Introduction

Across biomes, leaf functional traits are important for predicting leaf and ecosystem functioning (Wright et al. 2004; Poorter et al. 2009). Coupled with large-scale relationships developed for plant albedo and leaf nitrogen (Ollinger et al. 2008; Hollinger et al. 2010), these patterns in leaf functional traits [leaf nitrogen per unit leaf area (Narea), leaf mass per area (LMA), and photosynthetic capacity] have been used to estimate gross primary productivity from local to global scales (Ryu et al. 2011). Vertical patterns in LMA and carbon:nitrogen (C:N) ratio have been incorporated into canopy process and land surface component models that predict carbon flux and forest response to changes in environmental conditions (Gutschick and Wiegel 1988; Raulier et al. 1999; Hanson et al. 2004; Medlyn 2004; Thornton and Zimmermann 2007). Consequently, leaf functional traits are useful for scaling from leaf- to ecosystem- to global-level processes when modeling carbon, water, and nutrient cycling.

A central theme when modeling forest canopy photosynthesis is the assumption that structural carbon and leaf nitrogen concentrations are optimally distributed with respect to light to maximize carbon gain (Field 1983; Hirose and Werger 1987; Sellers et al. 1992; Amthor 1994). During acclimation to high light availability, greater investment of carbon into leaf construction (high LMA) and nitrogen into RUBISCO and thylakoid proteins (high Narea) often results in higher rates of light-saturated photosynthesis and greater carbon isotope composition (δ13C) (Evans 1989; Hollinger 1989; Ellsworth and Reich 1993; Berry et al. 1997; Niinemets 1997; Livingston et al. 1998; Bond et al. 1999; Niinemets et al. 1999, 2001; Sack et al. 2006; Duursma and Marshall 2006). Thus, leaf functional traits (LMA, Narea, and photosynthetic capacity) and δ13C are often coordinated and scale with light within tree crowns and forest canopies, corroborating optimal patterns derived from models (Hirose and Werger 1987, Livingston et al. 1998; Ellsworth and Reich 1993; Duursma and Marshall 2006).

However, optimal patterns in photosynthetic capacity with respect to light may be constrained by other environmental factors, resulting in a decline in photosynthetic capacity per unit of irradiance that is often observed in field studies (Hollinger 1996; Bond et al. 1999; Niinemets and Valladares 2004; Buckley et al. 2013). For example, in tall Sequoia sempervirens and Pseudotsuga menziesii (e.g., up to 113 m; Koch et al. 2004; Woodruff et al. 2004; Burgess and Dawson 2007; Ishii et al. 2008) and in shorter tropical (e.g., up to 45 m; Cavaleri et al. 2010) and temperate deciduous (e.g., up to 18 m; Zwieniecki et al. 2004; Zhang et al. 2011b) trees, gravity and the length of the hydraulic pathway result in water potential gradients with height, potentially limiting leaf development as indicated by LMA. Hydraulic limitation on leaf structure can increase leaf tissue density and cell wall thickness and reduce mesophyll air-space, potentially restricting mesophyll conductance to CO2 and photosynthesis, as indicated by increasing δ13C with canopy height (Koch et al. 2004; Niinemets et al. 2004; Mullin et al. 2009; Oldham et al. 2010). The height at which water potential is limiting to leaf development is likely to vary among species depending on their hydraulic characteristics. Leaf Narea measured along vertical gradients has been linked to changes in LMA primarily due to the conversion of Nmass to Narea through LMA and the constant values of Nmass (Ellsworth and Reich 1993; Bond et al. 1999). Therefore, any constraints on LMA are likely to have similar constraints on Narea. Consistent with these observations, model-based approaches have attributed the discrepancy between theoretical patterns (optimal) and actual patterns (suboptimal) in photosynthetic capacity and leaf nitrogen to hydraulic constraints (Peltoniemi et al. 2012; Buckley et al. 2013). When hydraulic constraints are considered, leaf nitrogen and photosynthetic capacity are not directly proportional to light, likely due to the direct and indirect effects of greater xylem tension on either stomatal or mesophyll conductance to CO2 (Peltoniemi et al. 2012; Buckley et al. 2013). Evidence also suggests that the discrepancy among theoretical and actual patterns in leaf nitrogen may be an artifact of light models that do not account for diffuse and direct light (Hikosaka 2014).

