Abstract
Asymmetric competition for light may depress the growth rates (GRs) to different extents for different-sized tree individuals. Various responses of different functional groups to light availability result that tree individuals of different functional groups may experience different competition intensities, e.g. canopy and deciduous species grow faster and demand more light than understory and evergreen species. In this study, we estimated the effects of asymmetric competition for light using individual GRs and explored the effects of asymmetric competition on growth among different functional groups (e.g. canopy vs. understory species and deciduous vs. evergreen species).
We measured growth in circumference to determine the radial increments of a total of 2233 stems with diameter at breast height ≥ 5.0 cm in a permanent plot (140 × 80 m2) located in a typical evergreen and deciduous broadleaved mixed forest on Mt Shennongjia, China. All of the measurements were carried out at ~6-month intervals every April and October from 2012 to 2014, and biomass of each individual was calculated based on its diameter and species-specific allometry. We then calculated GRs of annual biomass growth (growth between October and the next October). Considering the hypothesis that asymmetric competition for light among trees of different sizes may result in a steeper allometric growth curve with increasing tree size, we further divided the sampled trees into different subsets according to their height, at intervals of 1 m, and then fitted the scaling relationship between the logarithm of the biomass GR (logGR) and the logarithm of diameter (logD) for each height class using standardized major axis regression. Finally, we used simple linear regression to test whether the scaling exponent was related to tree height. The above analyses were conducted for the annual growth of all tree species, canopy species, understory vs. treelets species and deciduous vs. evergreen species.
We observed a concave curve for the relationship between logGR and logD with an increase in the scaling exponent between logGR and logD with increasing tree height. This pattern held for the annual growth of canopy species and deciduous species but not for the annual growth of understory species, treelets or evergreen species. These results suggest that asymmetric competition for light is more important in regulating the GRs of the fast-growing species, such as canopy species and deciduous species, than those of shade-tolerant species, such as understory species, treelets and evergreen species.
INTRODUCTION
Tree height is one of the leading dimensions of ecological strategy (Falster and Westoby 2003; Kunstler et al. 2016; Westoby et al. 2002). Growing tall is costly for tree individuals; they must first invest in support structures and must continue to devote resources toward the maintenance of vasculature (Midgley 2003; Ryan and Yoder 1997; Westoby et al. 2002). However, being taller than neighbors gives a tree a competitive advantage in access to light (Coomes et al. 2011; Rüger et al. 2011). This difference in the ability to access light is called asymmetric competition for light. In general, species with a tall potential height tend to grow tall to win the lion’s share of light owing to their low tolerance to shade (Kunstler et al. 2016); when a species is unable to grow tall, high shade tolerance is an alternative strategy (Gommers et al. 2013). Therefore, competition for light is key in understanding strategies related to tree height (Falster and Westoby 2003). More specifically, tree species benefit more from growth in terms of height when they are more easily affected by asymmetric competition for light.
Growth is an important process representing the adaptation of an individual to its environmental conditions and the intensity of species interactions (Chi et al. 2015). This variable is widely used to test different competition models. For example, von Oheimb et al. (2011) found that growth in tree diameter was a function of local neighborhood size–asymmetric competition in a secondary evergreen broadleaved forest in eastern China. Within a species, the growth rate (GR) of a tree individual may increase with tree size (Chi et al. 2015, 2017; Coomes and Allen 2007; Coomes et al. 2011; Stephenson et al. 2014). The metabolic scaling theory (MST) proposes that the mass GR of an individual is proportional to the whole-tree sap flow (Enquist et al. 1999; Sperry et al. 2012), which is positively related to its stem basal area (Savage et al. 2010; West et al. 1999). However, other factors along with size, such as light and nutrient availability, functional traits and neighborhood interactions (e.g. Baker et al. 2003; Cavard et al. 2011; Feeley et al. 2007; Hahn et al. 2017; Li et al. 2017; Rüger et al. 2011; Scholten et al. 2017), also influence tree GRs; the relationship between GR and tree size cannot be simply described by a universal law under different conditions (Coomes et al. 2011; Muller-Landau et al. 2006; Russo et al. 2007; Rüger and Condit 2012). In closed forests, the observed GR may vary from the predicted GR because of asymmetric competition for light among different-sized trees (Coomes et al. 2011; Rüger and Condit 2012; Russo et al. 2007). Therefore, Coomes et al. (2011) hypothesized that asymmetric competition for light may result in a steeper allometric growth curve with tree size, i.e. the scaling exponents between the logarithm of the absolute biomass GR (logGR) and the logarithm of diameter (logD) increase with size rather than a constant, as the MST proposes. This mechanism provides a new quantitative method for estimating asymmetric competition for light, in which a steeper allometric growth concave curve (scaling exponent increases with tree size) indicates stronger asymmetric competitions for light among trees of different sizes. This idea was indirectly supported by the positive relationship between scaling exponent and productivity (Coomes et al. 2011). Given that tree height scales with tree size (Muller-Landau et al. 2006; West et al. 1999) and taller trees experience brighter environments and may absorb more light (Poorter et al. 2005; Rüger et al. 2011), it is reasonable to predict that the scaling exponents between logGR and logD increase with tree height and that stronger relationships between scaling exponents and tree height indicate stronger asymmetric competition for light.
