Author for correspondence: A. Pornon Fax: +33 5 61 55 61 96 Email: firstname.lastname@example.org
• Nitrogen transfers in mature (c. 40-yr-old) Rhododendron ferrugineum plants from mature leaves to growing shoots (Sh0) were estimated in the field at the branch scale.
• Before bud break, the N pool of either 1-yr-old (L1) or 2-yr-old (L2) leaves of selected branches was labelled with 15NH4Cl. The N dynamics were then investigated on three occasions during shoot growth. Mineral N uptake by roots was estimated by supplying seedlings with 15NH4+ or 15NO3− (glasshouse) and mature plants with 15NH4+ (field).
• Approximately 60% of the 15N supplied was recovered in the labelled branches (40% in L1 leaves and 19% in Sh0) 78 d after L1 labelling. A similar pattern was observed for L2 labelling. Only traces of 15N were detected in 1- and 2-yr-old-stems. Although changes in the isotopic contents showed the occurrence of N transfer from mature leaves to the shoot, total L1 and L2 leaf N content was not modified. Simultaneously, exogenous 15N uptake by the roots was very low.
• We conclude that, despite the internal N cycling, mature leaves do not constitute a net source of nitrogen supporting shoot growth. The latter depends mainly upon endogenous nitrogen stored in woody stems and roots.
Resorption of elements from senescing leaves is a major biological process since it reduces losses of elements through litter (Aerts, 1996) and enables the plant to reuse elements for growth and reproduction
Resorption efficiency is usually deduced from both the amount of nutrient in fallen leaves and in leaves before senescence (Aerts, 1996; Eckstein et al., 1998). Nutrient remobilization may, nevertheless, occur during the whole life-span of a leaf (Chabot & Hicks, 1982) and resorption from nonsenescing leaves could be particularly important in evergreen species for several reasons. (1) They usually allocate a higher proportion of biomass, and thus elements, to their leaves than deciduous species (Larcher, 1995); this enhances the storage function of assimilating tissues. (2) The amounts of various elements have been shown to steadily decrease with leaf age (Reader, 1978; Dambrine et al., 1991; Escudero et al., 1991; Oliveira et al., 1996) in such a way that all leaf cohorts, including nonsenescing leaves, are likely to be involved in resorption processes. This implies that resorption acts over a longer period in evergreen than in deciduous species and that the magnitude of the resorption may be underestimated by considering only senescing leaves. For example, studies have shown that remobilization of N from first-year needles of Pinus radiata (Nambiar & Fife, 1987) and Pinus sitchensis (Millard & Proe, 1993) provides a high proportion of N used by the current-year needle growth. (3) Numerous evergreen species exhibit gradual leaf fall so that the youngest leaf cohorts often include a greater number of leaves than the oldest ones and, as a consequence, contain larger reserves than the latter (Escudero et al., 1991).
Although in many evergreen species, leaf shedding and shoot growth are partly or fully synchronous, there is no real evidence that the two processes are physiologically linked. Pornon et al. (1998) have shown in some alpine populations of Rhododendron ferrugineum that a significant fraction, and in certain cases, all 2-yr-old leaves (L2) fell during shoot growth (June–July). Simultaneously, N levels decreased strongly in 1-yr-old leaves (L1). These observations suggest that N resorption from all leaf cohorts may be closely linked to shoot production and that direct transport of nitrogen from mature leaves to growing shoots can occur. This is thought to be a major difference between evergreen and deciduous species. Indeed, the deciduous species store the nitrogen from senescing leaves in autumn in stems and roots, essentially as proteins or free amino acids, until it is required for new growth early in spring the following year (Chapin et al., 1980; Deng et al., 1989; Millard & Proe, 1991). The role of leaves as reserve organs supporting shoot growth has been inferred in evergreen plants mainly by removing old leaves (Reader, 1978; Jonasson, 1989; Karlsson, 1994; Eckstein et al., 1998), but this may alter how plants function in general (Vanderklein & Reich, 2000). Moreover, in numerous studies, resorption from nonsenescing leaves has been either ignored or not clearly taken into account. The quantification of fluxes between different leaf cohorts may be approached through isotopic labelling. To our knowledge, attempts to study direct transfer of N from leaves of different generations to growing shoots have seldom been made with this method.
