Review

Tansley reviews A worldwide analysis of within-canopy variations in leaf structural, chemical and physiological traits across plant functional types Author for correspondence: € Niinemets Ulo Tel: +372 53457189 Email: [email protected] Received: 1 May 2014 Accepted: 4 September 2014

€ Niinemets1,2, Trevor F. Keenan3 and Lea Hallik1,4 Ulo 1

Estonian University of Life Sciences, Kreutzwaldi 1, 51014 Tartu, Estonia; 2Estonian Academy of Sciences, Kohtu 6, 10130 Tallinn,

Estonia; 3Department of Biological Sciences, Macquarie University, North Ryde, NSW 2109, Australia; 4Tartu Observatory, T~oravere 61602, Estonia

Contents Summary

973

I.

Introduction

973

II.

Defining the structural, chemical and partitioning controls on foliage photosynthetic potentials

V. VI.

IV.

981

Trait scaling with light, plasticity and quantitative limitations

982

Conclusions: the economics spectrum for the within-canopy plasticity

989

Acknowledgements

989

References

990

974 VII.

III.

Global variation in leaf characteristics through the canopies

Construction of a global database on within-canopy variation in leaf structural, chemical and photosynthetic characteristics

976

Methodology for analysis of within-canopy leaf trait variation: concepts and standardizations

978

Summary Article reproduced from New Phytologist (2015) 205: 973–993 doi: 10.1111/nph.13096

Key words: acclimation, economics spectrum, leaf structure, light gradients, nitrogen (N) allocation, nitrogen content, photosynthetic capacity, plasticity.

Extensive within-canopy light gradients importantly affect the photosynthetic productivity of leaves in different canopy positions and lead to light-dependent increases in foliage photosynthetic capacity per area (AA). However, the controls on AA variations by changes in underlying traits are poorly known. We constructed an unprecedented worldwide database including 831 within-canopy gradients with standardized light estimates for 304 species belonging to major vascular plant functional types, and analyzed within-canopy variations in 12 key foliage structural, chemical and physiological traits by quantitative separation of the contributions of different traits to photosynthetic acclimation. Although the light-dependent increase in AA is surprisingly similar in different plant functional types, they differ fundamentally in the share of the controls on AA by constituent traits. Species with high rates of canopy development and leaf turnover, exhibiting highly dynamic light environments, actively change AA by nitrogen reallocation among and partitioning within leaves. By contrast, species with slow leaf turnover exhibit a passive AA acclimation response, primarily determined by the acclimation of leaf structure to growth light. This review emphasizes that different combinations of traits are responsible for within-canopy photosynthetic acclimation in different plant functional types, and solves an old enigma of the role of mass- vs area-based traits in vegetation acclimation.

I. Introduction Variation in daily integrated quantum flux density is the most striking feature of plant canopies. In dense canopies, the gradient Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

in light availability from canopy top to bottom can be > 50-fold (Hirose et al., 1988; Koike et al., 2001; Valladares, 2003; Niinemets & Anten, 2009), whereas relatively open canopies exhibit light gradients as large as 10–20-fold (Hirose et al., 1988; New Phytologist (2015) 205: 973–993 973 www.newphytologist.com

974 Review

New Phytologist

Tansley reviews

Werger & Hirose, 1988; Rambal, 2001; Joffre et al., 2007). Even in free-standing plants, the top foliage shades the leaves below and in the interior of the canopy, leading to strong light gradients (e.g. Le Roux et al., 1999). As a key acclimation response, foliage photosynthetic capacity (the light-saturated net assimilation rate per unit area, AA) increases with increasing long-term light availability within the canopy. Thus, leaves in different canopy positions with different physiological potentials are fine-tuned to their past and contemporary light conditions (Hirose & Werger, 1987a; Ellsworth & Reich, 1993; Anten, 1997; Pons & Anten, 2004; Niinemets & Anten, 2009). As a result of AA acclimation, whole-canopy integrated carbon gain is greater than if AA were constant throughout the canopy (Gutschick & Wiegel, 1988; Baldocchi & Harley, 1995; Lloyd et al., 2010; Dewar et al., 2012). Although the importance of the consideration of within-canopy acclimation of AA in the modeling of vegetation carbon gain has been well recognized at single plant, ecosystem and landscape scales (e.g. Baldocchi & Harley, 1995; Cescatti & Niinemets, 2004; Anten, 2005; Lloyd et al., 2010), recent world-scale modeling efforts have further demonstrated that the incorporation of within-canopy AA gradients can even result in major improvements to world-scale carbon gain estimations by the Earth System models (Bonan et al., 2012). Thus, a detailed understanding of within-canopy photosynthetic acclimation has broad implications for the prediction of vegetation productivity through organismal to global scales. In understanding the within-canopy acclimation responses of AA, it is relevant to consider that a wide range of leaf structural and chemical traits vary between canopy top and bottom, including leaf dry mass per unit area, leaf nitrogen content and nitrogen partitioning among proteins of the photosynthetic machinery (Hirose & Werger, 1987a; Ellsworth & Reich, 1993; Anten, 1997; Pons & Anten, 2004; Niinemets & Anten, 2009). Within-canopy changes in these key functional traits ultimately determine the within-canopy photosynthetic modifications, but there is currently limited understanding of the overall degree of within-canopy plastic modification in AA, of the extent to which different traits control photosynthetic acclimation and how the controlling traits differ among plant functional types. In this review, we first define the functional relationships between the key traits underlying the within-canopy variation in foliage photosynthetic potentials. We then describe the construction of a worldwide database on within-canopy trait variation and the derivation of standardized light availability estimates for all light gradients in the database. Because light is one of the most variable environmental drivers, we deem that standardization of light availability across the studies is absolutely essential for any meta-analysis intending to gain an insight into the light effects on plant functioning. We further use this database to cast light on structural, chemical and allocational controls on AA, and demonstrate a huge variation among traits in their overall light-dependent variability and in their control on AA. This review highlights that variations in trait plasticity, and in their relative controls on AA, are strongly dependent on plant New Phytologist (2015) 205: 973–993 www.newphytologist.com

functional type. In particular, species in the fast return end of the leaf economic spectrum (Wright et al., 2004; Shipley et al., 2006), with high rates of leaf turnover and canopy development, are characterized by modifications in AA as a result of nitrogen reallocation among leaves and changes in nitrogen partitioning within leaves. However, within-canopy acclimation in species in the slow return end of the spectrum, with low rates of leaf turnover and canopy development, is primarily driven by the acclimation of leaf structure to the light environment during leaf growth. Thus, we contend that foliage photosynthetic acclimation to within-canopy light gradients is fundamentally different among plant functional types. There is a debate over which leaf traits provide informative insight into leaf functioning with emphasis on the usefulness of area- vs mass-based expressions of leaf photosynthetic capacity (Lloyd et al., 2013; Westoby et al., 2013; Poorter et al., 2014). We address this debate by arguing that, in order to understand withincanopy photosynthetic acclimation, we must distinguish between the effects of leaf structure, nitrogen content and partitioning on foliage biochemical potentials, and therefore, consider the variations in both mass- and area-based leaf traits.

II. Defining the structural, chemical and partitioning controls on foliage photosynthetic potentials Modifications in leaf structural, chemical and allocational traits are directly related to the maximum potential photosynthetic capacity of leaves per unit leaf area, AA (lmol m
2 s
1). In this section, we review established relationships between different leaf traits and AA. To separate the effects of modifications in the photosynthetic capacity of single leaf cells, and the number of cells per unit leaf area, AA can be expressed as the product of the photosynthetic capacity per unit leaf dry mass, AM (lmol g
1 s
1), and leaf dry mass per unit area, MA (g m
2): A A ¼ A M MA :

Eqn 1

Dry mass per unit area is the key integrated trait characterizing the biomass cost of formation of a unit of leaf area (Poorter et al., 2009) and, as such, is strongly driven by the availability of carbon for the construction of photosynthetic machinery (Gutschick & Wiegel, 1988; Kull & Kruijt, 1999). The photosynthetic capacity of single leaf cells, in turn, depends on both the cellular nitrogen content and the nitrogen investment in photosynthetic machinery, which is commonly characterized by the photosynthetic nitrogen use efficiency (EN (lmol (g N)
1 s
1), that is, the rate of photosynthesis per unit leaf nitrogen content (Hirose & Werger, 1987b, 1994; Hikosaka et al., 1998; Hirose & Bazzaz, 1998; Yasumura et al., 2002; Escudero & Mediavilla, 2003; Pons & Westbeek, 2004): A M ¼ E N NM

Eqn 2

where NM (g g
1) is the leaf nitrogen content per leaf dry mass. Converting to a leaf area basis, and considering that leaf nitrogen content per leaf area, NA (g m
2), is: Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist NA ¼ NM M A

Tansley reviews

Eqn 3

we obtain: A A ¼ EN N M M A ¼ EN N A :

Eqn 4

EN is derived from measurements of photosynthetic rates and nitrogen content, but this does not mean that EN is necessarily dependent on or correlated with photosynthetic rate and/or nitrogen content expressed per unit area or mass. Indeed, classical studies demonstrate that EN can vary widely at a given rate of photosynthesis and a given leaf nitrogen content (Field & Mooney, 1986; Evans, 1989). Although correlative relationships among the traits defined by Eqns 2–4 have been reported (Wright et al., 2004; Westoby et al., 2013), any such correlation would not invalidate the functional dependences of photosynthetic capacity on the stacking of photosynthetic cells per unit leaf area, cellular nitrogen content and/or the allocation of cellular nitrogen to photosynthetic proteins. A key disadvantage with the concept of AA is that, at current ambient CO2 concentrations, it can vary with changes in stomatal diffusion conductance to CO2 (gs,CO2). This is important as the limitation of photosynthesis by gs,CO2 can vary through the canopy, especially during stress conditions (e.g. Aasamaa et al., 2004; Niinemets et al., 2004b). In addition, photosynthetic nitrogen use efficiency does not provide a direct measure of nitrogen investment in rate-limiting components of photosynthesis. These limitations of the use of AA can be elegantly addressed using the C3 plant biochemical photosynthesis model of Farquhar et al. (1980). In particular, two key model parameters, the maximum carboxylase activity of ribulose-1,5-bisphosphate (RubP) carboxylase/ oxygenase (Rubisco) (Vcmax) and the capacity for photosynthetic electron transport rate (Jmax), allow for alterations in foliage biochemical photosynthesis potentials to be analyzed directly without interfering stomatal controls. Vcmax characterizes the capacity of the dark reactions of photosynthesis (CO2 fixation) and Jmax the capacity of the light reactions of photosynthesis (RubP regeneration by ATP and NADPH produced in the photosynthetic electron transport chain in light). In practice, both light and dark reactions take place in light, and light-saturated photosynthesis is driven by both Vcmax and Jmax. Occasionally, a third limitation, the capacity for triose phosphate utilization (Sharkey, 1985), can constrain photosynthesis, but there are few data on within-canopy variation in this trait (for an exception, see Urban et al., 2003). Analogous to AA, Vcmax and Jmax per unit area (Vcmax,A and Jmax,A, both in lmol m
2 s
1) can be expressed as composites of independent variables. As Eqn 1 states, both the area-based potentials can be expressed as the products of MA and mass-based potentials Vcmax, M and Jmax, M (both in lmol g
1 s
1). Vcmax, M can be further expressed as: Vcmax; M ¼ 6:25Vcr jFR NM ;

Eqn 5

where Vcr is the specific activity of Rubisco, that is, the maximum rate of RubP carboxylation per unit Rubisco protein (lmol g
1 s
1), j is the fraction of active Rubisco sites, FR is the Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

Review 975

fraction of leaf nitrogen in Rubisco, NM is the leaf nitrogen content per dry mass (g g
1) and 6.25 (g g
1) is the nitrogen content of Rubisco protein (Niinemets & Tenhunen, 1997). The activation state of Rubisco, j, can vary as a function of environmental characteristics (e.g. Ethier et al., 2006; Yamori et al., 2006), although Vcmax in studies exploring functional leaf trait relationships is characteristically estimated in non-stressed plants with maximum activation state of Rubisco. Jmax,M is further given as: Jmax;M ¼ 8:06Jmc FB NM ;

Eqn 6

where Jmc is the capacity for photosynthetic electron transport per unit cytochrome f, FB is the fraction of nitrogen in rate-limiting proteins of photosynthetic electron transport and the factor 8.06 considers the nitrogen content of rate-limiting proteins and the molar stoichiometry relative to cytochrome f (Niinemets & Tenhunen, 1997). The use of cytochrome f as the key limiting step assumes that the capacity for linear electron transport rate is determined by the amount and turnover rate of electron carriers between photosystems I and II (for a discussion, see Evans, 1988; Evans & Seemann, 1989; Niinemets & Tenhunen, 1997). On an area basis, Vcmax,A is further expressed as: Vcmax;A ¼ 6:25Vcr jFR NM MA ¼ 6:25Vcr jFR NA ;

Eqn 7

and Jmax,A is given analogously. The specific activities, Vcr and Jmc have been widely considered to be conserved for C3 plants (e.g. Niinemets & Tenhunen, 1997; Le Roux et al., 2001; Walcroft et al., 2002; Urban et al., 2003; Rogers, 2014). Although there is evidence of a certain interspecific variability in Vcr (e.g. Galmes et al., 2014a,b), Vcr and Jmc can be considered to be invariable through the canopy of any given species. In principle, the Rubisco activation state could vary through the canopy, but to our knowledge, such a variation has not been studied so far. Considering Vcr and Jmc as constants and assuming that j = 1 (Rubisco is fully active), FR and FB are necessarily the apparent estimates of leaf nitrogen investment in rate-limiting components of photosynthetic machinery. In the absence of independent estimates of protein content in the rate-limiting components of photosynthetic machinery, FB and FR need to be estimated from the measurements of Jmax, Vcmax and nitrogen content, but this does not mean that the estimated nitrogen allocation is autocorrelated with the source variables (Niinemets et al., 1998; Evans & Poorter, 2001; Le Roux et al., 2001; Walcroft et al., 2002). With the outlined assumptions, the functional Eqns 1–7 define altogether 12 key structural, chemical, allocational and photosynthetic traits that make it possible to separate the sources of variation in leaf photosynthetic capacity (six variables, Eqns 1–4) and biochemical photosynthesis potentials (nine variables, Eqns 5–7). The approach based on the model by Farquhar et al. (1980) is more mechanistic, but there are far fewer data available on within-canopy variation in foliage biochemical photosynthesis potentials (Table 1). Nevertheless, the approaches are broadly analogous, sharing structural (MA), nitrogen (NM, NA) and allocation controls (EN vs FR and FB). All of the 12 defined traits can vary through plant canopies, but to our knowledge, there has been no comprehensive New Phytologist (2015) 205: 973–993 www.newphytologist.com