Different species respond to light availability in distinct ways because of their varied tolerance to shade (Kelly et al. 2009; Lusk et al. 2008; Onoda et al. 2014). The varied tolerance to shade may result in different effects of light competition, suggesting that the benefits of tree height growth vary among different species with different shade tolerance. For example, canopy trees have a higher photosynthetic capacity than understory trees, even compared with similar sized saplings exposed to similar light environments (Thomas and Bazzaz 1999). The photosynthetic capacity of canopy species may be more easily limited by light availability than that of understory species. Therefore, it is necessary to take functional groups into consideration in the study of asymmetric competition for light in multispecies communities. moreover, in a global study of competition, Kunstler et al. (2016) found that tree species with a lower potential tree height are more tolerant to competition, suggesting that understory species are less affected by asymmetric competition for light than canopy species. As a result, the scaling exponents between the logGR and logD of canopy species would be more strongly correlated with tree height than those of understory species.
The leaf life span may also influence asymmetric competition for light. For example, the leaf life span is often longer in shaded than in sunlit individuals (Ackerly and Bazzaz 1995; Reich et al. 2004). Besides, species with longer leaf life span always have a lower photosynthetic rate (Reich et al. 1992), suggesting the photosynthetic capacity of deciduous species may be more easily limited by light availability than that of evergreen species. Moreover, studies in evergreen forests suggest that the leaf mass per area (LMA, the inverse of the specific leaf area) is higher in sunlight than in shaded individuals of the same species, but at the species level, less sunlight benefits shade-tolerant evergreen species, which have higher LMAs than light-demanding evergreen species under the same conditions (Lusk et al. 2008). In deciduous forests, the LMA increases with light supply both at the individual and at the species level (Janse-Ten Klooster et al. 2007; Lusk and Warton 2007). Because the LMA is positively correlated with GR (Chaturvedi et al. 2011; Poorter et al. 2008;), it is reasonable to predict that the growth of deciduous species would be more easily affected by asymmetric competition for light than that of evergreen trees.
In this study, we focus on evergreen and deciduous broadleaved mixed subtropical forests and explore how asymmetric competition for light shapes tree growth. Because tall trees may benefit from their ability to attain prior access to light, asymmetric competition for light among different-sized trees may result in a concave growth curve (Coomes et al. 2011). We therefore hypothesize that the scaling exponent between logGR and logD increases as tree height increases (H1); that canopy species are more strongly affected by asymmetric competition for light than understory species, and that the scaling exponent of canopy species therefore exhibit a more significant trend with tree height than those of understory species (H2); and that deciduous species are more strongly affected by asymmetric competition for light than evergreen species, and that the scaling exponents of deciduous species therefore exhibit a more significant trend with tree height than those of evergreen species (H3).
