Contribution of previous year ’ s leaf N and soil N uptake to current year ’ s leaf growth in sessile oak

The origin of N which contributes to the synthesis of N reserves of in situ forest trees in autumn and to the growth of new organs the following spring is currently poorly documented. To characterize the metabolism of various possible N sources (plant N and soil N), six distinct 20-yearold sessile oaks were 15N labelled by spraying 15NH15 4 NO3: (i) on leaves in May, to label the N pool remobilized in the autumn for synthesis of reserves, (ii) on soil in the autumn, to label the N pool taken up from soil and (iii) on soil at the beginning of the following spring, to label the N pool taken up from soil in the spring. The partitioning of 15N in leaves, twigs, phloem, xylem, fine roots, rhizospheric soil and microbial biomass was followed during two growing seasons. Results showed a significant incorporation of 15N into the soil– tree system; more than 30 % of the administered 15N was recovered. Analysis of the partitioning clearly revealed that in autumn, roots’ N reserves were formed from foliage 15N (73 %) and to a lesser extent from soil 15N (27 %). The following spring, 15N used for the synthesis of new leaves came first from 15N stored during the previous autumn, mainly from 15N reserves formed from foliage (95 %). Thereafter, when leaves were fully expanded, 15N uptake from the soil during the previous autumn and before budburst contributed to the formation of new leaves (60 %).


Introduction
Tree carbon metabolism associated with photosynthesis, C allocation and remobilization of C storage is well documented (Barbaroux et al., 2003;Dickson, 1989), but tree nitrogen metabolism is less known.Nevertheless, seasonal N cycling is a determinant of plant fitness in perennials, particularly long-lived perennials such as forest trees (Cooke and Weih, 2005).In early spring, trees' nitrogen demand for growth can be satisfied either by uptake of external sources such as ammonium, nitrate and organic N available from the soil (Gessler et al., 1998a) or by remobilization of internal stores (Bazot et al., 2013;Coleman and Chen, 1993;Cooke and Weih, 2005;El Zein et al., 2011b;Gilson et al., 2014;Millard, 1996;Taylor, 1967).In many species, N remobilization for growth in spring occurs before utilization of N taken up by roots, typically during the 20-30 days before the roots actively take up N.These species include deciduous species, such as Quercus petraea (El Zein et al., 2011a), Malus domestica (Guak et al., 2003;Neilsen et al., 2001), Populus trichocharpa (Millard et al., 2006), Prunus avium (Grassi et al., 2003), Pyrus communis (Tagliavini et al., 1997) and Sorbus aucuparia (Millard et al., 2001); marcescent/evergreen species, such as Nothofagus fusca (Stephens et al., 2001) and coniferous evergreens, such as Picea sitchensis (Millard and Proe, 1993).In a few species (e.g. S. aucuparia), remobilization has completely finished before any root uptake of N occurs, even if trees are supplied with an adequate supply of mineral N in the soil.In contrast, other species have been shown to begin taking up soil N through their roots concomitantly with N remobilization.These include deciduous Juglans nigra × regia (Frak et al., 2002), Pyrus communis (Tagliavini et al., 1997), Betula pendula and evergreen Pinus sylvestris (Millard et al., 2001).All of these studies were conducted on young trees and/or under controlled conditions.Few studies have applied 15 N-labelled mineral fertilizer to larger, undisturbed trees growing in the field (El Zein et al., 2011a), and even those only evaluated the contribution of Published by Copernicus Publications on behalf of the European Geosciences Union.
spring N uptake to leaf and twig growth, while the contribution of stored N was indirectly estimated.However, in autumn, the process of N storage (N translocation from leaves to sink compartments), which starts concomitantly with leaf yellowing (Bazot et al., 2013), is associated with a stimulation of soil nitrogen uptake (Gessler et al., 1998b;Jordan et al., 2012;Kim et al., 2009).In the present study we proposed to investigate the contribution of N storage and that of N taken up from soil during autumn and spring to the development of new leaves of 20-year-old sessile oaks in the field, after budburst during the following spring.Does soil N or foliar N contribute most to the storage of N compounds in autumn?Does soil N or stored N contribute most to the synthesis of new leaves in spring?Soil 15 N labelling is a suitable tool to quantify autumn and spring uptake of N by roots.Labelling of foliage allows quantification of N remobilized from leaves to reserve compartments.During three distinct labelling campaigns, 3 × 2 distinct 20-year-old sessile oaks received 15 NH 15  4 NO 3 applied to their foliage (May) or on adjacent soil (September and March of the following year). 15N partitioning in all tree-soil compartments, i.e. leaves, twigs, trunk, roots, rhizospheric soil and microbial biomass, was analysed regularly.The contribution of assimilated 15 N to storage and remobilization was investigated.