In the present study, we investigated the nitrogen source-sink relationships between mature leaves (source) and the current shoot production (sink) in Rhododendron ferrugineum. The objectives were to quantify N resorption from 1-yr-old (L1) and 2-yr-old (L2) leaves during shoot growth and direct N transport from them to the growing shoots. Sink–source interactions were studied at the individual branch scale by labelling leaf nitrogen with 15N. In an additional experiment, we also studied nitrogen uptake by roots to specify the involvement of soil nitrogen in shoot growth.
Materials and Methods
Species and study site
Rhododendron ferrugineum L. (Ericaceae) is an evergreen shrub with well-branched trailing stems that reaches a height of 70–80 cm. It is widely distributed in the Alps and Pyrenees at altitudes between 1600 m and 2200 m (Ozenda, 1985). It reproduces both sexually by selfed and outcrossed seeds, and vegetatively through layering (Escaravage et al., 1997; Pornon et al., 1997). It is able to dominate many subalpine landscapes on north- to west-facing slopes (frequently reaching 90–100% of the cover) by outcompeting other species.
For outdoor experiments, the study was conducted at a subalpine heath site just above the timberline (‘Herbe Soulette’ pass, 42°51′ N, 0°52′ E, c. 1560 m above sea level) located in the central Pyrenees about 150 km south of Toulouse, France. The experiments took place on a north-west facing slope (75% inclination). The average annual precipitation is 1750 mm and the snow cover often persists from November until May The site was dominated by R. ferrugineum, Vaccinium myrtillus and Festuca eskia. Soils are Typical Haplorthod to Umbric Dystrochrept with a pH (H2O) of 4.3 and C : N ratio of 14.
Leaf 15N labelling
The experiment was carried out in the field during the 1998 growing season (from May to July). Before bud break, 16 individuals of similar size (i.e. with about 30 branches, corresponding to roughly 40-yr-old plants) were randomly selected. Eight branches bearing both L1 and L2 leaves were tagged per individuals (128 branches). The term ‘branch’, as used here, refers to the unit composed of the stems and leaves of the last two growth seasons, and the apical bud. Half of the 16 individuals were used for 15N labelling of 1-yr-old leaves (L1) and the other half for 15N labelling of 2-yr-old leaves (L2). Labelling was performed as follows: on 6 May 1998 (time T0′), all the 1-yr-old or alternatively 2-yr-old leaves (approximately 10 and 8 leaves per branch, respectively, Table 1) from the tagged branches of an individual were labelled by deposition of 4 µl of 15NH4Cl (50 mm, 15N abundance of 99 atom per cent) on the abaxial face of the leaves. Two microlitres of solution was deposited on each side of the main vein, gently abraded with a scalpel blade over roughly 2 mm2. We used 15NH4+ rather than 15NO3− because ericaceous plants do not have a nitrate reductase activity in their leaves (Smirnoff & Stewart, 1985). The volume and the concentration of the 15N solution supplied to the leaves were determined and tested to satisfy the following requirements: the volume of solution must be sufficiently small to entirely and rapidly penetrate the leaf; the amount of NH4+ supplied must be sufficiently small to prevent any toxic effects (Chaillou & Lamaze, 2001); the alteration of internal nitrogen status must be negligible (0.2 µmol exogenous 15N supplied vs c. 20–30 µmol endogenous N in a leaf); but the amount of tracer must be sufficiently great to still be detectable after isotopic dilution.
Table 1. Changes, during the growing period, in the main characteristics of the current shoots (mean ± SD) and the two generations of leaves of mature Rhododendron ferrugineum growing in field conditions. Shoots (Sh0) represent new stem, leaves and bud of the current-year tissues. The column ‘Mean dry mass per leaf’ for 1-yr-old leaf (L1) and 2-yr-old leaf (L2) represents the mean value of one leaf and total leaf biomass of a specified leaf type is given by the product of the columns with the number of leaves per branch1
Total dry mass (mg) n = 16
N concentration (%) n = 16
Number of leaves per branch n = 16
Mean dry mass per leaf (mg)
N concentration (%) n = 16
Number of leaves per branch n = 16
Mean dry mass per leaf (mg)
N concentration (%) n = 16
The differences between phenological stages were statistically tested by a one-way anova followed by a Tukey HDS multiple range test. Values sharing the same letter are not significantly different at P < 0.05.
