A comparative analysis of phenylpropanoid metabolism, N utilization, and carbon partitioning in fast- and slow-growing Populus hybrid clones

Abstract

The biosynthetic costs of phenylpropanoid-derived condensed tannins (CTs) and phenolic glycosides (PGs) are substantial. However, despite reports of negative correlations between leaf phenolic content and growth of Populus, it remains unclear whether or how foliar biosynthesis of CT/PG interferes with tree growth. A comparison was made of carbon partitioning and N content in developmentally staged leaves, stems, and roots of two closely related Populus hybrid genotypes. The genotypes were selected as two of the most phytochemically divergent from a series of seven previously analysed clones that exhibit a range of height growth rates and foliar amino acid, CT, and PG concentrations. The objective was to analyse the relationship between leaf phenolic content and plant growth, using whole-plant carbon partitioning and N distribution data from the two divergent clones. Total N as a percentage of tissue dry mass was comparatively low, and CT and PG accrual comparatively high in leaves of the slow-growing clone. Phenylpropanoid accrual and N content were comparatively high in stems of the slow-growing clone. Carbon partitioning within phenylpropanoid and carbohydrate networks in developing stems differed sharply between clones. The results did not support the idea that foliar production of phenylpropanoid defence chemicals was the primary cause of reduced plant growth in the slow-growing clone. The findings are discussed in the context of metabolic mechanism(s) which may contribute to reduced N delivery from roots to leaves, thereby compromising tree growth and promoting leaf phenolic accrual in the slow-growing clone.

Cellulose, chemical defence, development, lignin, poplar, root, salicortin, vascular

Introduction

Despite their fast growth rates, Populus species exhibit a propensity toward the accumulation of large foliar reserves of non-structural phenylpropanoid derivatives (NSPs) (Lindroth and Hwang, 1996). NSPs are metabolically costly to synthesize, relatively stable, and less available for growth than more labile carbon (C) forms (Lindroth and Hwang, 1996; Kleiner et al., 1999; Ruuhola and Julkunen-Tiitto, 2000; Kandil et al., 2004). In the context of overall tree fitness, putative metabolic costs (to tree growth) of NSPs are regarded as a potential trade-off in favour of enhanced fitness in stressful environments (Lindroth and Hwang, 1996; Koricheva, 2002; Glynn et al., 2007). Whether or how foliar NSP biosynthesis comprises a trade-off to growth in Populus or other taxa of fast-growing, early-successional tree species remains largely unresolved (Hamilton et al., 2001; Stamp, 2003; Glynn et al., 2007). In annual crop species where secondary metabolite levels often do not exceed a few percent of leaf dry weight, the cost to growth is estimated to be negligible (Foyer et al., 2007). However, the flavonoid-derived condensed tannins (CTs) and the salicin-containing phenolic glycosides (PGs) commonly comprise 1 to >25% of leaf dry weight in Populus (Lindroth and Hwang, 1996; Harding et al., 2005), and therefore may represent a significant cost to growth.

The regulation of NSP accumulation has not been reported for Populus, but in herbaceous taxa such as tobacco, foliar NSP concentrations increase as a function of photosynthetic carbon fixation and diminishing cellular nitrate-N levels (Matt et al., 2002; Fritz et al., 2006). In accordance with the tobacco findings, foliar NSP accrual is promoted in Populus grown under high light, low N, or elevated carbon dioxide levels (Bryant et al., 1983; Agrell et al., 2000; Osier and Lindroth, 2001, 2006). Because Populus are perennial species, they must cope with periodic changes in a range of environmental factors that affect growth. As they mature, they exhibit a number of ontogenetic, adaptive changes in growth rate, shoot–root ratio, and leaf chemistry (Wullschleger et al., 2005; Donaldson et al., 2006b). Due to their lengthy and complex growth cycle, a number of factors are expected to contribute to the regulation of growth and chemical defence, and a number of rationalizing hypotheses have been developed (reviewed in Stamp, 2003).