MATERIALS AND METHODS
Study site
This study was conducted in a permanent plot (140 × 80 m2) on Mt Shennongjia (31°19′4″N, 110°29′44″E), Hubei Province, China. Located at the elevation of 1650–1750 m, the plot has an average slope of 30° with an annual precipitation of 1330 mm and an annual mean temperature of 10.6°C. The vegetation is a typical subtropical evergreen and deciduous broadleaved mixed forest, dominated by a deciduous broadleaved species, Fagus engleriana, and an evergreen broadleaved species, Cyclobalanopsis multinervis (Ge et al. 2013), together with another 47 species in the tree layer. The soil type is montane yellow brown soil (Ge et al. 2013). This forest is characterized by a total stem basal area of 37.69 m2 ha−1, a mean stem diameter at breast height (DBH) of 13.2 cm, and a density of 1347 ha−1, with 71% of the stems belonging to late-successional species as defined by Ge et al. (2013).
Data collection
We installed steel dendrometer bands at breast height for all 2233 living stems with DBH ≥ 5.0 cm in the summer of 2011. We measured the changes in the gaps of the dendrometer bands (changes in tree circumference) using digital calipers (precision of ± 0.01 mm). These measurements were carried out every 6 months in April and October from 2012 to 2014. The diameter increments were further calculated by dividing the circumference growth by ‘pi’. Before these measurements, we measured the initial DBH of each stem with flexible tape for the measurement of DBH and the height of each stem using radar distance measurement (Vertex IV, Haglöf Sweden AB) in October 2011. The growth between April and October was defined as summer growth and that between October and the next April was defined as winter growth; the sum of the summer and winter growth made up the annual growth. See Chi et al. (2015) for more detailed information.
Defining functional groups
In this study, all stems were divided into different functional groups according to their canopy stature and leaf habit (Table 1). All species were grouped into treelets (8 species, 113 individuals), understory (14 species, 1258 individuals) or canopy species (27 species, 862 individuals) according to their canopy statures (Chi et al. 2015). In particular, the canopy stature categorization was based on the (approximately) largest stem of each species (Chi et al. 2015; King et al. 2006). The 95th percentile of the DBH of all stems ≥0.1 × DBHmax (the largest DBH) for a particular species was defined as D950.1, which was independent of the species abundance and highly correlated with its maximum height (King et al. 2006). Species with D950.1 ≥ 5 cm but D950.1 < 12 cm, D950.1 ≥ 12 cm but D950.1 < 25 cm or D950.1 ≥ 25 cm were defined as treelets, understory and canopy species, respectively (Chi et al. 2015; King et al. 2006). See Chi et al. (2015) for more detailed information on the classification of canopy stature. All species were also divided into evergreen (34 species, 1106 individuals) or deciduous species (15 species, 1127 individuals) according to leaf habit. The canopy stature and leaf habit of each species are also available in appendix S1 of Chi et al. (2015).
description of functional groups
| Functional groups | No. of species | No. of stems | DBH (cm) | Height (m) | ||||
|---|---|---|---|---|---|---|---|---|
| Min | Max | Mean | Min | Max | Mean | |||
| All species | 49 | 2233 | 4.10 | 72.35 | 13.15 | 2.00 | 38.00 | 11.77 |
| Canopy | 27 | 862 | 4.22 | 72.35 | 18.33 | 5.00 | 38.00 | 15.68 |
| Understory | 14 | 1258 | 4.12 | 32.17 | 10.20 | 2.00 | 25.50 | 9.51 |
| Treelets | 8 | 113 | 4.10 | 12.97 | 6.54 | 3.50 | 18.00 | 7.19 |
| Evergreen | 34 | 1106 | 4.10 | 40.49 | 10.35 | 2.00 | 30.40 | 9.06 |
| Deciduous | 15 | 1127 | 4.12 | 72.35 | 15.91 | 5.00 | 38.00 | 14.43 |