A major difficulty in understanding the effects of light conditions on leaf functional traits is confounding water potential and light gradients with height. Experimental shading can be a useful approach in teasing apart the effects of light and leaf water potential, testing mechanistic hypotheses, and providing insight into interrelated leaf traits and chemistry. Branch-level shading has shown that leaf function (i.e., photosynthesis, respiration, and leaf nitrogen) can acclimate to shading despite no structural changes (Brooks et al. 1994), and that leaf age has similar effects on photosynthetic capacity as shading (Brooks et al. 1996). Manipulation of branch-level and whole-plant light availability has also provided insight into leaf functional and morphological acclimation to light (Goulet and Bellefleur 1986; Naidu and DeLucia 1998; Bloor and Grubb 2004; Jones and Thomas 2007; Ishii and Ohsugi 2011), branch autonomy (Yamamoto et al. 1999; Brooks et al. 2003; He and Dong 2003; Lacointe et al. 2004; Kawamura 2010), and light effects on branch growth and carbon allocation (Claussen 1996; Henriksson 2001). However, little is known about shading effects on leaves in the presence of gravitational water potential gradients in tall trees.

The main objectives of this study were to identify the effects of shading on leaf functional traits at various heights within a sugar maple (Acer saccharum) canopy to tease apart the effects of height (hydraulic limitation) and light environment on leaf functional traits and shoot growth. Using an experimental approach, we tested the following hypotheses for A. saccharum: (1) Narea is optimally distributed within A. saccharum tree crowns as a result of the strong influence of light on LMA (but not on Nmass); (2) shading reduces environmental stress on leaves (i.e., reduced light and leaf temperature), resulting in reduced stomatal closure and lower leaf δ13C; (3) leaves growing in higher light availability experience greater increases in Narea over time because LMA changes while Nmass does not; (4) leaf δ13C increases over time due to stomatal sensitivity to drier conditions that develop during mid- to late summer.

Methods

Site and experimental design

The study was conducted in closed-canopy sugar maple (Acer saccharum Marshall) forest at the Michigan Technological University Ford Forestry Center near Alberta, Michigan, USA (46.65°N, 88.48°W). Mean annual temperature and precipitation at the Ford Forestry Center are 4.8 °C and 810 mm, respectively (NOAA, WS ID 15608). This stand consisted mostly of Acer saccharum, but also included Betula alleghaniensis, Ostrya virginiana, Tilia americana, and Ulmus americana. In 2009, the mean height of the stand was 23.0 m, the basal area was 33 m2 ha−1, and the density was 267 trees ha−1. The tree density of Acer saccharum was 259 tree ha−1 or 97 % of the tree density. A cable zip-line system (Coble and Cavaleri 2014) provided crown access along a two-dimensional plane below three 15-m-high cables, and arborist-style climbing techniques were used to access the canopy above 15 m. More information about the site history and methodology can be found in Coble and Cavaleri (2014).

Prior to bud burst in the spring of 2013, shade structures were installed at four heights (1–3, 7–9, 12–14, and 17–20 m) along three vertical transects (see “Appendix A” in the Electronic supplementary material, ESM). Seven trees were used in the overall design, and each vertical transect contained 2–3 trees. Shade structures were constructed with PVC pipe to form a 0.8 × 0.8 m frame. Shade cloth (50 %) was draped over the frames and tightly fastened using zip ties. The shade structures were suspended from aluminum bars, which were either screwed into or clamped onto large branches with a stainless steel padded repair clamp. Branches below the shade structure and paired branches next to shade structures were flagged for leaf sampling and for branch measurements after leaf senescence.

