SEARCH

SEARCH BY CITATION

Keywords:

  • aboveground biomass;
  • age sequence;
  • belowground biomass;
  • ecophysiology;
  • isometric pattern;
  • root:shoot ratio;
  • space;
  • time

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

1. Knowledge of biomass partitioning is essential for estimating spatial patterns and temporal dynamics of root biomass in terrestrial ecosystems. The isometric hypothesis predicts that aboveground biomass scales isometrically with belowground biomass across both individual plants and community types (i.e. the slope of the log–log relationship between above- and belowground biomass is not significantly different from 1.0), and that the isometric pattern is independent of variation in environmental conditions. However, current evidence primarily comes from observations over space. It is still unknown whether biomass partitioning patterns occurring over time are similar to those observed over space.

2. In this study, we explored biomass partitioning patterns in forest ecosystems over space and time by synthesizing biomass measurements made in 112 stands extracted from 16 age sequences around the world. We characterized biomass partitioning patterns in forest ecosystems using both root: shoot ratio and the relationship between above- and belowground biomass. Data across various individual studies were organized to reflect biomass partitioning patterns over space, while data within each individual study were used to illustrate biomass allocation patterns over time.

3. Our results showed that root:shoot ratio did not exhibit any significant trend with stand age over space. Similarly, root:shoot ratio remained relatively constant over time in 10 out of 16 age sequences.

4. Reduced major axis analysis indicated that the slope of the log–log relationship between above- and belowground biomass did not reveal a significant difference from 1.0 over space, supporting the isometric hypothesis. Likewise, the slope was not statistically different from 1.0 in 10 out of 16 age sequences, nor did it show any significant change with climatic factors and stand age. Thus, the isometric hypothesis also most probably holds true during forest stand development.

5.Synthesis. Our results demonstrate that biomass partitioning patterns occurring over time are consistent with those observed over space, suggesting that belowground biomass dynamics in forest ecosystems may be reliably estimated from aboveground biomass using biomass partitioning patterns generated over space.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Biomass partitioning, commonly described by the root:shoot (R:S) ratio or the allometric function relating belowground biomass (MB) and aboveground biomass (MA), can be a critical indicator of plant physiological processes (Wilson 1988; Cannell & Dewar 1994; van Wijk et al. 2003; Mokany, Raison & Prokushkin 2006). It reflects the different investment of photosynthates in above- and belowground organs (Shipley & Meziane 2002), and its variation is a response to differential selection for adaptations to different environmental conditions (Bazzaz & Grace 1997). For instance, growth-limiting conditions usually lead to greater biomass partitioning to those organs that are constrained (Reynolds & Thornley 1982; Luo, Field & Mooney 1994; Luo, Hui & Zhang 2006; Wang & Taub 2010). Biomass partitioning is also an important parameter for estimating MB from the more easily measured MA (Cairns et al. 1997; Mokany, Raison & Prokushkin 2006; Yang et al. 2010) and has been incorporated into terrestrial ecosystem modelling (Friedlingstein et al. 1999; Landsberg 2003). Therefore, quantitative assessments of biomass partitioning are essential for evaluating the responses of vegetation growth to ambient environmental conditions (Niklas 2005) and for understanding root biomass distributions in terrestrial ecosystems (Hui & Jackson 2005; King et al. 2007; Yang et al. 2009, 2010).

Biomass partitioning has been widely examined across various ecosystems (e.g. Jackson et al. 1996; Titlyanova et al. 1999; Hui & Jackson 2005; Mokany, Raison & Prokushkin 2006; Litton, Raich & Ryan 2007; Li, Luo & Lu 2008; Fan et al. 2009; Yang et al. 2009, 2010; Ma et al. 2010; Wang et al. 2010). More importantly, an isometric hypothesis has been developed to characterize the allometric relationship between MA and MB (Enquist & Niklas 2002; Cheng & Niklas 2007). This hypothesis predicts that MA scales isometrically with MB across both individual plants and community types (i.e. the slope of the log–log relationship between MA and MB is not significantly different from 1.0), and that the scaling slope does not vary with environmental factors (Enquist & Niklas 2002; Cheng & Niklas 2007). Specifically, the isometric model states that standing leaf biomass (ML) scales as the 3/4 power of standing stem biomass (MS), ML scales as the 3/4 power of standing root biomass (MR), and MR scales isometrically with respect to MS (Enquist & Niklas 2002; McCarthy & Enquist 2007). Considering this point, inline image andinline image where β1 andβ are allometric constants (i.e. the y-intercept of the log–log regression relationship) (Niklas 2005; Cheng & Niklas 2007). Given that MA is the sum of leaf and stem biomass (i.e. inline image) and MB is only comprised of roots (i.e.MR = MB), the relationship between MA and MB can be expressed asinline image. If inline imagecan be expected to scale isometrically with respect to MB (Niklas 2005; Cheng & Niklas 2007).

