Community assembly and functional leaf traits mediate precipitation use efficiency of alpine grasslands along environmental gradients on the Tibetan Plateau

Introduction

Precipitation determines spatial and temporal variability in productivity of diverse temperate and alpine grasslands worldwide (Bai et al., 2008; Knapp & Smith, 2001; Sala et al., 2012; Sala et al., 1988; Shi et al., 2014; Yang et al., 2008). It is important to elucidate the mechanisms underlying the relationships between productivity, plant diversity, and precipitation with ongoing global change (Fay et al., 2011; Huxman et al., 2004; Knapp et al., 2008; Knapp et al., 2002; Shen et al., 2015; Weltzin et al., 2003). That is because changing precipitation regimes and other anthropogenic disturbances likely alter community composition and ecosystem functionality, and consequently affect ecosystem services and human welfare (Carmona et al., 2012; Christensen et al., 2004; Fay et al., 2011; Fay et al., 2008; Fernandez-Going, Anacker & Harrison, 2012; Osem, Perevolotsky & Kigel, 2002; Varnamkhasti et al., 1995).

Indeed, aboveground net primary productivity (ANPP), a key integrative ecosystem function, has been well documented to increase across vegetation types with increasing annual or seasonal precipitation gradient (Hu et al., 2007; Huxman et al., 2004; Jiang et al., 2015; Knapp & Smith, 2001; Sala et al., 1988; Yang et al., 2008; Yang et al., 2009). A large number of studies have accepted the term of precipitation use efficiency (PUE, the ratio of ANPP to precipitation) as a proxy of the sensitivity of productivity response to precipitation, because it has normalized the effect of water availability by placing productivity on a per-unit-of-precipitation basis (Bai et al., 2008; Gao et al., 2011; Hu et al., 2010; Huxman et al., 2004; Le Houerou, 1984; Paruelo et al., 1999; Varnamkhasti et al., 1995; Vermeire, Heitschmidt & Rinella, 2009; Yang et al., 2010). Most previous studies focused to examine the manner of productivity in response to changes in precipitation regimes (Bai et al., 2008; Gao et al., 2011; Hu et al., 2007; Paruelo et al., 1999; Sala et al., 1988; Yang et al., 2010); however, rare research has examined the potential and inherent regulating mechanisms of community assembly on ecosystem productivity.

As reported by Paruelo et al. (1999), differences in PUE among biomes can be attributed primarily to vegetation constraints under the given biogeochemical background. For example, O’Connor, Haines & Snyman (2001) and Knapp et al. (2002) found that PUE was influenced mostly by community composition and that species complementarity ensured greater and more stable productivity in species rich grasslands. In the American mixed-grass prairies, Vermeire, Heitschmidt & Rinella (2009) also found that PUE was likely more responsive to community composition of plant functional groups than to amount and seasonal distribution of precipitation. In the temperate sandy grasslands of Inner Mongolia, China, Zuo et al. (2012) even confirmed that productivity was driven indirectly by community composition, which was determined by or co-evolved with abiotic environmental variables. However, the literature on PUE across different alpine grasslands on the Tibetan Plateau, especially focusing on the mechanistic understanding of the effects of species assembly and/or functional traits, is relatively limited.

Relative growth rates and water use strategies that generally differ among species are likely regulated by both plant morphological and physiological traits. Foliar functional traits, such as specific leaf area (SLA, leaf area per unit dry mass), stable carbon isotopic composition (δ¹³C, ¹³C/¹²C isotope discrimination) and leaf area index (LAI, leaf area per unit land area), are often used to examine the response of productivity to precipitation in grassland studies (Fernandez-Going, Anacker & Harrison, 2012; Hu et al., 2008; Luo et al., 2004; Orwin et al., 2010; Violle et al., 2007; Wei et al., 2011; Wellstein et al., 2011). For example, Hu et al. (2008) reported that shifts of community composition induced by climate change, and consequent changes in community-level LAI, can significantly affect water and carbon cycles in temperate grasslands. A lot of studies found that SLA can serve as a good predictor of plant strategies on resource capture, usage and availability (Poorter & Evans, 1998; Wilson, Thompson & Hodgson, 1999). It is a prevailing view that δ¹³C can be used as an indicator of plant intrinsic water use efficiency, because it relates to leaf stomatal behavior in photosynthesis (Caldeira et al., 2001; Flanagan & Farquhar, 2014; Seibt et al., 2008; Song et al., 2008). Therefore, any change in community species assembly or plant functional traits can consequently mediate the response of grassland productivity to climate change at a given temporal and spatial scale (Reich et al., 2003; Wang et al., 2012).

