|Home | About | Journals | Submit | Contact Us | Français|
How the remobilization of S and N reserves can meet the needs of seeds of oilseed rape subject to limitation of S fertilization remains largely unclear. Thus, this survey aims to determine the incidence of sulphate restriction [low S (LS)] applied at bolting [growth stage (GS) 32], visible bud (GS 53), and start of pod filling (GS 70) on source–sink relationships for S and N, and on the dynamics of endogenous/exogenous S and N contributing to seed yield and quality. Sulphate restrictions applied at GS 32, GS 53, and GS 70 were annotated LS32, LS53, and LS70. Long-term 34SO42− and 15NO3− labelling was used to explore S and N partitioning at the whole-plant level. In LS53, the sulphur remobilization efficiency (SRE) to seeds increased, but not enough to maintain seed quality. In LS32, an early S remobilization from leaves provided S for root, stem, and pod growth, but the subsequent demand for seed development was not met adequately and the N utilization efficiency (NUtE) was reduced when compared with high S (HS). The highest SRE (65±1.2% of the remobilized S) associated with an efficient foliar S mobilization (with minimal residual S concentrations of 0.1–0.2% dry matter) was observed under LS70 treatment, which did not affect yield components.
Sulphur (S) is an important nutrient for plant growth and development. In comparison with other crops such as cereals, oilseed rape (Brassica napus L.) requires a relatively large amount of mineral S (Zhao et al., 1997). During the last two decades, the reduced atmospheric pollution by industries has resulted in a major reduction in S emissions and, as a consequence, S deposition into the soil has strongly declined, particularly in Western Europe (McNeill et al., 2005). A deficiency in S can reduce yield, and impacts on the quality of harvested products (Janzen and Bettany, 1984; McGrath and Zhao, 1996; Scherer, 2001). The technical centre for oilseed production in France (CETIOM) recommends systematic S fertilization for oilseed rape crops with ~30kg S ha−1. Therefore, more attention should be paid to S fertilization practices that need to be optimized to fulfil plant S requirements whilst minimizing cost. Similarly to nitrogen (N) uptake (Rossato et al., 2001), the S requirement of oilseed rape would depend on the stage of plant development and environmental conditions. Indeed, the S requirement is not stable during the growth cycle of oilseed rape: S uptake increased from stem extension to the start of flowering, whereas little S uptake was generally (but not exclusively) observed during pod filling (McGrath and Zhao, 1996; Postma et al., 1999).
Winter oilseed rape can be used to reduce N leaching during the autumn–winter period because of its high capacity to take up nitrate from the soil. N and S nutrition are tightly linked during the growth cycle (Reuveny et al., 1980; Fismes et al., 2000). N and S are both involved in amino acid and protein synthesis. Restriction of S supply has been shown to depress the nitrate uptake and nitrate reductase activity in maize and spinach (Friedrich and Schrader, 1978; Prosser et al., 2001), and can result in nitrate accumulation in leaves of oilseed rape (McGrath and Zhao, 1996). Fismes et al. (2000) reported that the S and N use efficiency of oilseed rape are synergistic at optimum rates and antagonistic at excessive levels of one of the elements. S fertilization is required to improve N use efficiency and thereby maintain a sufficient oil content and fatty acid quality of seeds (Fismes et al., 2000).
During vegetative development, winter oilseed rape is at the rosette stage in winter and the leaves represent a major store of nutrients which can be remobilized thereafter to sustain growth of reproductive tissues, as shown specifically for N (Schjoerring et al., 1995; Rossato et al. 2001; Noquet et al. 2004; Malagoli et al., 2005a). For instance, nearly 75% of the N content in reproductive tissues of oilseed rape is derived from N mobilization occurring mostly in leaves and stems (Malagoli et al., 2005b). Therefore, leaves emerging during the rosette stage would play a crucial role in seed filling and contribute to the maintenance of seed yield (Noquet et al., 2004). Thus, optimizing S fertilization requires a better understanding of (i) source–sink relationships for S at the whole-plant level; and (ii) processes of S mobilization by evaluating plant S partitioning in relation to the plant growth stage and N status from stem extension to harvest.
