Introduction

Peatlands are an important component of terrestrial carbon storage and represent a major carbon sink. Peatlands cover only approximately 3% of the world”s land area, but they potentially store approximately 30% of the global terrestrial ecosystem carbon, equivalent to 455 Pg C (Gorham 1991; Blodau 2002). Both the belowground and aboveground components of plants are primary sources of labile C to peatland soil. The decomposition of plant material is a key component of nutrient cycling and a major contributor to soil CO2 flux (Murphy and Moore 2010). Understanding plant aboveground biomass (AGB) and belowground biomass (BGB) allocation patterns across natural gradients is necessary for an in-depth understanding of how plants may respond to future climate change.

Compared with the considerable number of studies on AGB in peatlands, little is known about the belowground component (Weltzin et al. 2000; O’Driscoll et al. 2011; Byrd et al. 2011). Currently, using default values of root:shoot (R/S) ratios and AGB of different vegetation types to estimate large-scale root carbon storage is a practical interim method (IPCC 2003). This method is also permitted and applied by nations to estimate BGB and carbon stocks for national greenhouse gas inventory purposes (Snowdon et al. 2000; Australian Greenhouse Office 2002; Eamus et al. 2002; Mokany et al. 2006). Many peatland C models also rely on R/S ratios to estimate belowground biomass (Murphy et al. 2009a). However, some vegetation types do not have the IPCC default R/S values; such as tundra, cool temperate arid shrublands and alpine peatlands (Mokany et al. 2006). According to Murphy (2009), the BGB may be equal to or greater than the AGB in peatland ecosystems. A wide variation in R/S ratios (3:1–30:1) and a mean value of 11:1 have been suggested for arctic tundra ecosystems (Dennis and Johnson 1970). Relatively few researchers have considered the variations in R/S ratios across gradients of plant types and environments within and between peatland ecosystems (Finér and Laine 2000; Murphy et al. 2009a). Miller (2011) examined the effects of long-term drainage on plant community composition and biomass in boreal continental peatlands (bogs and fens) but only considered the AGB. In summary, we still do not know the R/S ratios in alpine peatland environments or understand their relationships with natural gradients within alpine peatlands.

The optimal partitioning theory and the allometric biomass partitioning theory are two important biomass allocation hypotheses. Under the optimal partitioning theory, plants allocate biomass to the organ that acquires the most limiting resource. This means that plants should allocate more biomass to roots when water or nutrients are limiting and shift more biomass aboveground in higher nutrient or moisture conditions (McConnaughay and Coleman 1999; McCarthy and Enquist 2007; Kobe et al. 2010;). Allometric partitioning theory states that plant allocation between components is mainly regulated by total plant size and follows a scale relationship between compartments (West et al. 1999; Genet et al. 2011). These two hypotheses have been studied extensively in woody plants and grasslands, mainly at the individual level (Enquist and Niklas 2002; Murphy et al. 2009b; ). Recently, Murphy and Moore (2010) reported that the relationships between AGB and BGB of shrubs in an ombrotrophic peatland complied with the allometric relationships, whereas herbs did not. Moreover, variations in water conditions may change the allometric relationship between AGB and BGB (Murphy et al. 2009). In comparison with information at the individual level, little evidence is available at the community level. The results of Yang et al. (2009) support the isometric allocation hypothesis for the AGB and BGB relationship in Tibetan grasslands. However, whether this relationship holds true across alpine peatlands types is unknown.

The Zoige wetland of the Tibetan plateau is an alpine peatland at permanent low temperatures. The existing studies are limited and mainly focused on methane fluxes (Chen et al. 2009; Chen et al. 2011) and landscape pattern changes (Bai et al. 2008), and little work has been conducted to directly compare the biomass of vegetation communities, vertical distribution of roots and effects of environmental factors on biomass allocation. The major objectives of this study were to 1) characterize differences in different plant communities with respect to both AGB and BGB, 2) quantify belowground root distribution with depth, and 3) determine the relationship between biomass and environmental factors.

