Remobilized old-leaf nitrogen predominates for spring growth in two temperate grasses


Author for correspondence: Ursula Bausenwein Tel: +44 (0)1224 318 611 Fax: +44 (0)1224 498 206 Email:


  • • The importance of N derived from internal remobilization and root uptake to growth in spring is reported here for the perennial grasses Festuca rubra and Agrostis capillaris derived from seminatural grasslands in Scotland, UK.
  • • Plants grown in sand culture, received 15N-enriched nutrient solution during the first year of growth and nutrient solution with N at natural abundance during the subsequent spring and summer when destructive harvests were taken.
  • • Labelled N was recovered in new growth of overwintering tillers and new tillers. Remobilized N contributed 70% and 82% for F. rubra and A. capillaris, respectively, to the total N in new above-ground growth in early spring, declining to 34% and 45%, respectively, by mid June. Species showed similar patterns of remobilization on a new growth biomass basis. The root system did not remobilize N to support new above-ground growth. Labelled N was derived from senescing leaves present on overwintering tillers. Net balances of labelled N suggest that N was translocated between tillers; reproductive tillers acted as sinks, vegetative tillers as the source of N.
  • • Initial growth in spring is largely independent of N uptake from the soil, provided that overwintering leaves are present on the plants.


In early spring, plant-available soil N can be low due to reduced N mineralization rates at low soil temperatures (Floate, 1970). Shoot meristems, however, can experience conditions favourable for growth, that is warm temperatures, adequate water supply and unobstructed illumination. Discrepancies between supply of N from mineralization and plant demand for N can be circumvented by the utilization of N stored within perennial plant parts (Chapin et al., 1990). Remobilization of N out of storage structures into the new spring growth is a process well studied for trees, shrubs and agricultural species (Chapin et al., 1980; Limami et al., 1996; Millard, 1996), while information for grassland species is limited. Remobilization has been quantified using 15N labelling for the deciduous grass Molinia caerulea (Thornton & Millard, 1993) and the herbaceous dicot Rumex acetosa (Bausenwein et al., 2001). In the first 2 months of growth, c. 40–50% of the N in new leaves was remobilized out of the roots and basal internodes of M. caerulea or the taproots of R. acetosa. In evergreen grass species, the contribution of N taken up the previous year to spring growth has not been assessed. Also, it is not clear from which tissues N would be remobilized. In an early review on below-ground structures of grasses, Weinmann (1948) suggested that the roots or rhizomes of grasses act as a storage organ for N during winter. This was concluded from seasonal changes in %N, or in total N, of the root systems, without any demonstration that these changes were related to storage and the reutilization for new growth. The same criticism can apply for later studies postulating winter N storage in grass roots (Power, 1986; Nordin & Näsholm, 1997).

In this study we wanted to clarify the importance of N remobilization during spring growth and the location of N storage in Agrostis capillaris and Festuca rubra, two perennial wintergreen grass species frequently coexisting in upland pastures of intermediate fertility (Grime et al., 1988). We grew plants of both species in sand culture for 2 yr. In the first year the plants were supplied with 15NH415NO3, while in the second year, plants received N at natural abundance. A series of destructive harvests were taken to address if N taken up in the previous year contributes to the new growth of the following year, and whether species-specific differences exist, and from which plant compartment N remobilization occurs.

Materials and Methods

Plant material and harvests

Festuca rubra and Agrostis capillaris plants were collected from two seminatural grazed grasslands in Scotland, one being on the Kirkton farm (56°25′ N 4°38′ W) in West Perthshire, the other being on the Cleish Hills (56°7′ N 3°28′ W) in Fife. From each site, 10 accessions were selected and vegetatively propagated. The experiment was set up in April 1996, with individual tillers planted in 12.5 cm diameter pots containing a mixture (by mass) of nine parts sand and one part limestone chips (1–2 mm diameter). The pots were placed in an open sided glasshouse where they were protected from rain but otherwise exposed to natural variations in environmental conditions. From the end of November 1996 to mid March 1997 plants were protected from frost by a polythene tunnel, containing thermostatically controlled electric heaters, constructed within the open-sided glasshouse. Pots were randomly arranged in a block structure consisting of seven blocks. From April 1996 until 31 January 1997 plants were fed two or three times weekly with 100 cm3 aliquots (reduced volume in winter) of a nutrient solution containing 1.5 mol m−315NH415NO3 enriched to 2.5 atom percentage excess, 2.1 mol m−3 CaCl2, 0.75 mol m−3 MgSO4, 0.5 mol m−3 K2SO4, 0.307 mol m−3 NaH2PO4, 26 mmol m−3 Na2HPO4, 50 mmol m−3 H3BO3, 10 mmol m−3 FeC6H5O7, 8.6 mmol m−3 MnSO4, 2 mmol m−3 ZnSO4 and 1 mmol m−3 CuSO4, pH 5.6.