The most widely cited theories that have been used to model NSP accumulation and growth in Populus and related taxa (Bryant et al., 1983; Herms and Mattson, 1992; Jones and Hartley, 1999) differ largely in their emphasis on the importance of N and C use in controlling the partitioning of assimilate between NSP and growth. According to the ‘Protein Competition Model’ (PCM) of Jones and Hartley (1999), phenylalanine is the N currency shared by phenylpropanoid and protein biosynthesis. At the cellular level, however, tissue nitrate concentration is now known to be an important regulator of phenylpropanoid synthesis (Fritz et al., 2006). In addition, the amino N of phenylalanine can be recovered for other uses (vanHeerden et al., 1996; Singh et al., 1998), making it unclear whether phenylpropanoid synthesis would interfere with, or be suppressed by, protein synthesis.

Other reports are in support of a less prominent role for phenylalanine, and illustrate the importance of C fixation and use as posited in the ‘growth–differentiation balance hypothesis’ (GBDH) of Herms and Mattson (1992). According to the GBDH, NSP synthesis depends on the availability of labile carbohydrates in relation to overall growth demand (Mattson et al., 2005; Glynn et al., 2007). Indeed, biosynthesis of NSP is metabolically costly in terms of C use, and requires substantial sucrose catabolism following localized induction in Populus (Arnold et al., 2004). C from acetate groups is sufficient to sustain flavonoid biosynthesis even when the phenylalanine ammonia-lyase (PAL) enzyme is inhibited by chemical means in Salix (willow) (Mattson et al., 2005; Keski-Saari et al., 2007). Similarly, in Populus, foliar CT content was found to be correlated with flavonoid pathway gene expression, while expression of the upstream PAL gene remained relatively constant (Harding et al., 2005). Overall, phenylpropanoids appear to be more directly linked to growth via a trade-off-based mechanism than do other C-rich secondary metabolites that are not derived from N-containing precursors (e.g. terpenoids; Honkanen et al., 1999)

The present investigation employed a comparative, whole-plant approach for the analysis of phenylpropanoid metabolism and its possible effects on growth of two Populus genotypes. Genotypic differences in the control of leaf N supply appeared to be important to foliar NSP accrual. N concentration was lower in leaves of the slow-growing clone, although it was similar in roots and actually higher in stems compared with the fast-growing clone. A hypothetical scenario is presented by which stem effects on N delivery to expanding leaves could modulate leaf NSP accrual and plant growth.

Materials and methods

Plant materials

The two cottonwood clones compared in this study originate from a suite of P. fremontii × angustifolia hybrids that have been reported on previously (Schweitzer et al., 2004; Harding et al., 2005; Fischer et al., 2006; Woolbright et al., 2008). The subject clones, one slow-growing (SG) with relatively high foliar NSP concentrations, and the other fast-growing (FG) with lower NSP levels, have been maintained by vegetative propagation at Michigan Technological University (Houghton, MI, USA). FG is an F₁ hybrid and SG is a hybrid backcrossed with the P. angustifolia parent (TG Whitham, personal communication). For this study, rooted cuttings, 10–15 cm in height, were transferred into aerated nutrient solution formulated for temperate tree species and modified for Populus (Harding et al., 2005). Individual plants were maintained in perlite-filled pots (12 cm×8 cm×8 cm) bathed in a continuous flow of nutrient solution which received a daily augmentation of 0.36 mmol l⁻¹ N (2.5 mmol l⁻¹ per week) with a molar ratio of nitrate-N to ammonium-N of 4. Distilled water was added daily to the nutrient solution reservoir as needed to maintain a constant volume. Nutrient solution was replaced weekly, and a pH of 5.8±0.3 was maintained by daily additions of 2 N KOH or 2 N HCl.

Plant growth monitoring and tissue processing

At the start of the growth monitoring period, 15 FG and 17 SG plants ∼1 m in height were distributed into four large (1 m² surface area) ebb and flow hydroponics tubs. The selection of similar sized plants was achieved by the use of staggered plantings of the populations from which experimental plants were selected. Established cuttings of SG plants were 4–6 weeks older than FG plants at the time of selection. Approximately 30 cm separated neighbouring plants in each tub, and each tub contained only one genotype. Tubs were positioned 30 cm apart and 3.2 m below a bank of 1000 W, ventilated, metal halide lamps providing supplemental lighting of 600 μmol m⁻² s⁻¹ at plant tops. A photoperiod of 16 h was maintained. Daytime temperatures in the greenhouse ranged between 25 °C and 32 °C, and night-time temperature was 22 °C. Humidity was not controlled. Plant height from the base of the stem to leaf plastochron index (LPI)-0, of the youngest unfurled leaf, 3 cm in length (Larson and Isebrands, 1971), was recorded, and the petiole of LPI-0 was marked. Two of the tubs, containing seven and eight plants of FG and SG, respectively, were harvested 4 weeks after the start of the experiment. Those plants were used for chemical analysis only. The remaining plants were harvested 8 weeks after the start of the experiment for both chemical and biomass measurements.