Site description
The experiment was conducted in an area of 20-year-old naturally regenerated oak in the Barbeau forest (48 • 29 N, 02 • 47 E), 60 km southeast of Paris, France, at an elevation of 90 m on a gleyic luvisol.The average air temperature is 10.5 • C and the annual rainfall in this temperate location is 690 mm.Six 20-year-old sessile oaks (Quercus petraea L.) were selected, their height ranged between 8 to 10 m and their average diameter at breast height was 10 cm.In order to limit possible interference of root cutting with nitrogen allocation, at least 5 months before labelling a 0.5-0.6 m deep trench was dug around each tree, then the trench was lined with a polyethylene film and backfilled.All roots and root exudates inside this perimeter therefore originated from the isolated tree, and were contained in this trench volume.The area delimited by the trench was about 5 m 2 .The distance between each tree was at least 20 m.

15 N pulse labelling
Three labelling campaigns were carried out: the first (L 1 ) on the foliage at the end of May (27 May 2009), the second (L 2 ) on the soil at the beginning of September (9 September 2009) and the third (L 3 ) on the soil the following March (20 March 2010).All labelling campaigns were conducted on sunny days.Two oaks were labelled during each campaign: trees 1 and 2 during L 1 , trees 3 and 4 during L 2 and trees 5 and 6 during L 3 .Of buds showing leaf unfolding (Vitasse et al., 2009), 50 % occurred in those sessile oaks on 20 April 2010; this date was defined as budburst.The L 1 campaign consisted of homogenous spraying on all foliage of 5 g 15 NH 15 4 NO 3 (98 at.%), i.e. 1.82 g of 15 N, dissolved in 2.5 L distilled water.Prior to L 1 , the soil of the surrounding trenches was protected with a plastic tarpaulin covering the whole area of the trenched plot to avoid soil pollution with 15 N.The tarpaulin was sealed to the trunk at 50 cm height with Terostat-VII (Teroson, Henkel, Germany).It remained on the soil for 2 weeks after labelling.Before removing the plastic tarpaulin, crowns were sprayed with distilled water in order to avoid any soil contamination after the removing of the tarpaulin.
This first campaign aimed to label foliage and subsequently the N reserves developed from the remobilization of leaf N the following autumn.The L 2 campaign consisted of homogenous spraying of 5g 15 NH 15 4 NO 3 (98 at.%), i.e. 1.82 g of 15 N, dissolved in 20 L distilled water on the soil of the trench plot of two other selected oak trees (3 and 4).With this procedure, N reserves developed from autumnal soil N uptake were expected to be labelled.The third and last labelling campaign, L 3 , consisted of homogenous spraying of 5 g 15 NH 15  4 NO 3 (98 at.%), i.e. 1.82 g of 15 N, dissolved in 20 L distilled water on the soil of the trench plot of trees 5 and 6, thus labelling their spring N uptake.