Material and Methods

Site Description

The high-altitude Zoige peatland (32° 100 N–34° 100 N, 101° 450 E–103° 250 E) is located on the eastern margin of the Qinghai–Tibet Plateau (Fig. 1). It is the highest and largest peat marsh in China and was, formed during the Early Holocene (9355 ± 115 BP) (Chai and Jin 1963). The Zoige peatland contains 26% of the total peatland area of China and 45% of the total peat reserves (Joosten 2004). Mean annual temperature in the area is 0.6–1.0 °C because of the high altitude (3400–3600 m), and the majority of the 580–860 mm of annual precipitation falls in summer (Chen and Bloemendal 1999, Zhang and Jiang 2008). The dominant vegetation in the majority of this area is perennial herb (Carex muliensis, Heleocharis uniglumis, Blysmus sinocompressus). The distribution of species is related to water table position, with Carex muliensis and Heleocharis uniglumis found mostly in wetter areas of the peatland while Kobresia pusilla and Potentilla fulgens prefer drier areas (Han et al. 2011, 2012). In this study, we explore biomass allocation and relationships between biomass allocation and environmental factors using data surveyed from 32 sites across the Zoige peatland during 2011–2012. The 32 sites essentially represent the major plant communities and surface water status variation of the Zoige alpine peatland.

Fig. 1
figure 1

Spatial distribution of sampling sites across alpine peatlands in the Zoige wetland, Qinghai–Tibetan Plateau

Full size image

Field Biomass Survey and Soil Features

The AGB and BGB samples were collected at 32 sites on the Zoige peatland in 2011 and 2012 during July and August, approximately the time of peak biomass production. At each 10 × 10 m site, the site surface bare spot area, water table and plant species composition were investigated. Within each of the 32 sample plots, five replicates of vegetation were harvested at the ground surface from 50 × 50 cm squares.

The BGB was estimated from three replicate 5 cm diameter soil cores collected at depth intervals of 10 cm down to the maximum root depth using a soil auger. The maximum root depth was determined by soil carbon storage estimation (0–2 m) and as described in another paper (Ma 2013). Dates from the upper 80 cm of soil were used here because almost all plant roots were concentrated within this zone in all samples from the Zoige peatland. Cores were washed through a 0.5 mm sieve to remove soil. Both AGB and BGB samples were oven-dried at 65 °C until a constant mass was reached. Total biomass was the sum of AGB and BGB.

Soil samples were collected at depth intervals of 10 cm to the maximum root depth using a soil auger. Six replicate 5 cm diameter soil cores were collected from each site. Three of these were oven-dried at 105 °C until a constant mass was reached to measure gravimetric soil water content and soil bulk density (BD). The remaining soil cores were air-dried for analysis of soil physicochemical properties. Soil organic carbon (SOC) was determined with the K2Cr2O7 titration method after digestion (Nelson and Sommers 1975). Total nitrogen (TN) was determined using the semi-micro-Kjeldahl method (Lu 1999). Total phosphorus (TP) was determined colorimetrically after wet digestion with H2SO4 plus HClO4 (Parkinson and Allen 1975).

Data Analysis

We calculated the mean values of AGB, BGB, total biomass and R/S ratio for all sampling sites. In addition, differences in AGB, BGB and R/S ratio between the seven types of plant communities were evaluated using ANOVA. To avoid pseudoreplication (Hurlbert 1984; Weishampel 2009), data from subplots of the same cover type within a given plot were pooled to form a single unit rather than treated as independent replicates.

Vertical root distributions were modeled using the asymptotic equation described by Gale and Grigal (1987) and Jackson (1996) (Eq. 1):

$$ \mathrm{Y}=1-{\beta}^{\mathrm{d}} $$
(1)

where Y is the cumulative percentage of root biomass from the soil surface to depth d (cm) and β is the estimated parameter. Values of β can range from 0.1 to 1, where 1 indicates that all production is at depth and 0.1 that all root production is at the surface (Murphy et al. 2009). We then calculated the percentage of root biomass found in the upper 30 cm of soil for each biome, based on their respective β values.

To study the relationships between AGB and BGB, the data were analyzed with a linear regression function. The analysis was conducted using SPSS software (SPSS version 18.0; SPSS Inc., Chicago, IL).