The first destructive harvest was taken on 31 January 1997. In the remaining pots 15N-enriched solution was flushed out with several changes of deionized water followed by two changes of a nutrient solution identical to that supplied in 1996, except that all N was now at natural isotopic abundance. Plants then received nutrient solution with N at natural isotopic abundance throughout 1997. The remaining harvests were taken on 21 March, 17 April, 7 May, 27 May and 19 June 1997.

At each harvest, one replicate of each of the 20 accessions was removed. Plants were washed free from the sand and separated into roots and shoots. From the shoot, 10 tillers which had grown the previous year (old vegetative tillers (OVT); ‘old’ defining tiller establishment in the first year of the experiment) were selected for separation. Once tillers, which were going to flower, were distinguishable from vegetative tillers (old flowering tillers (OFT)), a subsample of five of those tillers was chosen for separation. At later harvests, OVTs were consisting of either very small or dying tillers, or vigorous ones. The latter ones were sampled, though it was not always possible to obtain 10 tillers. Tillers were separated into dead leaves, old living leaves (present at the switch from 15N to 14N on 31 January 1997, referred to as ‘old leaves’), new living leaves (growth had occurred after 31 January 1997, ‘new leaves’) and reproductive parts (flower stems and flowers produced in 1997). New tillers (appearance after the change to N at natural abundance) were also collected without further separation as they were solely consisting of new leaves. The above-ground data are expressed for individual tillers. As it was not possible to separate the roots attached to a particular tiller, the root data are for whole plants. Samples were frozen at −20°C, then freeze-dried, weighed and ball-milled.

The separation between old and new leaves was achieved by monitoring the leaf development on a reference plant. On 10 and 11 February 10 tillers were selected and the youngest leaf marked with nontoxic paint on one replicate pot of each genotype. Before every harvest, the number of new leaves was counted on those tillers, and this genotype-specific number was used for the separation of harvested tillers into new and old leaves.

Total N and 15N enrichment

The total N and 15N concentrations were determined using a TracerMAT continuous flow mass spectrometer (Finnigan MAT, Hemel Hempstead, UK). The 15N enrichment was used to calculate the uptake of N from the 15N labelled nutrient solution in 1996 (Millard & Neilsen, 1989). Labelled N in plant material harvested in the second season therefore gave a measure of the N in the plant taken up during the first year, and its recovery in new shoot material provided a measure of remobilization of N. It was assumed that the difference between the total and labelled N content of a tissue was derived from current root uptake in the second season.

Statistical analysis

To test for effect of harvest date, a one-way ANOVA was used on log-transformed data, for determination of start of root growth, t-tests were performed between consecutive harvests (Genstat 5 Committee 1993). The significance level was P < 0.05.


Festuca rubra

Growth on old tillers of F. rubra consisted of the production of new leaves and reproductive parts on OFTs, and the production of new leaves only on OVTs (Figs 1a, 1c). Growth of new leaves on OFTs was rapid until 7 May, by which time development of inflorescences was occurring. Both N from remobilization and current uptake was used for the growth of new leaves, which had their highest N content on 7 May and which declined thereafter (Fig. 1d). Growth of flower stems and flowers on the OFTs occurred between 17 April and 19 June (Fig. 1a), while their total N and labelled N content were greatest on 7 May, and declined in the following harvests (Fig. 1b). In the new leaves of OVTs no d. wt increases were observed after 17 April, and total as well as labelled N declined thereafter (Figs 1c, 1d). The old leaves of old tillers decreased in d. wt and N content due to senescence and death (Figs 1e, 1f). On the last harvest date, the biomass of the new tillers was 0.02 g tiller−1. They were also supported by labelled N (Table 1), and their isotopic composition was identical to that of the new growth on OFTs (data not shown).

Figure 1.

Biomass, total N and labelled N content of individual tillers and the whole root system of Festuca rubra plants against harvest date. Old vegetative tillers (OVTs), circles; old flowering tillers (OFTs), squares; total N, closed symbols; labelled N, open symbols. Values are the mean and SE for 20 replicates.