| Functional groups | No. of species | No. of stems | DBH (cm) | Height (m) | ||||
|---|---|---|---|---|---|---|---|---|
| Min | Max | Mean | Min | Max | Mean | |||
| All species | 49 | 2233 | 4.10 | 72.35 | 13.15 | 2.00 | 38.00 | 11.77 |
| Canopy | 27 | 862 | 4.22 | 72.35 | 18.33 | 5.00 | 38.00 | 15.68 |
| Understory | 14 | 1258 | 4.12 | 32.17 | 10.20 | 2.00 | 25.50 | 9.51 |
| Treelets | 8 | 113 | 4.10 | 12.97 | 6.54 | 3.50 | 18.00 | 7.19 |
| Evergreen | 34 | 1106 | 4.10 | 40.49 | 10.35 | 2.00 | 30.40 | 9.06 |
| Deciduous | 15 | 1127 | 4.12 | 72.35 | 15.91 | 5.00 | 38.00 | 14.43 |
description of functional groups
| Functional groups | No. of species | No. of stems | DBH (cm) | Height (m) | ||||
|---|---|---|---|---|---|---|---|---|
| Min | Max | Mean | Min | Max | Mean | |||
| All species | 49 | 2233 | 4.10 | 72.35 | 13.15 | 2.00 | 38.00 | 11.77 |
| Canopy | 27 | 862 | 4.22 | 72.35 | 18.33 | 5.00 | 38.00 | 15.68 |
| Understory | 14 | 1258 | 4.12 | 32.17 | 10.20 | 2.00 | 25.50 | 9.51 |
| Treelets | 8 | 113 | 4.10 | 12.97 | 6.54 | 3.50 | 18.00 | 7.19 |
| Evergreen | 34 | 1106 | 4.10 | 40.49 | 10.35 | 2.00 | 30.40 | 9.06 |
| Deciduous | 15 | 1127 | 4.12 | 72.35 | 15.91 | 5.00 | 38.00 | 14.43 |
| Functional groups | No. of species | No. of stems | DBH (cm) | Height (m) | ||||
|---|---|---|---|---|---|---|---|---|
| Min | Max | Mean | Min | Max | Mean | |||
| All species | 49 | 2233 | 4.10 | 72.35 | 13.15 | 2.00 | 38.00 | 11.77 |
| Canopy | 27 | 862 | 4.22 | 72.35 | 18.33 | 5.00 | 38.00 | 15.68 |
| Understory | 14 | 1258 | 4.12 | 32.17 | 10.20 | 2.00 | 25.50 | 9.51 |
| Treelets | 8 | 113 | 4.10 | 12.97 | 6.54 | 3.50 | 18.00 | 7.19 |
| Evergreen | 34 | 1106 | 4.10 | 40.49 | 10.35 | 2.00 | 30.40 | 9.06 |
| Deciduous | 15 | 1127 | 4.12 | 72.35 | 15.91 | 5.00 | 38.00 | 14.43 |
Data analysis
The biomass of each stem was first calculated based on diameter, with different scaling relationships used for different species according to Wang et al. (2007). Combined with diameter increments, the biomass increment (dM = Mt – M0), where M0 is the beginning biomass calculated using the initial diameter and Mt is the biomass at the time of measurement using the diameter plus the diameter increment. And the absolute biomass GR (GRm = dM/dt) of each stem were calculated. Based on the diameter increment used at different periods, the summer (GRms) and winter (GRmw) biomass growth were calculated: GRms = (MO – MA)/dt, where MO (MA) is the biomass in October (April) in the same year; GRmw = (MA – MO)/dt, where MO is the biomass in October, while MA is the biomass in the next April. By adding both together, we obtained the annual biomass growth (GRma). All GRs were calculated as average GRs across years.
Asymmetric competition may result in a positive correlation between the scaling exponent of logGRma and logD and tree height (Coomes et al. 2011). To test this idea (H1), we fitted the relationship between logGRma and logD with a locally weighted polynomial regression (LOWESS). To describe this concave shape quantitatively, we further divided the trees into different height classes using an interval of 1 m (e.g. the tree height class of 5 m contained trees with heights from 5 to 5 + 1 m) and then fitted the logGRma ~ logD for each height class with a standardized major axis (SMA) regression. The significant scaling exponents of the SMA regression were selected for subsequent analysis. Finally, we used a simple linear regression analysis to test whether the scaling exponent of the logGRma ~ logD (the scaling exponents of the SMA regression for each height class) significantly changes with the tree height class. An increase in the scaling exponent with tree height indicates asymmetric competition for light favoring larger stems, whereas no trends indicate no significant asymmetric competition for light.