We then investigated the relationships between each environmental variables and AGB, BGB, total biomass, R/S ratio and β value. Multiple factor analysis (MFA) (BARALOTO 2011; Le et al 2008) was used to obtain an overview of the plots and the variables describing them. The advantage of MFA is that variables are separated into groups, each of which is given equal weight in the analysis. Environmental variables are separated into three groups: soil physical and chemical properties (soil water content, BD, SOC, soil TP, soil TN, C:N ratio and N:P ratio), site surface bare spot area and water table. The MFA analyses were conducted using the package FactoMineR (Le et al. 2008) in the R language and environment for statistical computing version 2.11.1 (R Core Development Team 2009). Finally, the relationships between AGB, BGB, total biomass, R/S ratio, β and environmental variables (soil water content, water table, SOC, TN, TP, C:N, N:P) were tested individually by regression analyses, and the correlations among environmental variables in the Zoige peatland were studied using correlation analyses. Regression analyses and correlation analyses were performed using SPSS software (SPSS version 18.0; SPSS Inc., Chicago, IL).

Results and Discussion

AGB, BGB and R/S Ratio

The dominant vegetation community types in the majority of this area are Carex muliensis, Heleocharis uniglumis, Blysmus sinocompressus, Carex lasiocarpa, Kobresia pusilla and others. The distribution of species is related to water table position, with Carex muliensis and Heleocharis uniglumis found mostly in permanent water areas of the peatland while Kobresia pusilla, Elymus nutans and Potentilla fulgens prefer drier areas. The environmental characteristics of the seven vegetation communities are given in Table 1.

Table 1 Environmental characteristics of different community types in the Zoige peatland. For convenience, the first letter abbreviations of dominant species are given instead of communities in the rest of the paper (e.g., CM instead of Carex muliensis + Myriophyllum spicatum)
Full size table

The mean AGB, BGB, total biomass and R/S ratio values by total and major plant communities in the Zoige peatland are given in Fig. 2. The average AGB, BGB and total biomass in the Zoige alpine peatland were 341.01, 3262.93 and 3620.36 g m−2, respectively, and the average R/S ratio was 10.32. The CM and EPC communities had the largest BGB and total biomass followed by the CK and BCP communities, and finally the KP, EPA and PA communities. Total biomass and BGB were smallest in the PA community and largest in the CM community. Our survey also indicates that the CM, EPC, BCP and CK communities were usually located in areas with a permanent or seasonal surface water regime, and KP, EPA and PA communities were usually located at the humid soil surface.

Fig. 2
figure 2

The mean AGB, BGB, TB (Total biomass) and R/S ratio values in the Zoige peatland. Different letters indicate significant differences between the seven plant communities (one-way ANOVA, p < 0.05)

Full size image

We found that roots represent a quantitatively important biomass in the Zoige peatland. The R/S (BGB/AGB) ratio (10.32) for the Zoige peatland (Fig. 2) was higher than values of 0.28–1.38 observed in temperate peatland by Murphy (2009), 0.72–1.27 in an ombrotrophic bog by Moore et al. (2002) and 7.5 (1.5–21.9) in an average wetland by Cˇížková (1999). Mokany et al. (2006) reported that shrublands and grasslands possessed a much greater range in R/S ratios (0.34–26.03) than forests and woodlands. According to a global analysis of root distributions in terrestrial biomes conducted by Jackson (1996), R/S ratios were highest for tundra, grasslands and cold deserts (ranging from 4 to 7) out of all terrestrial ecosystems. Our estimated R/S ratio is comparable to estimates made for semi-wet arctic tundra ecosystems (R/S ratio of 11) by Dennis and Johnson (1970). This result is in accordance with the significant increase in the R/S ratio with declining temperature (Hui and Jackson 2005; Mokany et al. 2006) and increasing altitude (Leuschner et al. 2007; McCarthy and Enquist 2007). The higher R/S ratio in alpine peatland could be due to the relatively slow depletion of root carbohydrates in response to low respiration rates in cold environments (Yang et al. 2009) and might be associated with slower root turnover in colder regions (Gill and Jackson 2000; Yang et al. 2009).