Sampling and analytical methods
Leaves, twigs, trunk phloem and xylem and soil monoliths (15 cm depth, very few fine roots were present below 15 cm depth) of each labelled trees (1,2,3,4,5,6) were sampled after labelling until the end of 2010 (Table 1).At each sampling date 20 leaves and 20 twigs were collected randomly throughout the crown.Sampling was always performed between 10:00 and 12:00 UTC.The leaves were rinsed with distilled water to remove any excess 15 N.At each sampling date, two small disks of bark (14 mm diameter, 10 mm depth) were collected at 1.3 m height using a corer.Thereafter, phloem and xylem tissues were separated by hand with a cutter blade.The leaf mass per area (LMA) was measured at each sampling date.Fine roots were hand picked from the soil monoliths and washed with a 0.5 M CaCl 2 isotonic solution.Soil adhering to roots was removed with a brush and sieved at 2 mm.All plant tissues and soil samples were brought to the laboratory in a cooler.Plant tissues were lyophilized and ground to a fine powder with a ball mill before analyses.At each sampling date, one aliquot of each plant powder (1 mg) was transferred into tin capsules (Elemental Microanalysis, UK, 6 × 4 mm, ref.D1006, BN/139877).On some dates (day after labelling (DAL) 1, 126, 337, 460 for leaves and twigs of L 1 ; DAL 126, 337, 460 for roots of L 1 ; DAL 227 and 350 for leaves and twigs of L 2 ; DAL 49 and 350 for roots of L 2 and DAL 40 and 166 for leaves, twigs and roots of L 3 ), four aliquots of powder were transferred into tin caps in order to test the repeatability of the analysis.The total N concentration of plant and soil samples was analysed by dry combustion using an N autoanalyser (Flash EA 1112 series, Thermofinnigan). 15N abundance was quantified in the same plant and soil fine powder aliquots with a mass spectrometer (PDZ Europa, University of Davis, Isotopes Facility, California).
Microbial N contents of fresh soil samples were determined using the chloroform fumigation-extraction method (Vance et al., 1987).Two fresh soil subsamples of 10 g were prepared.One subsample was fumigated for 24 h with chloroform vapour, while the other was not fumigated.Nitrogen extraction was performed using 50 mL of 0.5 M K 2 SO 4 for 30 min under vigorous shaking.The extracts (fumigated and not fumigated) were filtered, then analysed for N content using an N analyser (TNM-1, Shimadzu, Champs-sur-Marne, France).The microbial 15 N abundance was estimated using the same procedure, except that the extraction solution was 0.03 M of K 2 SO 4 in order to avoid any alteration of the mass spectrometer with the K 2 SO 4 salt during 15 N analysis.

Calculations
All 15 N enrichments were corrected for the background natural abundance of this isotope, using control values determined in plants and soils just before labelling.The seasonal variations of the natural 15 N abundance of each compartments were also followed throughout the season; those variations were very weak, consequently we chose to use the 15 N natural abundance of the labelled trees just before labelling.The total weight of each analysed compartment (i.e.leaves, twigs, trunk phloem and xylem and fine roots) was extrapolated from those of six equivalent trees (same size and same diameter) grown on the same site and under the same conditions.Those trees were felled as follows: two in October of the first labelling year (2009), two in the following May (2010) and two the following February (2011).Total leaf biomass was corrected according to the LMA.All data were expressed as proportion of recovered 15 nitrogen (PRN) in a specific compartment using the following calculation Eq. ( 1): where Q 15 N was the quantity of 15 N recovered from a compartment on a specific date, and Max Q 15 N was the maximum quantity of 15 N recovered from all the sampled compartments during the experiment.The % contribution of each 15 N source (L 1 : leaves, L 2 : autumn soil N, L 3 : spring soil N) to the 15 N recovered in the roots in autumn or in the leaves of the second year was determined according to the following calculation Eq. ( 2): (2)