976 Review

New Phytologist

Tansley reviews

Table 1 Number of within-canopy gradients for, and overall variation and light-dependent plasticity of, the studied structural, chemical and physiological traits

Trait1

Number of light gradients

Light-dependent plasticity (fold change for integrated light range of 3–40 mol m
2 d
1)2

Overall variation Min–Max

Structural traits 532 9.15–673.8 MA (g m
2) Mass-based chemical and physiological traits NM (%) 374 0.317–6.54 AM (nmol g
1 s
1) 140 0.0–1155 AM,V (nmol g
1 s
1) 102 5.38–1101 86 36.5–2065 Vcmax,M (nmol g
1 s
1) Jmax,M (nmol g
1 s
1) 84 124–7187 Partitioning traits EN (lmol mol
1 s
1) 158 0–551.8 EN,V (lmol mol
1 s
1) 100 5.53–431.9 FR (g g
1) 87 0.0102–0.7981 FB (g g
1) 87 0.0079–0.125 Combined (area-based) chemical and photosynthetic traits NA (g m
2) 535 0.143–8.93 AA (lmol m
2 s
1) 276
0.19–49.1 AA,V (lmol m
2 s
1) 135 0.495–53.3 113 1.64–176.7 Vmax,A (lmol m
2 s
1) Jmax,A (lmol m
2 s
1) 97 3.42–350.0

Median

Average  SE

n

Min–Max

Median

Average  SE

n

73.8

97.7  0.9

7390

0.839–3.53

1.75

1.834  0.030

292

2.156 100.0 124.1 313.4 1003

2.236  0.012 155.4  4.9 171.7  4.3 405  10 1367  34

5757 1295 1137 825 873

0.686–2.73 0.547–6.00 0.603–6.26 0.599–3.58 0.598–6.23

1.08 1.13 1.22 1.21 1.26

1.156  0.023 1.48  0.13 1.44  0.13 1.33  0.09 1.45  0.16

171 57 54 43 39

81.4 84.4 0.1325 0.0444

104.0  2.1 102.7  1.9 0.1501  0.0032 0.0481  0.0007

1730 1254 951 930

0.533–4.22 0.619–2.71 0.721–2.16 0.732–2.55

1.20 1.13 1.12 1.12

1.39  0.07 1.26  0.06 1.25  0.05 1.27  0.07

65 53 44 39

1.514 8.17 10.69 31.2 96.8

1.635  0.009 10.16  0.15 12.21  0.20 35.9  0.6 106.1  1.7

9231 2780 1501 1159 1013

1.01–4.75 0.784–10.9 1.21–6.33 1.12–6.98 1.30–11.1

1.90 1.99 2.09 2.07 2.10

2.021  0.039 2.44  0.13 2.36  0.12 2.31  0.13 2.53  0.25

268 130 66 57 40

Altogether, 831 within-canopy gradients were included in the analysis (n = 121 for annual forbs, n = 60 for perennial forbs, n = 46 for annual C3 grasses, n = 42 for perennial C3 grasses, n = 12 for C4 grasses, n = 280 for winter-deciduous woody broad-leaved species, n = 15 for winter-deciduous woody needle-leaved species, n = 62 for evergreen tropical and subtropical woody broad-leaved species, n = 84 for evergreen temperate and Mediterranean woody broad-leaved species and n = 110 for evergreen woody needle-leaved species). 1 AA, photosynthetic capacity (light-saturated net assimilation rate at current ambient CO2 concentration) per unit area; AM, photosynthetic capacity per unit dry mass; EN, photosynthetic nitrogen use efficiency (photosynthetic capacity per unit nitrogen content); FB, apparent fraction of leaf nitrogen in bioenergetics (Eqn 6); FR, apparent fraction of leaf nitrogen in ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) (Eqns 5, 7); Jmax,A, capacity for photosynthetic electron transport per unit area; Jmax,M, capacity for photosynthetic electron transport per unit dry mass; MA, dry mass per unit area; NA, dry mass per unit leaf area; NM, nitrogen content per dry mass; Vmax,A, maximum carboxylase activity of Rubisco per unit area; Vcmax,M, maximum carboxylase activity of Rubisco per unit dry mass. FB and FR are denoted as the apparent fractions because they have been derived from estimates of Jmax and Vcmax (Eqns 5–7), and provide the minimum estimates of nitrogen in rate-limiting components of photosynthesis needed to reach given levels of Jmax and Vcmax under the assumption that mesophyll diffusion conductance is infinite. The variables with the subscript ‘V’ (AA,V, AM,V and EN,V) were calculated from the estimates of Vcmax or Jmax according to the photosynthesis model of Farquhar et al. (1980) (section III.3). A detailed explanation of the estimation and definition of leaf traits is provided in sections II and III. 2 Light-dependent plasticity for the within-canopy average integrated light (Qint) range of 3–40 mol m
2 d
1 was calculated for each trait by Eqns 9–10. Only light gradients with the data spanning a threshold light range of at least 5–15 mol m
2 d
1 were included in this analysis (for the distribution of light gradients according to certain minimum and maximum Qint estimates, see Fig. S1). Qint is defined as the average daily integrated incident quantum flux density over 50 d after bud burst. The values of light-dependent plasticity of less than one indicate reduction of the trait value with increasing light availability.

study analyzing the overall variation in these traits through plant canopies, examining how these variations differ among plant functional types, or quantitatively evaluating the extent to which the variations in the constituent variables ultimately contribute to within-canopy variations in leaf photosynthetic capacity.

III. Construction of a global database on within-canopy variation in leaf structural, chemical and photosynthetic characteristics 1. Data sources, gradient types, units and species classification An extensive literature survey was carried out to find published data on within-canopy gradients in leaf morphology, chemistry and leaf photosynthetic properties. Only studies reporting within-canopy variation in at least three different canopy positions and with New Phytologist (2015) 205: 973–993 www.newphytologist.com

explicit measurements of within-canopy light conditions or cumulative leaf area index were considered. The survey identified 292 studies providing altogether 831 light gradients for 304 taxa belonging to different plant functional types (Supporting Information Table S1). The data on leaf functional traits were extracted from the studies, together with key meta-information. The metainformation included in the database was the plant growth environment (artificial vs field), the location of the study, year and time of sampling, the age of the sampled foliage, plant size, details of light measurements and, whenever pertinent, experimental treatment and the start of the treatment and/or planting date. For compound-leaved species, it was also noted whether the data referred to leaflets or entire leaves. Several studies reported multiple light gradients in plant stands subject to various experimental treatments, including density and drought treatments, defoliation, thinning, pruning, extra illumination, shading, fertilization and atmospheric [CO2] elevation Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist (Table S1). In addition, replicate light gradients reported in the same studies included different sites, multiple leaf age classes and plant ages and sizes (Table S1). Replicates in time were often also associated with differences in plant size (e.g. DeJong et al., 1989; Vapaavuori & Vuorinen, 1989; Vapaavuori et al., 1989; Rosati et al., 2000; Palmroth et al., 2002; Han et al., 2006; Han & Chiba, 2009), warranting a consideration of these gradients separately. However, to reduce the effects of extreme treatments, rigorous routines were developed for the inclusion of gradients from studies including treatments (Table S1). On the other hand, individual light gradients consisted of different-sized or aged individuals in some studies, whereas other studies pooled replicates in time or different leaf ages. In addition, data ranges for individual replicate gradients in some multiple-gradient studies were too narrow and the consideration of single gradients was impractical. Various treatments applied in the original studies, the presence of replicate gradients and pooled treatments are explicitly stated in Table S1. For all leaf-level traits, the units were harmonized, and all data were expressed per unit projected leaf area, as explained in Notes S1. The species included in the database spanned a wide range of vascular plant functional types (Table S1). Although a large number of species were included in the database, some plant functional types were only represented by a few species, and the number of species available for different traits also varied. For the purposes of statistical analyses, plant species were classified among the functional types to include normally at least 20 light gradients for each functional type (Table S1). Thus, functionally convergent or related plant functional types with a limited number of species were merged with more species-rich plant functional types as explained in Table S1. The functional type classification was also varied depending on the availability of observations for any given trait. Finer scale plant functional types, such as annual and perennial forbs and grasses, and C3 and C4 grasses for herbaceous species and woody evergreens, separated according to climatic origin (Mediterranean and temperate evergreens vs subtropical and tropical evergreens), could be used for more data-rich traits. For traits with fewer gradients (Table 1), plant functional types with fewer observations were merged to form a broader functional grouping. 2. Measurements of foliage photosynthetic capacity For foliage photosynthetic characteristics, only studies explicitly stating that the measurements were conducted at saturating or close to saturating quantum flux density (Q), providing the photosynthetic capacity, AA, were incorporated in the database. In all cases, measurement conditions, temperature, Q and ambient CO2 concentration (Ca) were included in the database. The estimates of net assimilation rate conducted at ambient CO2 concentrations of 267–420 lmol mol
1 (average  SE = 354.2  0.3 lmol mol
1) and temperatures of 15–35°C (average  SE = 26.7  0.1°C with the bulk of the measurements, 75%, conducted between temperatures of 20 and 30°C) were defined as AA, and used directly in developing the correlations with light availability. This definition of AA is inclusive and corresponds to that used commonly in global databases of leaf structure and function (Niinemets, 1999; Wright Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

Tansley reviews

Review 977

et al., 2004; Kattge et al., 2011). Although temperature significantly alters AA, the temperature response of AA has a broad optimum over this temperature range (Cannell & Thornley, 1998; Way & Yamori, 2014). We also argue that as long as the measurement temperature is constant for different leaves measured within the canopy light gradient, valid estimates of within-canopy photosynthetic plasticity can be derived and these plastic estimates are much less sensitive to variations in environmental drivers than is AA itself (for the estimation of relative plasticity, see section IV.3). Depending on whether the leaf dry mass per unit area (MA) and/or the leaf nitrogen content were available, mass-based photosynthetic capacity (AM, Eqn 1) and/or nitrogen use efficiency (EN, Eqn 4) were also calculated. 3. Standardized estimates of leaf biochemical potentials and photosynthetic capacity Whenever the C3 photosynthesis model parameters of Farquhar et al. (1980), Jmax,A and Vcmax,A, were reported, they were included in the database. In this analysis, all model parameters are based on intercellular CO2 concentration (Ci), as only one study included in the analysis (Niinemets et al., 2006) provided estimates of mesophyll diffusion conductance needed to calculate chloroplastic CO2 concentration. As a result of the significant CO2 drawdown between intercellular air space and chloroplasts (for a review, see Flexas et al., 2012), the use of Ci somewhat underestimates Vcmax and Jmax as discussed in Notes S2. The estimates of Jmax,A and/or Vcmax,A measured at a leaf temperature other than 25°C were converted to 25°C equivalent estimates as explained in Notes S2. Mass-based estimates of biochemical potentials at 25°C and apparent fractional nitrogen allocations to Rubisco (FR) and bioenergetics (FB) were further calculated according to Eqns 5–7 for datasets reporting NA or MA and NM. As different Rubisco kinetic characteristics have been used across the studies (Table S2) and this can potentially result in a major bias (Rogers, 2014), we have converted all the Vcmax estimates to a common set of Rubisco kinetic characteristics as described in Notes S2. Given that the temperature relationships of Rubisco kinetic characteristics from the study of Jordan & Ogren (1984), as fitted by Harley et al. (1992) and Niinemets & Tenhunen (1997), were used by the majority (60.5%) of the studies, all other estimates were converted to this set of kinetic characteristics. We also provide a procedure to convert Jmax estimates (Notes S2), but note that the Jmax values reported here were not converted, as different Rubisco kinetics had only a minor impact on Jmax estimation. Overall, data on within-canopy variations in Vcmax (or the slope of the A vs Ci response curve) were reported in 42 studies, providing information for 69 species, whereas Jmax estimates were reported in 44 studies, providing information for 45 species. Apart from the CO2 response curves, inverse modeling techniques can be used to estimate Vcmax from AA (e.g. Niinemets & Tenhunen, 1997; Niinemets, 1999; Niinemets et al., 1999c; Patrick et al., 2009). However, for the inverse modeling, estimates of dark respiration (Rd) and Ci are necessary, and, when available, these data, together with estimates of stomatal conductance to water vapor (gs), were New Phytologist (2015) 205: 973–993 www.newphytologist.com

978 Review

Tansley reviews

also included in the database. In addition, Jmax can be estimated from the CO2-saturated net assimilation rate reported in some studies or from the effective quantum yield of photosystem II measured by chlorophyll fluorescence at saturating light (Evans, 1993; Schreiber et al., 1994; Niinemets et al., 1999a). Thus, additional estimates of Vcmax and/or Jmax were derived from foliage photosynthetic characteristics as described in Notes S2, altogether resulting in estimates of foliage photosynthetic potentials (Vcmax or Jmax or both) for 85 species from 92 studies. When foliage biochemical potentials, Jmax and Vcmax, were ultimately available, a standardized estimate of AA, AA,V, was determined for a leaf temperature of 25°C and a Ci of 251 lmol mol
1 using the C3 photosynthesis model of Farquhar et al. (1980), as described in Notes S2. This value of Ci used for standardization corresponded to the average Ci estimated across the studies reporting it. Standardized estimates of mass-based photosynthetic capacity, AM,V, and nitrogen use efficiency, EN,V, were also estimated. We note that these standardized estimates are based on the C3 photosynthesis model, and do not have an analogous meaning for controls in C4 photosynthesis. However, none of the studies conducted with C4 species reported estimates of Ci or stomatal conductance, and thus non-standardized estimates of AA are always reported for C4 species through the article. Although some of the AA,V values were derived from the measured estimates of AA (Notes S2), many studies reporting estimates of Jmax and Vcmax did not report actual photosynthesis data. Thus, a large proportion of AA,V estimates were complementary to the measured AA values. Of the total of 3513 unique individual photosynthesis measurements in the entire dataset combining measured and standardized data, 729 were from AA,V (48 unique within-canopy gradients from, altogether, 324 photosynthetic capacity gradients). Analogously, 528 AM,V values from a total of 1823 (35 light gradients from 175) and 681 EN,V measurements from a total of 2411 photosynthetic nitrogen use efficiency estimates (43 gradients from 201) were unique.

