E, about 300 km; 43°30′–45°30′ N, about 200 km) that was sampled in most years, the same behavior could be observed as for the whole area (open circles in Figure 2), while the outlier almost disappeared (P_A4: R² = 0.84, p = 0.010; P_P4: R² = 0.73, p = 0.032).

[IMAGE OMITTED. SEE PDF]

In contrast, spatial and temporal patterns of all C3 plants, P_A3, and P_P3 were rather consistent in the whole study region and in the small region with high-frequently sampling (Figure 2c–e). P_P4 was the highest in 2002, which was warmer and drier than in the other years. The growing season precipitation (gP) in 2002 was on average 157 mm across all sampling sites, while it was on average 201 mm in the other years. The relative biomass of C4 annuals/perennials within C4 communities showed an almost binomial behavior. The C4 community was either dominated by C4 perennials or by C4 annuals, while locations, where both life forms coexist, were rather scarce (Figure 3). Averaged over sites and time, both life forms had similar shares in the community: P_A4 was 11%, with an SD of 19%, and P_P4 was 13%, with an SD of 19%.

[IMAGE OMITTED. SEE PDF]

Abiotic and biotic factors affecting the abundance of C4 annuals and perennials

Results from regression analysis showed that P_A4 was significantly positively correlated with MJulT and JulT, but negatively correlated with MAP, gP, and year (a proxy for successional time, denoted as the number of years since 2000) (p < 0.001), while P_P4 was only positively correlated with successional time (p < 0.05; Table S2 in the Supporting Information). For both P_A4 and P_P4, the effect of Year × MAP was also highly significant (Table S2). Among these factors, the effects of successional time (i.e., number of years since 2000) and MJulT explained 22% and 13% of the total variance in P_A4, respectively (Table S2). Other variables explained less than 10% of the total variance in P_A4 and P_P4. The low explanatory power was at least partly due to the zero values of P_A4 and P_P4, which resulted from the quasi-binominal behavior of both life forms that cannot be described well by continuous variables, such as environmental properties.

Among the 280 sites in our study, 63 sites were sampled in dry years, 183 sites were sampled in normal years, and 34 sites were sampled in wet years. The results showed that, across all sites examined, P_P4 was significantly higher in dry and normal years than in wet years (Figure 4). P_A4, in contrast, was significantly greater in wet years than in dry years (Figure 4). When all C4 plants were pooled together, P_C4 did not differ between dry, normal, and wet years (Figure 4).

[IMAGE OMITTED. SEE PDF]

SEM further showed that temperature and P_P4 had predominant influences on P_A4 based on the significant standardized path coefficients (Figure 5). The effect of P_P4 was even stronger than that of temperature. Together, temperature and P_P4 explained 41% of the total variance in P_A4 (Figure 5). The effect of temperature and precipitation on P_P4 was not significant. Overall, SEM revealed that the increase of temperature (i.e., MJulT and JulT) had a significantly positive effect on P_A4, while the successional stage (i.e., increase of P_P4) had a significantly negative effect on P_A4 (Figure 5).

[IMAGE OMITTED. SEE PDF]

Effect of grazing on abundance of C4 plants

Given the dominant land use of grazing in the study area, we only surveyed 32 ungrazed sites, but 248 grazed sites in this study. To improve our analysis and data interpretation, we only selected 32 pairs of the ungrazed (reference) and corresponding grazed sites to examine the effect of grazing on the relative biomass of P_C4, P_A4, and P_P4. Our results showed that the relative biomass of C4 annuals was significantly higher in grazed sites than in recently ungrazed sites (Figure 6). For the relative biomass of all C4 plants or that of C4 perennials, no significant differences between the grazed and ungrazed sites were observed at the regional scale (Figure 6).

[IMAGE OMITTED. SEE PDF]

DISCUSSION

Spatiotemporal patterns of C4 annuals and C4 perennials

Two recent studies demonstrated that the abundance of C4 plants has increased substantially in the past decades in grasslands on the Mongolian Plateau (Auerswald et al., 2009; Wittmer et al., 2010), and this phenomenon has been attributed to the rapid increase of temperature. Indeed, the mean growing temperature in July (most of the annual precipitation falls) was above 22°C (the crossover temperature for the photosynthesis of C4 plants) around 1995, and warming even increased after 1995 on the Mongolian Plateau. In the context of improved C4 habitat quality in this region, our study found that the patterns of relative abundance differed between C4 annuals and C4 perennials, based on data collected systematically from 2001 to 2009.