Table 1.  Change in the amount of labelled N in above-ground tissues of individual tillers between 31 January and the harvest date of maximum labelled N in reproductive parts
SpeciesDateTissue15N tiller−1 (mg)
  1. OFT, old flowering tillers; OVT, old vegetative tillers.

Festuca rubra7 MayNew leaves on OVT+0.26 (± 0.02)
  New growth on OFT+2.23 (± 0.31)
  New tiller+0.12 (± 0.01)
  Old leaves on old tiller−0.80 (± 0.06)
Agrostis capillaris19 JuneNew leaves on OVT+0.33 (± 0.03)
  New growth on OFT+0.62 (± 0.10)
  New tiller+0.29 (± 0.09)
  Old leaves on old tiller−0.30 (± 0.02)

On 31 January F. rubra had 74 (± 5) tillers, with an estimated labelled N content of 72 mg. At the time of maximum N content in the reproductive tillers (7 May), total tiller number had increased to 134 (± 9) tillers. By then, 26 (± 3) flowering tillers were present, which were estimated to contain 56 mg labelled N.

The root system of F. rubra began to grow between 31 January and 21 March. The most rapid root biomass increases were between 7 May and 19 June (Fig. 1g), after the period of highest rate of above-ground growth. The amount of labelled N in the root system was affected by harvest date (P = 0.03) (Fig. 1h): there were no significant differences between 31 January and 27 May, but the amount measured on 19 June was statistically higher than on the three preceding harvest dates. N derived from current uptake significantly increased in the roots between 31 January and 19 June, resulting in an overall increase in the total N of that compartment (Fig. 1h).

Agrostis capillaris

The above-ground growth and development in A. capillaris peaked later than in F. rubra. New leaves on OFTs as well as on OVTs increased their d. wt until the end of the observation period (Fig. 2c), with total N and labelled N increasing in the new leaves of OFTs until 19 June, and in those of OVTs only until 27 May (Fig. 2d). In A. capillaris growth of and N allocation to flower stems and flowers became apparent by 19 June (Fig. 2a, 2b), with the peak of reproductive growth occurring after the final harvest (observation on the reference plants, which were not harvested). D. wt, total and labelled N of old leaves declined due to senescence (Fig. 2e, 2f). New tillers had a biomass of 0.002 g tiller−1 on 27 May, and acted as a sink for labelled N (Table 1). Due to the high tiller numbers per plant, tillers were not counted.

Figure 2.

Biomass, total N and labelled N content of individual tillers and the whole root system of Agrostis capillaris plants against harvest date. Old vegetative tillers (OVTs), circles; old flowering tillers (OFTs), squares; total N, closed symbols; labelled N, open symbols. Values are the mean and SE for 20 replicates.

There were statistically significant increases in the weight of the root system between 21 March and 27 May, while no significant changes in root biomass occurred between 31 January and 21 March or between 27 May and 19 June (Fig. 2g). By contrast to F. rubra, root growth in A. capillaris had a later start in spring and peaked before reproductive development. Total N in the roots increased until 27 May, with no further rise in N content thereafter (Fig. 2h). Harvest date had significant effects (P < 0.001) on the labelled N content of the roots, which was lowest on 31 January and highest on 27 May (Fig. 2h).

In F. rubra and A. capillaris, labelled N was recovered in new leaves of both old and new tillers, and lost from the old leaves. Table 1 summarizes these changes in the amount of labelled N in above-ground parts at the harvest date with highest allocation to the flowering tillers. At this time, labelled N in the new leaf material on OFTs of both species was greater than the losses that had occurred from the old leaves.

Contribution of labelled N to total N in new growth

Labelled N was initially the main source of N for the new growth on OFTs, accounting for 70% of the total N for F. rubra and 82% for A. capillaris by 21 March. Thereafter, uptake of N at natural abundance diluted the labelled N (Fig. 3a). Over the entire time course of the experiment, the contribution of labelled N to the total N was significantly higher (P < 0.001) for A. capillaris than F. rubra. If, however, this contribution was considered in relation to the biomass produced, the differences between species became negligible (Fig. 3b).

Figure 3.

Contribution of labelled N to the total N in new above-ground growth of old flowering tillers (OFTs) for Festuca rubra (open squares) and Agrostis capillaris (closed squares) against harvest date (a) or against the d. wt of new growth (new leaves and reproductive parts) (b). Values are the mean and SE for 20 replicates, grouped for harvest dates in (b).