IV. Methodology for analysis of within-canopy leaf trait variation: concepts and standardizations 1. Standardized estimates of light availability We use the daily average incident integrated photosynthetically active quantum flux density on a horizontal surface during leaf development as an estimate of leaf light conditions (Qint). As a standardized estimate, Qint was averaged for 50 d after the start of foliage development, or for the actual number of days since the start of foliage development for leaves either younger than 50 d or with an average leaf life-span of less than 50 d. This definition is based on observations indicating that foliage trait vs integrated average light relationships become stable 30–60 d after leaf formation in both woody and herbaceous canopies (Niinemets et al., 2004a; Niinemets & Keenan, 2012). Although foliage formation is initially rapid and light conditions within the canopy can undergo fast changes in developing herbaceous canopies (Niinemets & Keenan, 2012), leaf growth stops after the onset of generative growth, and leaves are exposed to New Phytologist (2015) 205: 973–993 www.newphytologist.com

New Phytologist relatively stable light conditions for the rest of the growing season (for a discussion on the possible effects of leaf turnover and reacclimation, see section VII.1 ). From a practical perspective, the estimation of light availability for shorter time periods in broad meta-analyses is associated with inherent uncertainties, in particular as a result of the estimation of dates of leaf formation (Notes S3), as well as day-to-day variations in cloudiness conditions at specific sites that cannot be captured well because of the relatively crude resolution of global databases of incident solar radiation (Notes S3). Details of the conversion of the different light measurements to Qint are reported in Notes S3. As the incident quantum flux density is typically estimated by horizontal sensors and sampling locations in within-canopy studies characteristically corresponded to a cluster of leaves with a certain variation in foliar angle among the leaves within the given leaf cluster, in this analysis, the definition of Qint refers to the quantum flux density on a horizontal surface. Indeed, very few within-canopy studies have reported the actual leaf angles or leaf angle distributions in specific canopy positions (for some exceptions, see Hollinger, 1989, 1996; Niinemets, 1998; Ishida et al., 2001; Niinemets & Fleck, 2002; Niinemets et al., 2002; Cescatti & Zorer, 2003; Fleck et al., 2003). The use of the quantum flux density on a horizontal surface could be considered as an inherent limitation of the Qint concept. However, as the solar angle varies during the day, the effect of leaf angle on daily integrated leaf light receipt is relatively moderate (although it depends on the gap distribution in the canopy), with the largest effects on the share of light interception between the leaf upper and lower surface and on the distribution of the timing of interception of solar radiation with different intensities (Gutschick & Wiegel, 1988; Cescatti & Niinemets, 2004; Valladares & Niinemets, 2007). Given this reasoning, four studies in which the light sensor was positioned parallel to the leaf surface (Oberbauer & Strain, 1986; Oberbauer et al., 1987; Lynch & Gonzalez, 1993; Ackerly & Bazzaz, 1995) and eight studies in which the sensor was attached to the leaf surface (Chazdon & Field, 1987; Ackerly, 1992; Tjoelker et al., 1995; Traw & Ackerly, 1995; Osada et al., 2001; Han et al., 2006; Han & Chiba, 2009; Posada et al., 2009) were also incorporated into this analysis. Although many within-canopy studies in the past have used the relative light intensity as the predictor variable in light availability vs leaf trait relationships (e.g. Hirose & Werger, 1987a; Anten et al., 1998b; Niinemets et al., 1998), we argue that, at a given relative light availability, there are strong variations in absolute incident light within and between seasons, and between different geographic locations. Indeed, significant fractions of light gradients were positioned at the lower (maximum Qint < 5 mol m
2 d
1) and upper (minimum Qint > 15 mol m
2 d
1) ends of the whole light range (Fig. S1), implying that the relative light estimates calculated for such truncated light ranges would be incomparable with broader light ranges. Furthermore, relative light availability is particularly inappropriate in glasshouse and growth chamber studies, where there can be major reductions in light as a result of absorption by enclosure materials and differences in the distance between the light source(s) and the plants (Niinemets & Keenan, Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist 2012). Thus, we argue that having a standardized light estimate over the same time period is absolutely essential for a rigorous comparison of light-driven plasticity of plant traits. 2. Fitting of leaf trait vs Qint relationships Typically, leaf trait vs Qint relationships are non-linear (for sample relationships for representative forb, grass and broad-leaved deciduous woody species, see Fig. 1). Either non-linear functions in the form y = axb or logarithmic functions in the form y = alog (x) + b have been used to fit the data (e.g. Hirose & Werger, 1987a; Hirose et al., 1989; Anten et al., 1995, 1998a; Poorter et al., 2009, 2010; Niinemets & Keenan, 2012). However, initial data fitting demonstrated that these two-parameter functions did not always provide the best fits to the data. Thus, we have used a threeparameter monomolecular equation (Ceulemans et al., 1980; Causton & Dale, 1990):
y ¼ ymax 1
e d
fx ; Eqn 8 where ymax is the asymptotic value of the trait, and d and f are the empirical parameters. The monomolecular equation is more flexible than the two-parameter equations and can realistically fit strongly curved to almost linear relationships. Equation 8 was fitted to the data iteratively by minimizing the sum of the squares between the measured and the predicted trait values. In the case of variables

120

4

r2

3

r2

80

r

40

r2 (a)

2 r2

r

20

30

40

0

(c)

20

2

r2

15

2

r

10

0.1 (d)

r

2

25 2

0.2

40

r

2

0

r

r2

r2

1 (b)

0.3

10

3

2

0.4

120

0

Absolute trait values differ for different functional traits, and any given functional trait strongly varies among plant functional types. Furthermore, as a result of non-linearity, the trait plasticity also varies at different parts of the Qint gradient. This is relevant, as the

0

r2

r2

80

3. Definition and estimation of trait plasticity

1

0 160

declining asymptotically with increasing Qint, this equation or a slightly modified equation (Eqn S4.1 in Notes S4) was employed. Further details of data fitting and adjustment of the statistical thresholds for the significance of relationships depending on the availability of the data are provided in Notes S4. After parameterization, Eqn 8 or S4.1 was used to estimate the trait values at fixed values of Qint of 1, 3, 6, 12, 20, 30 and 40 mol m
2 d
1 to allow for comparison of trait estimates among plant functional types at standardized Qint. Values of Qint of 1– 3 mol m
2 d
1 correspond to typical light environments in the understory, 6–12 mol m
2 d
1 to moderately high light levels observed in medium-sized gaps and in mid-canopy, and 20– 40 mol m
2 d
1 to high light environments observed in large gaps and in the upper canopy. Extrapolation beyond the data in calculating the trait estimates was constrained and did not exceed  0.7–5 mol m
2 d
1, depending on the position in the Qint gradient and on the light range available for the given light gradient (Notes S5).

r

2

2

0

Review 979

Tansley reviews

r

(e)

10

r

5

2

0

2

20

30

40

0

(f)

0

10

20

30

40

Fig. 1 Illustration of representative light-dependent variations in leaf dry mass per unit area (a), nitrogen content per dry mass (b) and area (c), photosynthetic nitrogen use efficiency (photosynthetic capacity per unit nitrogen) (d), and light-saturated net assimilation rate at ambient CO2 concentration (photosynthetic capacity) per dry mass (e) and area (f) in the perennial forb Solidago altissima (open circles; data of Hirose & Werger, 1987a,b; Werger & Hirose, 1988), perennial grass Phragmites australis (closed circles; data of Hirose & Werger, 1994, 1995) and broad-leaved winter-deciduous tree Fagus crenata (triangles; data of Iio et al., 2005). Data were fitted by Eqn 8, and the negative relationship in F. crenata in (b) by Eqn S4.1 (Supporting Information Notes S4). All relationships are significant at least at P < 0.02, except for the non-significant correlations in F. crenata in (b) and (e) shown by dashed lines. The seasonal average daily integrated quantum flux density (Qint) is defined as the average for 50 d since the start of leaf development or for the actual number of days for leaves younger than 50 d. Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist (2015) 205: 973–993 www.newphytologist.com

980 Review

New Phytologist

Tansley reviews (a)

(d)

(b)

(e)

(c)

(f)

Fig. 2 (a–f) Variation in key foliage structural, chemical, allocational and photosynthetic traits among plant functional types (Eqns 1–5), and correlations among functional trait-specific average trait values (insets in b, e and f). All trait values were standardized to a constant moderately high integrated seasonal average daily quantum flux density (Qint, see Fig. 1 for the definition) of 12 mol m
2 d
1. These Qint-standardized values were derived from leaf trait vs Qint relationships fitted by Eqns 8 and S4.1 (see Fig. 1 for sample relationships). The distribution of trait values for different plant functional types is characterized by the box plots, where the box length provides the interquartile range (IQR), the bottom of the box the 25th percentile (first quartile, q1), the top of the box the 75th percentile (third quartile, q3) and the horizontal line within the box the median value. The lower whisker corresponds to q1
1.5IQR, or to the minimum estimate, and the upper whisker corresponds to q3 + 1.5IQR, or to the maximum estimate. The dots denote values outside the whisker limits. For each trait, the width of the box was set proportional to the square root of the number of gradients available for each trait (see Table 1 for the total number of gradients). Plant species were classified among the functional types to include normally at least 20 light gradients for each functional type, but for traits with fewer gradients (Table 1), similar plant functional types with fewer observations were merged to form a coarser classification. The herbaceous plant functional types distinguished were annual (Ann.) and perennial (Per.) forbs (forbs for the merged group), and annual and perennial C3 grasses and C4 grasses (grasses for the merged group). For woody species, the plant functional types separated were winter-deciduous (Winter-decid.), needle-leaved winter-deciduous (Needle-leaved decid.), subtropical and tropical broad-leaved evergreens (Evergr. trop.), temperate and Mediterranean broad-leaved evergreens (Evergr. temp) and needle-leaved evergreens (Needle-leaved everg.) (see Table S1 for plant functional types for any given species). Average trait values in the insets were fitted by non-linear regressions in the form y = axb in (b) and (f) and y = cedx in (c) (all significant at P < 0.001). Table S3 provides significance estimates of log-transformed mean comparisons among plant functional types according to ANOVA followed by Tukey tests.

ranges of Qint available for different gradients varied among and within studies (Fig. S1). As a result of the non-linearity of light responses of foliage traits (Fig. 2; Poorter et al., 2009; Niinemets & New Phytologist (2015) 205: 973–993 www.newphytologist.com

Keenan, 2012), it is essential that trait comparisons are conducted over the same light range. To directly compare the light-dependent plasticity of any given trait among plant functional types at a Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist

Tansley reviews

specified range of Qint, the relative light-dependent plasticity (PL) of a given trait over the given light range l, PL,l, was calculated as: PL;l ¼

viþx
vi ; DQint ðviþx þ vi Þ=2

Eqn 9

where vi is the trait value corresponding to a seasonal average quantum flux density of i (Qint,i) and vi+x is the trait value corresponding to Qint,i + x (Qint,i+x). PL,l is defined as a relative plasticity because it is normalized with respect to the average trait value across the given light range, (vi+x + vi)/2. Such a normalization makes it possible to consider differences in absolute trait values between traits and between plant functional types for any given trait. The fold change (fL) of a given trait over the given light range l, vi+x/vi, and PL,l are related as: fL;l ¼

DQint PL;l þ 2 : 2
DQint PL;l

Eqn 10

However, we note that fL is not defined when vi ≤ 0. The relative light-dependent plasticity was calculated over the Qint ranges 1–3, 3–6, 6–12, 12–20, 20–30, 30–40 and 3–40 mol m
2 d
1 using the trait values, vi, estimated from the parameterized trait vs Qint responses (Eqn 8, S4.1) employing the constraints specified in Notes S5. 4. Analysis of the quantitative importance of variations in constituent traits on photosynthetic plasticity Equations 1–7 specify the functional relationships among structural, chemical and allocation traits and foliage photosynthetic potentials. The sources of light-dependent variation in composite traits, such as nitrogen content per area and photosynthetic potentials per unit leaf area, can be partitioned among the variations in constituent variables using a response coefficient analysis (Poorter & Nagel, 2000). The response coefficients for a given composite trait determined along a light gradient provide the fractions of variation caused by light-dependent variations of all the constituent traits (for the mathematics of the response coefficient analysis, see Notes S6). For example, the lightdependent variation in AA as the product of AM and MA (Eqn 1) can be quantitatively ascribed to variations in AM (AA response coefficient for AM, RAAMA ) and MA (AA response coefficient for MA, AA AA RM ) such that RAAMA þ RM = 1. The response coefficient for a A A given variable can be positive, when an increase in this characteristic is associated with an increase in the response variable. The response coefficient can also be zero, when a change in the trait does not affect the response variable, and it can be negative, when the trait increase is associated with a decrease in the response variable. The response coefficient analysis cannot be applied when the response variable is essentially invariable. Such an invariability in the response trait can reflect the invariability of the constituent traits, but can also result from opposite light responses of the constituent traits. For all light gradients in the current study, we calculated the response coefficients for all dependent variables in Eqns 1–7 using the trait values estimated Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

Review 981

at the maximum, vi(Qint,max), and minimum, vi(Qint,min), Qint values present in the given dataset (section IV.1). Notes S6 provide further details of the estimation of the response coefficients. 5. Statistical analyses In the case of fitting of the Qint responses (section IV.2), the explained variance, r2, was used as the goodness of fit. Kolmogorov–Smirnov tests were used to check the normality of traits and trait plasticity at specific Qint values and the normality of trait response coefficients. Depending on the degree of positive skewness, square root or logarithmic transformations were used. In the case of negatively skewed distributions, the distributions were first reflected, and then transformed by square root or logarithmic transformations, and reflected again to restore the ranking. Finally, the traits, their plasticity and response coefficients were compared among plant functional types by one-way ANOVA followed by Tukey’s test, which is a conservative approach for unbalanced data. Different from multiple-factor ANOVA models, lack of balance does not affect the calculation of the sum of squares in one-way ANOVA (Montgomery, 2012). In addition, correlation and regression analyses were used to test for statistical relationships among traits and functional trait-specific means. All statistical relationships were considered to be significant at P < 0.05.