The isometric hypothesis has been validated across both individual plants (e.g. Enquist & Niklas 2002; Niklas 2005) and community types (e.g. Cheng & Niklas 2007; Yang et al. 2009, 2010); however, current evidence is primarily derived from observations over space, with little information available over time. The ecological patterns occurring over time may differ from those observed over space since geographic patterns usually reflect the contribution of vegetation types, while temporal patterns within site may be more complicated, incorporating interannual variation in climatic and other site-specific factors (Lauenroth & Sala 1992; Jobbágy, Sala & Paruelo 2002; Fisher, Frank & Legget 2010). Thus, it remains unknown whether the isometric pattern between MA and MB observed over space also exists over time and whether the scaling slope over time is insensitive to variation in environmental conditions.

In this study, we examined partitioning patterns between MA and MB across various forest ecosystems around the world by synthesizing biomass measurements made in 112 stands and described in 16 published studies which documented changes in MA and MB during stand development. More specifically, this study aimed at testing the following four hypotheses: (i) the R:S ratio exhibits similar trends with stand age over space and time, (ii) MA scales isometrically with MB across various individual studies, (iii) an isometric pattern between MA and MB also exists during stand development and (iv) the slope of the log–log relationship between MA and MB over an age sequence is independent of environmental conditions.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Data sources

We collected biomass measurements made in 112 stands from 16 published studies that quantified aboveground biomass (MA) and belowground biomass (MB) in forest ecosystems during stand development (see Appendix S1, Table S1 in Supporting Information). These papers were assembled from a range of literature sources according to the following three criteria: (i) both coarse and fine roots were included in MB, (ii) stand age was quantitatively described and (iii) at least five age stages were involved. We were then able to statistically examine the relationship between MA and stand age, between MB and stand age, between root:shoot (R:S) ratio and stand age, and between MA and MB over an age sequence. The raw data were either extracted from published tables or obtained by digitizing graphs using the software Originpro 7.5 (OriginLab, Northampton, MA, USA). Our database consisted of geographic location (longitude and latitude), climatic information (mean annual temperature and mean annual precipitation), stand age, forest type and three target variables (MA, MB and R:S ratio). The age range exhibited large differences among various studies, varying from 1–5 to 9–316 years (see Table S1).

Data analyses

Data were processed by the following four steps. First, we examined changes of MA, MB and R:S ratio with respect to stand age across two levels: across site (a test of site and age effects) and within site (a test of age effect alone). As in other studies (e.g. Lauenroth & Sala 1992; Jobbágy, Sala & Paruelo 2002; Hui & Jackson 2005), the ‘across site’ analysis was performed to explore biomass partitioning patterns over space, while the ‘within site’ analysis was conducted to investigate biomass partitioning patterns over time. Specifically, the ‘across site’ analysis combined all observations at each age stage from each sequence to examine the relationships of MA, MB and R:S ratio with stand age across the 16 chronosequence studies. By contrast, the ‘within site’ analysis investigated the relationships of MA, MB and R:S ratio with stand age for each individual sequence.

Second, to examine whether MA and MB scale isometrically, we performed reduced major axis (RMA) regression (e.g. Enquist & Niklas 2002; Niklas 2005; Yang et al. 2009, 2010; Wang et al. 2010) to examine MAMB relationships across various sites and within each site. The regression relationship of the form log =  b (log x) was used to describe the allometric relationship between MA and MB, where x is MB (Mg ha−1), y is MA (Mg ha−1), a is the intercept, and b is the scaling slope (Enquist & Niklas 2002; Niklas 2005, 2006; Yang et al. 2009, 2010; Wang et al. 2010). The scaling slope and y-intercept of the allometric function were determined using the software package ‘Standardized Major Axis Tests and Routines’ (Falster, Warton & Wright 2003). If 95% confidence interval of the scaling slope covered 1.0, the relationship between MA and MB was considered to be isometric.