T0, 6 May; T1, 18 May; T2, 18 June; T3, 21 July.
42.3 ± 20.97a
1.15 ± 0.14a
10.18 ± 1.98a
33.5 ± 8.91a
1.50 ± 0.15a
7.71 ± 1.34a
32.02 ± 8.01a
1.06 ± 0.11a
n = 333
n = 204
42.3 ± 20.97a
1.15 ± 0.14a
9.94 ± 1.6a
36.92 ± 5.34ab
1.38 ± 0.10ab
6.97 ± 1.35a
34.95 ± 12.62a
1.07 ± 0.15a
n = 236
n = 243
97 ± 39.66a
2.76 ± 0.29c
9.83 ± 1.29a
41.52 ± 8.62b
1.37 ± 0.15ab
6.74 ± 1.27a
34.36 ± 5.12a
1.11 ± 0.17a
n = 285
n = 207
403 ± 176.5b
1.99 ± 0.28b
9.77 ± 1.14a
35.85 ± 4.31ab
1.25 ± 0.14b
5.96 ± 1.83b
33.42 ± 13.22a
1.01 ± 0.15a
n = 340
n = 158
In order to estimate the amount of 15N effectively loaded into the leaves, a first harvest was performed the day after T0′ (time T0). For the study of N dynamics, three harvests were made throughout the period of shoot development: before bud break (18 May, T1), during current shoot growth (18 June, T2) and at the end of shoot growth (21 July, T3). Two branches per individual were collected at each harvest. Ten additional individuals were used to determine the natural 15N abundance in the tissues.
Immediately after branch excision, the tissues were fixed and stored in liquid nitrogen. In the laboratory, the branches were separated into the various compartments studied: current shoots (Sh0), 1-yr-old leaves (L1), 2-yr-old leaves (L2), 1-yr-old stems (St1) and 2-yr-old stems (St2). The tissues were dried at 70°C for 48 h and ground to a fine powder (< 1 µm). The two harvested branches of a given individual were pooled by compartment (see section 15N-Nitrogen uptake by the plant) and carefully mixed before analysis of 15N abundance using a continuous-flow isotope ratio mass spectrometer coupled with an elemental analyser (model ANCA-MS, Europa Scientific, Crewe, UK; Clarkson et al., 1996). Leaves that fell during the experiment were not collected.
15N-Nitrogen uptake by the plant
At the end of April 1998, 18 young Rhododendron ferrugineum shrubs (c. 5-yr-old) were transferred from the field to the laboratory’s greenhouse. At this age, shrubs are about 10 cm high and usually bear one or two branches. Environmental conditions were natural light, 25°C/20°C and 50%/70% relative humidity (r.h.) (day and night, respectively). Plant roots were carefully removed from the soil, rinsed with tap water and transplanted into individual pots of sand (0.2 10−3 m3). The pots were placed in trays containing a complete nutrient solution with a pH of 4.5. The basal culture solution contained the following macronutrients: 0.1 mol m−3 KNO3, 0.05 mol m−3 (NH4)2SO4, 0.25 mol m−3 MgSO4, 0.4 mol m−3 CaSO4, 0.25 mol m−3 CaCl2, 0.25 mol m−3 KH2PO4, 1 mol m−3 KCl. The basal culture solution also contained the following micronutrients: 9 mmol m−3 MnSO4,H2O, 0.75 mmol m−3 ZnSO4,H2O, 0.3 mmol m−3 CuSO4(H2O)5, 45 mmol m−3 H3O3B, 0.5 mmol m−3 MoO3, 90 mmol m−3 EDTA Na2Fe. The solution was renewed three times a week to prevent nutrient depletion, nitrification and pH changes. The characteristics of the solution were not notably changed between two renewals. At the beginning of shoot growth (beginning of July), plants were supplied with the same nutrient solution as previously except for nitrogen which was present either as NO3− or NH4+ (0.2 mol m−3 enriched with 15N up to 50 atom per cent, nine plants per treatment). Although we suspected that Rhododendron ferrugineum preferentially uses NH4+ rather than NO3− as nitrogen source, both forms of nitrogen were tested. At the end of shoot growth (15 d later), plants were sampled, pooled in batches of three individuals and separated into four compartments (roots, stems, leaves and shoots).