Contribution of stored N to new growth

While seasonal storage of N has been reported for many herbaceous perennials on the basis of fluctuating N contents in storage structures such as (tap) roots or tubers (Rosnitschek-Schimmel, 1985; Cyr & Bewley, 1990; Fischer et al., 1995), the contribution of this store to the new growth has rarely been quantified.

On the basis of N budgets it was shown that the initial growth of the alpine herb Bistorta bistortoides was supported solely by remobilized N, and that uptake of N only occurred after growth had started (Jaeger & Monson, 1992). By contrast, new leaves of R. acetosa, an evergreen herb, and of the deciduous grass M. caerulea contained c. 40–50% of N derived from current uptake as early as mid March when grown in sand culture, indicating that N remobilization and uptake can occur simultaneously even in early spring (Bausenwein et al., 2001; Thornton & Bausenwein, 2000). The present study showed that F. rubra and A. capillaris behaved similarly to R. acetosa by utilizing both internal and external N by mid March. We observed significant differences between A. capillaris and F. rubra in the percentage of remobilized N in the new growth over the entire observation period, with A. capillaris having consistently higher values. As the two species differed in the timing of their growth, the contribution of remobilization in relation to d. wt gave a more meaningful basis for comparison. Under these circumstances, A. capillaris and F. rubra have similar contributions of remobilization. The temporal differentiation in shoot development between A. capillaris and F. rubra has been recognized before (Grime et al., 1988), and, as our data show, can be expanded to root growth.

Origin of remobilized N

Net balances of labelled N show that in F. rubra and A. capillaris the turnover of old leaves and the concomitant release of N out of the senescing tissues were responsible for the recovery of labelled N in new above-ground plant material. The root system was not a winter storage compartment of N contributing to above-ground growth in spring. Instead, there was a slight increase in the content of labelled N in the roots of both species towards the end of the experiment. We know from both F. rubra and L. perenne seedlings (Schulte auf’m Erley et al., 2000) and our own data from more short-term labelling experiments with F. rubra and A. capillaris (unpublished) that a proportion of the N remobilized from senescing leaves can be incorporated into the roots.

The sequential use of leaf nitrogen through translocation of N from older leaves to young expanding leaves is well-known for a variety of species, for example the sedge Eriophorum vaginatum (Jonasson & Chapin, 1985), seedlings of L. perenne and F. rubra (Schulte auf’m Erley et al., 2000) or vegetative Triticum aestivum plants (Vouillot & Devienne-Barret, 1999). For overwintering plants, data are more limited and – to our knowledge – not available for temperate grasses. It has been reported that Eriophorum vaginatum, a sedge occurring in tundra ecosystems, stores N in overwintering leaves (Chapin et al., 1980), however, these leaves are also long-lived and continue elongation in the following spring. Also, it is known that evergreen trees such as Eucalyptus globulus or Pinus radiata use N from their foliage for leaf growth in spring (Nambiar & Fife, 1991; Wendler et al., 1995). In Picea sitchensis, 15N labelling has shown that overwintering needles provided nearly all, and the root system none, of the N derived from remobilization in young needles (Millard & Proe, 1993).

Many experiments determining N remobilization in grasses to new leaf material and its site of origin involve defoliation; in several cases it had been shown that both roots and remaining stubble remobilize N (Ourry et al., 1988; Bakken et al., 1998). In wheatgrass, Agropyron dasystachyum, new growth was supported equally by N remobilized out of old leaves and of roots (Li et al., 1992). In that experiment, all leaves had been removed from the plant at the start of the labelling period with 15N, with the chase period starting once plants had reached a 2-leaf stage, when N accumulation in the leaves might not have been at its maximum. These experimental findings suggest that N remobilization from the root to the new shoot occurs when the N pool above-ground is insufficient.

The fact that N derived from old leaves is important for spring growth can explain certain field observations: autumn cuts adversely affected spring growth in various grass species including F. rubra and A. capillaris (Grant & Hunter, 1968). Later cuts, leaving the plants insufficient time for regrowth before winter were more detrimental than earlier cuts, and might be due to the lack of a substantial pool of leaf N that could contribute to early growth.