Third, to test whether the slope of the relationship between MA and MB over time is independent of environmental factors, we first estimated the scaling slope of the MAMB relationship for each individual sequence, and then investigated the relationship between the scaling slope and mean annual temperature (MAT), mean annual precipitation (MAP) and mean stand age over each sequence. A significant relationship between the scaling slope and environmental factors would indicate that biomass partitioning patterns were sensitive to variations in environmental conditions. Otherwise, we could deduce that environmental factors did not exert profound effects on above- and belowground biomass partitioning patterns.

Fourth, we conducted ordinary least squares (OLS) regression to explore the relationship between MB and MA across sites. Specifically, all biomass measurements made in 112 stands were used to establish the allometric function between MB and MA across 16 chronosequence studies. Like other studies (e.g. Cairns et al. 1997; Mokany, Raison & Prokushkin 2006; Yang et al. 2009, 2010), a power function was used to fit the relationship between MB and MA. To test whether MB dynamics during stand development could be reliably estimated from MA using the allometric function generated over space, the power function between MB and MA generated across sites was then used to predict MB from MA within site. All statistical analyses, with the exception of RMA regression, were conducted using the software package R (R Development Core Team 2007).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Both MA and MB increased with stand age and then levelled off across sites (Fig. 1a,b). These changes could be well fitted by logarithmic functions of MA = 98.8 log (age) - 93.1 and MB = 23.5 log (age) - 18.5, respectively. The slopes of MA vs. stand age and MB vs. stand age were significantly larger than zero in most individual studies (< 0.05) (Fig. 1c,d), showing that both MA and MB exhibited increasing trends during stand development. However, R:S ratio did not show any significant trend with stand age across sites (r= 0.01, = 0.35) (Fig. 2a). Moreover, the trend was still not significant even if we excluded the three data points where stand age was over 150 years (> 0.05). Likewise, no significant trend was observed for the R:S ratio dynamics in 10 out of 16 individual studies (Fig. 2b), indicating that the R:S ratio also most probably remained relatively constant during stand development. Nevertheless, the R:S ratio declined with stand age in five sequences while it increased in one sequence (see Fig. S1).

image

Figure 1.  Changes in aboveground biomass (MA) and belowground biomass (MB) with stand age across site (a–b) and within site (c–d). The x-axis in panel c–d is the slope of the relationship between target variables (i.e. MA and MB) and stand age. The y-axis in panel (c–d), site number, denotes various individual studies used in this study as shown in Table S1. The dashed line in panel (c–d) denotes the constant slope (equal to zero). If zero is covered by the 95% confidence interval of the slope, the target variable does not change significantly over time. The slope is ordered to clearly illustrate whether it is significantly different from zero. The inset in panel c–d shows the percentage of published studies exhibiting various patterns of MA and MB over time. Not: no trend, Inc: increase.

Download figure to PowerPoint

image

Figure 2.  Changes in root:shoot (R:S) ratio with stand age across site (a) and within site (b). The change in R:S ratio over stand age within site is illustrated by the percentage of published studies exhibiting various patterns of R:S ratio over time. Inc: increase, Not: no trend, Dec: decrease.