Harvesting took place between late morning and mid-afternoon, and conditions were sunny on both the 4 week and 8 week dates. Prior to processing, stem lengths above and below the marked petiole were measured, and the number of leaves above the mark recorded. Stem internodes above the mark (new growth) and in the subjacent mid-stem region were split with a razor and vacuum-dried. Root tissue aliquots were snap-frozen in liquid nitrogen for storage at –80 °C. The remaining root tissue was rinsed, blotted, and vacuum-dried. Dried roots were divided into two fractions. The ‘elongating’ root fraction combined small lateral roots, <3 cm in length, with the distal 3–4 cm of longer lateral roots and primary roots. The ‘coarse’ root fraction comprised the remaining root tissue, excluding the basal 3–5 cm. As defined, elongating roots comprised ∼30–50% of total root mass. Leaves at LPI-3 (expanding) and LPI-6 (source) were snap-frozen in liquid N and stored at –80 °C. Powder aliquots were later freeze-dried for chemical analysis. Leaves for biomass determinations were weighed and vacuum-dried.

Chlorophyll fluorescence monitoring

Estimates of variable fluorescence (F_v′=F_m′–F_t, where F_m′ is the maximum fluorescence in the light-adapted state and F_t is the basal fluorescence in the light-adapted state), and photosystem II (PSII) quantum yield (QY=F_m′/F_v′) were obtained using a hand-held Fluorpen™ (Qubit Systems Ltd, Canada). For each determination, four readings were collected from each leaf and averaged. LPI-1 to LPI-5 of three plants of each genotype were monitored every third day over a 9 d period for the analysis.

Chemical analysis

Dried leaf and stem internode samples were ground through a Wiley mill to pass a 20-mesh (fibre analysis) or 40-mesh sieve. N concentrations were measured with a Thermo Finnigan Flash 1112 elemental analyser. The PGs salicin, salicortin, and hydroxycyclohexenone (HCH)-salicortin were measured via high-performance thin-layer chromatography (HPTLC) according to Lindroth (1993). PGs purified via flash chromatography served as standards. Total PG concentration was determined as the sum of the three constituent PGs. CTs were assayed using a modified butanol-HCl method (Porter et al., 1986). CTs purified from P. angustifolia via adsorption chromatography served as the standard. Sugar (hexose and sucrose) and starch levels were determined via the dinitrosalicylic acid method as modified by Lindroth (2002). Glucose served as the reference standard. Glucose released from salicin, salicortin, and HCH-salicortin was determined separately, and the value subtracted from total sugar. Acid-digestible fibre (ADF; used to estimate cellulose content) and Klason lignin were estimated gravimetrically using the Ankom 200 Fiber Analyzer, following sequential extractions in hot acid–detergent solution (100 °C for 1 h) and incubation in 72% H₂SO₄ (3 h).

Statistical analysis

Because only small amounts of tissue from young stem (internodes 1–6 of both clones) and LPI-3 (SG in particular) were available per plant for the entire suite of chemical analyses, samples from several individual plants were pooled for the assays that required larger amounts of tissue. As a result of pooling, the number of replicates was small in some cases, as specified in the figure legends.

It was clear from preliminary leaf disc screening that clonal differences in sugar, CT, and PG were essentially the same at both the 4 week and 8 week dates. Therefore, all 4 week and 8 week data were combined for statistical analysis and simplified presentation. Mixed-effects, nested analysis of variance (ANOVA) was performed using PROC MIXED (SAS 9.0), with genotype, position (organ), and (genotype×position) as fixed effects, and individual tree as a random effect in the following model: Y_ijk=μ+α_i+β_j+γ_ij+b_k(i)+ε_ijk. In this model, μ represents the overall means; α_i represents the effect due to genotype i; β_j represents the effect due to position j; γ_ij represents interaction between genotype i and position j; b_k(i) represents random tree k effect nested within genotype i; and ε_ijk represents error factors. Significance testing of linear correlations, and of selected means comparisons between clones, was performed by the two-sample t-test (data with normal distribution), and the Mann–Whitney rank sum test (data with non-normal distribution) using SigmaStat 3.5.