V. Global variation in leaf characteristics through the canopies 1. Worldwide database on within-canopy variation in leaf structural, chemical and photosynthetic traits We have constructed, to our knowledge, the most comprehensive database of within-canopy variations in 12 key foliage structural, chemical and physiological traits playing an important role in photosynthesis (see Table S1 and Notes S7 for data sources; Eqns 1–7). Altogether, the database includes 831 within-canopy gradients for 304 species belonging to 63 families and covering most vascular plant functional types forming dense vegetation canopies. To put the size of the database into perspective, the number of individual trait observations, 9231 for nitrogen content per area (NA), 7390 for leaf dry mass per unit area (MA), 5757 for nitrogen content per dry mass (NM) and 3513 for photosynthetic capacity per unit leaf area (AA), is, depending on the trait, 2.8–4.7-fold greater than that for the global Glopnet database (Wright et al., 2004), and of similar magnitude for some traits (c. 68% for NA and 115% for AA) to the TRY database at the time of its publication (Kattge et al., 2011). Thus, we consider this extensive database as an invaluable resource for understanding worldwide patterns in leaf functional differentiation through canopy light gradients. Standardized estimates of average incident integrated light availability during leaf growth (Qint) were determined for all leaves (section IV.1, Notes S3). Analysis of the distribution of all light gradients together (Fig. S1) underscores the fact that studies on New Phytologist (2015) 205: 973–993 www.newphytologist.com

982 Review

Tansley reviews

light-driven plasticity need to pay due attention to the withincanopy light range. A significant proportion of studies conducted so far include low to moderate light ranges, which can importantly limit the detection of the full light-driven plasticity. Nevertheless, c. 80% of light gradients included a minimum Qint value of at least 10 mol m
2 d
1, and c. 60% of studies included a maximum Qint value of at least 20 mol m
2 d
1 (Fig. S1). The limited light range can be a genuine limitation associated with particular stand growth conditions (open vs closed stands), species dispersal characteristics (understory vs overstory) and, in some species, lack of foliage formation or die-off in low light in the lower canopy when the upper canopy leaves have access to high light (e.g. Niinemets & Kull, 1998; Niinemets et al., 1998; Pons & Anten, 2004; Boonman et al., 2006). Poorter et al. (2009, 2010) coped with the limited light range by standardizing all measurements conducted at different light environments (understory–gap light gradient) to a common moderately high integrated light estimate of 8 mol m
2 d
1. However, all light gradients that do not include this light range, a significant fraction in our worldwide database (Fig. S1), would be left out when using this method. Thus, here, we have developed an alternative data distillation strategy by fitting the data over the available light range and deriving foliage lightdependent plasticity estimates (Eqn 9) for finite light ranges according to predefined rigorous criteria for available minimum and maximum Qint and Qint range (section IV.3 and Notes S5). This more inclusive strategy makes it possible to use most of the data, with only a minor fraction of the data rejected as a result of light ranges that are too narrow (between less than 1% to a few percent for different traits). 2. Overall variations in leaf traits among plant functional types Across the whole database, all the studied traits were characterized with extremely high variability of c. 20- to more than 100-fold for different traits (Table 1). Combinations of traits strongly varied across plant functional types with woody plant functional types being characterized by high MA and low NM, AM and EN (for an illustration of average trait values standardized to a moderately high Qint of 12 mol m
2 d
1 in different plant functional types, see Fig. 2) and with analogous patterns for Vcmax,M, Jmax,M, FB and FR (data not shown). For plant functional type averaged and light standardized (Qint = 12 mol m
2 d
1), there were strong negative correlations of AM vs MA (Fig. 2b), NM vs MA (Fig. 2f) and EN vs MA (r2 = 0.894, P < 0.001), and positive correlations among AM vs NM (Fig. 2e) and AM vs EN (r2 = 0.904, P < 0.001). These scaling relationships reflect the leaf economics spectrum, i.e. co-occurrence of combinations of high MA, low NM and AM in the low return end of the spectrum (woody) vs low MA, high NM and AM in the high return end of the spectrum (Wright et al., 2004). As a result of these compensatory relationships, NA (Fig. 2a) and AA (Fig. 2d) were much less variable among the plant functional types than were MA (Fig. 2b), NM (Fig. 2c), AM (Fig. 2e) and EN (Fig. 2f) (Table S3). For the broad herbaceous vs woody comparison, the average values of NM, AM, EN and AA were significantly greater for herbaceous New Phytologist (2015) 205: 973–993 www.newphytologist.com

New Phytologist species, whereas the average values for MA and NA were greater for woody species (P < 0.001 for all comparisons, except for the comparison of AA, where P < 0.01). Comparisons among individual plant functional types further corroborated these broad patterns. The averages for MA, NM, AM and EN differed significantly among all plant functional types, but NA differed among fewer herbaceous and woody plant functional types, and AA was significantly lower only in needle-leaved evergreens than in most other plant functional types (Table S3). This low variability in NA and invariability of AA among plant functional types is consistent with the trait variation in the worldwide leaf economics spectrum (Wright et al., 2004), and indicates that a given value of AA and NA can result from different combinations of underlying traits.

VI. Trait scaling with light, plasticity and quantitative limitations 1. Non-linear scaling of foliage traits with light and the strength of light relationships All of the traits analyzed here are characterized by strong lightdependent within-canopy variations, with particularly large lightdependent increases observed for the area-based traits AA, Jmax,A, Vcmax,A, NA and MA, followed by mass-based photosynthesis potentials and nitrogen allocation traits (Table 1). The extensive analysis carried out here demonstrates that the Qint vs structural, chemical and photosynthetic trait relationships are characteristically non-linear (for sample relationships, see Fig. 1), implying that, for incomplete light gradients, data fitting and analysis over finite light ranges is essential (Notes S4). Here, the data were fitted by non-linear three-parameter equations (Eqn 8 or S4.1) that improved the degree of explained variance compared with conventional logarithmic or power equations (for sample fits of key leaf traits, see Fig. 1). Low r2 values of regression fits typically truly reflected the trait invariability through the canopy (Fig. 1b,e) rather than extensive unexplained within-canopy fluctuations of the given trait. Analysis of the variations among the traits according to the degree of explained variance (r2) in leaf trait vs Qint relationships demonstrates that the area-based characteristics, NA (Fig. 3a) and AA (Fig. 3d), are universally strongly associated with Qint in all plant functional types. These two traits are followed by MA, which, however, is weakly dependent on Qint in grasses (Fig. 3b). The relationships are strong across plant functional types, although herbaceous species have, on average, higher r2 for both NA (P = 0.02) and AA (P = 0.002), and woody species for MA (P < 0.001), vs Qint relationships. In the case of AA and NA, this reflects lower r2 values in needle-leaved evergreens (Fig. 3a,d; Table S3), whereas, in the case of MA, the significant herbaceous vs woody contrast reflects primarily the greater r2 values in winterdeciduous species relative to annual forbs and grasses (Fig. 3b; Table S3). Different from these three traits, there are major differences in r2 for NM (Fig. 3c), AM (Fig. 3e) and EN (Fig. 3f) among herbaceous Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist

Fig. 3 (a–f) Variation in the degree of explained variance (r2) in leaf trait vs integrated light (Qint) relationships (Eqns 8 and S4.1 in Notes S4) among different foliage structural, chemical, allocational and photosynthetic traits and among different plant functional types illustrated by the box plots (see Fig. 2 for a detailed explanation). The first number next to each box provides the percentage of statistically non-significant relationships (see Notes S4 for the significance limits), whereas the second number provides the percentage of statistically significant negative relationships. Classification of data according to plant functional types as in Fig. 2. For significantly different r2 values among plant functional types, see Table S3.

and woody species (P < 0.001 for all three traits; for statistical differences among individual herbaceous vs woody plant functional types see Table S3), with woody species having much weaker Qint dependences for these traits (Fig. 3c,e,f). In addition, perennial forbs also had a lower average r2 than annual forbs for NM vs Qint relationships (Fig. 3c; Table S3). Another way to gain an insight into the strength of leaf trait vs Qint relationships is to look at the percentage of statistically nonsignificant relationships and at significant, but negative relationships. Of course, in the case of statistically non-significant relationships developed over limited canopy light range, we cannot ultimately say whether there could have been a significant response over a more extensive light range. Nevertheless, comparisons of the Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

Tansley reviews

Review 983

(a)

(d)

(b)

(e)

(c)

(f)

light response of different traits analyzed in the same study over the same light range, and across studies (e.g. Fig. 1), can still provide appropriate information on the overall differences in the strength of the light relationships. In the case of NA and AA, there were no statistically significant negative relationships, and the number of non-significant Qint responses was, in most cases, low, a few percent, with the exception of somewhat larger percentages of c. 10–15% in perennial C3 grasses and evergreen broad-leaved tropical and needle-leaved species for NA (Fig. 3a) and in evergreen broad-leaved tropical species for AA (Fig. 3d). In the case of MA (Fig. 3b), grasses had a large percentage of non-significant and negative relationships, almost 50%, whereas, in annual forbs, c. 13% of relationships were negative. New Phytologist (2015) 205: 973–993 www.newphytologist.com

984 Review

Tansley reviews

Different from these three traits, lack of statistical significance was much more frequent for NM (Fig. 3c), AM (Fig. 3e) and EN (Fig. 3f) vs Qint relationships, in particular for woody plant functional types. Among woody plant functional types, the percentages of negative or non-significant relationships were c. 49–70% for NM, c. 35–50% for AM and c. 50% for EN. Although the Qint responses for these three traits were much stronger for herbaceous plant functional types, 33% of NM relationships for perennial forbs and 16% for those in grasses were not significant (Fig. 3c). We note that because of the limited number of observations available for different plant functional types, not all comparisons among functional types can be made for all traits. Nevertheless, these data demonstrate important qualitative differences in leaf trait vs light relationships for different plant functional types, in particular among herbaceous vs woody species. 2. Negative trait plasticity: why does it occur? According to the functional relationships linking leaf structural, chemical and allocational traits to leaf photosynthetic capacity (Eqns 1–7), maximization of foliage photosynthetic capacity within the canopy light gradients suggests that all of these leaf traits should increase with increasing Qint. Such broad trends for all traits are apparent across all the data (Table 1), but, nevertheless, a certain number of negative Qint relationships were observed for MA, AM, NM and EN (Figs 1, 3). Although most of these statistically significant negative trait vs Qint relationships typically exhibited a minor negative trend, implying that the overall effects of the negative variations in these traits on the target variable AA were moderate, the question remains as to what are the sources of such puzzling negative scaling trends. In the case of MA, we believe that the negative Qint responses can be explained by three different mechanisms dependent on the location of growing leaf tissues in the canopy, the timing of foliage growth and the timing and rate of growth of neighboring individuals in the canopy. First, in the temperate grass Carex acutiformis, which is characterized by a particularly strong negative response of MA to Qint (Hirose et al., 1989; Pons et al., 1993), foliage extends from the base of the leaf. Thus, the basal parts of long extended leaves have to support the entire leaf. Therefore, the leaves increase in thickness and have a more pronounced midrib towards the base of the plant, resulting in a greater MA at a lower Qint in the bottom of the canopy. Second, as explained in Notes S4, in fast-growing canopies, characteristically in herbs, but also occasionally in woody species (Vapaavuori & Vuorinen, 1989; Vapaavuori et al., 1989; Dickmann et al., 1990; Medhurst & Beadle, 2005), continuously forming leaves at the top of the canopy, there can be important interactions among leaf age and Qint. Although all leaves may have been formed at high light, leaves in the extending upper canopy can be developmentally younger and exhibit a lower MA at the time of sampling than fully mature leaves in the lower canopy. For canopies in which the formation of new leaves extends into the late season, it is also plausible that abovecanopy Qint is higher for leaves formed in the lower canopy than for leaves forming later. Finally, in mixed fast-growing canopies with a New Phytologist (2015) 205: 973–993 www.newphytologist.com