Results
For each labelling, the two analysed trees displayed similar patterns of total recovered 15 N in each compartment (data not shown) and 15 N partitioning throughout the experiment.Moreover, the test of repeatability of the analysis revealed very little variability in the 15 N partitioning at a specific date or in a specific compartment (Figs. 1, 2, 3).Consequently, results were expressed as the mean of both trees (L 1 : 1 + 2; L 2 : 3 + 4, L 3 : 5 + 6).
3.1 15 N partitioning within the plant-soil system during the first leafy season  1).
The PRN was at maximum in leaves (96 %, Fig. 1a) 1 day after L 1 , then decreased continuously during the four following months (from 27 May to 30 September 2009, i.e. until the 126th day after labelling) with a mean decrease of 80 % between these two dates (Fig. 1a).The same pattern was observed in twigs, where the PRN decreased from 3 % on day 1 to 0.4 % on day 126 (Fig. 1a).
In the trunk phloem tissue and the fine roots, the PRN stayed relatively stable or slightly increased until day 57 (24 July 2009).They then increased until day 126 (30 September 2009), when they reached 4.75 % in the phloem and 16 % in the roots (Fig. 1b, c).The PRN from the rhizospheric soil and microbial biomass was less than 1 % (Fig. 1d).During winter (2 December 2009; day 189) the PRN reached 18.5 % in fine roots (Fig. 1c).

After the first soil labelling (L 2 , 9 September 2009)
The total balance for the administered 15 N demonstrated maximum recoveries within the plant-soil systems 3 days after L 2 of 70 %.By the end of October (49 days after labelling), recoveries from the sampled compartments decreased to 22 % of the administered 15 N (Table 1).Three days after labelling, 3 % of the recovered 15 N was present in the fine roots (Fig. 2c).Nine days after labelling (18 September 2009), the PRN showed that the majority of the 15 N was recovered from the soil, with 61 % of the 15 N recovered from the rhizospheric soil and 32.5 % from the microbial biomass (Fig. 2d).During the following 40 days (until 28 October 2009), the PRN from the soil decreased to 8.5 % in the rhizospheric soil and 9.5 % in the microbial biomass (Fig. 2d).On the same date, 6 % of the 15 N was recovered from the fine roots (Fig. 2c).Less than 1 % of the 15 N was recovered from the phloem, xylem and twigs (Fig. 2a, b).In December (day 84) the PRN from the soil was similar to that of the previous date and 4 % of the 15 N was recovered from the fine roots (Fig. 2c, d).
Eight days after budburst (28 April, i.e. 337 days after L 1 ), 25 % of the recovered 15 N was observed in new leaves.By 19 May this had decreased to 17 % (Fig. 1a).On 28 April twigs contained 3.5 % of the recovered 15 N (Fig. 1a), phloem 4 % (Fig. 1b) and fine roots 10 % (Fig. 1c).From then until September (i.e.460 days after labelling), the PRN from leaves remained relatively stable (22 %), whereas it largely decreased in fine roots (0.35 %) (Fig. 1a, b, c).Less than 0.2 % of the total 15 N recovered over the season was from the rhizospheric soil and microbial biomass (Fig. 1d).Just before budburst following the second labelling (L 2 , 8 April 2010, 208 days after labelling) 19 % of the administered 15 N were recovered from all the analysed compartments (Table 1).Most of it was from the rhizospheric soil (14.5 %, Fig. 2d).The microbial biomass contained 9.5 % of the recovered 15 N and the fine roots 2 % (Fig. 2d, c).The rest of the 15 N (less than 5 %) was distributed between the twigs, trunk phloem and xylem (Fig. 2a, b).The same pattern was observed 8 days after budburst (227 days after labelling): most of 15 N was recovered from soil microbial biomass and rhizospheric soil (12 %, Fig. 2d), 2.25 % was recovered from fine roots, 3.5 % of 15 N was recovered from phloem and xylem and only 0.5 % was recovered from new leaves (Fig. 2a).
From 8 April (208 days after labelling) to 19 May (247 days after labelling and 30 days after budburst), the PRN decreased in soil microbial biomass and rhizospheric soil (7 %), but increased in fine roots (9.5 %) (Fig. 2d, c).A noticeable increase of the PRN from leaves was also observed at this date (4.5 %, Fig. 2a).Thereafter, the PRN from soil microbial biomass and fine roots decreased slightly from May 19 to June 28 (i.e.247 to 287 days after labelling), then remained stable until the end of August (Fig. 2d, c).The PRN from leaves increased to 7 % in June (Fig. 2a).DAL: days after labelling.The two lines for each category (continuous and dotted) correspond to tree 5 and tree 6. Vertical bars indicate standard errors.
For trees with soils that were labelled in spring (L 3 , 20 March 2010), the maximum recovery of the administered 15 N occurred 40 days later: 51.5 % from the sampled compartments.Recovery decreased thereafter and stabilized at 19.5 % until autumn 2010 (Table 1).
Twenty days after labelling and before budburst, the soil microbial biomass contained 44.5 % of the recovered 15 N and the rhizospheric soil 39 % (Fig. 3d).The remaining 15 N was mainly located in the roots (2 % of recovered 15 N, Fig. 3c).Eight days after budburst, the PRN was quite similar: 61 % in microbial biomass and 32 % in rhizospheric soil (Fig. 3d). 15N recovered from fine roots followed a pattern similar to that observed on the previous sampling occasion (Fig. 3c).However, between 8 and 30 days after budburst (from 28 April to 19 May 2010, i.e. from 40 to 61 days after labelling), the PRN in microbial biomass and in rhizospheric soil decreased sharply to 3.2 % (Fig. 3d).On that date, 17 % of the 15 N was recovered from the fine roots (Fig. 3c) and 21.2 % from the leaves (Fig. 3a).The PRN from leaves remained stable until the beginning of June (74 days after labelling) (Fig. 3a).From that date until September the PRN from leaves and fine roots declined slightly (Fig. 3a,  c).The PRN from microbial biomass decreased continuously throughout the season and reached 2.5 % in September (day 166 after labelling) (Fig. 3d).

