Vertical leaf mass per area gradient of mature sugar maple reflects both height-driven increases in vascular tissue and light-driven increases in palisade layer thickness

Abstract

A key trait used in canopy and ecosystem function modeling, leaf mass per area (LMA), is influenced by changes in both leaf thickness and leaf density (LMA = Thickness × Density). In tall trees, LMA is understood to increase with height through two primary mechanisms: (i) increasing palisade layer thickness (and thus leaf thickness) in response to light and/or (ii) reduced cell expansion and intercellular air space in response to hydrostatic constraints, leading to increased leaf density. Our objective was to investigate within-canopy gradients in leaf anatomical traits in order to understand environmental factors that influence leaf morphology in a sugar maple (Acer saccharum Marshall) forest canopy. We teased apart the effects of light and height on anatomical traits by sampling at exposed and closed canopies that had different light conditions at similar heights. As expected, palisade layer thickness responded strongly to cumulative light exposure. Mesophyll porosity, however, was weakly and negatively correlated with light and height (i.e., hydrostatic gradients). Reduced mesophyll porosity was not likely caused by limitations on cell expansion; in fact, epidermal cell width increased with height. Palisade layer thickness was better related to LMA, leaf density and leaf thickness than was mesophyll porosity. Vein diameter and fraction of vascular tissue also increased with height and LMA, density and thickness, revealing that greater investment in vascular and support tissue may be a third mechanism for increased LMA with height. Overall, decreasing mesophyll porosity with height was likely due to palisade cells expanding into the available air space and also greater investments in vascular and support tissue, rather than a reduction of cell expansion due to hydrostatic constraints. Our results provide evidence that light influences both palisade layer thickness and mesophyll porosity and indicate that hydrostatic gradients influence leaf vascular and support tissues in mature Acer saccharum trees.

Introduction

Leaf mass per area (LMA, g m⁻²) is arguably the most important leaf functional trait (Poorter et al. 2009), because it is strongly correlated with ecosystem productivity (Reich et al. 1997), photosynthesis (Jurik 1986, Ellsworth and Reich 1993, Miyazawa et al. 1998, Wilson et al. 2000, Kitajima et al. 2002, Wright et al. 2004, Grassi et al. 2005, Montpied et al. 2009), respiration (Wright et al. 2004, Cavaleri et al. 2008), leaf lifespan (Reich et al. 1999) and leaf nitrogen (Ellsworth and Reich 1992, Rosati et al. 2000) across biomes and within forest canopies. Leaf mass per area can be used in modeling and scaling carbon fluxes from local to global scales (Gutschick and Wiegel 1988, Raulier et al. 1999, Hanson et al. 2004, Medlyn 2004, Thornton and Zimmerman 2007, Ryu et al. 2011). Leaf mass per area is influenced by both leaf thickness and density (LMA = Thickness × Density), which may respond independently to different environmental factors via adjustments in cellular structure (Witkowski and Lamont 1991, Niinemets 2001).

Increasing leaf thickness often corresponds with an increase in palisade layer thickness, which maximizes overall absorption of light at greater depths within the mesophyll (Cui et al. 1991, Vogelmann and Martin 1993, Evans 1999). Increasing palisade layer thickness accompanied by increasing leaf thickness, LMA (g m⁻²) and photosynthetic capacity with treatments of higher light is a common pattern observed in greenhouse experiments (Chabot and Chabot 1977, Chabot et al. 1979, Oguchi et al. 2005, Tosens et al. 2012). Similarly, vertical gradients in both leaf thickness and LMA scale with natural light gradients in temperate deciduous canopies (Hutchison et al. 1986, Hollinger 1989, Ellsworth and Reich 1993, Niinemets et al. 1999, Coble and Cavaleri 2014), which are likely influenced by adjustments in palisade layer thickness. Thicker palisade layers in response to high light also correspond with greater surface area of mesophyll palisade cells exposed to intercellular air spaces, and this response has led to greater photosynthetic capacity and mesophyll conductance of leaves (Nobel et al. 1975, Nobel 1977, Hanba et al. 2002, Kenzo et al. 2004).

Leaf density tends to vary across species depending on cell compactness and intercellular air space (Niinemets 1999). Leaf mesophyll porosity (proportion of air space in leaf mesophyll) is an important component to cellular and leaf function. Reduced air space can constrain mesophyll conductance, defined as diffusion of CO₂ from substomatal cavities to sites of carboxylation within chloroplasts (Parkhurst 1994, Syversten et al. 1995, Flexas et al. 2008, Marchi et al. 2008, Gu et al. 2010), which can be a major limitation to photosynthesis and forest carbon uptake (Keenan et al. 2010). For example, in tall Sequoia sempervirens trees, reduced mesophyll porosity higher in the canopy resulted in declining mesophyll conductance with height, which was an important factor in limiting photosynthesis (Mullin et al. 2009, Oldham et al. 2010). Constraints on leaf development and reduced mesophyll porosity in tall western conifer species are thought to be associated with decreasing turgor pressure with height and limited osmotic adjustments higher in the canopy (Koch et al. 2004, Woodruff et al. 2004, Meinzer et al. 2008, Oldham et al. 2010). Since leaf turgor pressure is necessary for cell elongation (Hsiao 1973), reduced turgor pressure can lead to the formation of small and densely packed cells and reductions in intercellular air space (Cosgrove 1993, 2000, Oldham et al. 2010).