New Phytologist developing size hierarchy of individuals (Anten & Hirose, 1998; Anten et al., 1998a; Hikosaka et al., 1999) or with differences in growth rate of constituent species (Hirose & Werger, 1994, 1995), the leaves formed early have been exposed to the same Qint. However, later during the course of the stand development, plant individuals growing more rapidly will shade the developing leaves in individuals growing more slowly. This implies that, in these less competitive individuals, early- and late-formed leaves have been exposed to different Qint values. We argue that negative Qint relationships for other leaf traits can also be explained, in many cases, by developmental effects. However, these mechanisms cannot explain negative Qint relationships for flush-type species either forming all leaves simultaneously at the beginning of the growing season (the bulk of winterdeciduous species) or having the leaf flush periods broken by sustained periods of no new leaf formation (evergreen needleleaved species, many tropical canopy species). In these species, it has been argued that a moderate reduction in NM accompanied by reductions in AM can reflect dilution of foliar nitrogen as a result of enhanced contents of structural carbohydrates and lignin (e.g. Niinemets & Kull, 1998) to cope with potentially more severe water stress in the upper canopy (Niinemets et al., 1999b, 2004b; Hubbard et al., 2002; Aasamaa et al., 2004; Sellin & Kupper, 2004). Although water stress can be especially significant in plant stands growing in inherently drier habitats, limited hydraulic conductivity of stem and branches can constrain water transport to the upper canopy in clear days, even when soil water availability is not limiting (Joyce & Steiner, 1995; Brodribb et al., 2005; Renninger et al., 2006; Ewers et al., 2007; Peltoniemi et al., 2012). Enhanced water stress is further expected to reduce intercellular CO2 concentration more strongly in the upper canopy (Aasamaa et al., 2004; Niinemets et al., 2004b), reducing the estimates of AM and EN. Furthermore, limited water transport also alters the propagation of root-borne hormonal signals by cytokinins and abscisic acid, which can importantly modify the

Fig. 4 Correlations of the median values of the relative light-dependent plasticity estimates (PL, Eqn 9) for the light-saturated net assimilation rate per unit area (photosynthetic capacity, AA; red squares), nitrogen content per area (NA; purple circles) and leaf dry mass per area (MA; green circles) among forbs and winter-deciduous trees across different ranges of integrated light (Qint in mol m
2 d
1, see Fig. 1 for the definition). As a result of the non-linear dependence of leaf traits on Qint, the plasticity decreases with increasing Qint (see Fig. S2 for the variation in PL values for these three traits with the Qint range used for PL calculation). Although the PL values for different traits vary with the Qint range, the figure demonstrates that trait PL rankings for different plant functional types are essentially insensitive to Qint range. Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist

Tansley reviews

development of leaf photosynthetic capacity (Pons et al., 2001; Boonman et al., 2009). Taken together, we suggest that a complex array of factors can be responsible for negative scaling of foliage traits with increasing Qint in the canopy. Some of these factors, for example, differences in leaf age, could sometimes be controlled by skilled sampling designs. However, this is not necessarily possible in canopies with fast leaf turnover, where there are inherent interactions between light environment, leaf age and leaf turnover. 3. Implications of trait non-linearity for trait plasticity As a result of non-linearity, the relative trait plasticity calculated over a finite light range (PL, Eqn 9) decreases with increasing Qint range (Figs 4, S2). The pertinent question is what is the most appropriate light range for comparison of the

Fig. 5 (a–f) Comparison of relative trait plasticity (PL, Eqn 9) estimated for a moderately high integrated light (Qint) range of 6–12 mol m
2 d
1 within and among key plant functional types. As defined, PL values are directly comparable for different traits and for the same traits for light gradients with varying average trait values. The box plots for each trait and each plant functional type characterize the distribution of the given trait. Plant functional types and details of box plots are given in Fig. 2. The total number of gradients for each trait is reported in Table 1, and statistically significant plasticity differences among plant functional types are demonstrated in Table S3. Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

Review 985

plasticities of different traits and plasticities of any given trait among plant functional types. Although the lowest light range, 1–3 mol m
2 d
1, used for PL calculation is apparently the most plastic (Figs 4, S2), only slight inaccuracies in Qint estimation have a large effect on the plasticity estimate and therefore, can make this light range overly sensitive to such errors. However, the traits become light saturated at the higher light ranges of 20–30 and 30– 40 mol m
2 d
1 (Figs 4, S2), implying that the variability of plasticity over this light range is too low for any comparison among plant functional types. Furthermore, as a result of limitations in light range coverage (Fig. S1), the use of both the lower and higher light ranges, and the overall light range of 3–40 mol m
2 d
1, would lead to the neglect of a significant part of the light gradients. Thus, for the following analysis, we have used the PL estimates for the moderately high light range of 6–12 mol m
2 d
1. Over this

(a)

(d)

(b)

(e)

(c)

(f)

New Phytologist (2015) 205: 973–993 www.newphytologist.com

986 Review

New Phytologist

Tansley reviews

light range, just before flattening the trait vs Qint responses, foliage traits still respond strongly to Qint, and this light range was also present in the bulk of the light gradients. Indeed, analysis of differences in trait plasticity for AA, NA and MA among woody and herbaceous species indicated that the differences were in the same direction for these three traits for all light ranges, with the exception of only the highest light range of 30–40 mol m
2 d
1 for AA (Fig. 4). Thus, we conclude that the use of the selected light range 6–12 mol m
2 d
1 is representative to gain an insight into trait and functional type differences in trait plasticities. 4. Does plant functional type determine the light-dependent plasticity of different traits? The analysis of light-dependent plasticity over the moderately high light range of 6–12 mol m
2 d
1 highlights important variations in plasticity among different traits and for given traits across plant functional types (Fig. 5). First, this analysis underscores the overall greater light-dependent plasticity of the area-based traits AA (Fig. 5d), NA (Fig. 5a) and MA (Fig. 5b) compared with the mass-based traits NM (Fig. 5c), AM (Fig. 5e) and the allocation trait EN (Fig. 5f), confirming the evidence of overall differences in the variability of traits over the whole light range (Table 1). However, there are major differences among plant functional types in the plasticity of individual traits. For the broad herbaceous vs woody comparison, there is no clearcut difference in NA plasticity (P > 0.9), but a finer scale analysis demonstrates several significant differences in NA plasticity among plant functional types. In

(a)

particular, NA plasticity in annual C3 grasses tends to be greater, and that in needle-leaved evergreens smaller, than in most other plant functional types (Fig. 5a, Table S3). The broad herbaceous vs woody comparison suggests that herbaceous species are more plastic in all other traits, except for MA, which is more plastic in woody species (P < 0.001 for all comparisons). However, at a finer scale, MA (Fig. 5b) and NM (Fig. 5c) plasticities in perennial forbs are similar to those in woody species, and AM and EN plasticities for forbs (perennials and annuals pooled because of a limited sample size) are similar to those in woody plant functional types with a few exceptions (Table S3). In the case of AA plasticity, key differences are a greater plasticity in perennial C3 grasses than in all other woody plant functional types, and a greater plasticity in annual forbs than in evergreen tropical and needle-leaved species (Fig. 5d, Table S3). Surprisingly, there are very few significant differences in these key traits among woody plant functional types. Only NA plasticity of evergreen needleleaved species is significantly less than that for winter-deciduous and temperate evergreen broad-leaved species, and MA plasticity in winter-deciduous broad-leaved species is greater than that in evergreen needle-leaved and tropical broad-leaved species (Fig. 5a; Table S3). Clearly, the small number of observations for the plasticity of some traits, such as MA in grasses (Fig. 5b), limits the generality of our conclusions, but, nevertheless, the broad differences among herbaceous vs woody species and among more frequently sampled functional types stand out and emphasize that light-dependent plasticity truly varies among plant functional types.

(b)

Fig. 6 Quantitative partitioning of the light-dependent variations in nitrogen content per area (NA) as a result of leaf dry mass per unit area (MA) and nitrogen content per dry mass (NM, Eqn 3) within and among plant functional types according to the response coefficient analysis (a), and the correlations between the relative light-dependent plasticities in NA and MA for the light range of 6–12 mol m
2 d
1 (PL, Eqn 9) in different plant functional types (b). The response coefficient (Rc) for a given trait (section IV.4 and Notes S6, Eqns S6.1–S6.4) provides the fraction of light-dependent variance in the target variable as a result of the light-dependent variation in the given trait. The response coefficients of NA for MA and Nm sum up to 1. Positive Rc values indicate a positive effect of the given trait on the target variable, whereas negative Rc values indicate a negative effect. Rc values close to zero indicate that light-dependent variation in the given variable has a minor effect on the variation of the target variable. Figure 2 provides the specifics of the box plots as used in (a) and the logic of the separation of plant functional types in (a) and (b). Each symbol in (b) corresponds to a different light gradient. For better visual assessment, the outliers in the box plots were suppressed in (a), and the data for the perennial grass Carex acutiformis exhibiting strongly negative PL values between
0.027 and
0.063 for MA (see section VI.2 and Notes S4) were suppressed in (b). (a) Statistical significance among the response coefficients for the given plant functional type (paired ttests): ***, P < 0.001; **, P < 0.01; *, P < 0.05; ns, not significantly different. (b) The functional type-specific regressions were compared by SMATR version 2.0 which provides tests for differences among slopes, intercepts and elevations of standardized major axis regressions (Warton et al., 2006). A common slope test indicated that the NA vs MA plasticity relationship in herbs had a greater elevation (P < 0.001 for all data and P = 0.02 for a dataset in which C. acutiformis data exhibiting large negative values of MA plasticity were removed). New Phytologist (2015) 205: 973–993 www.newphytologist.com

Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist

Tansley reviews

(a)

(b)

(c)

Fig. 7 Variation in the response coefficients (Rc, section IV.4 and Notes S6, Eqns S6.1–S6.4) describing the light-dependent alteration in foliage photosynthetic capacity per area (AA) within and among plant functional types as illustrated by the box plots (see Fig. 2 for details of the box plots). As detailed in Fig. 6, the response coefficients provide a quantitative means to distribute the light-dependent variation in the target variable among the variations in the constituent traits. Here, the response coefficients for the three different functional forms for the expression of the composite trait AA are provided. Panels demonstrate: (a) the response coefficients for leaf dry mass per unit area (MA) and photosynthetic capacity per dry mass (AM) for the functional relationship AA = MAAM (Eqn 1); (b) the response coefficients for the nitrogen use efficiency (EN) and nitrogen content per area (NA) for the relationship AA = ENNA (Eqn 4); (c) the response coefficients for EN, MA and NM (Eqn 4). In all cases, the response coefficients for the given relationship sum up to 1. The latter relationship is the composite of that demonstrated in (b) and in Fig. 6(a). Because the response coefficients for measured (AA) and standardized (Notes S2, AA,V) variables were strongly correlated (r2 = 0.71–0.95), average response coefficients were calculated. The width of the boxes in each panel is proportional to the square root of the number of light gradients (n = 17–70 for all, except for the grasses in the functional relations including MA, where n = 9). The coding for the statistical significance among the mean values of the response coefficients for each plant functional type is the same as in Fig. 6a.

Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

Review 987

5. Quantitative partitioning of within-canopy photosynthetic variations among the constituent traits The quantitative limitation analysis (section IV.4 and Notes S6, Eqns S6.1–S6.4) makes it possible to separate the sources of variation of any given target trait among its components, thereby providing informative insight into the functional trait relations (Eqns 1–7). The response coefficients describing variation in NA NA (Eqn 3), the NA response coefficient for MA, RM , and that for NM, A NA RNM , strongly varied among plant functional types (Fig. 6a). In the case of annual forbs and grasses, RNNMA was significantly larger than NA RM , indicating that the light-dependent variation in NA was A primarily dependent on variation in RNNMA . In perennial forbs, RNNMA NA and RM did not differ statistically, suggesting that the control on A NA was shared by both constituent variables NM and MA. Different NA from the herbaceous species, RM was much larger than RNNMA in all A woody plant functional types, indicating that the variation in NA was mainly driven by changes in MA. Alternative analysis of covariation of light-dependent trait plasticity indicates that the relationship between MA and NA plasticities falls almost to a 1 : 1 line in woody species, whereas herbaceous species have a stronger increase in NA plasticity with increasing MA plasticity, demonstrating the additional role of NM plasticity in shaping NA plasticity in herbaceous species (Fig. 6b). By focusing on the quantitative light-dependent controls on AA by AM and MA (Eqn 1), we again indicate that the AA control is mainly attributable to changes in AM in herbaceous plant functional types, whereas changes in AA are dominated by MA in all woody plant functional types (Fig. 7a). In a similar manner, when the variation in AA is deconvoluted among the variations caused by NA and EN (Eqn 4), the contribution of the variations in EN is greater in herbaceous plant functional types and the contribution of the variations in NA is greater in woody plant functional types (Fig. 7b). Ultimately, when the changes in AA caused by alterations in EN and changes in the components of MA and NM (Fig. 7a, Eqn 4) are analyzed together, it becomes evident that, in forbs, AA changes primarily as a result of changes in both MA and EN, whereas, in grasses, as a result of changes in both EN and NM (Fig. 7c). In woody species, AA variation is most strongly determined by changes in MA, whereas the contributions of NM and EN are equally low in all cases, except for evergreen tropical woody species, where changes in EN contribute somewhat more to AA change than do changes in NM (Fig. 7c). Analyses of the sources of variation in biochemical photosynthesis potentials (Eqns 5–7) provide qualitatively similar evidence to the analyses of AA controls, although, because of the limited number of observations for several plant functional types, the results are less clearcut (Fig. 8). Especially in herbaceous species, when the effects of variations in all three contributors are analyzed together, variations in MA contribute somewhat more strongly to changes in the area-based photosynthetic potentials than do changes in the fractional nitrogen investment (Fig. 8c). This latter result apparently contradicts the greater role of variations in massbased photosynthesis potentials relative to MA changes in

New Phytologist (2015) 205: 973–993 www.newphytologist.com

988 Review

Tansley reviews

(a)

New Phytologist require the examination of functional controls among different herbaceous plant functional types as more data become available in the future. 6. Why is there invariability or variability of different traits within and among canopies?