Efficiency of labelling
Isotope labelling experiments are technically challenging and as a consequence are very scarce on trees growing in natural conditions.In this paper, field labelling campaigns were conducted on 20-year-old naturally regenerated oaks.For each campaign (only) two trees were labelled.Nevertheless the similarity of the results between them suggests that the observed 15 N partitioning in soil and tree is a representative view of the functioning of such systems During the first labelling procedure (L 1 ), a significant fraction of the added 15 NH 15 4 NO 3 was incorporated into the leaves of the sessile oaks.A significant proportion of the 15 N was allocated to the leaves: more than 90 % of the 15 N was recovered from this compartment.The total balance for the administered 15 N demonstrated maximum recoveries within the plant-soil systems of 32 %, 1 day after leaf labelling.The remaining 15 N was probably lost by leaf leaching.However, soil protection with plastic tarpaulins avoided all contamination of soil and roots as indicated by the 15 N recovered in the belowground compartments (Fig. 1d).Thereafter, the recovery of administered 15 N from the sampled compartments decreased to 14.5 %, probably due to allocation of 15 N to nonharvested compartments, such as old branches, coarse roots or the inner part of the trunk.Indeed, data currently available on woody plants show that nitrogen is re-translocated from leaves to storage sites such as old branches, trunk or coarse roots (Valenzuela Nunez et al., 2011;Bazot et al., 2013).The soil 15 NH 15  4 NO 3 labelling (L 2 ) conducted in September was also effective.Indeed, the total balance for the 15 N applied to the soil demonstrated maximum recoveries within the plantsoil systems of 70 %, 3 days after soil labelling.The rest of the 15 N was most probably lost by soil leaching (30 % of the 15 N provide).Thereafter the recovery of administered 15 N from the harvested compartments decreased to 22 %.As with the leaf-labelling experiment (L 1 ), this decrease was presumably due to allocation of 15 N to non-harvested compartments.Finally, the soil 15 NH 15 4 NO 3 labelling carried out the following March (L 3 ) was also effective, with maximum recoveries within the plant-soil systems of 51.5 %, 40 days after soil 15 N labelling.This recovery decreased to a mean of 19 % during the rest of the season.