On the other hand, hydrostatic constraints in tall trees may be offset by several mechanisms, including leaf anatomical adjustments that increase water storage and act as a hydraulic buffer to sustain turgor pressure (Ishii et al. 2014, Azuma et al. 2016). Whole-tree structural adjustments such as lower leaf area to sapwood area ratio may also sustain leaf-specific hydraulic conductivity (e.g., hydraulic efficiency) and minimum leaf water potential (Nabeshima and Hiura 2004). Finally, adjustments in osmotic potential can maintain greater turgor pressure in the tops of tall trees (Coble et al. 2016). Thus, reduced mesophyll porosity in upper canopy leaves may not be caused by reduced turgor. In an Acer saccharum Marshall forest, leaf density increased with light availability despite a concurrent increase in turgor pressure (Coble and Cavaleri 2014), suggesting that factors associated with light availability may influence mesophyll porosity. A plausible explanation for these findings is that increases in palisade layer thickness may occupy a larger volume of the mesophyll, reducing the amount of air space in leaves. While turgor is an important factor in regulating cell expansion, biophysical aspects of cell walls (extensibility and turgor threshold) may also limit cell expansion in water-stressed leaves of deciduous trees (Zhang et al. 2011a). In A. saccharum, leaf expansion rates were lower in upper canopy leaves despite having greater turgor (Coble et al. 2016), indicating biophysical constraints on leaf development. Thus, it is also possible that height may influence mesophyll porosity in tall temperate deciduous trees.

The main objectives of this study were to investigate environmental drivers of leaf anatomical traits (palisade layer thickness, mesophyll porosity, epidermal cell width, vein diameter and fraction of vascular tissue), and to identify potential mechanisms behind vertical gradients in leaf morphological traits (LMA, leaf thickness, leaf density) in a broad-leaved deciduous forest (A. saccharum). In this study we tested the following hypotheses: (i) light primarily drives palisade layer thickness, which corresponds with variation in leaf thickness and LMA; (ii) height constrains the expansion of epidermal cells, limiting the formation of intercellular air space resulting in reduced mesophyll porosity and denser leaves; and (iii) upper canopy leaves respond to reduced leaf water potential through greater investments in leaf vascular and support tissue.

Materials and methods

Study site

The study was conducted in an A. saccharum forest at the Michigan Technological University Ford Center and Forest near Alberta, MI, USA (46.65°N, 88.48°W). The mean annual temperature and precipitation at the Ford Forestry Center are 4.8 °C and 810 mm, respectively (NOAA, WS ID 15608). The A. saccharum stand has not been actively managed since 1956 (Erickson et al. 1990). This stand also included Betula alleghaniensis, Ostrya virginiana, Tilia americana and Ulmus americana. The tree density of A. saccharum was 259 trees ha⁻¹, which was 97% of the tree density of the stand (267 tree ha⁻¹). The mean height of the stand was 23.0 m and the basal area was 33 m² ha⁻¹.

The closed canopy portion of this stand was accessed with a cable zip-line system, and the exposed canopy was accessed with a tower. At the closed canopy, three cable zip-lines provided canopy access from 0 to 15 m in height along three two-dimensional planes. Arborist climbing techniques were used to access the canopy above the zip-lines above 15 m and up to 30 m. Two to three vertical transects were designated to each zip-line and an additional transect next to the zip-lines was added to obtain additional leaves above 15 m, with a total of nine vertical transects. Six to 21 sampling locations (dependent on the number of accessible branches) were assigned to each transect, for a total of 137 sampling locations in the closed canopy. A 19-m aluminum walk-up tower (Upright, Inc., Selma, CA, USA) was constructed 55 m from the zip-lines in the same stand in August 2012 in order to access portions of tree crowns that were exposed to brighter light conditions at lower heights due to a canopy opening. At both sites, we used a telescoping pole-pruner to collect leaves at the tops of trees (up to 30 m). The tower and zip-line sites will be herein referred to as the ‘exposed canopy’ and the ‘closed canopy’, respectively. More information about the site history and methodology can be found in Coble and Cavaleri (2014).

Height and light measurements

Height above the ground at each sampling location was measured using a tape measure. Light conditions or ‘canopy openness’ at the exposed and closed canopies were measured as diffuse non-interceptance (DIFN), the fraction of radiation that is transmitted through the canopy (Norman and Welles 1983), using two plant canopy analyzers (LAI-2000 and LAI-2200, LI-COR, Inc., Lincoln, NE, USA). Light measurements were made during overcast conditions or following sunrise until 1 h following sunrise. Open sky measurements (proxy for ‘above-canopy’ light conditions) were collected in a nearby open field (~400 m from the site) using the LAI-2000 mounted on a tripod at 30-s intervals, and below-canopy measurements were collected using the LAI-2200. Prior to and following below-canopy measurements, open sky measurements were collected with the LAI-2200 next to the LAI-2000 in order to calibrate open measurements collected by the LAI-2000. At each sampling point, we collected two light measurements and used a 180° view cap to prevent climbing ropes or the tower from obstructing the view. The average of both light measurements at each sampling point was used for further analysis. We used FV2200 software (LI-COR, Inc.) to adjust open sky measurements and to estimate DIFN by matching open sky and below-canopy readings closest in time. For leaves collected at the top of trees (pole-pruner collection), we assumed a DIFN of 1 (i.e., 100% canopy openness) because light measurements using the LAI-2200 were not possible for these leaves. From May 10 to October 19, we measured open sky photosynthetic photon flux density (PPFD_above, µmol m⁻² s⁻¹) at 10-min intervals using a light sensor (S-LIA-M003, Onset Computer Corporation, Bourne, MA, USA) in a nearby open field ~400 m from the site.