Light, leaf water potential, and morphology measurements

We measured light conditions as percent photosynthetic photon flux density (%PPFD) for paired shaded and unshaded branches in June and August of 2013 using a ceptometer (Sunfleck PAR ceptometer, Decagon Devices, Pullman, WA, USA). We collected ten measurements above paired shaded and unshaded branches between 12:00 and 15:00 h in June and August. Ten open-sky measurements were collected prior to light measurements in a nearby open field to estimate %PPFD (mean below-canopy PPFD divided by mean above-canopy PPFD × 100). Previous studies have used a similar technique where open-sky measurements were temporally and spatially offset from below-canopy measurements (Martens et al. 1993; Knapp et al. 2008). In order to minimize the error associated with temporal changes in open sky measurements, below-canopy and open-sky measurements were made under uniform overcast or cloudless conditions.

Immediately following light measurements (12:00–15:00 h) in June and August, three leaves from shaded and unshaded branches were cut near the base of the petiole, placed in sealed plastic bags with moist paper towels, and stored in an ice chest until measurements were taken. Midday leaf water potential (Ψmid) measurements were made using a pressure chamber (PMS Instrument, Co., Corvallis, OR, USA) within an hour of leaf collection. Leaves were then stored at 2 °C until leaf morphology measurements were made.

Leaves were scanned into digital format using an Epson Expression 10000XL flatbed color image scanner (Seiko Epson Corporation, Nagano, Japan), and images were digitally analyzed for one-sided leaf area using ImageJ v1.44j (Schneider et al. 2012). Using Archimede’s principle, we derived leaf volume by immersing fresh leaves in a beaker of water placed on a balance (Coble and Cavaleri 2014). Leaves were dried at 65 °C for 48 h and weighed to the nearest 0.1 mg. LMA was calculated as the leaf dry mass (g) divided by leaf area (m2), and density was calculated as the leaf dry mass (g) divided by leaf volume (cm3).

Leaf nitrogen and carbon isotope composition

Leaf samples used to estimate leaf nitrogen and carbon isotope composition were processed at the Michigan Technological University Forest Ecology Stable Isotope Laboratory. The set of three leaves collected from each of the shaded and unshaded branches from each month were combined and ground to a fine powder (8,000 M Mixer/Mill, Spex SamplePrep LLC, Metuchen, NJ, USA). Leaf nitrogen on a mass basis (Nmass, mg g−1) was determined using a Costech elemental combustion system 4010 connected to a Thermo Finnigan ConfloIII Interface and Delta+ continuous flow stable isotope ratio mass spectrometer (Thermo Scientific, Waltham, MA, USA). Leaf Narea was determined as the product of Nmass and LMA divided by 100. Leaf δ13C was calculated as δ13C = 1,000(R sample/R standard−1) (‰), where R sample is the 13C/12C ratio for the sample and R standard is the ratio for a standard.

Data analysis

We conducted an ANCOVA for shading, month, and height effects (independent variables) on light (%PPFD), Ψmid, leaf mass, leaf area, LMA, leaf density, Narea, Nmass, C:N ratio, and δ13C (dependent variables). Height (1–21 m) was the covariate, and the categorical variables—shading and month—both had two levels (shaded, unshaded; June, August). Regression analysis was used to determine significant relationships between height and light, morphology, and leaf nitrogen parameters as described above. We used three approaches to separate out the effects of light and height on LMA, Narea, and δ13C. First, we plotted LMA, Narea, and δ13C vs. height by month within a narrow band of light conditions (1–3 %). Second, we plotted the residuals of LMA, Narea, and δ13C vs. height against light and the residuals of LMA, Narea, and δ13C vs. light against height. Calculation of residuals accounted for month effects by plotting within each month. Third, we compared the contribution of light and height to the full model (y = β 0 + β 1light + β 2height) for predicting LMA, Narea, and δ13C using partial R 2 values. The interaction terms (height × light) were not significant and were not included in the full models. Light values were natural log transformed (ln) for relationships with LMA to satisfy regression assumptions and to develop linear models for ANCOVA and partial R 2 analyses. All statistical analyses were conducted using R statistical software (R Development Core Team 2013). The “lm” and “anova” R functions were used to define the linear model and to produce the ANCOVA output, respectively.