To avoid potential artefacts due to root damage during transplantation from soil to sand, a similar uptake experiment was performed with plants whose roots were disturbed but which remained in soil. Thus, in a second experiment (end of April 1998), 20 young plants were extracted from the field with a 0.2 × 10−3 m3 ball of earth left on the roots, placed in pots in the greenhouse and then treated as previously. In addition, in a third outdoor experiment, six c. 40-yr-old individuals randomly selected were uprooted with a ball of earth roughly 20 × 10−3 m3, placed individually in plastic bags to prevent leaching and replaced in the field (October 1997). Before the next growing period (end of May 1998), each plant was supplied with 60 ml of NH4Cl (23 mm, 15N abundance of 99 atom per cent) fractionated in injections of 10 ml regularly distributed in the soil around the plant. After cessation of shoot growth (mid July), shoots were sampled and assayed for 15N.
The amount of 15N in excess in each plant compartment was calculated as the product of m, the dry mass of the compartment, times c, the total nitrogen concentration (%), times e, the 15N in excess. Isotopic excess was calculated as the difference between the 15N abundance in the compartments of labelled plants and ‘natural’15N abundance in control plants (0.365%).
Differences between phenological stages were statistically tested for each parameter and each plant compartment by performing one-way anova (Systat Inc., 1997). Percentages of 15N were arcsine transformed before being analysed. When a significant variation among means was detected we performed a Tukey HDS multiple range test to determine the source of variation.
Branch growth and nitrogen content
Changes in the mean characteristics of leaves and shoots of 40-yr-old Rhododendron ferrugineum plants during the growing period (from late spring to early summer) are shown in Table 1. Before growth (times T0 and T1), Sh0 is composed of the apical bud, and then it corresponds to the entire current-year leaves, stem and new buds (T2 and T3). The current-year shoots (Sh0) started to grow in mid-June (after T1) and Sh0 leaves were fully expanded by late July (T3). During this period, some 2-yr-old leaves (L2) were shed whereas the number of 1-yr-old leaves (L1) was unaffected. The N concentration in Sh0 tissues was high during growth and was maximum at the beginning of the growth period (time T2 in Table 1). The average individual leaf mass of L1 leaves transiently increased significantly during shoot (Sh0) development whereas the mass of L2 leaves did not vary significantly during the same period. Between times T0 and T3, the N concentration decreased significantly in L1 leaves but remained constant in L2 leaves (attached leaves).
Changes in N levels in the various branch compartments reveal that during shoot development a large amount of N accumulated in Sh0 (about 7.5 mg per branch between T1 and T3, Fig. 1) while the amount of N in attached L1 and L2 leaves was not significantly modified. Thus, net remobilization of N from L1 and L2 leaves could only slightly contribute to shoot (Sh0) growth.
One day after loading (time T0), 86% and 76% of the amount of 15N theoretically provided to L1 and L2 leaves (T0′), respectively, was recovered in these compartments. The discrepancy may result from a loss of 15N in the external environment. Part of the 15N supplied could have already moved from the labelled compartment to unlabelled ones which were not analysed at this time. Moreover, owing to their location in the canopy of the shrubs it was often difficult to deposit accurately 2 µl of 15N solution on the leaves, especially L2 leaves. Thus, in this text, the amount of 15N is expressed as a percentage of the amount found at T0. At the end of the experiment (T3), about 60% of the labelled N present at T0 in L1 and L2 leaves, respectively, were recovered in the branch (compartments Sh0, L1, L2, St1 and St2, Fig. 2). In all cases, only traces of 15N were detected in the stem wood. More than half of the initial amount of 15N was lost by L1 leaves during the experiment (78 d). Approximately 19% appeared in the shoots but only traces were found in L2 leaves. Thus, consistent 15N flux occurred from L1 leaves to the growing shoots. In the case of L2 labelling, the same pattern was observed although 15N export to Sh0 was smaller (13%).