In order to account for seasonal, cumulative changes in leaf morphology due to light acclimation (Coble and Cavaleri 2015), we estimated time-integrated light availability that occurred from leaf expansion to leaf collection. We estimated integrated photosynthetic photon flux density (PPFD_INT, mol m⁻²) over the lifespan of leaves, defined here as beginning at 50% leaf expansion on 11 May 2012 and ending on the day of leaf collection. We estimated PPFD_INT at each sampling location with the following equation:

$${\mathrm{PPFD}}_{\mathrm{INT}}\text{=}\sum _{\text{Harvest}}^{\text{50\% Expansion}}\mathrm{DIF}\mathrm{N}\times {\mathrm{PPFD}}_{\mathrm{above}}$$

(1)

We assumed that DIFN was equal to the fraction of PPFD transmitted through the canopy, which has been shown in a previous study (Machado and Reich 1999).

Leaf morphology

We collected one leaf per sampling location following leaf expansion until the end of August at the closed canopy, and leaves within each vertical transect were collected on the same day. Leaf sampling occurred later (end of August) at the exposed canopy because the tower was constructed in August. We selected a subset of leaves (n = 87) for morphological and anatomical measurements that covered the broad range of light and height values at the closed and exposed canopy. Immediately following collection, leaves were placed in sealable, plastic bags with a moist paper towel and temporarily stored in an ice chest prior to bringing leaf samples to the lab. Leaves were then stored at 2 °C until leaf morphology measurements were made. Leaf area was measured by scanning leaves with a bench-top leaf area meter (LI-3100, LI-COR, Inc.). Leaf volume was derived by immersing fresh leaves in a beaker of water placed on a balance and using Archimedes’ principle (Coble and Cavaleri 2014). Leaves were dried at 65 °C for 48 h and weighed to the nearest 0.1 mg. Leaf mass per area 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).

In addition to measuring leaf density in the laboratory as described above, we estimated leaf density by dividing LMA by leaf thickness. Measurement of leaf thickness using anatomical images of leaf transverse sections is described below. To compare this estimation of density with our measured leaf density, we plotted leaf density values as derived from laboratory measurements (measured density) against leaf density values as derived from LMA and thickness (estimated density; Figure 1). There was a strong correlation between both leaf density values from the same leaves (r² = 0.65); however, the measured leaf density values appeared to be greater than estimated values, particularly at low leaf density (Figure 1). We used both values of leaf density for subsequent analysis because measured values may be biased due to small air-bubbles that form along the leaf cuticle (Coble and Cavaleri 2014) and estimated values may be biased because leaf veins in A. saccharum are thicker than photosynthetic tissue and were not included in leaf thickness measurements.

Figure 1.

Relationship between estimated density and measured density with the 1:1 line (dashed line). Estimated density was calculated as Density = LMA/Thickness, and measured density was derived from leaf volume and dry mass measurements.

Leaf anatomy

Prior to leaf morphology measurements, a small section of leaf (~16 mm × 8 mm) was cut from the right lobe of each leaf and temporarily stored in formaldehyde–acetic acid–ethanol–water solution (10:5:50:35, by vol.). Leaf sections were further divided into two to three sections using a scalpel and embedded in paraffin to create a block. Leaf specimens in each block were cut at 5 µm, perpendicular to the leaf adaxial surface, using a microtome (Finesse 325, Thermo Shandon, Pittsburgh, PA, USA). This was repeated twice for each block to produce six to nine leaf transverse sections. Between each leaf section, 200 μm of the block was sliced off. Leaf transverse sections were mounted on slides, stained with hematoxylin and eosin in an automatic stainer (Model Linistain GLX, Thermo Shandon), and covered with a coverslip. Ten images at 20× magnification were collected from three leaf transverse sections per leaf (i.e., one transverse section per microtome cut) using a microscope (Eclipse E400, Nikon, Inc., Melville, NY, USA) with a camera (Leica DFC295, Leica Microsystems, Buffalo Grove, IL, USA) mounted above the objective lenses. We randomly selected one image from each leaf transverse section (i.e., three images per leaf) used in the analysis.

ImageJ software (Schneider et al. 2012) was used for all image analysis. For each image, we created five evenly spaced vertical lines that were randomly offset. Leaf and palisade layer thickness (μm) and horizontal width of epidermal cells (μm) were measured (Figure 2a) at each of the vertical lines and averaged to obtain one leaf thickness, palisade layer thickness and epidermal cell width measurements per leaf. To estimate mesophyll porosity, the same set of images were first converted to a 32-bit gray-scale image, and the image threshold was adjusted to the point just prior to the rapid increase in the color histogram in order to convert to a binary image (Oldham et al. 2010). All cells were filled in with black using the ‘brush tool’ and the intercellular air space was kept white (Figure 2b). The image was again converted to a black and white image, and the area of the cells (black area) was measured. The image was inverted so that the black area became white, and we removed the black area that was not part of the leaf, so that only the air space was black (Figure 2c). The area of the air space was measured, and the area of the air space and the cells were added to estimate the total leaf cross-sectional area. Mesophyll porosity was then calculated as the area occupied by intercellular air space divided by the total area occupied by the leaf cross section (Oldham et al. 2010). We measured the diameter of leaf veins, which included the vascular bundle and bundle sheath (Figure 2a) in both the vertical and horizontal direction. Finally, we measured the cross-sectional area of the veins and support tissue (e.g., vascular bundle, bundle sheath and bundle sheath extension) using the ‘freehand selection’ tool in ImageJ. To estimate the proportion of vascular and support tissue relative to cellular area (herein referred to as ‘fraction of vascular tissue’), the cross-sectional area of the vascular and support tissue was divided by the cross-sectional area of all cells.