Results

Shading and seasonal effects on environmental drivers and leaf functional traits

Light availability decreased as a result of shading, did not change from June to August, and increased exponentially with height for both shaded and unshaded leaves (Fig. 1a, b). Leaf midday water potential (Ψmid) was unaffected by shading, was lower in August compared with June, and decreased linearly with height for June (Table 1; Fig. 1c, d). Total cumulative precipitation was 9.0 cm the month prior (May 18 to June 18) to June measurements and 0.4 cm the month prior (July 13 to August 13) to August measurements (NOAA, WS ID 15608).

Fig. 1
figure 1

Relationships between height and light (a, b) and ΨMid (c, d) for shaded and unshaded leaves (a, c) and for leaves collected in June and August (b, d). Nonsignificant relationships (P > 0.05) with height are indicated by ns

Full size image
Table 1 Summary of ANCOVA results for tests of height (Ht), shade treatment (ST), month (M), and all two-way interaction effects on light (%PPFD), Ψmid, leaf morphological parameters (mass, area, LMA, density), shoot growth, area- and mass-based leaf nitrogen (Nmass and Narea), C:N ratio, and δ13C
Full size table

Leaf mass and area, two components of LMA, both decreased under shading but displayed opposite trends with height (Table 1; Fig. 2). Leaf mass increased linearly with height for unshaded leaves and was greater at higher heights compared with shaded leaves (Table 1; Fig. 2a). Leaf mass also increased linearly with height for leaves collected in June but not for August, and leaf mass was greater for leaves collected in August (Table 1; Fig. 2b). Leaf area was lower for shaded leaves, decreased linearly with height for shaded leaves only, and did not change with height within June and August (Table 1; Fig. 2c, d),

Fig. 2
figure 2

Relationships between height and leaf mass (a, b), area (c, d), LMA (e, f), and density (g, h) for shaded and unshaded leaves (a, c, e, g) and for leaves collected in June and August (b, d, f, h). Nonsignificant relationships (P > 0.05) with height are indicated by ns

Full size image

Both LMA and density decreased in response to shading, but shading did not affect branch growth (Table 1; Fig. 2e–h). LMA increased linearly with height among unshaded and shaded leaves, but the slope of the LMA–height relationship was greater for unshaded leaves (Fig. 2e). In contrast, slopes were similar for LMA–height relationships in June and August (Fig. 2f). Similar to leaf mass, leaf density increased linearly with height for unshaded leaves and for leaves collected in June (Table 1; Fig. 2g, h). Finally, height, shade treatment, and the height × shade treatment interaction did not have an effect on 2013 branch growth (cm year−1; Table 1).

Leaf nitrogen among shaded and unshaded branches along vertical gradients was measured to identify potential shading effects at different heights. Mean Nmass of shaded leaves was greater than that of unshaded leaves (Table 1; Fig. 3a), and Nmass of leaves collected in June was greater than that of leaves collected in August (Table 1; Fig. 3b). Narea decreased under shade treatment, particularly at higher heights (Table 1; Fig. 3c). Narea increased linearly with height for unshaded and shaded leaves (Fig. 3c) and for leaves collected in June and August (Fig. 3d). The C:N ratio decreased under shade treatment and increased from June to August (Table 1; Fig. 3e, f).

Fig. 3
figure 3

Relationships between height and Nmass (a, b), Narea (c, d), C:N ratio (e, f), and δ13C (g, h) for shaded and unshaded leaves (a, c, e, g) and for leaves collected in June and August (b, d, f, h). Nonsignificant relationships (P > 0.05) with height are indicated by ns

Full size image

Leaf δ13C gradients were compared between shaded and unshaded branches to identify acclimation responses to light, but leaf δ13C did not change under shade treatment (Table 1; Fig. 3g). Leaf δ13C decreased from June to August at all heights (Table 1; Fig. 3h). Leaf δ13C increased linearly with height for unshaded and shaded leaves and for leaves collected in June and August (Fig. 3g, h).