Results

Height and biomass

During the 8 week monitoring period, shoot incremental height gain was larger in FG than SG (Table 1). Biomass of terminal shoot organs emerging during the monitoring period was greater in FG, primarily due to more rapid leaf growth. In general, leaf expansion was more rapid and of shorter duration in FG (Fig. 1A and Supplementary Table S1 available at JXB online). Photosynthetic chlorophyll fluorescence parameters were measured several times over a 9 d period to compare the development of photosynthetic competence in the clones (Fig. 1B and Supplementary Table S2). The increase in PSII quantum yield and variable fluorescence with leaf development was similar for both clones (Fig. 1B). Dry mass of lower stem internodes did not differ between clones (Table 1), although it is important to note that this was expected since lower stems of SG were 4–6 weeks older than those of FG. The stem biomass trends are consistent with the interpretation that, relative to SG, FG exhibited more rapid height growth in elongating internodes, and more rapid or more sustained radial growth in mature internodes. Root biomass and the root-to-shoot ratio calculated from averaged root and shoot biomass data were higher in FG than SG (Table 1).

Table 1.

Height and biomass accrual of hydroponically grown plants

	SG	FG	t	df	P
Height
Initial height (IH, cm)	106±8.8	118±10.3	–	–	–
New height growth (NH, cm)	78±3.6	103±7.5	8.38	14	<0.001
Incremental change (NH/IH)	0.74±0.06	0.88±0.1	3.28	14	0.005
Upper stem (new growth) biomass
Leaf (g)	6.4±0.9	11.9±2.2	*	*	<0.001
Stem (g)	7.6±1.4	7.4±0.9	−0.46	14	0.656
Total (g)	14.1±2.1	19.2±2.9	4.09	14	0.001
Lower stem biomass
Leaf (g)	23.4±8.1	18.1±7.9	−1.27	14	0.224
Stem (g)	28.4±6.8	34.0±4.6	1.94	14	0.073
Root biomass
Root (g)	12.2±3.3	19.2±4.7	3.59	15	0.003
Root-to-shoot ratio	0.18	0.24	–	–	–

Means and standard deviations of biometric data were determined using n=8 plants of each genotype. Lower stem leaf biomass included twigs. Statistical significance of the difference between clone means was determined using the two-sample t-test for parametric data, and the Mann–Whitney rank sum test for non-parametric data (*).

Fig. 1.

Leaf expansion and chlorophyll fluorescence. (A) Dry masses of fully expanded leaves that occupied LPI positions +1, –1, and –3, denoted as pre(+1), pre(–1) and pre(–3), respectively, at the start of the 8 week experiment, and of expanding leaves that occupied LPI positions 2, 4, 5, 8, and 10 at the time of harvest. Data represent the means and SD of eight plants. (B) Quantum yield of PSII and variable fluorescence of expanding leaves. Data represent the means and SD of three plants for each LPI position, each derived from the mean of 36 instrument readings.

Phenylpropanoid and carbohydrate concentrations

Non-structural components:

In all organs analysed, NSPs were more abundant in SG than FG (Fig. 2; Supplementary Tables S3, Supplementary Data at JXB online). In leaves, sharply higher CT concentrations accounted for most of the NSP difference between the two clones (Fig. 2B). In upper internodes, concentrations of both CT and PG were 3- to 5-fold higher in SG than in FG (Fig. 2). Soluble sugar concentrations were generally similar in source leaves (LPI-6) of both clones, but were higher in expanding leaves (LPI-3) of FG than SG (Fig. 3A; Supplementary Tables S3, Supplementary Data). In contrast, soluble sugar contents were lower in the upper stem internodes of FG than SG (Fig. 3A). Foliar starch concentrations were higher in leaves and upper internodes in SG than FG (Fig. 3B, C). NSP and non-structural carbohydrate concentrations in elongating and coarse root fractions, though substantial, were smaller than in leaves, and differed little between clones (Figs 2, 3). Starch concentrations were nearly identical in the elongating root fraction of both clones, but were lower in the coarse root fraction of FG than SG (Fig. 3B, C; Supplementary Tables S3, Supplementary Data).