Nitrogen uptake by the plants
In the greenhouse, and whatever the substrate (sand or natural soil) the 15N uptake by plants supplied with either 15NH4+ or 15NO3− during the growth period was extremely low (Table 2). The amount of 15N recovered in the shoots at the end of the growing period was negligible compared with the total N content of this compartment. For the experiment where an inert substrate (sand) was used, the enrichment in 15N-NH4+ or 15N-NO3− was 50 atom per cent in the nutrient solution so that the amount of exogenous N in the plants can be estimated by doubling the amount of 15N in excess in the tissues. As observed for plants in the greenhouse, the amount of 15N in excess in 40-yr-old plants (in field experiment) over the natural 15N content was extremely low (64 ± 60 µg 15N compared with 7340 ± 3760 µg total N).
Table 2. Amount of total N and 15N (mean SD) in different plant parts of 5-yr-old Rhododendron ferrugineum growing on sand or in natural soil supplied with a complete nutrient solution containing 200 µm of either 15NO3− or 15NH4+ (15N isotopic excess of 50 atom per cent) during shoot growth. Culture conditions in the glasshouse were natural light, 25°C/20°C and 50%/70% r.h. (day/night, respectively)
Total N (µg) n = 38
15N-NH4+n = 9
15N-NO3−n = 9
15N-NH4+n = 10
15N-NO3−n = 10
The first column represents total N in plants growing on sand and natural soil.
2605 ± 40
7.17 ± 0.58
5.34 ± 4.03
1.26 ± 1.57
0.16 ± 0.09
4666 ± 157
3.41 ± 0.68
1.52 ± 1.01
0.12 ± 0.18
0.06 ± 0.04
3141 ± 381
10.68 ± 1.6
5.04 ± 2.02
0.19 ± 0.28
0.11 ± 0.10
3031 ± 858
43.50 ± 7.33
13.54 ± 3.46
3.30 ± 3.19
1.08 ± 0.90
The role of leaves as reserve organs supporting shoot growth has been inferred in evergreen plants mainly by removing old leaves (Reader, 1978; Jonasson, 1989; Karlsson, 1994; Eckstein et al., 1998), a method which potentially alters various plant functions, including photosynthesis (Vanderklein & Reich, 2000). In this study, we quantified the direct N transfers from each of the two leaf generations to the growing shoots in an evergreen shrub, Rhododendron ferrugineum, by direct labelling of the leaf N pools. Except for yellowing of the small abraded area, leaf function did not seem to be affected by the experimental procedure. To our knowledge, this method has rarely been used in field experiments although it has the advantage of allowing investigation of source–sink relationships at the branch scale by specific labelling of the source organs. This is not the case if the 15N is loaded in the plant through root uptake.
Because the nutrient content of older leaves is often lower than that of younger leaves (Reader, 1978; Dambrine et al., 1991; Escudero et al., 1991; Oliveira et al., 1996), it has been suggested that nutrients could be withdrawn from foliage with low productivity and allocated to new leaves of high productivity (Chabot & Hicks, 1982). Studies on coniferous trees have shown that remobilization of N from 1-yr-old needles (Nambiar & Fife, 1987; Millard & Proe, 1993) provided a high proportion of the N used by the growth of current-year needles. Our results showed that in Rhododendron ferrugineum the N pool was much greater in L1 leaves than in L2 leaves and that N concentrations in L1 leaves slightly but regularly decreased with leaf age, as already shown for other species (Reader, 1978; Dambrine et al., 1991; Escudero et al., 1991; Oliveira et al., 1996). Despite this, it clearly appeared that these leaves did not constitute an important net source of N for the growing tissues since the total N content of L1 leaves was not significantly modified during the growth period. Indeed, the number of L1 leaves on a branch remained constant while the small decrease in foliar N concentration was partly compensated by an increase in leaf biomass. (Table 1 and Fig. 1).