Teasing apart the effects of light and height on leaf functional traits

We used a three-step approach (see “Methods” for the full description) to further tease apart the effects of light and height on three important variables: LMA, Narea, and δ13C. Within a narrow range of light conditions (1–3 %PPFD), LMA increased with height in June but not in August (Fig. 4a). The residuals of LMA vs. (ln)light increased linearly with height, and the residuals of LMA vs. height increased nonlinearly with height (Fig. 4b, c). The partial regression analysis showed that height contributed more to the full model for predicting LMA in June, but light contributed more to the full model in August (Table 2).

Fig. 4
figure 4

Relationship between height and LMA, Narea, and δ13C for leaves growing within a narrow range of light conditions (1–3 %PPFD; a, d, g). Residuals of LMA, Narea, and δ13C vs. light (%PPFD) plotted against height (b, e, h) and residuals of LMA, Narea, and δ13C vs. height plotted against light (c, f, i). Note that residuals were calculated from models developed for each month (June and August). Nonsignificant relationships (P > 0.05) with height are indicated by ns

Full size image
Table 2 Partial regression analysis of LMA and Narea vs height and light
Full size table

Narea increased with height in both June and August within the narrow range of light conditions (Fig. 4d). The residuals of Narea vs. light increased linearly with height and the residuals of Narea vs. height increased linearly with light (Fig. 4e, f). Partial regression analysis showed that height contributed more to the full model for predicting Narea in June, and light contributed more in August (Table 2).

Leaf δ13C increased linearly with height in June within a narrow range of light conditions, but not in August (Fig. 4g). The residuals of δ13C vs. light increased linearly with height, and the residuals of δ13C vs. height did not change with light (Fig. 4h, i). The partial regression analysis showed that height contributed more to the full model for predicting δ13C in both June and August (Table 2).

Discussion

Optimization of leaf functional traits

Our results provide partial support for our first hypothesis that Narea, but not Nmass, would be optimally distributed within the canopy as a result of the strong influence of light on LMA. Experimentally reduced light availability resulted in lower leaf mass, density, LMA, and Narea, all most apparent in upper canopy positions, whereas Nmass showed the opposite trend. Even though Nmass increased with shading, a larger decrease in LMA with shading resulted in a decrease in Narea. In a previous observational study of A. saccharum, Coble and Cavaleri (2014) concluded that changes in LMA, density, and thickness with height were primarily driven by light. Leaves that develop under high light availability tend to be thicker as a result of thicker palisade mesophyll cell layers, which maximize light capture (Oguchi et al. 2005; Zhang et al. 2011a) and denser leaf tissues (Niinemets et al. 1999; Coble and Cavaleri 2014), both of which lead to greater LMA. For A. saccharum, leaves growing in high light with greater LMA also have greater Narea and photosynthetic capacity (Ellsworth and Reich 1992a, 1993; Jones and Thomas 2007), likely due to the critical role of leaf nitrogen in protein-pigment complexes in the thylakoid membrane and in RUBISCO (Evans 1989). Leaves acclimated to high light also maintain higher leaf hydraulic conductance in A. saccharum (Sack et al. 2003), other temperate deciduous trees (Aasamaa et al. 2004; Lemoine et al. 2002; Sellin and Kupper 2007; Sellin et al. 2008), and conifer trees (Jerez et al. 2004; Burgess et al. 2006), suggesting that light availability, leaf hydraulic conductance, Narea, LMA, and photosynthetic capacity have co-optimal patterns with light. Overall, adjustments in LMA and Narea to shading and along light gradients as reported in this study provide evidence that the distributions of LMA and Narea become optimized through morphological acclimation to light over the course of the growing season.