Fig. 2.

Tissue concentrations of NSP constituents, PG and CT, in leaves, roots, and stem internodes of the two clones. (A) Phenolic glycosides. (B) Condensed tannins. Shown are data (% dry weight) from expanding leaf (LPI-3), fully expanded source leaf (LPI-6), upper stem internodes near expanding leaves (Int 1–6), upper stem internodes along a developmental gradient of maturing source leaves (Int 7–8 and 9–13), and elongating root and coarse root fractions. Internode number increases basipetally. Replicates (n) ranged from 11 to 15 plants for each LPI, from 7 to 11 plants for each subset of internodes, and from 15 to 16 plants for each root fraction. Comparisons between clones were made for each organ fraction shown. The two-sample t-test was used to determine significance of differences between clone means, indicated by asterisks above the SG histogram bars (*P <0.05; **P <0.01; ***P <0.001).

Fig. 3.

Tissue concentrations of (A) soluble sugars and (B) starch. Shown are data (% dry weight) from expanding leaf (LPI-3), fully expanded source leaf (LPI-6), upper stem internodes near expanding leaves (Int 1–6), and upper stem internodes along a developmental gradient of maturing source leaves (Int 7–8 and 9–13), and elongating root and coarse root fractions. Replicate numbers and significance testing are as in Fig. 2.

Structural components:

Lignin content reached a sustained maximum with respect to stem biomass much earlier in FG than in SG (Fig. 4A). A histological approach was used to compare lignification and vascular development of primary–secondary transitional internodes 3–6. UV autofluorescence of stem cross-sections revealed that fibre development and lignification were comparatively slow in xylem, but not phloem, of SG relative to FG (Fig. 5). The production of lignifying cells, based on the width of the autofluorescing zone in the xylem at both stages of internode growth, was more rapid in the FG clone (Fig. 5, arrows). Together, the Klason lignin and histological data were consistent with slowed xylem development relative to diameter growth of young internodes in SG. Lignin concentration increased between internode ∼10 and mid-stem internodes 20–25 in SG, but not FG (Fig. 4A).

Fig. 4.

Lignin and cellulose in developing stems. (A) Klason lignin and (B) ADF cellulose (% dry weight). Each of the upper internode data points (Int 7–8, Int 9–10) represents the mean and SD of 4–6 determinations. The mid-stem internode data points (Int 20–25) represent the means of eight (SG) and 16 (FG) determinations. In several cases, SD bars are present but are smaller than the data symbol. Exceptions are the Int 1–6 and Int 11–12 data points which represent single determinations from pooled stems with no SD. Internode replicates (n) differ from those in Figs 2 and 3 because pooling of stem internodes necessary for lignin and cellulose determinations differed from that for other stem internode assays.

Fig. 5.

Lignin UV autofluorescence of primary–secondary transitional stem internodes. Shown are internode 3 from FG (A) and SG (B), and internode 6 from FG (C) and SG (D). Images were obtained from 75 μm vibratome sections using an excitation wavelength of 365 nm. Scale bar=500 μm. The arrow placed across the xylem is to facilitate a comparison of secondary xylem width which was wider in FG than in SG. pf, phloem fibre; xy, xylem.

At all internodes analysed, cellulose was more abundant than lignin, and exhibited a more uniform accrual trajectory than did lignin (Fig. 4B). However, cellulose content was substantially lower in stem internodes of SG compared with FG. Of potential relevance to cellulose biosynthesis and accrual was the observation that starch concentrations were higher in stems and leaves of SG than FG (Fig. 3). Leaf tissues were used to investigate the possibility of a clonal difference in the metabolic relationship between starch and cellulose biosynthesis (Fig. 6). Due to sample limitation (see Materials and methods), leaf tissues, but not stem tissues, offered sufficient replication to conduct the correlation analysis. Strong individual plant entrainment of starch metabolism throughout leaf expansion was observed in both clones (Fig. 6). Source leaf starch and cellulose content were negatively correlated in FG (–0.908) but not in SG (Table 2). Clonal differentials in cellulose, lignin, and NSP were smaller in roots than in shoots (Figs 2, 7). A large increase in cellulose and a decrease in starch during the maturation of elongating into coarse roots were observed in both clones (Figs 3, 7).