Download figure to PowerPoint

Reduced major axis (RMA) analysis indicated that MA scaled isometrically with MB over space (Fig. 3a). Specifically, the scaling slope of the relationship between MA and MB across sites was estimated at 1.01, with 95% confidence intervals of 0.94–1.09 (Fig. 3a), indicating that the scaling exponent was not significantly different from 1.0 over space. Likewise, the slope of the MAMB relationship was not statistically different from 1.0 in 10 out of 16 individual studies (> 0.05) (Fig. 3b), consistent with the isometric pattern observed over space (Fig. 3a). Of those studies inconsistent with the isometric pattern, the scaling slope was observed to be significantly larger than 1.0 in five sequences while it was lower than 1.0 in one sequence (Fig. 3b). However, the scaling slope did not exhibit any significant trend with climatic factors (MAT: r= 0.07, = 0.33; MAP: r= 0.08, = 0.30) or stand age (r= 0.01, = 0.94) (Fig. 4). Conversely, the ordinary least squares (OLS) relationship between MB and MA across sites could be well characterized by a power function of M_B = 0.38M_A ^{0.92} (r= 0.86, < 0.001) (Fig. 5a). Interestingly, the MB estimated from MA using this power function accorded well with measured values, with a mean relative difference of 9.6% (Fig. 5b).

image

Figure 3.  Reduced major axis (RMA) regression between MA and MB across site (a) and within site (b). The x-axis in panel (b), the scaling slope, indicates the slope of the log–log relationship between MA and MB. The y-axis in panel b, site number, denotes various individual studies used in this study as shown in Table S1. The dashed line in panel b denotes the constant slope (equal to 1.0) predicted by the isometric hypothesis. If 1.0 is covered by the 95% confidence interval of the slope, biomass partitioning between MA and MB follows the isometric prediction. The inset in panel b shows the frequency distribution of the scaling slope within site.

Download figure to PowerPoint

image

Figure 4.  Relationships of the slope of MAMB relationship with mean annual temperature (a), mean annual precipitation (b) and mean stand age over each sequence (c).

Download figure to PowerPoint

image

Figure 5.  Ordinary least squares (OLS) regression between MB and MA across site (a), and frequency distribution of the relative difference between predicted and measured MB during stand development (b). MB within site is predicted from MA using the MBMA relationship generated across site. The dashed line denotes the relative difference between predicted and measured MB is equal to zero. The solid curve is a Gaussian distribution fitted to the frequency data.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The R:S ratio did not show significant changes with stand age over space (Fig. 2a) or over time at most sites (Fig. 2b). These results differed from those obtained by Mokany, Raison & Prokushkin (2006), who found that the R:S ratio exhibited a decreasing trend with stand age across forests and woodlands from around the world (r= 0.05, = 0.003). Our results also differed from those of Wang, Fang & Zhu (2008), who observed that the R:S ratio decreased with stand age in forests of northeastern China (r= 0.04, < 0.05). However, our results concurred with an earlier analysis by Cairns et al. (1997) which demonstrated that the R:S ratio did not change with stand age across forest ecosystems from around the world. Taken together, these comparisons suggest that stand age alone cannot explain a large proportion of variation in the R:S ratio across a broad geographic gradient (Cairns et al. 1997; Mokany, Raison & Prokushkin 2006; Wang, Fang & Zhu 2008) or a long-term age gradient.

Though R:S ratio was relatively stable across stand age when we examined it across sites, it is possible that our analysis hid small differences that existed due to forest types. To test this possibility, we examined trends in R:S ratio over time in temperate and tropical forests separately. Our results demonstrated that R:S ratio did not exhibit any significant trend with stand age in temperate forests (r= 0.01, = 0.63) or tropical forests (r= 0.01, = 0.74) (see Fig. S2 in Supporting information). These results were consistent with previous observations by Cairns et al. (1997), who reported no significant change of R:S ratio with stand age in tropical, temperate or boreal forests.

The slope of the relationship between MA and MB across various individual studies was not significantly different from 1.0 (Fig. 3a), supporting the isometric hypothesis (Enquist & Niklas 2002; Niklas 2005; Cheng & Niklas 2007). Moreover, the scaling slope did not reveal a significant difference from 1.0 in 10 out of 16 individual studies (Fig. 3b) and did not exhibit any significant change with environmental factors (Fig. 4). These results also accorded well with the isometric prediction (Enquist & Niklas 2002; Niklas 2005; Cheng & Niklas 2007). The isometric MAMB relationship observed in this study concurred with previous observations across various forest types (Cheng & Niklas 2007) and grassland biomes (Yang et al. 2009, 2010). Specifically, Cheng & Niklas (2007) demonstrated that the scaling slope of the relationship between MA and MB was indistinguishable from 1.0 in most forest types of China. Likewise, Yang et al. (2009) reported that MA scaled isometrically with respect to MB in the Tibetan alpine grasslands. Moreover, the isometric pattern observed in the Tibetan alpine grasslands also occurred in other grassland types of northern China (Yang et al. 2010). More interestingly, the isometric relationship did not differ significantly between temperate and alpine grasslands or between steppe and meadow (Yang et al. 2010). These isometric patterns between MA and MB observed in various ecosystems suggest that plants invest their annual growth proportionally above- and belowground with increasing body size (Enquist & Niklas 2002).