(b)

(c)

Fig. 8 Quantitative separation of light-dependent variation sources in foliage biochemical potentials by response coefficient (Rc) analyses within and among plant functional types illustrated by the box plots (see Fig. 2 for the explanation of box plot characteristics). Foliage biochemical potentials are characterized by the biochemical photosynthesis model characteristics of Farquhar et al. (1980), the capacity for ribulose-1,5-bisphosphate carboxylation, Vcmax, and the capacity for photosynthetic electron transport, Jmax. As Vcmax and Jmax plasticities (Table 1) and the corresponding response coefficients for individual Vcmax and Jmax relationships were strongly correlated, we have used average response coefficients for ‘biochemical photosynthesis capacity’ (CA). The response coefficients are shown for three different functional forms of expression of foliage biochemical potentials specified by Eqns 5–7. Panels show: (a) the response coefficients for MA and CM (average of the corresponding response coefficients for Vcmax,M and Jmax,M) for the functional relationship CA = MACM; (b) the response coefficients for NA and Fi (average of the response coefficients for the fractions of nitrogen in Rubisco, FR, and in bioenergetics, FB) for the functional relationship CA ~ FiNA; (c) the response coefficients for MA, NM and Fi for the functional relationship CA ~ FiMANM. The statistical significance among the response coefficients for each plant functional type is the same as in Fig. 6a, except for the last comparison in (b), where P = 0.08.

herbaceous species (Fig. 8a). As the light-dependent plasticity for different traits varies among different herbaceous plant functional types (Fig. 5), gaining an insight into this discrepancy would New Phytologist (2015) 205: 973–993 www.newphytologist.com

The extent to which light-dependent changes in foliage photosynthetic potentials depend on variations in different underlying traits, and how these controls vary among plant functional types, have been poorly characterized to date. Analysis of the global within-canopy trait variations highlights that AA and foliage biochemical photosynthesis potentials increase with increasing Qint similarly in canopies of all major plant functional types (Table 1; Fig. 5d). However, detailed analyses of all constituent traits based on differences in r2 values, the number of nonsignificant or negative relationships (Fig. 3), differences in trait plasticities (Figs 4, 5, S2) and quantitative limitation analyses (Figs 6–8) collectively demonstrate that a similar AA plasticity is associated with different combinations of plasticities of underlying traits in different plant functional types. In particular, in woody plant functional types, AA increases with increasing light availability mostly because of increasing MA, whereas, in herbaceous plant functional types, the increase in AA is shared between changes in NM and EN (Figs 7, 8). We argue that these differences reflect fundamental contrasts in the functioning of different types of plant canopies. In the case of herbaceous species with high leaf turnover, the whole canopy is formed during a single growing season and the canopy grows continuously until the formation of inflorescences. Thus, all leaves have been exposed to high light availability at least at some part during their development. In addition to canopy vertical expansion from the top, senescence is induced in older shaded leaves in the lower canopy. There is ample experimental evidence that shading in fast-growing species with fast leaf turnover, especially shading of individual leaves, can introduce programmed cell death, leading to rapid declines in leaf nitrogen content, leaf photosynthetic capacity and, ultimately, to leaf abscission (Burkey & Wells, 1991; Pons & Pearcy, 1994; Ackerly & Bazzaz, 1995; Ono et al., 2001; Vos & van der Putten, 2001; Boonman et al., 2006). Although MA increases with increasing Qint in the canopies of several herbaceous species (Figs 1a, 5b), the increase is much less than the corresponding change in NM through the canopy (Figs 1b, 5c), such that the increase in NA (Figs 1c, 5a) is mainly dependent on the within-canopy gradient in NM. Furthermore, AM (Figs 1e, 5e) and EN (Figs 1d, 5f) strongly vary through the canopies of herbaceous species. All of these changes are compatible with shade-induced reallocation of nitrogen from components of photosynthetic machinery to leaves in upper high-light-exposed leaves, as well as with partial reacclimation to changed light conditions in shaded leaves. Thus, the overall evidence suggests that the within-canopy AA acclimation in herbaceous canopies is a highly dynamic active process. The situation is different in flush-type woody species, such as most winter-deciduous species and evergreen needle-leaved species, temperate and Mediterranean evergreens and many tropical evergreens, where periods of intensive leaf formation are broken Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist by dormancy or rest periods. In these species, all leaves or the given leaf flush are formed essentially at the same time, and the leaves are exposed to different light environments during their entire lifespan. All leaves are maintained through the entire season and there is little change in leaf nitrogen content and photosynthetic capacity until the onset of leaf senescence, which occurs at the end of the growing season in deciduous species (Niinemets et al., 2004a; Grassi et al., 2005) or typically somewhat before new leaf flush in evergreen species (Grubb, 1996; Lusk et al., 2003; Portillo-Estrada et al., 2013). In these species, there is a strong variation in MA through the canopy (Figs 1a, 5b), whereas NM (Figs 1b, 5c), AM (Figs 1e, 5e) and EN (Figs 1d, 5f) vary much less, and, accordingly, the within-canopy changes in AA are primarily determined by changes in MA. This implies that the formation of the within-canopy gradient in AA in these species is largely passive, reflecting the acclimation of MA to the canopy light gradient during leaf development. Nevertheless, these species possess a certain capacity to acclimate to sudden modifications in the light environment that can occur on gap formation in the canopy, but these changes are typically slow and constrained by the inability for structural modifications once the leaves have developed fully (e.g. Naidu & DeLucia, 1998; Niinemets et al., 2003; Oguchi et al., 2005). Of course, there are mixed response patterns between these two extreme mechanisms of acclimation to within-canopy light gradients (Fig. 5). In particular, within-canopy acclimation in fast-growing deciduous and evergreen woody canopies with continuous canopy growth approximates the acclimation in herbaceous canopies. In such stands, foliage developed earlier becomes increasingly shaded by the expanding canopy, leading to strong leaf age and light gradients (Vapaavuori & Vuorinen, 1989; Vapaavuori et al., 1989; Noormets et al., 1996; Kull et al., 1998; Weih, 2009). Similar to herbaceous species, MA is relatively invariable in fast-growing woody canopies, and the within-canopy changes in NA are chiefly determined by a strong gradient in NM (Vapaavuori & Vuorinen, 1989; Vapaavuori et al., 1989; Noormets et al., 1996; Kull et al., 1998), whereas the within-canopy variations in AA are driven by Qint-dependent increases in AM, NM and EN (Vapaavuori & Vuorinen, 1989; Vapaavuori et al., 1989).

VII. Conclusions: the economics spectrum for the within-canopy plasticity Across all data incorporated in the database, a vast variability in key leaf functional traits was observed (Table 1, Fig. 2). This vast variability is consistent with the worldwide leaf economics spectrum linking plant species with fast leaf turnover and high NM, AM and low MA at the high return end of the spectrum, and plant species with low leaf turnover and low NM, AM and high MA at the low return end of the spectrum (Wright et al., 2004, 2005; Westoby et al., 2013). The concept of the leaf economics spectrum has been criticized on the grounds that within-canopy variation is primarily driven by changes in area-based leaf traits (Lloyd et al., 2013). However, the worldwide analysis here suggests that the concerns expressed with regard to the appropriateness of Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

Tansley reviews

Review 989

mass-based traits in the context of light acclimation of photosynthetic potentials are not valid. Indeed, there is a spectrum of trait responses to the within-canopy light gradient within the leaf economics spectrum. Although, in species at the low return end of the economics spectrum, the acclimation response of AA is driven by changes in MA, in the canopies of herbaceous species at the high return end, AA varies mainly as a result of changes in NM and in fractions of nitrogen invested in photosynthetic machinery. Indeed, compared with the worldwide variation in leaf traits, the variation in within-canopy acclimation is much compressed, reflecting the opposite scaling of leaf functional traits (e.g. the insets in Fig. 2b, f) and different combinations of traits varying through plant canopies in different plant functional types. These compensatory responses lead to bafflingly low variability in AA plasticity across plant functional types (Fig. 5d). However, this ‘invariability’ of AA plasticity is a result of fundamentally different structural, chemical and allocational controls in plant functional types with varying leaf turnover. According to the ‘full optimality’ hypothesis of within-canopy photosynthetic acclimation, AA should be directly proportional to Qint. However, such a full optimality has never been found. Lack of ‘full optimality’ has been suggested to reflect constraints on photosynthetic acclimation (Niinemets & Valladares, 2004; Dewar et al., 2012; Niinemets, 2012; Peltoniemi et al., 2012). We suggest that plant canopies fundamentally differ in the suite of operating constraints. A limited rate of nitrogen reallocation and a limited share of nitrogen investments among leaf proteins of the photosynthetic apparatus and the rest constitute such constraints in species with high leaf turnover, whereas the structural construction limits define the constraints for within-canopy photosynthetic acclimation in species with low leaf turnover. We argue that studies attempting to resolve the controls on within-canopy variations in AA should analyze variations in structural, chemical and allocational traits collectively.

Acknowledgements We thank Kaia Kask for assistance in data digitization and database management, and Anne Aan, Niels Anten, Ismael  Aranda, Jan Cerm ak, Robin Chazdon, Theodore DeJong, Tim Fahey, Manfred Forstreuter, Yuko Hanba, Hiromitsu Kisanuki, Patrick Meir, Domingo Morales, Serge Rambal, Maria Soledad Jimenez Parrondo, Ingmar Tulva and Charlie Warren for providing details of the experimental design and/or unpublished data accompanying published datasets. We also thank Takayoshi Koike, Yukihiro Chiba and Ichiro Terashima for help with Japanese papers, and Susanne von Caemmerer for clarifications with regard to Rubisco kinetic characteristics. The manuscript greatly benefited from pertinent suggestions and critiques of three anonymous reviewers. Financial support for this study was provided by the Estonian Ministry of Science and Education (institutional grant IUT-8-3) and the European Commission through the European Regional Fund (Center of Excellence in Environmental Adaptation), the European Social Fund (postdoctoral grant MJD122) and the European Research Council (advanced grant 322603, SIP-VOL+). New Phytologist (2015) 205: 973–993 www.newphytologist.com

990 Review

Tansley reviews

References € 2004. Drought acclimation of two Aasamaa K, S~ober A, Hartung W, Niinemets U. deciduous tree species of different layers in a temperate forest canopy. Trees: Structure and Function 18: 93–101. Ackerly DD. 1992. Light, leaf age, and leaf nitrogen concentration in a tropical vine. Oecologia 89: 596–600. Ackerly DD, Bazzaz FA. 1995. Leaf dynamics, self-shading and carbon gain in seedlings of a tropical pioneer tree. Oecologia 101: 289–298. Anten NPR. 1997. Modelling canopy photosynthesis using parameters determined from simple non-destructive measurements. Ecological Research 12: 77–88. Anten NPR. 2005. Optimal photosynthetic characteristics of individual plants in vegetation stands and implications for species coexistence. Annals of Botany 95: 495–506. Anten NPR, Hirose T. 1998. Biomass allocation and light partitioning among dominant and subordinate individuals in Xanthium canadense stands. Annals of Botany 82: 665–673. Anten NPR, Miyazawa K, Hikosaka K, Nagashima H, Hirose T. 1998a. Leaf nitrogen distribution in relation to leaf age and photon flux density in dominant and subordinate plants in dense stands of a dicotyledonous herb. Oecologia 113: 314–324. Anten NPR, Schieving F, Werger MJA. 1995. Patterns of light and nitrogen distribution in relation to whole canopy carbon gain in C3 and C4 mono- and dicotyledonous species. Oecologia 101: 504–513. Anten NPR, Werger MJA, Medina E. 1998b. Nitrogen distribution and leaf area indices in relation to photosynthetic nitrogen use efficiency in savanna grasses. Plant Ecology 138: 63–75. Baldocchi DD, Harley PC. 1995. Scaling carbon dioxide and water vapour exchange from leaf to canopy in a deciduous forest. II. Model testing and application. Plant, Cell & Environment 18: 1157–1173. Bonan GB, Oleson KW, Fisher RA, Lasslop G, Reichstein M. 2012. Reconciling leaf physiological traits and canopy flux data: use of the TRY and FLUXNET databases in the Community Land Model version 4. Journal of Geophysical Research – Biogeosciences 117: G02026. Boonman A, Anten NPR, Dueck TA, Jordi WJRM, van der Werf A, Voesenek LACJ, Pons TL. 2006. Functional significance of shade-induced leaf senescence in dense canopies: an experimental test using transgenic tobacco. The American Naturalist 168: 597–607. Boonman A, Prinsen E, Voesenek LACJ, Pons TL. 2009. Redundant roles of photoreceptors and cytokinins in regulating photosynthetic acclimation to canopy density. Journal of Experimental Botany 60: 1179–1190. Brodribb TJ, Holbrook NM, Zwieniecki MA, Palma B. 2005. Leaf hydraulic capacity in ferns, conifers and angiosperms: impacts on photosynthetic maxima. New Phytologist 165: 839–846. Burkey KO, Wells R. 1991. Response of soybean photosynthesis and chloroplast membrane function to canopy development and mutual shading. Plant Physiology 97: 245–252. Cannell MGR, Thornley JHM. 1998. Temperature and CO2 responses of leaf and canopy photosynthesis: a clarification using the non-rectangular hyperbola model of photosynthesis. Annals of Botany 82: 883–892. Causton DR, Dale MP. 1990. The monomolecular and rectangular hyperbola as empirical models of the response of photosynthetic rate to photon flux density, with applications to three Veronica species. Annals of Botany 65: 389–394. € 2004. Sunlight capture. Leaf to landscape. In: Smith WK, Cescatti A, Niinemets U. Vogelmann TC, Chritchley C, eds. Photosynthetic adaptation. Chloroplast to landscape. Berlin, Germany: Springer Verlag, 42–85. Cescatti A, Zorer R. 2003. Structural acclimation and radiation regime of silver fir (Abies alba Mill.) shoots along a light gradient. Plant, Cell & Environment 26: 429–442. Ceulemans R, Impens I, Moermans R. 1980. The response of CO2 exchange rate to photosynthetic photon flux density for several Populus clones under laboratory conditions. Photosynthesis Research 1: 137–142. Chazdon RL, Field CB. 1987. Determinants of photosynthetic capacity in six rainforest Piper species. Oecologia 73: 222–230. DeJong TM, Day KR, Johnson RS. 1989. Partitioning of leaf nitrogen with respect to within canopy light exposure and nitrogen availability in peach (Prunus persica). Trees: Structure and Function 3: 89–95. New Phytologist (2015) 205: 973–993 www.newphytologist.com