N dynamics in soil-tree systems during the first leafy season
Following the first labelling procedure, the 15 N was quickly incorporated into leaves; more than 90 % of the 15 N applied was accounted for in leaves 1 day after labelling.Thereafter this portion decreased continuously throughout the season.The unaccounted-for fraction of the 15 N had presumably been transferred to other compartments, including those which were not sampled, i.e. branches and coarse roots.This important foliar N remobilization was observed to continue in leaf-labelled trees until yellowing, i.e. the end of September.Data currently available on woody plants show that nitrogen is mainly retranslocated from leaves to storage sites during the autumn (Coleman and Chen, 1993;Cooke and Weih, 2005;Dong et al., 2002;Taylor, 1967), due to the predominant role of leaf senescence in the constitution of N stores.Leaf senescence leads to the breakdown of leaf proteins, the transfer of their nitrogen to the perennial plant parts and the formation of N storage compounds (vegetative storage proteins and amino acids) (Dong et al., 2000;Tromp, 1983).In this study, a noticeable increase in the percentage of recovered 15 N in fine roots was observed on 30 September (16 %).This compartment could be defined as a storage compartment in young sessile oaks.Such an observation has been already reported for oaks of the same pole stand (Gilson et al., 2014), and similar findings were reported for field-grown adult peach trees by Tagliavini et al. (1997), being typical of other young deciduous trees (Millard and Proe, 1991;Salaün et al., 2005;Tromp and Ovaa, 1979;Wendler and Millard, 1996).On this date (end of September), branches and coarse roots could also have contributed significantly to N storage, as previously described (Bazot et al., 2013).
At the same time, root uptake can also contribute directly to storage, as proposed by Millard (1996).Indeed, 49 days after labelled 15 N had been applied to surrounding soil (L 2 ) in September, 5.75 % was recovered from the trees' fine roots.It can be underlined that at the end of September, foliage 15 N made up 73 % of the 15 N recovered in roots, whereas soil 15 N uptake contributed to 27 % of the 15 N recovered in roots (Eq.2, Fig. 4).The soil N uptake in this period was mainly recovered in the root system; there was little labelled N in the rest of the trees.This is consistent with the results of Tagliavini et al. (1997) and Jordan et al. (2012), who found a significant fraction of labelled N in fine root samples of peach trees with 15 N applied on soil before fruit harvest in September.
Concomitantly with root N uptake for storage, notably in fine roots, a strong immobilization of N in microbial biomass was observed.Indeed, on October 7 (i.e.28 days after labelling), when yellowing was well advanced, 12.5 % of the applied 15 N was recovered in microbial biomass and 21.5 % in rhizospheric soil: there was a competition for soil N between microbial N immobilization and reserve synthesis by root N uptake at that time.This is consistent with the idea that soil microorganisms are strong short-term competitors for soil N due to their high surface area to volume ratio, wide spatial distribution in the soil and rapid growth rates compared with plants roots (Hodge et al., 2000).Thereafter, root N uptake was still efficient during late yellowing (between 7 and 28 October), since 15 N recovered from the fine roots slightly increased from 3.5 to 5.5 %, whereas the 15 N that was recovered from microbial biomass decreased from 12.5 to 10 %.This could be explained by microbial mortality and turnover, which releases N to the soil, combined with the capacity of plants to sequester N for longer (Barnard et al., 2006;Bloor et al., 2009;Hodge et al., 2000).
After leaf fall, trees may have a significant capacity for nitrate uptake in the fine roots in midwinter (i.e. in the absence of leaves), as already shown in Japan oak (Ueda et al., 2010).However, in our case, N soil uptake was limited by low soil temperature, which affected the mineralization rate and root activity, since the 15 N recovered from roots slightly decreased between October 28 and December 2 (5.5 to 4 %) and then declined to 1.75 % between 2 December and 8 April.