All the above analyses were conducted using R 3.0.3 (http://www.R-project.org/). Analyses were conducted for the annual growth of all tree species, as well as canopy species, understory and treelets species, and deciduous and evergreen species. Moreover, we also analyzed the data using an interval of 0.5 m and 2 m for all those mentioned functional groups (see supplementary Figs S1–S3).
RESULTS AND DISCUSSION
The LOWESS regression fitted a concave curve for the relationship between logGRma and logD (Fig. 1), providing empirical support for the inclusion of asymmetric competition into models of tree growth advocated by Coomes et al. (2011). Furthermore, this result also showed asymmetric competition for light contributed to the discrepancy between the observed scaling exponent and the prediction, which was consistent with previous studies (Coomes et al. 2011; Rüger and Condit 2012). Specifically, consistent with our first hypothesis (H1), the scaling exponent of the logGRma ~ logD relationship for all the sampled trees increased with tree height (Fig. 2A, R2 = 0.53, P < 0.001, 17 height classes). This result suggested that taller trees held an advantage over neighbors in improving whole-tree photosynthetic because of their ability to attain prior access to light (Coomes et al. 2011; Rüger et al. 2011).
relationships between the annual growth in biomass and the initial DBH for all trees using LOWESS. Black dashed line shows the theoretical scaling exponent (slope = 2) according to the metabolic scaling theory (MST).
relationships between tree height and the scaling exponents between initial DBH and annual growth of (A) overall, (B) canopy, (C) understory and (D) treelets species. Each point represents a height class.
Taking the canopy stature into consideration, we found that a similar pattern held true for canopy species (Fig. 2B, R2 = 0.79, P < 0.001, 16 height classes) but not for understory species (Fig. 2C, R2 = 0.01, P = 0.78, 15 height classes) or treelets (Fig. 2D, R2 = 0.25, P = 0.50, 4 height classes). These results were consistent with our second hypothesis (H2), indicating that asymmetric competition for light played a more important role in regulating the growth of canopy species than that of understory species or treelets. The tree species in forests can be arranged along an axis between two extreme strategies: fast growth with high light requirements and slow growth with high shade tolerance (Grime 1977; Janse-Ten Klooster et al. 2007; Kunstler et al. 2016). Owing to their tall heights and high light exposure, canopy species grow faster than understory species and treelets in this area (mean diameter GR: 0.13 vs. 0.04 vs. 0.03 cm year−1, respectively, Chi et al. 2015). As a consequence of the decrease in light use efficiency with tree height (Onoda et al. 2014) and the fast growth strategy demonstrated by canopy species, it is reasonable to infer that canopy species demand more light than understory species and treelets. In general, light-demanding species are more sensitive in their response to changes in light availability (Kelly et al. 2009; Turnbull et al. 1993). For example, the light-saturated rates of photosynthesis (Amax) in an early successional species, Eucalyptus grandis, increased by 73%, while a late-successional species, Flindersia brayleyana, only increased by 36% when light levels increased from 30% to 60% of full sunlight (Kelly et al. 2009). Therefore, canopy species are more strongly affected by asymmetric competition for light and benefit more from a taller height than understory species and treelets.