Alpine grasslands on the Tibetan Plateau have increasingly gained attention for their sensitivity and vulnerability to climate change and overgrazing because severe habitat conditions constrain species to recruit from small local species pool there. Considerable studies have examined impacts of climate change and grazing disturbance on species diversity (Chen et al., 2008; Wu et al., 2012), above- and below-ground biomass (Li et al., 2011; Sun, Cheng & Li, 2013; Yang et al., 2009; Zeng, Wu & Zhang, 2015), and the plant diversity-productivity relationship (Ma et al., 2010; Wang et al., 2013; Wu, Shen & Zhang, 2014). On this plateau, water stress is one of the main factors constraining productivity of alpine grasslands (Shen et al., 2015; Shi et al., 2014). Therefore, determining how plants use precipitation may help understanding the underlying mechanisms associated with the plant diversity-productivity relationship. However, to our knowledge, only Hu et al. (2010) and Yang et al. (2010) have recently documented the spatial PUE pattern across diverse alpine grasslands, reporting a linear PUE-MAP (mean annual precipitation) relationship and a hump-shaped PUE-MAP relationship, respectively, along the precipitation gradient across the Qinghai-Tibetan Plateau. Indeed, both studies provided valuable examinations of soil nutrients, species richness (SR) and canopy coverage in controlling the spatial variation in PUE, however, they did not clarify whether and how community intrinsic structural properties, such as species composition and plant traits assembly, can also affect PUE of the Tibetan alpine grasslands. To better manage alpine pastures, it is necessary to estimate the relative contribution of climatic conditions, soil nutrients and vegetation regimes to the variability in response of alpine grassland productivity to precipitation.

In this study, we conducted a multi-site transect survey across the northern Tibetan Plateau in summer 2010, with maximum aboveground biomass (AGB), species diversity indices and community leaf traits measured, and climate and soil variables collected. Here, we had three objectives for the analyses of PUE variations across alpine grassland types at the regional level. The first is to clarify the spatial variation of PUE along climatic and soil nutrient gradients across zonal alpine grasslands on the northern Tibetan Plateau. In previous studies, PUE was often regressed against mean annual or seasonal precipitation. This did not make sense because PUE was also derived from precipitation. In this study, in addition to soil influences, we mainly focused on the independent effect of temperature alone, and the combined effect of both temperature and precipitation on PUE at the community level. Our second objective is to detect whether and how species composition and leaf functional traits can influence the spatial variation in PUE along climatic and resource gradients across the northern Tibetan Plateau. Finally, the third objective is to disentangle the relative contribution of climate, soil, and community properties to PUE across alpine grasslands on the northern Tibetan Plateau.