Winter N storage in grasses had been assigned to the root system on the basis of increases of %N and of N content in the root system in winter. There are two problems with this. First, %N (in relation to d. wt) fluctuations of roots are not necessarily an indication of storage as other compositional changes of the root will influence its d. wt. Second, increases in the N content of the root system in winter might be related to root growth, which according to Weinmann (1948), continued, when, above-ground, growth, had ceased; Stuckey (1941) reported significant production root systems starting in October. Garwood (1967) measured the appearance of new roots of L. perenne, Dactylis glomerata and Phleum pratense, which were produced in increasing numbers through late winter to early spring. Allocation of N to the roots before and during winter is more likely to be associated with tissue production than the formation of storage pools.

A factor limiting the potential of grass roots as winter storage organs might be their longevity. Long-term seasonal storage requires structures with a sufficient lifetime to allow for filling in autumn and withdrawal in spring. In a recent investigation, using minirhizotrons for direct observations of specific root cohorts of A. capillaris accessions identical to those used in our experiment, the longevity of roots was short with a medium life-span of 38 d (J. A. Baddeley, unpublished). In L. perenne only 84% or 38% of the roots survived after 21 d, depending on the climatic conditions of the study site (Watson et al., 2000). Even though these observations were on root cohorts initiated in summer, and it is known for both grasses and trees that summer roots have a shorter life-span than roots initiated towards the end of the year (Garwood, 1967; Eissenstat & Yanai, 1997), they give an indication that parts of the root system of temperate grasses might be too short-lived for winter storage. Species-specific differences exist for root longevity, and these had been attributed to variations in tissue-density and root diameter, with low tissue-density (as observed in species with high relative growth rate from fertile environments) and small root diameter resulting in shorter life-spans (Ryser, 1996; Eissenstat & Yanai, 1997). The deciduous grass M. caerulea, which remobilizes N for spring growth predominantly out of its roots (Thornton & Millard, 1993), has a root system dominated by main axes with big diameters and high physical strength (Jefferies, 1915). By contrast, both F. rubra and A. capillaris produced root systems with few main axes and many fine laterals, and so were probably not equipped with sufficiently long-lived roots suitable as storage organs.

Integration of tillers

Until 17 April, the labelled N decline in old leaves nearly matched the labelled N increase in new leaves, whereas in later harvests a net import of labelled N into the reproductive tiller occurred. Not all the N of old leaves would have been available for retranslocation as some N was retained in the dead leaves. We were not able to sample completely the dead leaves because affiliation to an individual tiller was often not clear. In his review on nutrient resorption from senescing leaves, Aerts (1996) gives 59% of the initial leaf N as average resorption efficiency for a wide range of graminoids. Even though resorption efficiencies can be as high as 75% (Festuca pratensis; Hauck et al., 1997) or 86% (Triticum aestivum; Williams, 1955), the amount of labelled N available from old leaves of OFTs would have been insufficient to account for the labelled N recovered in the new leaves. Moreover, growth of new tillers was also supported by remobilized N. In the case of F. rubra, approx. 2–3 new tillers were associated with a OFT. In the absence of any net losses of labelled N from the root we conclude that N is translocated between tillers, with those becoming reproductive acting as the sink and those remaining vegetative as the source of (labelled) N. This is more apparent for F. rubra as the measurements did not adequately cover the reproductive development of A. capillaris. The movement of N out of vegetative tillers can be concluded from the decrease in their N content while their d. wt remained constant. Our sampling of OVTs at later harvests was biased towards vigorous tillers, neglecting those that were very small or (nearly) dead. We might therefore have underestimated the source potential of the average OVT. In clonal vegetative plants, movements of N are predominantly from older to ontogenetically younger tillers, though in Schizachyrium scoparium (bunchgrass) the reverse movement of 15N had also been detected (Welker et al., 1991). Flowering might alter this pattern. In Hordeum vulgare, flower development was correlated with the death of some secondary tillers despite the fact that they had established their own root system, indicating a hierarchical integration of tillers in favour of reproductive ones (Anderson-Taylor & Marshall, 1983).

To conclude, undefoliated A. capillaris and F. rubra plants are largely independent of N uptake from the soil in early spring, because N requirements for above-ground growth are met by N remobilization from the old leaves. In late spring, N from the soil is an important N source; however, flower development can result in further internal N translocations, with vegetative tillers providing N to the flowering tillers.


We thank B. Thornton and R. Wendler for their advice and comments, M. Tyler, A. Sim and C. MSmithsonian for their skilful assistance, and the Scottish Executive Rural Affairs Department for funding this work.