Table 2.

Correlation analysis of starch and cellulose concentrations in leaves

Parameter	R	P
LPI-8 cellulose versus LPI-6 starch
SG, n=11	−0.449	0.16
FG, n=15	−0.908	<0.001
LPI-3 starch versus LPI-6 starch
All, n=26	0.938	<0.001

Fig. 6.

Regression analysis of starch concentrations between expanding (LPI-3) and fully expanded source (LPI-6) leaves. Starch data were collected from leaves of 26 plants for the analysis.

Fig. 7.

Lignin and cellulose concentrations in root fractions. (A) Klason lignin of elongating and coarse root fractions and (B) ADF cellulose of elongating and coarse root fractions (both in % dry weight). Histogram means and SD were determined from n=14–15 replicates. The two-sample t-test was used to determine significance of differences between clone means, indicated by asterisks above the SG histogram bars (**P <0.01; ***P <0.001).

Total nitrogen and carbon

After N is taken up, it is transported to expanding leaves through root and stem tissues where NSP, lignin, sugar, and cellulose metabolism differed between the two clones. Not surprisingly, the plant-wide distribution of N from roots to leaves differed between the clones (Table 3). Total N of elongating and coarse roots differed little between clones, but was slightly lower in SG than FG (Table 3). Nitrogen concentrations in mid-stem (internodes 20–25) and upper stem internodes subtending the source leaves (internodes 7–13) were substantially higher in SG than FG (Table 3, Figs 2, 4). The N concentration difference between clones in young internodes (1–6) was less striking, although the tendency for higher %N in SG persisted. In contrast, foliar N concentrations were substantially lower in SG than in FG (Table 3). When the leaf and internode %N data were expressed in the form of a leaf:internode N ratio, the ratio was higher in FG, for both expanding and recently expanded (source) leaves (Table 3). The N ratio illustrates the comparative reduction in N distribution to leaves of SG.

Table 3.

Leaf, internode and root total N concentration and internode-to-leaf ratios in new growth

	SG	FG	t	df	P
Expanding leaf	3.5±0.30	4.2±0.46	−4.10	24	<0.001
Expanding leaf internodes (1–6)	2.3±0.24	1.9±0.49	*	*	0.04
Leaf-to-internode ratio	1.5	2.1	–	–	–
Source leaf	2.9±0.19	3.9±0.45	−7.12	24	<0.001
Source leaf internodes (7–13)	2.1±0.30	1.6±0.44	3.89	31	<0.001
Leaf-to-internode ratio	1.4	2.4	–	–	–
Mid-stem internodes	1.6±0.2	1.2±0.2	4.42	21	<0.001
Elongating root	3.1±0.24	3.3±0.3	−2.29	29	0.03
Coarse root	2.2±0.15	2.4±0.2	−1.92	29	0.06

Total N (% dry weight) data were pooled from both harvests of FG and SG plants. Means and standard deviations were obtained from at least n=8 SG and n=15 FG plants. Statistical significance of the differences between clone means was determined using the two-sample t-test for parametric data, and the Mann–Whitney rank sum test for non-parametric data (*).

Total C was slightly higher in expanding and source leaves of SG than FG (Table 4). The C:N ratio also was higher in SG leaves. Thus, the decrease in C:N ratio between source and expanding leaves of SG was not due to a decrease in total C.

Table 4.

Total leaf C content and C:N ratio

	SG	FG
Total C
Expanding leaf	46.71±0.29	45.88±0.67
Source leaf	46.63±0.97	45.42±0.51
C:N ratio
Expanding leaf	13.31±1.04	11.07±1.25
Source leaf	16.25±1.08	11.76±1.27

Data were pooled from both harvests of FG and SG plants. Means and standard deviations were obtained from n=8 (SG) or n=15 (FG) plants. As determined by two sample t-test, differences between genotypes were significant (P <0.001) for all comparisons, and the C:N ratio differed significantly between expanding and source leaf in SG (P <0.001).