Constraints on leaf functional traits

Height effects on LMA and Narea were most apparent early in the growing season (June), when leaf water potential showed linear decreases with height. Height has been found to directly drive LMA gradients in forest canopies, where height effects have been detected under saturating light conditions (Burgess and Dawson 2007; Ishii et al. 2008; Cavaleri et al. 2010; Coble et al. 2014). Height-related limitations to leaf development in tall conifer trees (Pseudotsuga menziesii and Sequoia sempervirens) include water potential gradients (Koch et al. 2004; Burgess and Dawson 2007; Ishii et al. 2008) and subsequent reductions in turgor pressure (Woodruff et al. 2004; Meinzer et al. 2008). Reduced leaf water potential can constrain cell turgor pressure if no osmotic adjustments are made, which may result in denser leaf tissue because turgor pressure is necessary for cell expansion and division (Lockhart 1965; Hsiao 1973). However, recent evidence suggests that leaf water storage in tall trees may compensate for the limitations of hydraulic transport (Ishii et al. 2014). In temperate deciduous species such as Robinia pseudoacacia, greater turgor pressure in water-stressed leaves of seedlings can be achieved by osmotic adjustment through most of the day, but midday depressions in turgor pressure that fall below the yield pressure of cell wall extension can lead to reduced leaf expansion (Zhang et al. 2011b). If hydraulic constraints are accounted for (e.g., low leaf hydraulic conductance), the simulated optimal Narea distribution is not proportional to light (Peltoniemi et al. 2012), which may explain why field-based observations of photosynthetic capacity indicate that it saturates at high light (Buckley et al. 2013). Decreasing leaf hydraulic conductance with height often occurs in conifer or evergreen species (e.g., Ryan et al. 2006), but it has been found to increase with height in the deciduous species Tilia cordata and Betula penula (Sellin and Kupper 2007; Sellin et al. 2008). Also, sun leaves in A. saccharum can maintain higher leaf hydraulic conductance than shade leaves (Sack et al. 2003). Thus, biophysical limitations on leaf growth associated with lower leaf water potentials early in the growing season may constrain the optimal distributions of LMA which, in turn, likely constrain the optimal Narea distributions given the relative insensitivity of Nmass to vertical environmental gradients.

Our results do not support our second hypothesis that shading reduces environmental stress, resulting in lower δ13C. In the same stand as used in this study, Coble and Cavaleri (2014) found that predawn water potential decreased linearly with height, which was likely due to the gravitational component of water potential (Scholander 1965), since transpiration is negligible at night. Thus, leaves at the top of the canopy maintain greater tension in the water column due to height alone. When trees are transpiring, however, this underlying gravitational tension in the water column is amplified by a combination of frictional resistances and greater evaporative demand, potentially leading to decreases in stomatal aperture (Bauerle et al. 1999; Koch et al. 2004; Niinemets et al. 2004; Ishii et al. 2008). In addition to gravitational potential gradients observed in this stand, VPD increased with height during the same study period (data not shown). The stomatal conductance of A. saccharum is particularly sensitive to leaf water potential and high VPD under high light conditions (Ellsworth and Reich 1992b). Collectively, these studies indicate that gradients in leaf water potential and/or VPD may impose constraints on stomatal conductance, resulting in greater δ13C higher in the canopy. In contrast, Duursma and Marshall (2006) attributed vertical gradients in δ13C to fractionation in the conductance from intercellular air space to the chloroplast, which generally scales with photosynthetic capacity. We suspect that this was not the case in our study because Narea and LMA both strongly correlate with photosynthetic capacity in A. saccharum (Ellsworth and Reich 1993; Jones and Thomas 2007), and Narea and LMA both responded to shading whereas δ13C did not.

Seasonal effects on optimization and constraints on leaf function

Our results do not fully support our third hypothesis that leaves growing in higher light would experience a greater increase in Narea over time due to changes in LMA. Due to a simultaneous increase in LMA and decrease in Nmass, the combination of opposite changes over time neutralized any effects of time on Narea. Although there was no effect of time on Narea at any height, we found that height was more important earlier in the growing season whereas light was more important later in the growing season. Migita et al. (2007) suggested that Narea in Quercus serrata is optimized both spatially and temporally where optimal distributions occurred later in the growing season. However, these conclusions were based on data collected only during the late growing season (September) through leaf senescence (November). Extending the work of Migita et al. (2007), we present multiple lines of evidence indicating that the Narea distribution was constrained by height early in the growing season and became functionally optimized later in the growing season through the acclimation of LMA to the light conditions.