Because most 2-yr-old leaves fell during shoot growth (June –July) in some populations of Rhododendron ferrugineum in the Alps Pornon et al. (1998) suggested, in agreement with other authors (Reader, 1978; Shaver, 1981), that a net translocation of nutrients from senescing old leaves to young expanding leaves may occur. By contrast, in the present study, we found that the number of L2 leaves only slightly decreased during shoot growth and, finally, that the amount of nitrogen in this leaf generation was not significantly altered (Table 1 and Fig. 1). Indeed, the amount of N lost through leaf shedding during the period of study was quite low compared with the other pools. In this condition, L2 leaves cannot constitute an important net source of N for growing tissues, as suggested by Pornon et al. (1998), in alpine populations. Jonasson (1989), studying other evergreen species, also reached the same conclusion that there was no net nutrient translocation from old to new developing leaves. These results suggest that the processes of nutrient remobilization from senescing leaves to new shoots vary between Rhododendron ferrugineum populations and is not a general characteristic of all evergreens. It depends on a close synchronization between leaf senescence and shoot growth.
Our results show that, whatever the mineral form supplied (NO3− or NH4+), root uptake provided current shoots with very little exogenous N. This was demonstrated by quantitative (with a known 15N excess in the medium) and nonquantitative (with natural soil enriched with 15N) experiments carried out in the glasshouse and outdoors on seedlings and 40-yr-old plants. Since neither leaf N nor exogenous N consistently support shoot (Sh0) growth, it appears that the major N source for the growing shoots was likely endogenous nitrogen stored in woody tissues (old stem and root) and that there was total asynchrony between N requirements for growth and N root uptake. Usually, shoot growth and nutrient uptake are at least partly synchronous (Millard & Proe, 1991; Tagliavini et al., 1997; Proe et al., 2000). Thus, the shrub could delay N uptake until the mineralization peak occurs, which could be a major adaptive feature of this species to the low and episodic N availability of the subalpine ecosystems.
Although the amount of total leaf N remained almost unchanged, a large portion of labelled N provided to the leaves moved out of this compartment (60%). This could have resulted from a different biochemical (metabolism) or biophysical (compartmentation and fluxes) fate between labelled N and the endogenous (natural) N. However, in other experiments in which we drew a complete picture of internal transfers of N in smaller individuals, we were able to show that labelled N and endogenous N followed the same turnover during N cycling. This strongly supports the hypothesis that the bulk of the 15NH4+ provided to the leaves was entirely assimilated into organic N and was then homogeneously diluted in the endogenous leaf N pool. Under these assumptions, the loss of 15N by the leaves while no net change in leaf N content occurred suggests the existence of fluxes of N entering and leaving the leaves at an equivalent rate, leading to isotopic dilution. In the hypothesis of steady-state fluxes, this rate, F, is given by the following ratio: F = (Ln2 × QN)/T1/2 where QN is the amount of total nitrogen in L1 or L2 leaves and T1/2, the time required for half of the 15N to move out of L1 or L2 leaves (Devienne et al., 1994). The total N pool in L1 and L2 leaves was approximately 5 mg and 2.5 mg, respectively, and the kinetics of 15N loss from the leaves allowed us to estimate the value of T1/2 as 50 d. Thus, F reached c. 70 µg d−1 and 35 µg d−1 for L1 and L2 leaves, respectively. This indicates that intense internal cycling of N took place in the branch. Such a general cycling of amino-N and of other nutrients has been demonstrated in cereals (Cooper & Clarkson, 1989). A significant portion of labelled N provided to the leaves (about 13% and 19% for L2 and L1 labelling, respectively) was transferred to the current shoots (Sh0). In a recent study using multiple stable isotopes, Proe et al. (2000) reported that the use of net changes in N content leads to an underestimation of the contribution of remobilization in Pinus sylvestris (see also Mead & Preston, 1994 in Pinus contorta). In the present work, in agreement with these authors, we observed that in Rhododendron ferrugineum, although leaves do not constitute a net source of N, remobilization (sensuProe et al., 2000) of leaf N occurs to support shoot (Sh0) growth.
We thank Dr A. Gojon and P. Tillard (INRA-Montpellier) for the 15N analyses. This work was supported by the Parc National des Pyrénées.