New Phytologist Dewar RC, Tarvainen L, Parker K, Wallin G, McMurtrie RE. 2012. Why does leaf nitrogen decline within tree canopies less rapidly than light? An explanation from optimization subject to a lower bound on leaf mass per area. Tree Physiology 32: 520–534. Dickmann DI, Michael DA, Isebrands JG, Westin S. 1990. Effects of leaf display on light interception and apparent photosynthesis in two contrasting Populus cultivars during their second growing season. Tree Physiology 7: 7–20. Ellsworth DS, Reich PB. 1993. Canopy structure and vertical patterns of photosynthesis and related leaf traits in a deciduous forest. Oecologia 96: 169– 178. Escudero A, Mediavilla S. 2003. Decline in photosynthetic nitrogen use efficiency with leaf age and nitrogen resorption as determinants of leaf life span. Journal of Ecology 91: 880–889. Ethier GJ, Livingston NJ, Harrison DL, Black TA, Moran JA. 2006. Low stomatal and internal conductance to CO2 versus Rubisco deactivation as determinants of the photosynthetic decline of ageing evergreen leaves. Plant, Cell & Environment 29: 2168–2184. Evans JR. 1988. Acclimation by the thylakoid membranes to growth irradiance and the partitioning of nitrogen between soluble and thylakoid proteins. In: Evans JR, von Caemmerer S, Adams WW III, eds. Ecology of photosynthesis in sun and shade. Melbourne, Vic., Australia: CSIRO, 93–106. Evans JR. 1989. Photosynthesis and nitrogen relationships in leaves of C3 plants. Oecologia 78: 9–19. Evans JR. 1993. Photosynthetic acclimation and nitrogen partitioning within a lucerne canopy. I. Canopy characteristics. Australian Journal of Plant Physiology 20: 55–67. Evans JR, Poorter H. 2001. Photosynthetic acclimation of plants to growth irradiance: the relative importance of specific leaf area and nitrogen partitioning in maximizing carbon gain. Plant, Cell & Environment 24: 755–767. Evans JR, Seemann JR. 1989. The allocation of protein nitrogen in the photosynthetic apparatus: costs, consequences, and control. In: Briggs WR, ed. Photosynthesis. Proceedings of the C.S. French Symposium on Photosynthesis held in Stanford, California, July 17–23, 1988. New York, NY, USA: Alan R. Liss, Inc., 183–205. Ewers BE, Oren R, Kim H-S, Bohrer G, Lai C-T. 2007. Effects of hydraulic architecture and spatial variation in light on mean stomatal conductance of tree branches and crowns. Plant, Cell & Environment 30: 483–496. Farquhar GD, von Caemmerer S, Berry JA. 1980. A biochemical model of photosynthetic CO2 assimilation in leaves of C3 species. Planta 149: 78–90. Field C, Mooney HA. 1986. The photosynthesis–nitrogen relationship in wild plants. In: Givnish TJ, ed. On the economy of plant form and function. Proceedings of the Sixth Maria Moors Cabot Symposium, “Evolutionary constraints on primary productivity: adaptive patterns of energy capture in plants”, Harvard Forest, August 1983. Cambridge, UK: Cambridge University Press, 25–55. € Cescatti A, Tenhunen JD. 2003. Three-dimensional lamina Fleck S, Niinemets U, architecture alters light harvesting efficiency in Fagus: a leaf-scale analysis. Tree Physiology 23: 577–589. Flexas J, Barbour MM, Brendel O, Cabrera HM, Carriquı M, Diaz-Espejo A, Douthe C, Dreyer E, Ferrio JP, Gago J et al. 2012. Mesophyll diffusion conductance to CO2: an unappreciated central player in photosynthesis. Plant Science 193–194: 70–84. Galmes J, Andralojc PJ, Kapralov MV, Flexas J, Keys AJ, Molins A, Parry MAJ,  2014a. Environmentally driven evolution of Rubisco and improved Conesa MA. photosynthesis and growth within the C3 genus Limonium (Plumbaginaceae). New Phytologist 203: 989–999.  Dıaz-Espejo A, Mir A, Perdomo JA, Niinemets U, € Flexas J. Galmes J, Conesa MA, 2014b. Rubisco catalytic properties optimized for present and future climatic conditions. Plant Science 226: 61–70. Grassi G, Vicinelli E, Ponti F, Cantoni L, Magnani F. 2005. Seasonal and interannual variability of photosynthetic capacity in relation to leaf nitrogen in a deciduous forest plantation in northern Italy. Tree Physiology 25: 349–360. Grubb PJ. 1996. Rainforest dynamics: the need for new paradigms. In: Edwards DS, Booth WE, Choy SC, eds. Tropical rainforest research. Dordrecht, the Netherlands: Kluwer Academic Publishers, 215–233. Gutschick VP, Wiegel FW. 1988. Optimizing the canopy photosynthetic rate by patterns of investment in specific leaf mass. The American Naturalist 132: 67– 86. Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist Han Q, Araki M, Chiba Y. 2006. Acclimation to irradiance of leaf photosynthesis and associated nitrogen reallocation in photosynthetic apparatus in the year following thinning of a young stand of Chamaecyparis obtusa. Photosynthetica 44: 523–529. Han Q, Chiba Y. 2009. Leaf photosynthetic responses and related nitrogen changes associated with crown reclosure after thinning in a young Chamaecyparis obtusa stand. Journal of Forest Research 14: 349–357. Harley PC, Thomas RB, Reynolds JF, Strain BR. 1992. Modelling photosynthesis of cotton grown in elevated CO2. Plant, Cell & Environment 15: 271–282. Hikosaka K, Hanba YT, Hirose T, Terashima I. 1998. Photosynthetic nitrogen-use efficiency in leaves of woody and herbaceous species. Functional Ecology 12: 896– 905. Hikosaka K, Sudoh S, Hirose T. 1999. Light acquisition and use by individuals competing in a dense stand of an annual herb, Xanthium canadense. Oecologia 118: 388–396. Hirose T, Bazzaz FA. 1998. Trade-off between light- and nitrogen-use efficiency in canopy photosynthesis. Annals of Botany 82: 195–202. Hirose T, Werger MJA. 1987a. Maximizing daily canopy photosynthesis with respect to the leaf nitrogen allocation pattern in the canopy. Oecologia 72: 520– 526. Hirose T, Werger MJA. 1987b. Nitrogen use efficiency in instantaneous and daily photosynthesis of leaves in the canopy of a Solidago altissima stand. Physiologia Plantarum 70: 215–222. Hirose T, Werger MJA. 1994. Photosynthetic capacity and nitrogen partitioning among species in the canopy of a herbaceous plant community. Oecologia 100: 203–212. Hirose T, Werger MJA. 1995. Canopy structure and photon flux partitioning among species in a herbaceous plant community. Ecology 76: 466–474. Hirose T, Werger MJA, Pons TL, van Rheenen JWA. 1988. Canopy structure and leaf nitrogen distribution in a stand of Lysimachia vulgaris L. as influenced by stand density. Oecologia 77: 145–150. Hirose T, Werger MJA, van Rheenen JWA. 1989. Canopy development and leaf nitrogen distribution in a stand of Carex acutiformis. Ecology 70: 1610–1618. Hollinger DY. 1989. Canopy organization and foliage photosynthetic capacity in broad-leaved evergreen montane forest. Functional Ecology 3: 53–62. Hollinger DY. 1996. Optimality and nitrogen allocation in a tree canopy. Tree Physiology 16: 627–634. Hubbard RM, Bond BJ, Senock RS, Ryan MG. 2002. Effects of branch height on leaf gas exchange, branch hydraulic conductance and branch sap flux in open-grown ponderosa pine. Tree Physiology 22: 575–581. Iio A, Fukasawa H, Nose Y, Kato S, Kakubari Y. 2005. Vertical, horizontal and azimuthal variations in leaf photosynthetic characteristics within a Fagus crenata crown in relation to light acclimation. Tree Physiology 25: 525–536. Ishida A, Nakano T, Uemura A, Yamashita N, Tanabe H, Koike N. 2001. Light-use properties in two sun-adapted shrubs with contrasting canopy structures. Tree Physiology 21: 497–504. Joffre R, Rambal S, Damesin C. 2007. Functional attributes of Mediterranean-type ecosystems. In: Pugnaire FI, Valladares F, eds. Handbook of functional plant ecology. Boca Raton, FL, USA: CRC Press, 285–312. Jordan DB, Ogren WL. 1984. The CO2/O2 specificity of ribulose 1,5-bisphosphate carboxylase/oxygenase. Dependence on ribulose bisphosphate concentration, pH and temperature. Planta 161: 308–313. Joyce BJ, Steiner KC. 1995. Systematic variation in xylem hydraulic capacity within the crown of white ash (Fraxinus americana). Tree Physiology 15: 649–656. Kattge J, Dıaz S, Lavorel S, Prentice C, Leadley P, B€onisch G, Garnier E, Westoby M, Reich PB, Wright IJ et al. 2011. TRY – a global database of plant traits. Global Change Biology 17: 2905–2935. Koike T, Kitao M, Maruyama Y, Mori S, Lei TT. 2001. Leaf morphology and photosynthetic adjustments among deciduous broad-leaved trees within the vertical canopy profile. Tree Physiology 21: 951–958. Kull O, Koppel A, Noormets A. 1998. Seasonal changes in leaf nitrogen pools in two Salix species. Tree Physiology 18: 45–51. Kull O, Kruijt B. 1999. Acclimation of photosynthesis to light: a mechanistic approach. Functional Ecology 13: 24–36. Le Roux X, Sinoquet H, Vandame M. 1999. Spatial distribution of leaf dry weight per area and leaf nitrogen concentration in relation to local radiation regime within an isolated tree crown. Tree Physiology 19: 181–188. Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

Tansley reviews

Review 991

Le Roux X, Walcroft AS, Daudet FA, Sinoquet H, Chaves MM, Rodrigues A, Osorio L. 2001. Photosynthetic light acclimation in peach leaves: importance of changes in mass : area ratio, nitrogen concentration, and leaf nitrogen partitioning. Tree Physiology 21: 377–386. Lloyd J, Bloomfield K, Domingues TF, Farquhar GD. 2013. Photosynthetically relevant foliar traits correlating better on a mass vs an area basis: of ecophysiological relevance or just a case of mathematical imperatives and statistical quicksand? New Phytologist 199: 311–321. Lloyd J, Pati~ no S, Paiva RQ, Nardoto GB, Quesada CA, Santos AJB, Baker TR, Brand WA, Hilke I, Gielmann H et al. 2010. Optimisation of photosynthetic carbon gain and within-canopy gradients of associated foliar traits for Amazon forest trees. Biogeosciences 7: 1833–1859. Lusk CH, Matus F, Moreno-Chacon M, Salda~ na A, Jimenez-Castillo M. 2003. Seasonal variation in leaf litter nutrient concentrations of Valdivian rainforest trees. Gayana. Bota nica 60: 35–39. Lynch J, Gonza lez A. 1993. Canopy nutrient allocation in relation to incident light in the tropical fruit tree Borojoa patinoi (Cuatr.). Journal of the American Society for Horticultural Science 118: 777–785. Medhurst JL, Beadle CL. 2005. Photosynthetic capacity and foliar nitrogen distribution in Eucalyptus nitens is altered by high-intensity thinning. Tree Physiology 25: 981–991. Montgomery DC. 2012. Design and analysis of experiments. New York, NY, USA: Wiley. Naidu SL, DeLucia EH. 1998. Physiological and morphological acclimation of shade-grown tree seedlings to late-season canopy gap formation. Plant Ecology 138: 27–40. € 1998. Adjustment of foliage structure and function to a canopy Niinemets U. light gradient in two co-existing deciduous trees. Variability in leaf inclination angles in relation to petiole morphology. Trees: Structure and Function 12: 446– 451. € 1999. Research review. Components of leaf dry mass per area – Niinemets U. thickness and density – alter leaf photosynthetic capacity in reverse directions in woody plants. New Phytologist 144: 35–47. € 2012. Commentary. Optimization of foliage photosynthetic capacity Niinemets U. in tree canopies: towards identifying missing constraints. Tree Physiology 32: 505– 509. € Anten NPR. 2009. Packing the photosynthesis machinery: from leaf Niinemets U, to canopy. In: Laisk A, Nedbal L, Govindjee, eds. Photosynthesis in silico: understanding complexity from molecules to ecosystems. Berlin, Germany: Springer Verlag, 363–399. € Cescatti A, Lukjanova A, Tobias M, Truus L. 2002. Modification of Niinemets U, light-acclimation of Pinus sylvestris shoot architecture by site fertility. Agricultural and Forest Meteorology 111: 121–140. € Cescatti A, Rodeghiero M, Tosens T. 2006. Complex adjustments of Niinemets U, photosynthetic capacity and internal mesophyll conductance to current and previous light availabilities and leaf age in Mediterranean evergreen species Quercus ilex. Plant, Cell & Environment 29: 1159–1178. € Fleck S. 2002. Petiole mechanics, leaf inclination, morphology, and Niinemets U, investment in support in relation to light availability in the canopy of Liriodendron tulipifera. Oecologia 132: 21–33. € Keenan TF. 2012. Measures of light in studies on light-driven plant Niinemets U, plasticity in artificial environments. Frontiers in Plant Science 3: 156. € Kollist H, Garcıa-Plazaola JI, Herna ndez A, Becerril JM. 2003. Do the Niinemets U, capacity and kinetics for modification of xanthophyll cycle pool size depend on growth irradiance in temperate trees? Plant, Cell & Environment 26: 1787–1801. € Kull O. 1998. Stoichiometry of foliar carbon constituents varies along Niinemets U, light gradients in temperate woody canopies: implications for foliage morphological plasticity. Tree Physiology 18: 467–479. € Kull O, Tenhunen JD. 1998. An analysis of light effects on foliar Niinemets U, morphology, physiology, and light interception in temperate deciduous woody species of contrasting shade tolerance. Tree Physiology 18: 681–696. € Kull O, Tenhunen JD. 2004a. Within canopy variation in the rate of Niinemets U, development of photosynthetic capacity is proportional to integrated quantum flux density in temperate deciduous trees. Plant, Cell & Environment 27: 293– 313. € Oja V, Kull O. 1999a. Shape of leaf photosynthetic electron Niinemets U, transport versus temperature response curve is not constant along canopy light New Phytologist (2015) 205: 973–993 www.newphytologist.com