Seasonal increases in leaf mass, LMA, leaf density, and C:N and decreases in Nmass and δ13C suggest that leaves accumulate carbon-rich structural compounds or other compounds low in nitrogen, which may also indirectly affect the seasonal progression toward Narea optimization with light. First, studies have suggested that seasonal changes in LMA correspond with an accumulation of structural proteins (Yasumura et al. 2006) and calcium and silicon (Kitajima et al. 2002), indicating that seasonal patterns in LMA may be due to changes in cell wall structure and chemical composition. Calcium is particularly important in the deposition of lignin and non-cellulosic polysaccharides in cell walls (Eklund and Eliasson 1990). Second, a decrease in Nmass over time has been observed in other studies and was attributed to the accumulation of carbon (Reich and Walters 1994), which is further supported by the increase in the C:N ratio over time in this study. Previous investigations into Nmass have found that, across and within species, thicker, denser leaves tend to have lower Nmass (Reich and Walters 1994; Niinemets 1999; Wright et al. 2004). Consistent with this finding, LMA and density in this study were both negatively correlated with Nmass; however, density explained 54 % of the variation in Nmass, whereas LMA explained only 23 % of the variation in Nmass (data not shown). These studies and our experiment indicate a greater investment in cell wall structure or lignification over time, possibly allowing leaves to tolerate low Ψmid later in the growing season. We hypothesize that early-season constraints of leaf water potential on leaf morphology and nitrogen are more apparent because leaf cell wall thickening and lignification are not fully developed. Later in the growing season, however, leaves in high-light conditions are able to invest more into cell wall structure, thus offsetting constraints associated with gradients in leaf water potential later in the growing season (i.e., increasing importance of light over time).

Contrary to our expectations (hypothesis 4), we observed a decrease in δ13C from June to August despite drier conditions in July and August. Seasonal declines in δ13C have been reported in other studies and generally showed similar patterns among upper and lower canopy leaves (Damesin et al. 1997; Niinemets et al. 1999; Helle and Schleser 2004; Damesin and Lelarge 2003). During drier conditions, leaves tend to become more enriched in 13C as stomatal aperture decreases and as 12CO2 becomes depleted in substomatal chambers (Farquhar et al. 1989). However, the declines in δ13C and precipitation during the growing season indicate that soil water availability was not a limitation on photosynthesis in July and August, since δ13C represents the integrated photosynthetic activity (Dawson et al. 2002). We would also expect midday declines in stomatal conductance with decreasing leaf water potential as previously observed in A. saccharum seedlings growing in high light (Ellsworth and Reich 1992b). However, δ13C values were lower in August despite the lower Ψmid observed in August compared with June, suggesting that the Ψmid was not low enough to initiate midday stomatal closure. Overall, we speculate that leaf acclimation to light during leaf maturation reduced stomatal sensitivity to reduced leaf water potential.

Conclusions

We show that multiple interacting effects (light, height, and time) play roles in both optimizing and constraining distributions of leaf functional traits in A. saccharum. An underlying assumption behind current “big leaf” models that integrate leaf photosynthesis and functional traits over the canopy is that leaf nitrogen distribution is optimal with respect to light. Based on our results, we contend that constraints on leaf functional traits should be accounted for when integrating leaf functional traits with these models. Early-season constraints on leaf functional traits appear to be associated with gradients in leaf water potential. We show that LMA, Nmass, C:N ratio, and δ13C can change substantially over the growing season, suggesting that the highly regulated processes inherent to leaf maturation involve a combination of cell wall thickening and carbon and calcium accumulation. We speculate that cell wall thickening or lignification associated with leaf maturation contributes to the optimization of Narea and LMA with respect to light. Overall, our results indicate that light acclimation with increasing leaf age optimized the leaf functional traits of a broadleaf deciduous tree, despite the underlying height-related constraints that were more pronounced in the early growing season.