Theoretically, the isometric relationship between MA and MB during stand development is consistent with the relatively constant R:S ratio observed over age sequences (McConnaughay & Coleman 1999). Then, an interesting question arises why the R:S ratio does not show significant changes with stand age in most sequences? First, the non-significant change of the R:S ratio may be due to diverse patterns of biomass partitioning among various individual tree species. For instance, McConnaughay & Coleman (1999) reported that biomass allocation patterns in various plant species exhibited diverse ontogenetic drifts, including increasing, decreasing and non-significant change of the R:S ratio with increases in plant age. Consequently, the diverse patterns of biomass partitioning among different individual species over time may lead to a non-significant change in the R:S ratio in forest ecosystems during stand development. Second, the non-significant change of the R:S ratio may be driven by changes in nutrient availability over age sequence. Nutrient availability in mineral soils is usually high at the beginning of stand development due to accelerated mineralization after a stand-replacing disturbance (Covington 1981; Zak et al. 1990), but may decrease with stand age following forest regrowth (Kutsch et al. 2009). The decrease of nutrient availability may stimulate biomass partitioning in favour of MB (Bloom, Chapin & Mooney 1985; Chapin et al. 1987; Shipley & Meziane 2002), thus obscuring the ontogenetic decrease of the R:S ratio over time. Third, the non-significant change in the R:S ratio may also be ascribed to the altered species composition over time. During stand development, forest ecosystems usually experience significant changes in species composition (Wirth & Lichstein 2009). It is well known that different species may have different R:S ratios (Bazzaz & Grace 1997). Therefore, changes in species composition over time may also obscure the pattern of the R:S ratio dynamics with stand age.

In summary, this study examined biomass partitioning patterns in forest ecosystems over space and time by synthesizing biomass measurements made in 112 stands from 16 age sequences. Our results indicated that the R:S ratio did not exhibit any significant change with stand age both across 16 chronosequence studies (a test of site and age effects) and within 10 out of 16 individual studies (a test of age effect alone). Moreover, MA scaled isometrically with MB both across various individual studies and within 10 out of 16 individual studies. These results demonstrated that biomass partitioning patterns over time were similar to those observed over space, suggesting that MB dynamics during stand development may be reliably estimated from MA using the allocation function generated over space. The results observed in this study, together with those of previous studies (Enquist & Niklas 2002; Cheng & Niklas 2007; Yang et al. 2009, 2010) provide a framework for estimating spatial patterns and temporal dynamics of MB from MA in terrestrial ecosystems.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We are grateful to Dr Robert Jones and three anonymous reviewers for their thoughtful comments on earlier drafts of the manuscript. We also thank Dr Lindsay Haddon, Rebecca Sherry and Christina Schädel for checking language. This study was financially supported by the US National Science Foundation (NSF) under DEB 0743778, DEB 0840964, DBI 0850290 and EPS 0919466; by the Office of Science (BER), Department of Energy, Grants No.: DE-FG02-006ER64319 and through the Midwestern Regional Center of the National Institute for Climatic Change Research at Michigan Technological University, under Award Number DE-FC02-06ER64158.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Appendix S1. A list of papers from which data were extracted for this study.

Table S1. Characteristics of the 16 age sequences collected in this study.

Figure S1. Changes in root:shoot (R:S) ratio with stand age in six forest ecosystems.

Figure S2. Relationships between root:shoot (R:S) ratio and stand age in temperate and tropical forests.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

FilenameFormatSizeDescription
JEC_1774_sm_appS1.doc35KSupporting info item
JEC_1774_sm_figS1.doc356KSupporting info item
JEC_1774_sm_figS2.doc216KSupporting info item
JEC_1774_sm_tabS1.doc61KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.