N dynamic in soil tree system the following spring
In April (before budburst), for trees with leaves labelled in the previous year (L 1 ), the most part of 15 N was recovered in their roots (11.5 %).On the other hand, at the same date, most of the labelled N applied to soil in September (L 2 ) was recovered from the rhizospheric soil (14.5 %).When soil (and hence spring N uptake) was labelled (L 3 ) at the beginning of March, a month later most of the 15 N was recovered from microbial biomass and rhizospheric soil (81 %), but a small proportion of 15 N was recovered from the fine roots (1.5 %).The latter demonstrated a small N uptake before budburst, as has previously been observed in Japan oak (Ueda et al., 2010).This early N uptake from the soil could be related to sessile oak's hydraulic properties.As a ring-porous species, sessile oak achieves 30 % of its annual radial stem growth before leaf expansion in spring (Breda and Granier, 1996).Water flow pathways are then restored each spring before the onset of transpiration (Breda and Granier, 1996).This enables early root N uptake from soil as soon as a threshold soil temperature is reached.
Eight days after budburst, most of the 15 N applied to leaves (L 1 ) was recovered from new leaves (25.2 %) and new twigs (mean of 3.5 %).This clearly underlined that a significant proportion of 15 N used to synthesize new leaves came from 15 N stored during the previous autumn, as shown for Ligustrum (Salaün et al., 2005).Moreover, this N came from foliar N of the previous year, not from soil N uptake during the previous autumn.Indeed, trees labelled the previous autumn on soil (L 2 ) showed a similar partitioning of 15 N in leaves and twigs before budburst (208 days after labelling) and 8 days after budburst (227 days after labelling), there was no mobilization of 15 N for the new leaves and twigs synthesis for those trees.Less than 1 % of 15 N taken up from soil before budburst was recovered in leaves and twigs 8 days after budburst.A distinction might be made between stored N sourced from leaves and that sourced from soil, stored mainly in roots.N from leaves could be stored as amino acids in branches, trunk and coarse roots, whereas N taken up from soil could be stored in roots as NO − 3 .This N was not converted into amino acids by glutamine synthetase/glutamate synthase enzymes during winter, most probably due to low enzymatic activity in roots during winter (Bazot et al., 2013).As a consequence, in the following spring trees first remobilized easily, circulating forms of N and N to be stored nearer to demands.Indeed in trees, NO − 3 is hardly transported to their leaves but rather turned into amino acids in their roots (Morot-Gaudry, 1997).Indeed roots were the main site of NO − 3 reduction (Gojon et al., 1991).Consequently, soil 15 N was not the main contributor to the synthesis of new twigs and new leaves during the first 8 days after budburst.Eight days after budburst, 95 % of new leaves 15 N came from 15 Nlabelled reserves, 2 % from soil labelled the previous autumn and only 3 % from soil labelled in the current spring (Eq.2, Fig. 4).Previous studies have also found that N reserves contribute significantly to leaf expansion in young trees: in white birch (Wendler and Millard, 1996), sycamore maple (Millard and Proe, 1991), Japan oak (Ueda et al., 2009), pedunculate oak (Vizoso et al., 2008) and sessile oak (El Zein et al., 2011a).
Considering trees with soil that had been labelled in autumn (L 2 ), 8 days after budburst the proportion of recovered 15 N in microbial biomass decreased slightly, whereas it slightly increased in fine roots compared to the previous sampling date.One can suppose that the increased soil temperature and the first flux of C from plant to soil (rhizodeposition)-stimulated microbial biomass turnover, making 15 N available for root uptake.Very little 15 N was recovered from the other compartments of the trees.
Soil N uptake became really effective between 8 and 30 days after budburst.Indeed, whatever the date of the soil labelling (autumn or the current spring), 30 days after budburst, a sharp decrease in 15 N in the microbial biomass was observed, depending on an increase of 15 N in fine roots and in young leaves.In 28 June (at leaf maturity), 40 % of the 15 N recovered from leaves came from stored 15 N, 10 % came from 15 N applied to soil the previous autumn and 40 % came from 15 N applied on soil in March, 1 month before budburst (Eq.2, Fig. 4).This pattern of contribution was maintained throughout the season.Similar findings have been reported for other species.For example, 20-30 % of shoot-leaf N was supplied by spring-applied fertilizer for mature pear trees (Sanchez et al., 1990) and mature almond trees (Weinbaum, 1984), while only 13 % of a solution of nitrate-N and ammonium-N that was applied to soil contributed to total leaf N of apple trees (Neilsen et al., 1997).Sorbus aucuparia had remobilized half the N from storage before any was taken up by the roots (Millard et al., 2001).Finally, there is a concomitant/concurrent remobilization and uptake of N from the soil by some other species, as shown for Scots pine (Millard et al., 2001) and walnut (Frak et al., 2002).