C4 annuals and C4 perennials had clear spatial niche separation as indicated by the spatial patterns revealed in this study: (1) C4 annuals dominated the C4 community within an east–west belt region along 44° N across the study area; (2) the C4 communities were either dominated by C4 perennials or by C4 annuals, while sites, where both life forms coexisted equally, were scarce; and (3) C4 annuals dominated around the centers of gravity of C4 plants. Moreover, C4 annuals tend to spread toward northern latitudes (about 0.5°) and higher altitudes in the east mountainous area (Figure 1), compared with the map of recent C4 abundance taken from Auerswald et al. (2009). Our results support the findings from stable carbon isotope studies that C4 plants have recently expanded their ranges into northern and high-elevation areas on the Mongolian Plateau (Auerswald et al., 2009; Wittmer et al., 2010). It should be noted that, in our study, the field surveys were conducted in late July or in August each year, with no year-round samplings scheme conducted during 2001–2009. Therefore, for C4 species with different growth phenologies, their relative abundance may have been underestimated or overestimated to some degree. However, this problem does not have a major influence on our analysis for patterns of abundances of C4 annuals and C4 perennials, which does not rely on total annual growth. Moreover, our measurements used a similar point in time and we only analyzed whether the relative abundance of a certain species changed at this point in time.

Drivers and mechanisms controlling shifts in the dominance of C4 annuals and C4 perennials

The following reasons are likely responsible for the observed patterns of C4 annuals and C4 perennials. First, climate warming, which caused nearly 2°C increase in growing season temperature in Mongolian Plateau during the study period (NOAA NCDC Climate Data Online, 2015), favors C4 plants relative to C3 plants (Cowles et al., 2016; Z. Y. Li et al., 2015; Wittmer et al., 2010) because net photosynthesis of C4 plants increases more with increasing temperature than that of C3 plants (Ehleringer et al., 1997; Sage & Kubien, 2003). Second, the canopies of the Mongolian grassland typically leave 30%–70% bare soil depending on the climate and grazing intensity (Han et al., 2008; Schönbach et al., 2011; Steffens et al., 2009). The increased temperature has caused a warmer and drier habitat on the plateau in recent decades as compared with a 30-year (1961–1990) baseline (Hu et al., 2015; Wittmer et al., 2010). The open space and the dry and warm habitats allow the fast reaction of C4 annuals to colonize the available space, taking advantage of their rapid germination after effective precipitation events, high growth rates, short life spans, and large seed production (Coffin & Lauenroth, 1989; Ehleringer, 1985; Zhong et al., 2012). Third, C4 perennials that conservatively use resources increase their abundance in the long term, especially in years when resources are more limited. These inferences were supported by our SEM modeling and regression analysis, which showed that the growing season temperature had a positive effect on C4 annuals, while successional years had a positive effect on C4 perennials, but a negative effect on C4 annuals.

In spite of no spatial and temporal patterns reported for C4 life forms before our study, the above mechanism seems to be reasonable because it explains well some observed phenomena related to the colonization and succession of C4 after disturbances. For example, a 30-year mowing experiment showed that C4 annuals increased in the first 10 years after mowing (around 30% in the ninth year) and were then replaced by C4 perennials (Baoyin et al., 2015). After stopping vehicle travel, C4 annuals (mostly S. collina) rapidly recovered in the first several years and were then gradually replaced by C4 perennials (C. squarrosa) (S. G. Li et al., 2006). The mechanisms underpinning the burst and retreat of C4 annuals warrants more detailed studies.

Influences of precipitation and grazing on abundances of C4 annuals and C4 perennials

Water availability, which is controlled by temperature and precipitation, has been suggested as a secondary factor influencing C4 success in warm climate zones (Sage & Kubien, 2003). A recent study, however, demonstrates that the historical expansion of C4 plants in broad areas of Central Asia is driven by the increased precipitation seasonality and the enhanced long-term aridity (Shen et al., 2018). In our study, C4 species had higher relative abundance in dry and normal years than in wet years, which indicates that the competitive advantage of C4 photosynthesis is less apparent in wet years. In addition, high moisture availability during the early growing season favors C3 plants, while C4 plants benefit most from high temperature synchronically with high rainfall during the mid and late growing seasons (Auerswald, Wittmer, Bai, et al., 2012). A previous study also demonstrates that cold season precipitation has a positive effect on the abundance of dominant C3 grasses, but a negative effect on the abundance of C4 plants (Wan et al., 2015).

At a regional scale, the relative biomass of C4 annuals was greater than that of C4 perennials. These results clearly indicate that growing season precipitation is also a critical factor controlling the competitive advantage of C4 life forms. Variation in precipitation affects the ability of C4 annuals to quickly use favorable conditions in the initial stage of succession and to compete with C4 perennials in the later stage of succession (Knapp & Medina, 1999). For example, severe droughts lead to more empty space in the vegetation, which may benefit C4 annals in the following years (Knapp & Medina, 1999). Furthermore, C4 perennials seem to have a greater competitive advantage than C4 annuals in dry years due to their conservative use of resources, including high water and nitrogen use efficiency (Long, 1999; Yang, Auerswald, Bai, Wittmer, et al., 2011). Our findings are in agreement with several field studies that showed the burst of annuals in moderate-rainfall years following drought in grassland ecosystems (Q. Chen et al., 2011; Kinugasa et al., 2012; Loeser et al., 2007). Interestingly, Q. Chen et al. (2011) found that annual species, which took advantage of a drought event in 2005, became abundant in 2007 and 2008, and were replaced by perennials in the following dry year of 2009. Although these studies (Bao et al., 2004; Q. Chen et al., 2011; Kinugasa et al., 2012) have not separately considered C3 and C4 species in their functional classifications, the dynamic of community composition in their studies seems to be driven by the successive replacement between C4 annuals and C4 perennials.