Materials and Methods

Sampling design

To minimize the influences of complicating factors (e.g. livestock grazing and fencing duration) on measurements of species diversity, biomass, and leaf traits, we rigorously restricted our surveys to the 15 sites excluded from livestock grazing since 2006, five sites for each grassland type, to cover the represent natural communities of the three zonal alpine grassland types. Generally, the grazing exclosures were greater than 20 hectares in size. These ungrazed pastures were separated at intervals of approximately 50–80 km from each other across the northern Tibetan Plateau. At each site, we firstly set up a 200 × 200 m plot, approximately 200 m away from the exclosure edges, and then placed five 50 × 50 cm quadrats systematically at equal intervals along a 100-m transect line that was randomly laid out within the plot (Fig. 1). Species height (cm) and percent cover (%) of all the plants present at each quadrat were recorded before harvesting the living biomass. We measured the height of three individuals at least per species, thus an average height was used for further calculations. Plant biomass was cut to soil surface, sorted by species, and dried at 65 °C for 48 h in an oven prior to weighing. Most plants in alpine grasslands on this plateau generally reach their maximum coverage in August (Ma et al., 2010; Shi et al., 2014). Therefore, the maximum AGB can be viewed as ANPP, and the ratio of AGB to precipitation can be accepted as PUE in this study (Yang et al., 2010). In addition, thirty 0.1-m² circles were randomly placed within each plot to determine the local species pool and species relative frequency (Fig. 1).

Results

Spatial patterns of precipitation use efficiency along environmental gradients

PUE linearly decreased along the increasing AccT gradient, but linearly increased with increasing CMI (Figs. 2A and 2B). AccT and CMI were found to be negatively correlated with each other, and to be negatively and positively, respectively, correlated with GSP across sites (Table 2). PUE was found to marginally linearly increased with increasing SOC in the topsoil, but had no relation with STN (Figs. 2B and 2C). No correlation was found between SOC and STN across sites (Table 2).

Table 2:

Pearson correlation coefficients between growing season precipitation (GSP), accumulated active temperature for daily values above 5 °C (AccT), climatic moisture index (CMI, equaling to GSP/AccT, mm °C⁻¹), topsoil organic carbon (SOC, %), soil total nitrogen (STN, %), species richness in the 50 × 50 cm quadrat, the Shannon diversity index (H), functional group composition (FGC, sum score of species water ecological strategies, 1-xerophytes, 2-mesophytes and 3-hygrophytes, weighted with plant group coverage), community leaf area index (LAI_c, m² m⁻²) and community specific leaf area (SLA_c, defined as leaf area per leaf mass, cm² g⁻¹) at 15 sites across the northern Tibetan Plateau.

	GSP	AccT	CMI	SOC	STN	SR	H	FGC	LAI
AccT	−0.868**
CMI	0.914**	−0.975**
SOC	0.668**	−0.543*	0.661**
STN	0.075	−0.082	0.100	0.393
SR	0.770**	−0.563*	0.570*	0.415	−0.150
H	0.818**	−0.625*	0.629*	0.414	−0.129	0.974**
FGC	0.864**	−0.704**	0.775**	0.679**	0.039	0.781**	0.804**
LAI	0.900**	−0.704**	0.764**	0.689**	0.179	0.801**	0.839**	0.939**
SLA	0.164	0.018	0.046	0.143	−0.229	0.218	0.132	0.186	0.129

DOI: 10.7717/peerj.2680/table-2

Notes:

** correlation is significant at p < 0.01;

* correlation is significant at p < 0.05.

Spatial patterns of precipitation use efficiency along community structure variables

PUE linearly increased with SR, the Shannon diversity index, plant FGC, and community leaf area index (LAI_c) (Figs. 3A–3D). PUE had no relation with community specific leaf area (SLA_c) (Fig. 3E), but deceased with increasing foliar stable carbon isotope composition at the community level (Fig. 3F). Except of no correlation with STN, plant FGC and LAI_c were highly correlated with climatic variables and other vegetation properties (Table 2). Extreme correlations with absolute coefficient higher than 0.9 were found between GSP and CMI (0.914), between AccT and CMI (−0.975), between SR and the Shannon diversity index (0.974), and between plant FGC and LAI_c (0.939) (Table 2). Although SLA_c had no relation with climate variables, soil nutrients, or other community structural variables (Table 2), SLA was significantly related with foliar δ¹³C at the species level (Fig. 4).