Figure 2.

Image showing measurements of epidermal cell width and palisade layer thickness, as well as the location of the vascular bundle, bundle sheath and bundle sheath extension (a). All cellular area was filled-in and converted to a black-and-white, binary image (b). The image was then inverted so that the area occupied by intercellular air space was black for measurements of total intercellular air space (c). Mesophyll porosity was estimated by dividing the area occupied by intercellular air space by the total area of the leaf cross section (cellular area plus area of air space).

Data analysis

Statistical analyses were conducted using R statistical software (R Development Core Team 2013). Relationships between palisade layer thickness (µm), mesophyll porosity, epidermal cell width (µm), vein diameter (µm), fraction of vascular tissue, height (m) and light (PPFD_INT, mol m⁻²) were examined using regression analysis. Comparisons of these relationships between the two canopy types (closed and exposed canopy) were made using analysis of covariance (ANCOVA). We assessed the contribution of light and height to the full model for predicting palisade layer thickness, mesophyll porosity and epidermal cell width using partial R² analysis and calculating Akaike weights (Wagenmakers and Farrell 2004). Leaf anatomical parameters and light were natural log-transformed (ln) to satisfy regression assumptions of linearity and homoscedasticity and to develop linear models for ANCOVA, partial R² analysis and Akaike weights. We also assessed the contribution of palisade layer thickness and mesophyll porosity to the full model for predicting LMA and leaf density (estimated and measured) using partial R² analysis in order to identify morphological consequences of adjustments in anatomical structure. Finally, we compared a subsample of palisade thickness values collected only in August at the closed (n = 6) and exposed (n = 7) canopies at 17–21 m in height using one-way ANOVA to assess the timing of sample collection as a possible source of differences between canopy types.

Results

Leaf anatomy relationships with height and light

To assess the influence of environmental factors on leaf anatomy, we compared relationships among light, height, palisade layer thickness, epidermal cell width and mesophyll porosity between the exposed and closed canopy positions. Light (PPFD_INT) increased more rapidly with height at the exposed canopy compared with the closed canopy (Figure 3). Similarly, palisade layer thickness increased exponentially with height at the exposed and closed canopies (Figures 4 and 5a), and the slope of the height and palisade layer thickness relationship was steeper at the exposed canopy (Table 1). Palisade layer thickness increased non-linearly (log-linear) with light for both closed and exposed canopies, and there was a significant interaction between canopy type and light (Figure 5b, Table 1). The partial R² analysis showed that integrated light explained more variation in palisade layer thickness than height did, but the Akaike weights showed that neither height nor light was more important (Table 2). In our post hoc analysis of palisade layer thickness at the exposed and closed canopies at 17–21 m for August only, we found that palisade layer thickness at the exposed canopy was still significantly greater (P < 0.01) than at the closed canopy.

Table 1.

Summary of ANCOVA results for test of height, canopy type (CT) and height × CT effects on palisade layer thickness, epidermal cell width, mesophyll porosity, vein diameter and fraction of vascular tissue; and (ln)light (PPFD_INT; mol m⁻²), CT and light × CT effects on palisade layer thickness, epidermal cell width, mesophyll porosity, vein diameter and fraction of vascular tissue.

Source	df	(ln) Palisade layer thickness (μm)		(ln) Mesophyll porosity		(ln) Epidermal cell width (µm)		(ln) Vein diameter (µm)		(ln) Fraction of vascular tissue
		Mean square	F-ratio	Mean square	F-ratio	Mean square	F-ratio	Mean square	F-ratio	Mean square	F-ratio
Height	1	4.84	259.4***	3.60	51.4***	3.57	178.2***	4.77	186.3***	4.76	74.5***
CT	1	0.75	40.1***	0.40	5.7*	0.01	0.3	0.00	0.0	0.32	5.0*
Height × CT	1	0.76	41.0***	0.06	0.8	0.06	2.8	0.46	0.5***	0.00	0.0
Error	83	0.02		0.07		0.02		0.03		5.30
(ln)Light	1	5.56	233.56***	3.61	50.4***	1.89	56.6***	2.72	64.9***	3.70	51.2***
CT	1	0.22	9.41**	0.22	3.0	0.63	18.9***	1.08	25.8***	0.32	4.4*
(ln)Light × CT	1	0.15	6.21*	0.08	1.1	0.00	0.0	0.08	1.9	0.36	4.9*
Error	83	0.02		0.07		0.03		0.04		0.07

Degrees of freedom (df), mean square, F-ratio and level of significance are listed for main, interaction and error terms. *P < 0.05, **P < 0.01, ***P < 0.001.

Table 2.

Partial R² analysis for relationships among palisade layer thickness (μm), epidermal cell width (μm), mesophyll porosity, vein diameter (μm), fraction of vascular tissue, light (PPFD_INT, mol m⁻²) and height (m). Sample size, regression coefficients, standard error (±SE), R² values, and Akaike weights (w_i) are displayed for the log-linear regressions.