The communities located in areas with occasional surface saturation have been shown to allocate more biomass to roots than the communities in areas where the surface is permanently wet and in areas where the surface is never saturated. Similar results have been reported by Olsrud and Christensen (2011) for a subarctic mire ecosystem, where the semi-wet ecosystem had a higher R/S ratio compared with the wet minerotrophic ecosystem. This shift in biomass allocation may arise from waterlogging (McFarlane et al. 2003) or adverse soil chemical conditions (Leuschner et al. 2007; Lambers et al., 1998).

Vertical Distribution of Roots

Asymptotic modeling of the vertical root distribution in the Zoige peatland produced an average biomass depth distribution parameter (β) of 0.921 (Fig. 3). The mean for the whole Zoige peatland was approximately 86% of roots were in the upper 30 cm of soil. Higher β values denote a greater proportion of roots at depth. Plant communities varied in their rooting depth distribution (Table 2). This difference in the seven communities could be attributed to their different species composition. Although absolute differences between the average β values of the seven communities appear to be small, these differences lead to large differences in vertical root distributions. For example, mean β values from the E community (β = 0.8383) and F community (β = 0.9442) indicate that approximately 99% of the roots of the E community are located above 30 cm depth but only 82% of roots of the F community are located above 30 cm depth. Differences in the parameter β do not reflect differences in root biomass but only in the vertical root distributions relative to depth.

Fig. 3
figure 3

Mean cumulative root fraction in relation to soil depth in the Zoige peatland. Error bars show standard deviation. The trend line is the modeled cumulative root fraction by depth using the equation y = 1 − βdepth with an average β parameter of 0.9210

Full size image
Table 2 Mean cumulative root fraction by soil depth in the seven communities
Full size table

The mean β in our alpine peatlands showed shallower rooting profiles (β = 0.921, r2 = 0.96) than those observed in global assessments of grasses (β = 0.952, r2 = 0.88) and values were closer to those observed in tundra ecosystems (β = 0.914, r2 = 0.91) (Jackson 1996). Plants are likely to concentrate their root production in the surface layers of the soil profile in response to high concentrations of N, P, and K (Jobbágy and Jackson 2001; Murphy and Moore 2010). In addition, oxygen deficiencies are least likely to occur in shallow soil layers of peatlands (Schenk and Jackson 2002). Our results indicate that there are some similarities in vertical root distribution profiles between high-altitude peatland ecosystems and ecosystems at high latitudes. The shallower rooting profiles in these ecosystems are partly caused by the physical barriers inhibiting root growth in cold regions, such as permafrost (Jackson 1996; Kane et al. 1992). In addition, waterlogging usually limits root growth (Kane et al. 1992) by reducing water absorption and transpiration (Ladiges et al. 1981). These, among other factors, contribute to the fact that tundra ecosystems are the most shallowly rooted of all biomes.

Allometric Relationship between AGB and BGB

There was a positive allometric relationship (P < 0.01) between AGB and BGB. The relationship between BGB and AGB in the Zoige peatland could be characterized by a linear regression function of BGB =4.81 AGB + 1749.36 (r2 = 0.19, P < 0.01) (Fig. 4). There have been numerous studies on the allometric relationships between aboveground biomass and belowground biomass (for example, Feliciano et al. 2014; Njana et al. 2015; Ward 2015). Our results suggested that belowground biomass could be estimated from the aboveground biomass in the Zoige peatland. This information is valuable given the difficulty in measuring belowground biomass.

Fig. 4
figure 4

Bivariate plots of data for AGB and BGB in the Zoige peatland. The solid line is the linear regression of the data

Full size image

Biomass Allocation and Environmental Factors

MFA, regression analyses and correlation analyses were used to examine the relationship between environmental variations in addition to their relationships with biomass. Two major gradients in environmental variables were observed and they explained 61.2% of the variance in the dataset (Fig. 5). The first MFA dimension has strong contributions of the principal components of the soil physical and chemical properties, water table and surface conditions group. In particular, this dimension represents a gradient of increasing N:P ratios, C:N ratios and soil water content. The second dimension has weak contributions of soil chemical properties. Mean total biomass and BGB were positively correlated with soil water content, water table, SOC, and soil C:N ratios, and tended to increase along the first dimension of soil water content and soil chemical properties and to decrease with increasing surface bare spot area (Dimension 2) (Fig. 5).