Discussion

Analysis of the six major carbohydrate and phenylpropanoid pools (soluble sugar, starch, CT, PG, lignin, and cellulose) in developing leaves, stems, and roots of two hybrid lines was undertaken as a systems approach to determine whether foliar NSP biosynthesis causes, or directly contributes to, growth reduction in Populus. A trade-off would be expected based on published data that growth often correlates negatively with foliar NSP accrual in Populus (Lindroth and Hwang, 1996). A metabolic basis for such correlations has been offered in Salix (willow), where application of PAL-specific inhibitors reduced NSP accrual while enhancing plant growth (Ruuhola and Julkunen-Tiitto, 2003).

Recently it has become evident that negative effects of high foliar NSP on Populus growth may only be significant under conditions of limiting external N (Donaldson et al., 2006a). This finding is in line with a central tenet of GBDH that when growth is limited, and C is not, NSP accrual is favoured (Herms and Mattson, 1992). The results showed further that higher rates of foliar NSP accrual can indeed pose a greater cost to growth when N becomes limiting (Donaldson et al., 2006a). Based on the present findings, GBDH can also be used to rationalize the foliar NSP differences between genotypes, in this case SG and FG, under N-replete conditions. Total leaf %C was slightly higher in SG (Table 4), as were the contents of metabolically active sugar and starch (Fig. 2). Together, the C and N results are consistent with the interpretation that leaf growth was comparatively N-limited in SG. In nearby stem internodes, neither biomass growth nor %N was reduced in SG, relative to FG, during the same period. Because it is not clear how C availability limited SG leaf growth, the results potentially provide a degree of validation for the PCM tenet that phenylpropanoid competition for N within expanding SG leaves directly led to their reduced growth rate (Jones and Hartley, 1999). However, molecular evidence supports nitrate-N sensing rather than consumption of phenylalanine-N as the basis for phenylpropanoid stimulation in higher plants (Fritz et al., 2006).

It is worth noting the adaptive benefit that may be linked with how foliar N supply, foliar NSP accrual, leaf growth, and overall plant growth are integrated in SG. Foliar CT abundance is a characteristic that appears to vary with soil N mineralization rates in the ancestral habitats of SG, FG, and related hybrid genotypes (Schweitzer et al., 2004; Fischer et al., 2006). This pattern may be of genetic significance in that CT and other phenolics can affect soil N cycling, and promote habitation of leach-prone environments by N-demanding plants (Northup et al., 1995; Chapman et al., 2006). In such environments, slower plant growth may be favoured in exchange for improved soil N sequestration. In other words, reduced provision of N to leaves would slow plant growth, coupling growth rate with the rate of soil N mineralization. In response to lower leaf N, CT becomes a quantitatively important leaf C sink. In turn, the CT-enriched detritus of such leaves continues to promote soil N sequestration and to modulate soil N mineralization into leach-susceptible forms (Northup et al., 1995; Chapman et al., 2006). Because nitrate reductase activity is generally low in roots of Populus (Black et al., 2002), nitrate-N is an important N transport form in these species (Siebrecht and Tischner, 1999). In turn, variation in nitrate availability due to soil/climate conditions could be an important modulator of phenylpropanoid metabolism and, perhaps, biomass accrual during Populus ontogeny.

The observations that soluble sugar concentrations were not reduced and that both starch and CT were elevated in SG source leaves argue against the idea that leaf demand for N in that genotype was limited by low rates of C fixation. This is supported by data showing that the C:N ratio of SG leaves was ∼20–40% higher than that of FG leaves (Table 4). Chlorophyll fluorescence parameters used to gauge clonal differences in the efficiency of light utilization according to Genty et al. (1989) did not provide any indication that photosynthetic development followed a different trajectory in SG than in FG (Fig. 1). For both genotypes the increase in variable fluorescence (F_v) with leaf maturation was significant, but there was not a significant effect of genotype on F_v (Supplementary Table S2 at JXB online). The quantum yield of PSII did not differ by leaf age or genotype. From the chlorophyll fluorescence data it does not appear that the genotype differences in leaf C partitioning resulted from, or fundamentally altered, light harvesting and photosynthesis.