Conclusion
This paper completes knowledge of internal and external nitrogen cycles in a forest ecosystem.We highlighted that in autumn, N reserves are formed from N remobilized from leaves and N uptake by roots.This N is stored in roots, principally most probably in the form of amino acids and nitrate.Those reserves, especially N coming from leaves, contributed significantly to new tissue synthesis the following spring.Nevertheless, N uptake was also observed in spring before budburst; this N was not transferred to new twigs and new leaves during the first days following budburst.N uptake from soil only contributed significantly to the synthesis of new tissues when leaves were fully expanded.Two months after budburst the relative contributions of 15 N originating from leaves and 15 N uptake from soil were 40 : 60, whereas they were 95 : 5 on the 8 days after budburst.
Author contributions.Stephane Bazot and Laure Barthes conceived and designed the experiments.Stephane Bazot, Chantal Fresneau and Laure Barthes conducted all field and laboratory analyses.Stephane Bazot carried out data analysis, wrote most of the manuscript and prepared the figures.Chantal Fresneau, Claire Damesin and Laure Barthes contributed to the writing of the manuscript.

Figure 1 .
Figure 1.Partitioning of recovered 15 N (PRN%) from the sampled compartments following the first labelling campaign, i.e. from 27 May 2009 to 20 October 2010.(a) Leaves ♦ and twigs ×, (b) phloem , (c) fine roots ○, (d) rhizospheric soil ■ and micro- bial biomass + (for those compartments the Y axis was adjusted to 1).DAL: days after labelling.The two lines, continuous and dotted, correspond to tree 1 and tree 2. Vertical bars indicate standard errors.

Figure 2 .
Figure 2. Partitioning of recovered 15 N (PRN%) from the sampled compartments following the second labelling campaign, i.e. from 9 September 2009 to 20 October 2010.(a) Leaves ♦ and twigs ×, (b) phloem , (c) fine roots ○, (d) rhizospheric soil ■ and micro- bial biomass +.DAL: days after labelling.The two lines for each category (continuous and dotted) correspond to tree 3 and tree 4. Vertical bars indicate standard errors.

Figure 4 .
Figure 4. Conceptual scheme representing percentage contributions of 15 N (Eq.2) from each labelling campaign (L 1 : white, L 2 : light grey, L 3 : dark grey) in roots in the autumn and in new leaves in the season following the first labelling campaign.

Table 1 .
Labelling characteristics and recovery of 15 N administered in each labelling campaign from the sampled compartments of each tree, on each sampling occasion (DAL: days after labelling, JD: Julian day number).
3.1.1After the foliar labelling in spring (L 1 , 27 May 2009) The total balance for the administered 15 N demonstrated maximum recoveries of 15 N within the plant-soil system of www.biogeosciences.net/13/3475/2016/Biogeosciences, 13, 3475-3484, 2016 32 %, 1 day after leaf labelling.It decreased to 13.5 % of the administered 15 N recovered in the sampled compartments at the end of September (126 days after labelling) (Table