Fig. 5
figure 5

Ordination of variables and sample plots using multiple factor analysis (MFA), in which biomass variables are illustrated along the resulting dimensions. The panel shows the correlation circle with groups of soil, water, and surface variables. The meanings of the acronyms are as follows: Biomass depth distribution parameter (β); Total phosphorus (TP); Aboveground biomass (AGB); Total nitrogen (TN); Soil organic carbon (SOC); Water table (WT); Soil water content (SWC); Soil N:P ratio (N:P); Soil C:N ratio (C:N); Total biomass (TB); Belowground biomass (BGB); Belowground biomass: Aboveground biomass (R:S); Bare spot area (BSA); Soil bulk density (BD)

Full size image

Both AGB and BGB were positively correlated with the water table (r2 = 0.167, P < 0.05 for AGB, r2 = 0.320, P < 0.001 for BGB) and soil C:N ratio (r2 = 0.138, P < 0.05 for AGB, r2 = 0.418, P < 0.001 for BGB) (Table 3). However, the R/S ratio did not show any significant change along the gradient of the water table but did show a significant change with soil C:N ratio. This indicates that the rising water table did not initiate the reduction in the aboveground biomass.

Table 3 Relationship between biomass (AGB, BGB, total biomass, R/S and biomass depth distribution parameter (β)) and nutrient supply and water availability variations in the Zoige peatland
Full size table

Water conditions were positively correlated with total nitrogen (r2 = 0.45, P < 0.05 for water table, r2 = 0.59, P < 0.01 for soil water content) and soil organic carbon (r2 = 0.65, P < 0.01 for water table, r2 = 0.78, P < 0.01 for soil water content) (Table 4). Similar results from previous studies have shown that the increase in soil water availability may accelerate soil N cycling (Lü et al. 2014; Van Groenigen et al. 2014). Moreover, water availability might increase fine root production, allowing plants to explore more of the soil volume for available N (Li et al. 2011; Lü et al., 2014).

Table 4 Correlation among environmental factors in the Zoige peatland
Full size table

Moreover, both SOC (r2 = 0.15, P < 0.05) and soil water content (r2 = 0.485, P < 0.001) were positively correlated with BGB (Table 3). In addition, soil water content showed a significant positive correlation with soil organic carbon content (r2 = 0.78, P < 0.01) (Table 4). This result indicated that soil water content, soil organic carbon and their interaction could influence belowground biomass. Correlation analysis further confirmed that the water conditions, soil organic carbon and soil nitrogen were the main factors that influenced the plant biomass and biomass allocation in the Zoige peatland. These results suggested that allocation patterns were different in response to differences in water and nutrient availability.

Conclusion

This study is the first to document information on biomass allocation and its relationship with environmental factors in alpine peatlands on the Qinghai–Tibetan Plateau. In the Zoige alpine peatland, the average AGB, BGB and total biomass were 341.01, 3262.93 and 3620.36 g m−2, respectively, and the R/S ratio was 10.32. The CM and EPC communities had the largest BGB and total biomass, followed by the CK and BCP communities, and finally the KP, EPA and PA communities. Total biomass and BGB were smallest in the PA community and largest in the CM community. We found that the overall R/S ratio (10.32) in alpine peatlands was higher than that observed in temperate peatland (0.28–1.38) and is comparable to estimates made in a wetland arctic tundra ecosystem (R/S ratio of 11). Zoige peatlands have a much shallower root distribution than temperate grasslands, with 86% of roots in the top 30 cm of soil. This unique distribution of roots in alpine peatlands should, therefore, be incorporated into biogeochemical models that examine the feedbacks of alpine peatland vegetation to climatic change, because current global biogeochemical models never consider such a pattern of root distribution (Jackson 1996; Schenk and Jackson 2002). There was a positive allometric relationship (P < 0.01) between AGB and BGB. The water conditions, soil organic carbon and soil nitrogen were the main factors that influenced plant biomass and their biomass allocation in the Zoige peatland.