992 Review

Tansley reviews

gradients in temperate deciduous trees. Plant, Cell & Environment 22: 1497– 1514. € S~ober A, Kull O, Hartung W, Tenhunen JD. 1999b. Apparent Niinemets U, controls on leaf conductance by soil water availability and via light-acclimation of foliage structural and physiological properties in a mixed deciduous, temperate forest. International Journal of Plant Sciences 160: 707–721. € Sonninen E, Tobias M. 2004b. Canopy gradients in leaf intercellular Niinemets U, CO2 mole fractions revisited: interactions between leaf irradiance and water stress need consideration. Plant, Cell & Environment 27: 569–583. € Tenhunen JD. 1997. A model separating leaf structural and Niinemets U, physiological effects on carbon gain along light gradients for the shade-tolerant species Acer saccharum. Plant, Cell & Environment 20: 845–866. € Tenhunen JD, Canta NR, Chaves MM, Faria T, Pereira JS, Reynolds Niinemets U, JF. 1999c. Interactive effects of nitrogen and phosphorus on the acclimation potential of foliage photosynthetic properties of cork oak, Quercus suber, to elevated atmospheric CO2 concentrations. Global Change Biology 5: 455–470. € Valladares F. 2004. Photosynthetic acclimation to simultaneous and Niinemets U, interacting environmental stresses along natural light gradients: optimality and constraints. Plant Biology 6: 254–268. Noormets A, Kull O, Koppel A. 1996. Nitrogen dynamics in Salix leaves during the first production year. In: Perttu K, Koppel A, eds. Short rotation willow coppice for renewable energy and improved environment. Proceedings of a joint Swedish– Estonian seminar on energy forestry and vegetation filters held in Tartu 24–26 September 1995. Uppsala, Sweden: Swedish University of Agricultural Sciences, 51–59. Oberbauer SF, Strain BR. 1986. Effects of canopy position and irradiance on the leaf physiology and morphology of Pentaclethra macroloba (Mimosaceae). American Journal of Botany 73: 409–416. Oberbauer SF, Strain BR, Riechers GH. 1987. Field water relations of a wet-tropical forest tree species, Pentaclethra macroloba (Mimosaceae). Oecologia 71: 369–374. Oguchi R, Hikosaka K, Hirose T. 2005. Leaf anatomy as a constraint for photosynthetic acclimation: differential responses in leaf anatomy to increasing growth irradiance among three deciduous trees. Plant, Cell & Environment 28: 916–927. Ono K, Nishi Y, Watanabe A, Terashima I. 2001. Possible mechanisms of adaptive leaf senescence. Plant Biology 3: 234–243. Osada N, Takeda H, Furukawa A, Awang M. 2001. Leaf dynamics and maintenance of tree crowns in a Malaysian rain forest stand. Journal of Ecology 89: 774–782. Palmroth S, Stenberg P, Smolander S, Voipio P, Smolander H. 2002. Fertilization has little effect on light-interception efficiency of Picea abies shoots. Tree Physiology 22: 1185–1192. Patrick LD, Ogle K, Tissue DT. 2009. A hierarchical Bayesian approach for estimation of photosynthetic parameters of C3 plants. Plant, Cell & Environment 32: 1695–1709. Peltoniemi M, Duursma R, Medlyn B. 2012. Co-optimal distribution of leaf nitrogen and hydraulic conductance in plant canopies. Tree Physiology 32: 510–519. Pons TL, Anten NPR. 2004. Is plasticity in partitioning of photosynthetic resources between and within leaves important for whole-plant carbon gain in canopies? Functional Ecology 18: 802–811. Pons TL, Jordi W, Kuiper D. 2001. Acclimation of plants to light gradients in leaf canopies: evidence for a possible role for cytokinins transported in the transpiration stream. Journal of Experimental Botany 52: 1563–1574. Pons TL, Pearcy RW. 1994. Nitrogen reallocation and photosynthetic acclimation in response to partial shading in soybean plants. Physiologia Plantarum 92: 636– 644. Pons TL, van Rijnberk H, Scheurwater I, van der Werf A. 1993. Importance of the gradient in photosynthetically active radiation in a vegetation stand for leaf nitrogen allocation in two monocotyledons. Oecologia 95: 416–424. Pons TL, Westbeek MHM. 2004. Analysis of differences in photosynthetic nitrogen-use efficiency between four contrasting species. Physiologia Plantarum 122: 68–78. Poorter H, Lambers H, Evans JR. 2014. Trait correlation networks: a whole-plant perspective on the recently criticized leaf economic spectrum. New Phytologist 201: 378–382. New Phytologist (2015) 205: 973–993 www.newphytologist.com

New Phytologist Poorter H, Nagel O. 2000. The role of biomass allocation in the growth response of plants to different levels of light, CO2, nutrients and water: a quantitative review. Australian Journal of Plant Physiology 27: 595–607. € Poorter L, Wright IJ, Villar R. 2009. Tansley review. Poorter H, Niinemets U, Causes and consequences of variation in leaf mass per area (LMA): a meta-analysis. New Phytologist 182: 565–588. € Walter A, Fiorani F, Schurr U. 2010. A method to Poorter H, Niinemets U, construct dose–response curves for a wide range of environmental factors and plant traits by means of a meta-analysis of phenotypic data. Journal of Experimental Botany 61: 2043–2055. Portillo-Estrada M, Korhonen JFJ, Pihlatie M, Pumpanen J, Frumau KFA, € 2013. Inter- and intra-annual variations in Morillas L, Tosens T, Niinemets U. canopy fine litterfall and carbon and nitrogen inputs to the forest floor in two European coniferous forests. Annals of Forest Science 70: 367–379. Posada JM, Lechowicz MJ, Kitajima K. 2009. Optimal photosynthetic use of light by tropical tree crowns achieved by adjustment of individual leaf angles and nitrogen content. Annals of Botany 103: 795–805. Rambal S. 2001. Productivity of Mediterranean-type ecosystems. In: Mooney HA, Saugier B, Roy J, eds. Terrestrial global productivity: past, present, and future. San Diego, CA, USA: Academic Press, Inc, 315–344. Renninger HJ, Meinzer FC, Gartner BL. 2006. Hydraulic architecture and photosynthetic capacity as constraints on release from suppression in Douglas-fir and western hemlock. Tree Physiology 27: 33–42. Rogers A. 2014. The use and misuse of Vc,max in Earth System Models. Photosynthesis Research 119: 15–29. Rosati A, Day KR, DeJong TM. 2000. Distribution of leaf mass per unit area and leaf nitrogen concentration determine partitioning of leaf nitrogen within tree canopies. Tree Physiology 20: 271–276. Schreiber U, Bilger W, Neubauer C. 1994. Chlorophyll fluorescence as a nonintrusive indicator for rapid assessment of in vivo photosynthesis. In: Schulze E-D, Caldwell MM, eds. Ecophysiology of photosynthesis. Berlin, Germany: Springer Verlag, 49–70. Sellin A, Kupper P. 2004. Within-crown variation in leaf conductance in Norway spruce: effects of irradiance, vapour pressure deficit, leaf water status and plant hydraulic constraints. Annals of Forest Science 61: 419–429. Sharkey TD. 1985. Photosynthesis in intact leaves of C3 plants: physics, physiology and rate limitations. The Botanical Review 51: 53–105. Shipley B, Lechowicz MJ, Wright IJ, Reich PB. 2006. Fundamental tradeoffs generating the worldwide leaf economics spectrum. Ecology 87: 535–541. Tjoelker MG, Volin JC, Oleksyn J, Reich PB. 1995. Interaction of ozone pollution and light effects on photosynthesis in a forest canopy experiment. Plant, Cell & Environment 18: 895–905. Traw MB, Ackerly DD. 1995. Leaf position, light levels, and nitrogen allocation in five species of rain forest pioneer trees. American Journal of Botany 82: 1137–1143. Urban L, Le Roux X, Sinoquet H, Jaffuel S, Jannoyer M. 2003. A biochemical model of photosynthesis for mango leaves: evidence for the effect of fruit on photosynthetic capacity of nearby leaves. Tree Physiology 23: 289–300. Valladares F. 2003. Light heterogeneity and plants: from ecophysiology to species coexistence and biodiversity. In: Esser K, L€ uttge U, Beyschlag W, Hellwig F, eds. Progress in botany. Berlin, Germany: Springer Verlag, 439–471. € 2007. The architecture of plant crowns: from design Valladares F, Niinemets U. rules to light capture and performance. In: Pugnaire FI, Valladares F, eds. Handbook of functional plant ecology. Boca Raton, FL, USA: CRC Press, 101–149. Vapaavuori EM, Nurmi A, Vuorinen AH, Kangas T. 1989. Seasonal variation in the photosynthetic capacity of a willow (Salix cv. Aquatica gigantea) canopy. 2. Comparison of the structure and function of chloroplasts at different levels in the canopy. Tree Physiology 5: 445–457. Vapaavuori EM, Vuorinen AH. 1989. Seasonal variation in the photosynthetic capacity of a willow (Salix cv. Aquatica gigantea) canopy. I. Changes in the activity and amount of ribulose 1,5-bisphosphate carboxylase-oxygenase and the content of nitrogen and chlorophyll at different levels in the canopy. Tree Physiology 5: 423–444. Vos J, van der Putten PEL. 2001. Effects of partial shading of the potato plant on photosynthesis of treated leaves, leaf area expansion and allocation of nitrogen and dry matter in component plant parts. European Journal of Agronomy 14: 209– 220. Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist Walcroft A, Le Roux X, Diaz-Espejo A, Dones N, Sinoquet H. 2002. Effects of crown development on leaf irradiance, leaf morphology and photosynthetic capacity in a peach tree. Tree Physiology 22: 929–938. Warton DI, Wright IJ, Falster DS, Westoby M. 2006. Bivariate line-fitting methods for allometry. Biological Reviews of the Cambridge Philosophical Society 81: 259–291. Way DA, Yamori W. 2014. Thermal acclimation of photosynthesis: on the importance of adjusting our definitions and accounting for thermal acclimation of respiration. Photosynthesis Research 119: 89–100. Weih M. 2009. Genetic and environmental variation in spring and autumn phenology of biomass willows (Salix spp.): effects on shoot growth and nitrogen economy. Tree Physiology 29: 1479–1490. Werger MJA, Hirose T. 1988. Effects of light climate and nitrogen partitioning on the canopy structure of stands of a dicotyledonous, herbaceous vegetation. In: Werger MJA, van der Aart PJM, During HJ, Verhoeven JTA, eds. Plant form and vegetation structure. Adaptation, plasticity and relation to herbivory. The Hague, the Netherlands: SPB Academic Publishing, 171–181. Westoby M, Reich PB, Wright IJ. 2013. Understanding ecological variation across species: area-based vs mass-based expression of leaf traits. New Phytologist 199: 322–323. Wright IJ, Reich PB, Cornelissen JHC, Falster DS, Garnier E, Hikosaka K, Lamont BB, Lee W, Oleksyn J, Osada N et al. 2005. Assessing the generality of global leaf trait relationships. New Phytologist 166: 485–496. Wright IJ, Reich PB, Westoby M, Ackerly DD, Baruch Z, Bongers F, Cavender-Bares J, Chapin T, Cornelissen JHC, Diemer M et al. 2004. The world-wide leaf economics spectrum. Nature 428: 821–827. Yamori W, Suzuki K, Noguchi K, Nakai M, Terashima I. 2006. Effects of Rubisco kinetics and Rubisco activation state on the temperature dependence of the photosynthetic rate in spinach leaves from contrasting growth temperatures. Plant, Cell & Environment 29: 1659–1670. Yasumura Y, Hikosaka K, Matsui K, Hirose T. 2002. Leaf-level nitrogen-use efficiency of canopy and understorey species in a beech forest. Functional Ecology 16: 826–834.

Supporting Information Additional supporting information may be found in the online version of this article. Additional references can be found here. Fig. S1 Cumulative proportion of within-canopy light gradients with the maximum average seasonal integrated quantum flux density (Qint) equal to or greater than the given threshold value Qint,max, and the cumulative proportion of light gradients with the minimum Qint equal to or smaller than the given threshold value Qint,min. Fig. S2 Sample relationships of the variation of relative lightdependent plasticity with the integrated light range in forbs and winter-deciduous trees.

Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

Tansley reviews

Review 993

Table S1 List of species, species classification, trait coverage in the database, light gradient characteristics and data sources of the global within-canopy trait database Table S2 Kinetic characteristics of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) used across the studies and standardization factors used for the maximum Rubisco carboxylase activity Table S3 Results of statistical comparisons among plant functional types Notes S1 Leaf area conversions for needle-leaved species. Notes S2 Estimation of biochemical photosynthesis model parameters, their standardization, consideration of gross photosynthesis data and estimation of standardized photosynthetic capacity. Notes S3 Additional methods for the estimation of the integrated quantum flux density (Qint). Notes S4 Constraints employed in fitting non-linear relationships, consideration of non-asymptotic responses and statistical thresholds. Notes S5 Constraints applied in the calculation of relative plasticity and trait values at given light levels, determination of normalized plasticity, and consideration of plasticity estimates for measured and standardized photosynthetic traits. Notes S6 Quantitative limitation analysis: definition of response coefficients and computation details. Notes S7 Additional references cited in the online supporting information. These references form an integral part of the article. Please note: Wiley Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

New Phytologist (2015) 205: 973–993 www.newphytologist.com

This document is a scanned copy of a printed document. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material.