A decrease of soluble sugars in the expanding leaf compared with the source leaf was observed in SG but not FG (Fig. 2). The sugar concentration gradient observed between source and sink leaves of SG could be due to a combination of slowed secondary xylem development in the connecting internodes (Figs 4, 5) and/or enhanced levels of sugar-demanding metabolic activities there. Mass transport of water through the xylem is an important facilitator of phloem sugar transport in taxa such as Salix and Populus where phloem loading is passive (Munch, 1930; Turgeon and Medville, 1998). Therefore, a reduced rate of secondary xylem growth in younger internodes could affect the phloem transport of sugars. In addition, source leaf sugars destined for expanding leaves were routed through a region of developing internodes that was highly NSP enriched in SG. Primary internodes are one likely site of PG biosynthesis in Populus and related taxa (Clausen et al., 1989) and, therefore, the internodes could exhibit enhanced sink strength relative to developing leaves. The predominant PG in both clones is salicortin (Rehill et al., 2005) which is 42% glucose by weight. Therefore, the relatively high PG content of young SG internodes (∼12% dry weight versus ∼3% in FG) may reflect a substantial demand for soluble sugars, in addition to the basal C demand for phenylpropanoid skeletons. The high level of stem NSP biosynthesis could thus affect growth by competing with younger, more sink-like leaves, for soluble sugars in the transport stream. It is also possible that demand by expanding SG leaves for imported sugars was low due to an effect of slower xylem development on N delivery.

A unifying and interpretive scenario of the present findings is shown in Fig. 8. A cumulative effect of some metabolic process(s) in roots and stems on N distribution is posited to reduce the provision of N to expanding leaves while promoting pooling of N in SG stems. While starch and CT were both elevated in leaves of SG, consistent with a perceived N limitation, cellulose accrual was lower than in FG. In contrast to FG, there was not a correlation between foliar starch and cellulose contents in SG. The absence of a correlation could reflect strong interference by competing synthetic activities for starch-derived glucose in SG. Under the circumstances of lower foliar N in SG, CT synthesis could be an important alternative destination for starch glucose. Because cellulose precedes lignin deposition during secondary cell wall growth (Mellerowicz et al., 2001), reduced cellulose synthesis could have interfered with lignification of vascular traces, explaining the reduced xylem fluorescence in elongating internodes of SG (Figs 4, 5). A loop would thereby be constituted in which reduced vascular growth in upper stems further impeded N delivery to leaves. In general, the scenario illustrates how the relatively low amino acid concentrations commonly observed in leaf metabolic profiles of slower growing, high foliar NSP clones (Harding et al., 2005) could reflect a foliar N limitation, due to an inefficient transport system and/or an N-sequestering process in secondary stems and roots.

Fig. 8.

Proposed phenylpropanoid effects on N distribution and vascular development in SG and FG. Enhanced lignification in roots and lower stems of SG is proposed to reduce N distribution, represented by the term ‘flux’, into upper internodes. A decrease in foliar %N results, which favours starch and CT accrual at the expense of cellulose deposition in developing vascular traces. With less cellulose scaffolding, lignification is reduced and NSPs such as the PGs become the predominant phenylpropanoids. Ultimately, both reduced vascular development and high sugar demand for NSP biosynthesis interfere with the provision to expanding leaves in SG.

The use of labelled N will be necessary to determine unequivocally the nature of N flux differences between SG and FG referred to in Fig. 8. Besides phenylpropanoid synthesis, important determinants of variable N flux in trees also include N remobilization from storage reserves, and reallocation of N between roots and shoots according to seasonal growth demands (Millard, 1996; Grassi et al., 2003, and references therein). The present findings that nutrient distribution to leaves may be conditioned by metabolic processes related to stem N use are partially analogous to findings of others that vascular architectural heterogeneity affects leaf provision and growth (Orians et al., 2002, and references therein). Integration of gene networks that control vascular architecture and chemistry remains a relatively unexplored topic for future investigation.

The authors express their gratitude to Adam Gusse for excellence in tissue chemical analysis, and to Professor Thomas Whitham (Northern Arizona University) for providing the cottonwood genotypes. This research was supported by the US Department of Energy's Biological and Environmental Research Program DE-FG02-05ER64112 and by the National Science Foundation Plant Genome Program DBI-0421756.