Response variable	n	(ln) Light only				Height only				Light and height					Partial R2 for adding
		β0	β1	R2	wi	β0	β1	R2	wi	β0	β1	β2	R2	wi	Light	Height
(ln)Palisade layer thickness	86	2.45*** ± 0.09	0.21*** ± 0.01	0.70	0.00	3.26*** ± 0.04	0.03*** ± 0.00	0.61	0.00	2.64*** ± 0.09	0.14*** ± 0.02	0.01*** ± 0.00	0.77	1.00	0.16	0.07
(ln)Mesophyll porosity	86	−0.15*** ± 0.15	−0.17*** ± 0.02	0.37	0.03	−1.78*** ± 0.06	−0.03*** ± 0.00	0.36	0.03	−1.34*** ± 0.16	−0.10** ± 0.03	−0.02** ± 0.01	0.43	0.93	0.06	0.06
(ln)Epidermal cell width	86	2.27*** ± 0.11	0.12*** ± 0.02	0.36	0.00	2.62*** ± 0.03	0.03*** ± 0.00	0.67	0.73	2.61*** ± 0.09	0.00 ± 0.02	0.03*** ± 0.00	0.67	0.27	0.00	0.32
(ln)Vein diameter	86	2.68*** ± 0.13	0.15*** ± 0.02	0.37	0.00	3.12*** ± 0.04	0.03*** ± 0.00	0.65	0.68	3.05*** ± 0.10	0.02 ± 0.02	0.03*** ± 0.00	0.65	0.32	0.00	0.28
(ln)Fraction of vascular tissue	86	−3.27*** ± 0.15	0.17*** ± 0.02	0.36	0.00	−2.69*** ± 0.06	0.03*** ± 0.00	0.46	0.25	−2.98*** ± 0.15	0.07* ± 0.03	0.02*** ± 0.01	0.48	0.75	0.03	0.13

*P < 0.05, ***P < 0.001.

Figure 3.

Relationship between PPFD_INT (mol m⁻²) and height (m) at the exposed and closed canopy for sampled leaves. PPFD_INT represents the time-integrated light conditions over the life-span of sampled leaves beginning at 50% leaf expansion and ending at leaf collection.

Figure 4.

Leaf anatomical images for leaves at the closed and exposed canopy at four heights (23, 17–18, 8–9 and 3 m). PPFD_INT is also listed next to the height of each leaf.

Figure 5.

Relationships between height, PPFD_INT, palisade layer thickness (a, b), mesophyll porosity (c, d) and epidermal cell width (e, f) at the exposed and closed canopy.

Mesophyll porosity exponentially decreased with height and displayed a negative log-linear decrease with light (Figure 5c and d). The slopes did not differ between canopy types for either height or light (Table 1). The partial R² analysis and Akaike weights showed that for mesophyll porosity, light and height equally contributed to the full model (Table 2).

Epidermal cell width exponentially increased with height, showing no differences across canopy type (Figure 5e, Table 1). Epidermal cell width non-linearly (log-linear) increased with integrated light (Figure 5f), and analysis showed differences between intercepts but no interaction with canopy type (Table 1). The partial R² analysis and Akaike weights showed that height explained more variation in epidermal cell width than light (Table 2).

Vein diameter exponentially increased with height (Figure 6a), and the slopes differed between canopy types (Table 1). Vein diameter non-linearly (log-linear) increased with integrated light at the closed canopy and linearly increased with light at the exposed canopy (Figure 6b). Analysis showed differences in intercepts by canopy type, but no difference in slopes (Table 1). The partial R² analysis and Akaike weights showed that height explained more variation in vein diameter than light (Table 2).

Figure 6.

Relationships between height, PPFD_INT, vein diameter (a, b) and fraction of vascular tissue (c, d) at the exposed and closed canopy.

The fraction of vascular tissue linearly increased with height (Figure 6c), and analysis showed differences in intercepts, but no differences in slopes (Table 1). The fraction of vascular tissue non-linearly (log-linear) increased with integrated light at the closed canopy and linearly increased with integrated light at the exposed canopy (Figure 6d). The intercepts and slopes of the relationship between integrated light and fraction of vascular tissue were significantly different between canopy types (Table 1). The partial R² analysis and Akaike weights revealed that height was more important than integrated light for the fraction of vascular tissue (Table 2).

Inter-related morphological and anatomical traits

We investigated relationships between anatomical and morphological traits to identify potential mechanisms behind gradients in LMA and leaf density using partial R² and regression analysis (Table 3, Figure 7). Palisade layer thickness was positively correlated with LMA, leaf density and leaf thickness, while mesophyll porosity was negatively correlated with all three morphological variables (Table 3). Palisade layer thickness explained substantially more variability in LMA, density and thickness than did mesophyll porosity, as indicated by the partial R² analysis and Akaike weights (Table 3). Vein diameter was also positively correlated with LMA, density and leaf thickness within each canopy type (Figure 7). In fact, vein diameter explained 71–84%, 50–74% and 76–88% of the variation in LMA, leaf density and leaf thickness, respectively (Figure 7). In contrast, the fraction of vascular tissue explained less variation in LMA (28–50%), leaf density (18–39%) and leaf thickness (34–39%) (data not shown). We also investigated mechanisms associated with the creation of intercellular air space (mesophyll porosity). Mesophyll porosity decreased with both increasing epidermal cell width and palisade layer thickness, regardless of canopy type (Figure 8).

