4.1 Large discrepancies between the R/S ratios from greenhouse experiments and field investigations
In this study, we found that the individual-level mean R/S values for the six common species from the temperate grasslands in northern China ranged from 0.04 to 0.92 (Figure 3), which were lower than those from global and regional field community investigations (Table 2). Specifically, our result is much lower than that (1.66~15.21) of a field community survey performed in the same region (Wang et al., 2008); the R/S of perennials in this study is close to the result (0.84) of individual-level field observations (Wang et al., 2010) but still lower than that (0.75~2.98) of genus-level surveys (Luo et al., 2013). Considering the field sampling process, we suggest that the R/S obtained in field community studies may be strongly overestimated for three main reasons. First, field sampling methods are usually not equipped to accurately identify dead and live roots and are more prone to including dead roots in root samples, resulting in an overestimation of root biomass and therefore of R/S (Wang et al., 2010). Indeed, the large discrepancy between the R/S results at the individual and community levels in the same area can also support this view. For example, Wang et al. (2010) reported that the median R/S at the individual level in northern China’s grasslands was only 0.84, which was much lower than that at the community level (Table 2). In fact, root sampling at the individual level is more accurate than sampling at the community level because root systems of the same species are relatively uniform in morphology and colour, which makes it easier to distinguish living roots from dead roots. Second, grazing by herbivores can reduce shoot biomass to some extent, resulting in an underestimation of shoot biomass and the subsequent overestimation of R/S (Mikola et al., 2009). Third, for mature communities in the field, the dominant species are mainly perennials, and there are much fewer annuals. Our study found that the RMFs of annual grasses were extremely low. Therefore, if the R/S derived from perennial plant samples is used to estimate the root biomass of annuals, it will be highly biased.
The overestimation of R/S can occur during field sampling, but greenhouse pot experiments can effectively circumvent some of the shortcomings of field sampling. First, the roots of greenhouse plants can be fully sampled and estimated. The cultivation substrate in this study was homogeneous and uncomplicated and could be easily separated from the root system. Second, the greenhouse growth environment is consistent and stable among replicates; shoot biomass is not lost due to animal grazing, and the influence of microenvironmental variability on biomass allocation is avoided. We also acknowledge that greenhouse cultivation experiments have some drawbacks, such as the possibility of underestimating root biomass due to the height limitation of the planting pots, which limits the root growth space to some extent. In addition, for perennials, cultivation for only one growing season may result in some underestimation of root biomass if the root turnover time is greater than one year (Gill & Jackson, 2000). However, despite certain drawbacks, greenhouse cultivation experiments may still be a more accurate method than field sampling for measuring the R/S of different species in the future.