Taking leaf habit into consideration, we found that the scaling exponent of the logGRma ~ logD relationship for deciduous species was positively related to tree height (Fig. 3A, R2 = 0.31, P = 0.02, 16 height classes), while that of evergreen species showed no significant relationship (Fig. 3B, R2 = 0.17, p = 0.12, 16 height classes), which is consistent with our third hypothesis (H3), suggesting that the growth of deciduous species was more easily influenced by asymmetric competition for light than that of evergreen species. One potential explanation is evergreen species, especially evergreen understory trees, might experience opposite light conditions during winter and during the growing season (spring, summer and fall) in seasonally deciduous forests, (Miyazawa and Kikuzawa 2005; Valladares and Niinemets 2008). Specifically, understory evergreen trees receive more solar radiation during winter after the fall of deciduous leaves. For example, in warm temperate deciduous forests, the carbon accumulated by evergreen (shade-tolerate) species from autumn to spring accounts for more than half of the annual carbon sequestration (Miyazawa and Kikuzawa 2005). Even for deciduous species, extra solar radiation due to earlier bud-burst contributes up to 30–60% of total light intercepted during the growth period (Augspurger et al. 2005). All of these lines of evidence suggest that light availability affects tree growth differently across seasons, especially for evergreen species. Accordingly, we divided the annual growth of evergreen species into summer and winter growth. As expected, we found that the scaling exponent of the logGRms ~ logD relationship had a marginal significant positive relationship with tree height (Fig. 3C, R2 = 0.24, P = 0.05, 16 height classes), while that for winter growth showed no significant relationship (Fig. 3D, R2 = 0.05, P = 0.58, 9 height classes). These results indicated a stronger effect of asymmetric competition in summer than in winter, partly because the studied forest community was more closed in summer than in winter owing to the leaf fall of deciduous trees in winter, e.g. Peng et al. (2017) found leaf area index increased significantly from April to August. Asymmetric competition for light is thought to be stronger in more closed forests (Coomes et al. 2011). The differences in the asymmetric competition for light in evergreen species in summer and winter helped us to understand why asymmetric competition for light had no significant impact on the annual growth of evergreen trees.
relationships between tree height and the scaling exponents between initial DBH and annual growth for (A) deciduous and (B) evergreen species and (C) summer growth and (D) winter growth of evergreen species. Each point represents a height class.
Compared with the significant relationship between the scaling exponents of the annual growth of deciduous trees and tree height, the scaling exponents of evergreen summer growth were marginally correlated with tree height, which suggests asymmetric competition for light had a weak effect on the summer growth of evergreen individuals, at least less than that on the growth of deciduous trees. Consistent with the former idea that fast-growing species demand more light, deciduous species grew faster (0.11 cm year−1) and had higher light requirements than evergreen species (0.04 cm year−1) (Chi et al. 2015). Moreover, this also agreed with the idea that differences in light better explain the variance in deciduous species than in evergreen species in terms of the net assimilation rate (Niinemets et al. 2015). Fast-growing species generally had a low wood density but a high specific leaf area (SLA), whereas the wood density was thought to be positively related to the better tolerance of competition (Falster and Westoby 2003; Kunstler et al. 2016). In this community, deciduous species was found to have a higher SLA (226.85 cm2 g−1) and a lower wood density (0.57 g cm−3) than evergreen species (111.34 cm2 g−1 and 0.65 g cm−3, respectively) (Q. Guo et al., unpublished data), which provided another explanation for the pattern of stronger asymmetric competition for light in deciduous species than in evergreen species.
In summary, using the relationship between the scaling exponents of MST within each height interval and tree height, we found an increase scaling exponent of logGRma ~ logD with tree height. This pattern held true for the annual growth of canopy species but not for understory species and treelets; it also held true for deciduous but not for evergreen species. Considering seasonal growth, this pattern also held for the summer growth of evergreen trees but not for winter growth. These results suggest that asymmetric competition for light is more important in regulating GRs of fast-growing and light-requiring species (such as canopy species and deciduous species) than those of the shade-tolerant species (such as understory species, treelets and evergreen trees) and is stronger in summer than in winter. Our findings directly show that the effects of asymmetric competition for light are closely related to life-history strategies. The strong correlation between scaling exponent and tree height among canopy species in particular suggests that greater effects of competition for light generally push species to grow taller. However, our study is limited in that it only considered functional groups. Future studies should conduct comparisons at the species level and explore the potential relationship between the sensitivity to tree height of each species and functional traits, such as wood density and potential maximum tree height.
FUNDING
This work was partly supported by the National Program on Key Basic Research Project (grant no 2014CB954004) and the National Natural Science Foundation of China (grant nos 31321061, 31470486).
ACKNOWLEDGEMENTS
We are grateful to the National Field Station for Forest Ecosystem in Shennongjia, Hubei, for help in the field work.
REFERENCES
Author notes
*Correspondence address. Department of Ecology, College of Urban and Environmental Sciences and Key Laboratory of Earth Surface Processes of the Ministry of Education, Peking University, 5 Yiheyuan Road, Beijing 100871, China. Tel: +86-10-62-75-40-39; E-mail: zytang@urban.pku.edu.cn