Table 3.

Partial R² analysis for relationships among palisade layer thickness (μm), mesophyll porosity, LMA (g m⁻²), estimated and measured density (g cm⁻³) and thickness (µm). Sample size, regression coefficients, R² values and Akaike weights (w_i) are displayed for the log-linear regressions.

Response variable	n	Palisade layer thickness only				Mesophyll porosity only				Palisade thickness and mesophyll porosity					Partial R2 for adding:
		β0	β1	R2	wi	β0	β1	R2	wi	β0	β1	β2	R2	wi	Palisade thickness	Mesophyll porosity
(ln)LMA (g m−2)	86	2.79*** ± 0.04	0.03*** ± 0.00	0.90	0.24	4.53*** ± 0.09	−5.47*** ± 0.70	0.42	0.00	2.94*** ± 0.08	0.02*** ± 0.00	−0.74* ± 0.36	0.91	0.76	0.49	0.00
(ln)Leaf density (g cm−3) measured	86	−2.12*** ± 0.09	0.02*** ± 0.00	0.60	0.20	−0.51*** ± 0.10	−5.42*** ± 0.79	0.36	0.00	−1.77*** ± 0.18	0.02*** ± 0.00	−1.68* ± 0.77	0.63	0.80	0.27	0.02
(ln)Leaf density (g cm−3) estimated	86	−1.04*** ± 0.04	0.01*** ± 0.00	0.42	0.10	−0.51*** ± 0.04	−1.91*** ± 0.30	0.32	0.00	−0.86*** ± 0.08	0.01*** ± 0.00	−0.88* ± 0.35	0.46	0.90	0.14	0.04
Leaf thickness (µm)	86	17.36*** ± 3.38	2.05*** ± 0.08	0.89	0.62	151.09*** ± 8.01	−385.12*** ± 61.59	0.32	0.00	10.88 ± 7.22	2.12*** ± 0.10	31.45 ± 30.97	0.90	0.38	0.58	0.00

*P < 0.05, **P < 0.01, ***P < 0.001.

Figure 7.

Relationships between vein diameter and LMA (a), leaf density (b) and leaf thickness (c) at the exposed and closed canopy.

Figure 8.

Relationships between mesophyll porosity, epidermal cell width (a) and palisade layer thickness (b) for the exposed and closed canopy.

Discussion

Patterns and mechanisms associated with palisade layer thickness

Our results support our first hypothesis that cumulative light exposure primarily drives palisade layer thickness, which in turn drives vertical variation in LMA in this mature sugar maple forest canopy. We observed a steep increase in palisade layer thickness above 10 m at the exposed canopy where light availability increased rapidly compared with the closed canopy. Leaf mass per area, density and thickness were more strongly correlated with palisade layer thickness than with mesophyll porosity, highlighting the strong role of light in determining leaf morphology in this temperate deciduous canopy. In addition to the role of maximizing light absorption, thicker palisade layers may also play a role in protecting the photosynthetic apparatus from photoinhibition, which was demonstrated in studies that have observed a higher sensitivity to photoinhibition in leaves grown in shade (Langenheim et al. 1984, Ogren 1988, Kamaluddin and Grace 1992). Our results are consistent with other studies that have found light availability to strongly influence within-canopy variation in leaf morphology in A. saccharum and other broad-leaved temperate deciduous species (Ellsworth and Reich 1993, Sack et al. 2006, Jones and Thomas 2007, Coble and Cavaleri 2014). Studies of tropical forests, western conifer trees and in Eucalyptus plantations, however, have shown that light only influenced LMA lower in the canopy where light is limiting, while height appears to influence LMA higher in the canopy (Ishii et al. 2008, Cavaleri et al. 2010, Coble et al. 2014). Leaf mass per area may also be constrained within branch clusters along vertical gradients in temperate deciduous species (Osada et al. 2014). In A. saccharum, height-related constraints on leaf morphology were prevalent earlier in the growing season, but these constraints were overcome by morphological acclimation to light during leaf maturation over the course of the growing season (Coble and Cavaleri 2015, Coble et al. 2016).

Other factors that may influence palisade layer thickness include previous-year light conditions and seasonal changes in leaf morphology. There is evidence that, in the year prior to leaf expansion and growth, the number of palisade cell layers is determined by light conditions during bud development as the leaf primordium forms (Eschrich et al. 1989, Uemura et al. 2000, Jones and Thomas 2007). Contrary to these findings, leaf thickness and LMA were shown to be strongly influenced by light conditions during leaf development for species that develop only one palisade layer (e.g., Fagus japonica) (Kimura et al. 1998, Uemura et al. 2000). Our study shows that A. saccharum develops only one palisade layer, and a canopy shading experiment in the same stand revealed that LMA was strongly influenced by light conditions during the time of leaf development (Coble and Cavaleri 2015). Overall, this suggests that previous-year light conditions likely do not influence the results of this study.