Discussion

The data presented in this study illustrated that climate primarily constrains PUE of alpine grasslands on the Tibetan Plateau. Our results also indicated that vegetation structural properties including SR, community assembly, plant composition of different functional groups, and plant functional trait diversity can mediate the response of productivity to changing climate regimes. More importantly, we found that the explanatory variables that reflect community assembly of plant species, functional groups, and functional traits, can influence PUE as strongly as climatic variables. Therefore, we suggest that complex measures such as functional trait diversity, convergence and divergence of species niches, community assemblage of plant functional groups, are important predictors of ecosystem functioning with ongoing climate change for alpine grasslands.

Alpine grasslands have a long co-evolutionary history with severe abiotic environmental conditions on the Tibetan Plateau. Indeed, changes in precipitation regimes for the central Tibetan Plateau caused the spatial and temporal variation in grassland productivity and phenology (Shen et al., 2015; Shi et al., 2014). We accept the CMI to reflect the combining effect of precipitation and temperature together as did Wang et al. (2013). Our results showed that CMI alone explained 47.9% of the total variance in PUE across the sites in this study (Fig. 2; Table 3). Significant correlations of community structural and functional variables with CMI (Table 2) further confirmed that different PUE among alpine grassland types was the result of the environmental filtering effects on plant species composition and functional trait diversity. However, in our study, SOC and total nitrogen had no influence on PUE (Fig. 2), and both failed to enter the optimal GLM with multiple explanatory variables (Table 3). However, we cannot reject that that that loamy soils with higher organic carbon content in AM has a greater water-holding capacity that sandy soils in AS and desert-steppe (Yang et al., 2010). Knapp et al. (2002) reported that ANPP is more responsive to soil moisture variability than to mean soil water content. In addition, plant community composition and key carbon cycling process can be altered by increases in precipitation variability. A greater proportion of small rainfall events at arid desert-steppe may be more easily evaporated due to relatively sparser canopy structure, coarser soil texture and warmer soil temperature, whereas large rainfall events at the semi-humid meadow zone may be less evaporated and better hold up due to relative compacted canopy cover, finer soil texture and colder soil temperature (Table 1).

The alpine grasslands selected in this study cover a wide range of species diversity, productivity and environmental gradients, representing a good natural experiment platform from which to research climate-vegetation relationship on the Tibetan Plateau (Li et al., 2011). PUE for the three zonal alpine grassland types in our study (Table 1) were relatively lower than those reported in previous studies (Hu et al., 2010; Yang et al., 2010). For example, PUE for AMs and steppes reported by Hu et al. (2010) were nearly 2.7 times and twice the average values in our study. Although our sites cover a wide range in SR (from 2 to 16 species per 0.25 m²), average plant SR was lower when compared with AMs and steppes in the north-eastern and central Tibetan Plateau (Ma et al., 2010; Shi et al., 2014). It was reported that productivity was often greater and of lower annual variability for species-rich communities than species-poor ones across the Tibetan Plateau (Shi et al., 2014), possibly because the co-evolution of plant species assemblage with local climate condition and soil properties (Ma et al., 2010; Wang et al., 2013). Moreover, a positive linear relationship between productivity and SR has been reported in alpine grasslands on the Tibetan Plateau (Ma et al., 2010; Wang et al., 2013; Wu, Shen & Zhang, 2014). Our results also confirmed that SR, accounting for 17.8% of the total variance of PUE at the regional scale (GLM3 in Table 3), can be serve as a good predictor of PUE in this region (Fig. 3A). In the temperate sandy grasslands of Inner Mongolia, Zuo et al. (2012) argued that the positive correlation between plant diversity and productivity was indirectly driven by vegetation composition. The niche complementarity theory suggests that a species-rich community generally has more efficient resource utilization and more stable ecosystem functionality than species-poor communities (De Boeck et al., 2006; Marquard et al., 2009; O’Connor, Haines & Snyman, 2001). Therefore, the lower PUE in steppe and desert-steppe compared with that in AM in this study was likely caused by lower SR and the weaker complementary effect there.

