1. Introduction
The allocation of biomass to above- and belowground organs is one of the most important functional traits in plants (Dolezal et al., 2020; Enquist & Niklas, 2002). This trait not only reflects plant survival strategies but also has a significant influence on soil carbon (C) inputs (Poeplau, 2016; Rasse et al., 2005; Umaña et al., 2020). Grasslands are the most widespread terrestrial ecosystem and have important regulatory functions in the global C cycle (Dixon et al., 2014; White et al., 2000). The accurate assessment of the size of C pools in belowground biomass is an important issue for future research on grassland C cycling (Ahlstrom et al., 2015; Mokany et al., 2006). Due to the difficulty of root sampling, the belowground biomass of grasslands is usually indirectly estimated with the aid of the easier-to-measure aboveground biomass and empirical root to shoot ratio (R/S), which has become a key parameter for estimating the size of grassland C pools (Jackson et al., 1996; Mokany et al., 2006). Therefore, exploring the above- and belowground biomass allocation patterns of herbaceous plants and their modulating factors is pivotal for deepening our understanding of plant growth, survival, and carbon cycling in terrestrial ecosystems (Poeplau, 2016).
For grassland ecosystems, the R/S reported from previous field surveys is generally greater than 1 and varies widely among different studies. For example, Jackson et al. (1996) reported that the global mean value of R/S was 3.7; Mokany et al. (2006) combined data from community investigations and calculated a global mean R/S of 4.5 for temperate grassland communities. Yang et al. (2010) estimated a mean R/S value of 6.3 for grassland communities in China using field investigations from 265 sample sites, while Ma et al. (2006) obtained a mean R/S value of 61.3 for the Inner Mongolian grassland based on the synthesis of field-measured data; the R/S value at some of these sample sites reached over 400. In addition, results at the genus level differ greatly from those at the community level. For example, a study based on grasslands in northern China showed that the R/S of the four dominant genera (Stipa spp ., Cleistogenes spp ., Agropyron spp ., andLeymus spp .) ranged from 0.75 to 2.98 (Luo et al., 2013), while the median R/S at the individual level for this region was only 0.84 (Wang et al., 2010). Although a large amount of observational data has been accumulated globally over the past few decades, it is still difficult to reach a unified understanding at the global scale through data syntheses, because large unreliable datasets can deteriorate accurate statistical analyses and interpretations of the results (Mokany et al., 2006). In addition, community-level studies are restricted to reporting the weighted mean value of R/S for a given community without considering the large variations in R/S at the intra- and interspecific levels. To date, the reasons for the discrepancies in these results among different studies are unclear, which presumably results in large uncertainties in estimating the belowground C pools in grasslands. Therefore, accurate manipulation experiments are necessary to broaden our understanding of above- and belowground biomass allocation across different species.
Previous studies have overwhelmingly concluded that plant biomass allocation is jointly determined by both environmental and biotic factors (Agathokleous et al., 2019; Gedroc et al., 1996; Poorter et al., 2012). In terms of environmental factors, numerous studies have shown that water conditions (Eziz et al., 2017), temperature (Reich et al., 2014; Wang et al., 2016), nitrogen and phosphorus availability (Müller et al., 2000; Peng & Yang, 2016; Yan et al., 2019) and their ratios (Kumar et al., 2020), CO2 concentrations (Pei et al., 2020), and soil texture (Xie et al., 2012) can have significant effects on R/S. Among those biotic factors, planting density is not only an important selection pressure in nature but also an important tool for artificial pasture management (Boschma et al., 2019). It has been well recognized that planting density can alter the above- and belowground competition for light and nutrient among plants (Kira et al., 1953; Weiner, 1986; Weiner & Freckleton, 2010; Yoda et al., 1963) and thus influences biomass allocation. Through a global data synthesis analysis, Poorter et al. (2012) found that on average, plants tended to increase their R/S with increasing planting density. However, many site- or species-specific studies have also found very complicated density effects on R/S, with invariant (Casper et al., 1998), increasing (Berendse & Möller, 2009) and decreasing (Hecht et al., 2016; Weiner, 1990) trends with increasing planting density. However, the density effects on herbaceous plant biomass allocation, particularly across multiple species of different plant functional types, are still less well understood.
To address the knowledge gap mentioned above, we conducted a greenhouse experiment with a total of six common species planted in five density gradients. These species represent different plant functional types in the temperate grasslands in northern China. We cultivated these plants to the maturity stage and harvested the shoot and root biomass from individuals of each species. We aim to address the following two questions: (1) How does above- and belowground biomass allocation vary among different plant functional types? and (2) How does planting density regulate above- and belowground biomass allocation in these species?