Our conclusions regarding the strong influence of light on LMA (Coble and Cavaleri 2014) and palisade layer thickness (this study) were likely not biased by seasonal changes in leaf morphology and anatomy because we specifically tested for this and found that both traits were greater at higher light availability compared with lower light for leaves collected at similar heights and during the same week. Evidence for seasonal changes in LMA was observed by Coble and Cavaleri (2015) and Coble et al. (2016), where LMA increased throughout the growing season, especially in the upper canopy, and in this study, LMA was strongly correlated (R² = 0.90, Table 3) with palisade layer thickness. Vertical elongation of palisade cells following full leaf expansion has been reported in other species where leaves continue to thicken after achieving a maximum area (Miyazawa and Terashima 2001, Miyazawa et al. 2003, Yano and Terashima 2004). We also accounted for seasonal increases in leaf anatomy by using time-integrated irradiance (PPFD_INT), which showed strong correlations with palisade thickness. This suggests that PPFD_INT may serve as a useful parameter that incorporates light availability and seasonal changes in leaf morphology and function.

Patterns and mechanisms associated with mesophyll porosity

Our results do not support our second hypothesis that height constrains the expansion of epidermal cells, but provide partial support for the hypothesis that height influences mesophyll porosity. Mesophyll porosity decreased with both light and height within both canopy types. Our results were consistent with the findings of Oldham et al. (2010), who observed a strong linear decrease in mesophyll porosity with height in extremely tall (100+ m) S. sempervirens trees despite sampling leaves from the inner (shaded) and outer (exposed) crowns that differed in light availability. Reduced air space in leaves and thicker cell walls have been found to be important limitations to mesophyll conductance (Loreto et al. 1992, Parkhurst 1994, Syversten et al. 1995, Niinemets 1999, Flexas et al. 2008, Gu et al. 2010). Reduced mesophyll porosity of upper canopy leaves has been considered one of the primary reasons for decreasing mesophyll conductance with height in tall S. sempervirens trees (Mullin et al. 2009, Oldham et al. 2010). Height is often used as a proxy for gradients in leaf water potential because leaf water potential, in the absence of transpiration, declines linearly with height due to gravity in many tree species including A. saccharum (Scholander et al. 1965, Hellkvist et al. 1974, Bauerle et al. 1999, Koch et al. 2004, Coble and Cavaleri 2014). In tall western conifer trees such as S. sempervirens and Pseudotsuga menziesii, declining turgor pressure with height has been implicated as a key limitation to leaf development, leaf morphology and gas exchange (Koch et al. 2004, Woodruff et al. 2004, Ishii et al. 2008, Meinzer et al. 2008, Mullin et al. 2009).

In the same stand as this study, leaf expansion rates were lower in the upper canopy despite having greater turgor, indicating biophysical constraints on leaf development and cell expansion (Coble et al. 2016). In this study, however, upper canopy leaves had larger epidermal cells and lower mesophyll porosity, which seem to contradict our current understanding of the development of intercellular air-space in leaves. During leaf expansion, cells differentiate into epidermal, palisade and spongy mesophyll cells (Eschrich et al. 1989, Tosens et al. 2012) that increase in size during the development of intercellular air spaces (Dale 1988, Knight and Roberts 1994, Marchi et al. 2008, Zhang et al. 2011a, 2011b, Tosens et al. 2012). Given this observed pattern, a driving force would be required to separate spongy mesophyll cells. Avery (1933) considered intercellular air-space to be formed by expanding epidermal cells that separated spongy mesophyll cells, while Jeffree et al. (1986) provide evidence that this view is oversimplified, because initial signs of air-space and cell separation at the junction of cells is formed by the breakdown of cell walls. Regardless of the specific mechanisms involved in the initial separation of cells, the primary driving force of subsequent formation of larger intercellular air space is likely cell turgor pressure and/or mechanical forces in rapidly expanding leaves (Lockhart 1965, Jeffree et al. 1986, Jarvis et al. 2003). Overall, our results suggests that epidermal cell expansion was not responsible for increasing mesophyll porosity in A. saccharum. While height appeared to have some influence on mesophyll porosity, the decline in mesophyll porosity with height did not occur for the same reason that was expected.

Our results support an alternative hypothesis that thicker palisade layers in high light occupy more space and reduce mesophyll porosity. Lower mesophyll porosity higher in the canopy corresponded with greater palisade layer thickness, which comprised much of the total cross-sectional leaf area, particularly at high light. Hanba et al. (1999) observed a similar pattern of lower mesophyll porosity for thicker leaves with greater LMA across six temperate evergreen species. In our study, thicker leaves also appeared to have more cellular area occupied by vascular tissue and surrounding parenchyma tissue, which may also have contributed to reduced mesophyll porosity. This may partially explain why height and light were equally important in determining mesophyll porosity (Table 2). Similarly, Tosens et al. (2012) found that light and leaf water-stress were important in influencing mesophyll porosity in Populus tremula. Flexas et al. (2008) found that LMA set an upper limit to mesophyll conductance, where leaves with greater LMA constrained mesophyll conductance. Thus, there may be a trade-off between light capture and mesophyll conductance, if we assume that greater air space leads to greater CO₂ conductance in the mesophyll.