Based on the analysis of 32 pairs of the grazed and ungrazed (reference) sites, our study showed that grazing had no significant effect on the relative abundance of C4 species (C4 annuals and C4 perennials). This is in agreement with the results from observations on sites with contrasting long-term grazing intensities (Auerswald, Wittmer, Tungalag, et al., 2012) and the results from a controlled grazing experiment in a typical steppe (Ren et al., 2016). The insignificant grazing effect on C4 abundance might be explained by the buffering effect afforded by the differential responses of annuals and perennials to grazing. Indeed, the relative biomass of C4 annuals was significantly greater at grazed sites than at ungrazed sites, while the C4 perennials were more abundant at ungrazed sites than at grazed sites (although this was statistically insignificant). Moreover, grazing can have both positive and negative effects on C4 abundance. For instance, moderate grazing can stimulate photosynthesis of C4 annuals because of more open space, less light and resource limitations, and high surface radiation and temperature (Knapp & Medina, 1999). On the other hand, heavy grazing can prevent C4 annuals from completing their life circle, and thus decrease their reproductive output (e.g., seed production). Therefore, the effects of grazing intensity and grazing history on C4 annuals and C4 perennials need to be further examined in future studies.

Implications

It has been projected that climate change will further increase temperature and alter precipitation regimes in the coming years (Intergovernmental Panel on Climate Change, 2001). Our findings have important implications for predicting responses of C3–C4 composition and functioning and services of C3–C4 coexisting grasslands to climate change. Current models for identifying the C4 climate zone are mainly based on the quantum yield hypothesis (Collatz et al., 1998; Still et al., 2003), which assumes that a photosynthetic property is an overriding determinant for C3 and C4 distributions. Our study, however, provides new insight into the responses of C4 abundance in the global warming scenario. Annuals have the lowest relative biomass and the highest variability when compared to other plant functional types in grasslands on the Mongolian Plateau (Bai et al., 2004). Thus, the increase in C4 annuals contributes to the high variability and low stability of steppe ecosystems. The annuals allow using new (resource) opportunities quickly and thus contributing to the stabilization of the system under changing conditions, for example, recovery of primary production after drought. However, the high abundance of annuals will contribute to the uncertainty of annual primary production due to their high dependence on climatic conditions and resource availability during the growing season (Q. Chen et al., 2011), and thus affect livestock production. Our study shows that this uncertainty appears over thousands of square kilometers.

CONCLUSIONS

Our findings demonstrate that climate warming has modified the habitat conditions across vast areas of the Mongolian grassland. The spatiotemporal patterns of C4 annuals and C4 perennials were mainly controlled by increasing temperatures, interannual variations in precipitation, and dynamics of the two life forms. The behaviors of C4 annuals and C4 perennials followed the predictions for ruderal versus stress-tolerant groups in Grime's plant C–S–R strategy. Our findings highlight the necessity of considering the competition between C4 annuals and C4 perennials in studying the structure and functioning of C3/C4-dominated grasslands in the context of climate change.

AUTHOR CONTRIBUTIONS

Hao Yang: Data curation; formal analysis; investigation; writing – original draft. Karl Auerswald: Conceptualization; data curation; formal analysis; funding acquisition; investigation; resources; writing – original draft. Xiaoying Gong: Writing – original draft. Hans Schnyder: Writing – original draft. Yongfei Bai: Conceptualization; data curation; formal analysis; funding acquisition; investigation; writing – original draft; writing – review and editing.

ACKNOWLEDGMENTS

We thank Guo Jin (Grassland Research Station of Dongwu Qi, Inner Mongolia) and Tungalag R. (National University Mongolia, Republic of Mongolia) for their assistance with field surveys and Max Wittmer for statistical help. This study was supported by the DFG within the DFG research group 536 (MAGIM) and the National Natural Science Foundation of China (31630010 and 31320103916).

CONFLICT OF INTEREST

The authors declare no conflict of interest.

DATA AVAILABILITY STATEMENT

The original contributions presented in the study are included in the article/Supplementary Material; further inquiries can be directed to the corresponding author.

References

Auerswald, K., Wittmer, M. H. O. M., Bai, Y. F., Yang, H., Taube, F., Susenbeth, A., & Schnyder, H. (2012). C4 abundance in an Inner Mongolia grassland system is driven by temperature–moisture interaction, not grazing pressure. Basic and Applied Ecology, 13(1), 67–75. [DOI: https://dx.doi.org/10.1016/j.baae.2011.11.004]

Zhong, Y. K., Bao, Q. H., & Sun, W. (2012). Experimental research on mowing grassland dynamics in Inner Mongolia. Inner Mongolia University Press.