In addition to SR, we also examined the effects of the Shannon diversity index and plant functional groups. A significant linear relationship was observed between PUE and the Shannon diversity index (Fig. 2B). Because of the significant correlation between the Shannon diversity index and SR in our case (Table 2), their relative contributions were found comparable to variability in PUE, accounting for 22.3 and 17.8% of the total variance of PUE by the Shannon diversity index in the optimal model and SR in the secondary optimal model (Fig. 3; Table 3). Therefore, our results supported the idea that that the productivity of alpine grasslands on the Tibetan Plateau is dependent not only on SR but also on species composition. The relative contribution of community composition of plant functional groups that have differentially evolved to adapt to serve habitat conditions on this plateau was not strong as expected (Table 3). PUE was found to increase with increasing the value of FGC (Fig. 3C), which reflects an overall water ecological strategy weighted by plant functional group cover at the community level. Communities dominated by mesophytic grasses have been found to respond more rapidly to precipitation change than communities that were dominated by xerophytic grasses (Paruelo et al., 1999). This is partly consistent with the finding of Cantarel, Bloor & Soussana (2013) that climate-induced decrease in above ground biomass may be driven by changes in the relative abundance of plant functional groups.

The current research on the relationship between biodiversity and ecosystem functionality has increasingly moved the focus from plant species to trait-based approaches (Orwin et al., 2010; Violle et al., 2007; Wellstein et al., 2011). In this study, we also found that PUE significantly increased with increasing LAI_c (Fig. 3), which explained 17.3% of the total variance. This was consistent with the positive linear relationship between PUE and vegetation cover observed in alpine grasslands by Hu et al. (2010), who suggested that climate change-induced shifts in vegetation structure, and consequently LAI may have a significant impact on the relationship between ecosystem carbon and water cycles in grasslands (Hu et al., 2008). Semi-arid grasslands are projected to be among the most sensitive ecosystems to changes in precipitation (Cherwin & Knapp, 2012). Life history and biogeochemical mechanisms can interact to influence the production response of alpine grasslands to precipitation. The evolutionary history and ecological attributes of species present in the vegetation assemblage can influence production potential as a result of constrains on growth rate imposed by trade-offs with traits for stress tolerance. Recruitment limitation can thus constrain local-scale SR and productivity, either by a lack of seeds or by reduced seedling growth, likely due to competition from the established vegetation (Zeiter, Stampfli & Newbery, 2006) The severe water stress due to less precipitation and warmer in AS and desert-steppe communities may not only lead to self-thinning but also to greater relative dominance of plant species that are more tolerant to drought with lower SLA and deeper roots(Wu et al., 2013b; Zhu et al., 2015). Foliar δ¹³C data in plant tissues has increasingly been used to infer intrinsic water use efficiency (Farquhar, Ehleringer & Hubick, 1989; Seibt et al., 2008). Both Song et al. (2008) and Zhou et al. (2011) also confirmed that the leaf δ¹³C of dominant species can reflect plant responses to the environmental water gradients on the Qinghai-Tibetan Plateau. In our results, the significant relationship of foliar δ¹³C to SLA at the species level (Fig. 3) did not result in a significant relationship between PUE and SLA_c (Fig. 2E). However, PUE was closely related to plant foliar δ¹³C and declined linearly with increasing community foliar δ¹³C (Fig. 2F). Therefore, we ascribed this outcome to uncertainties resulting from the assembly of species with differential water ecology strategies.

In summary, the productivity of alpine grassland communities across the northern Tibetan Plateau is controlled primarily by rainfall during the plant growing months. Our hypothesis that community structural properties and plant functional traits would mediate the sensitivity of alpine grassland productivity in response to precipitation change is supported, but the limited amount of data in this study cannot clarify the causal networks among climate change, community assembly and functional trait diversity. Our short-term study design does not allow us to answer the question of how ecosystem stability in the response of productivity to precipitation either. Clearly, long-term observation of community structural (and functional) changes in species and trait assembly is needed for predicting ecosystem response under future climatic scenarios, and for policy-making about sustainable management of alpine grasslands on the Tibetan Plateau.

Supplemental Information