In leaves with lower mesophyll porosity, development of thicker palisade layers in high light also corresponds with greater surface area of mesophyll cells exposed to intercellular air spaces, which has been found to increase mesophyll conductance in numerous species (Hanba et al. 1999, 2002, Terashima et al. 2006, Tosens et al. 2012). Given that the resistance to CO₂ diffusion is much greater from intercellular air space to the stroma as compared with the pathway of substomatal to intercellular air space, greater surface area of mesophyll cells is required to maintain higher rates of photosynthesis (Terashima et al. 2006). Thus, thicker palisade layers and greater surface area of mesophyll cells exposed to air space corresponded with greater photosynthetic capacity in leaves of three Acer species and P. tremula growing in high light (Hanba et al. 2002, Tosens et al. 2012). Since LMA is strongly correlated with photosynthetic capacity (Ellsworth and Reich 1993, Jones and Thomas 2007) and palisade layer thickness (this study) in A. saccharum, the decline in mesophyll porosity at higher heights likely does not constrain photosynthesis via reduced mesophyll conductance. This hypothesis is further supported by Cano et al. (2013), who found that photosynthesis was consistently greater in upper canopy leaves, despite evidence of lower mesophyll conductance in upper canopy leaves of Fagus sylvatica compared with middle canopy leaves. Further research that investigates the role of mesophyll porosity in mesophyll conductance within forest canopies is required to understand limitations to photosynthesis in upper canopy leaves.

Patterns and mechanisms associated with vascular tissue

Our results support our third hypothesis that leaves in the upper canopy respond to lower leaf water potential through greater investment in leaf vascular and support tissue, which includes the vascular bundle, bundle sheath and bundle sheath extensions. While both vein diameter and fraction of vascular tissue increased with light, height was more important in explaining variation in both anatomical parameters. The leaf vein consists of a vascular bundle (i.e., xylem and phloem) surrounded by the bundle sheath. The bundle sheath, in particular, consists of parenchyma cells (Beck 2010) and plays an important role of sugar transport into and out of the mesophyll, as well as water transport from the xylem to mesophyll cells (Leegood 2008). In tall conifer trees Cryptomeria japonica and S. sempervirens, cross-sectional area of transfusion tissue (also parenchyma cells; Dickison 2000), leaf capacitance and saturated water content increased with height, suggesting that transfusion tissue may play an important role in compensating lower leaf water potential and permitting sufficient turgor pressure and leaf function for leaves at greater heights (Ishii et al. 2014, Azuma et al. 2016). In this study, the strong, positive correlation between vein diameter and height indicates that leaf water storage may compensate the decline in the leaf water potential with height. In addition to osmotic adjustments (Coble et al. 2016), the larger bundle sheath in upper canopy leaves may also aid maintaining greater turgor pressure during leaf development.

Acer saccharum leaves in this study had bundle sheath extensions that extended vertically from the bundle sheath to the upper and lower epidermis. Unlike conifer species, which lack bundle sheath extensions (homobaric), certain deciduous species contain bundle sheath extensions (heterobaric) (Leegood 2008) that transport water from leaf veins to the epidermis, inhibit the lateral transfer of CO₂ between intercellular spaces and lead to non-uniform photosynthesis (Terashima 1992). Lateral transfer of CO₂ in intercellular spaces in homobaric leaves is particularly beneficial during periods of drought stress and stomatal closure (Pieruschka et al. 2006). Bundle sheath extensions can also increase water storage and reduce water stress on the mesophyll (Terashima 1992). Consistent with this interpretation, Zwieniecki et al. (2004) found that upper canopy leaves invested more in vascular tissue (e.g., greater leaf density), and leaf expansion was greater in areas surrounding main veins as compared with the leaf perimeter. Overall, our analysis shows that leaves at greater heights invest more in structural and vascular tissue, which may contribute to reduced mesophyll porosity, help to compensate the effects of lower leaf water potential, and enhance transport of water and photosynthate through the xylem and phloem, respectively.

Conclusions

Our study highlights the strong influence of light availability on palisade layer thickness, and consequent effects on LMA, leaf thickness and leaf density. Results from this study and other studies in A. saccharum suggest that increasing palisade thickness with light may be the primary structural adjustment that increases light capture and photosynthetic capacity along natural light gradients. Also, our results suggest that integrated PPFD is a useful parameter for predicting leaf traits because it incorporates both light availability within the vertical canopy gradients and seasonal effects of light acclimation. Epidermal cell expansion did not appear to be constrained by height, and therefore did not explain decreasing mesophyll porosity with height. Rather, we show that as vascular tissue and palisade cells increase in size and number in the upper canopy, they occupy more leaf volume and decrease air space inside the leaves. Reduced mesophyll porosity may limit mesophyll conductance of CO₂, but this may be counter-balanced by increasing surface area of palisade mesophyll cells, which has been found to increase mesophyll conductance. Overall, the results from this study show that vertical gradients of LMA are strongly driven by light via its effects on both palisade layer thickness and mesophyll porosity, and support previous research that has investigated environmental factors on leaf morphology and function in A. saccharum canopies. We also present the trend of increasing vascular tissue in response to hydrostatic constraints (i.e., height) as another mechanism underlying the universal LMA–height gradient, as vascular and support tissue is denser than surrounding tissue types, and denser leaves have greater LMA.

Acknowledgments

We thank Bethany Blease, Samuel Clair, Dr Ashley Coble, Dr Alex Collins, Jennifer Eickenberry, Kayla Griffith, Dr Kevyn Juneau, Dr Mickey Jarvi, Dr Victor Busov and James Schmierer for their assistance in the lab and field.

Conflict of interest

None declared.

Funding

Research was sponsored by the US Department of Agriculture National Institute of Food and Agriculture McIntire-Stennis Cooperative Forestry Research Program (Grant # 32100-06098), Michigan Technological University Research Excellence Fund-Research Seed Grant and the Michigan Technological University Ecosystem Science Center.

Author notes

handling Editor Hiroaki Ishii