Oilseed rape may accumulate abundant amounts of sulphate (34SO42−), but this anion is not mobilized efficiently from vegetative to reproductive tissues: the S Harvest Index (SHI, i.e. the S amount in seeds divided by the total S in the whole crop) is only ~20% (McGrath and Zhao, 1996), indicating that a large proportion of S is retained in the vegetative tissues. Sulphate stored in the vacuoles is the main form of S reserve in vegetative tissues (Blake-Kalff et al., 1998; Scherer, 2001; Matula and Pechová, 2002). To sustain the S demand for growth of oilseed rape under S restriction occurring at the rosette stage, a strong S mobilization (mainly an 34SO42− mobilization), associated with an up-regulation of BnSultr4;1 and/or BnSultr4;2 expression (two transporters involved in efflux of sulphate from vacuoles; Kataoka et al., 2004; Parmar et al., 2007), was reported in leaves (Dubousset et al., 2009). Smith and Lang (1988) reported that 90% of the S transported via the phloem is inorganic in soybean. Sunarpi and Anderson (1998) described S redistribution in S-deficient vegetative soybean (with an 35S pulse–chase labelling method) and reported that ~25% of the mobilized S was recycled as 34SO42− via the root and the largest newly expanded leaf, which acts as an intermediary in the transport of S from the root to the youngest expanding leaves. S mobilization in suboptimal conditions of S fertilization was also examined in reproductive soybean (Sunarpi and Anderson, 1997; Naeve and Shibles, 2005). These authors reported that soybean leaves did not act as large reservoirs for S in conditions of suboptimal S fertilization. Nevertheless, under SO42−-sufficient conditions, it was shown that leaves of soybean supplied the seed with 20% of its total S requirement (Naeve and Shibles, 2005). Therefore, in soybean, the amount of S mobilized from leaves at the reproductive stage appears to be reliant on the amount previously stored in roots and leaves. In oilseed rape, the source–sink relationships for S, and more particularly the contribution of leaves in the S reallocation to seeds, remains unclear. The concentration of S in leaves at early flowering was suggested to be the best index in predicting S deficiency in terms of seed yield by McGrath and Zhao (1996).
Although mobilization of S and N from vegetative tissues is likely to be important for seed filling in oilseed rape, very little is known about the efficiency (dynamics and amounts) of S and N mobilization to the reproductive tissues. How the limitation of S fertilization impacts on the remobilization processes of S reserves and N reserves also remains largely unclear. To address these questions, the aim of this study was to determine the impact of sulphate restrictions [low S (LS) versus high S (HS)] applied at bolting (GS 32), visible bud (GS 53), and start of pod filling (GS 70) growth stages of winter oilseed rape on (i) the source–sink relationships for S and N at the whole-plant level; (ii) the remobilization of S reserves and N reserves and their contribution to developing seeds; and (iii) the seed yield and grain quality. To explore S and N reserve partitioning in oilseed rape, a greenhouse experiment was carried out for long-term steady-state labelling using stable isotopes as tracers, with 34SO42− and 15NO3− applied at the beginning of the stem elongation stage (GS 16) for different periods (17, 30, and 44d), before applying S restriction. In this way, the dynamics of the mobilization of S and N compounds in response to different levels of mineral S availability during the subsequent chase periods could be accurately estimated. Additionally, to determine if the foliar residual S and N concentrations were related potentially to an efficient mobilization of S and N to seeds, the S and N concentration in dead leaves in response to the different mineral S availabilities was examined in relation to their nodal positions.
The oilseed rape genotype chosen for this greenhouse experiment was cv. Capitol, a genotype well described in terms of N use efficiency (Malagoli et al., 2004, 2005a, b; Gombert et al., 2006; Etienne et al., 2007; Desclos et al., 2008, 2009). After surface sterilization, seeds were germinated on vermiculite in 20.0l tanks for 24 seedlings and grown with a thermoperiod of 20°C (day 16h) and 15°C (night 8h), on 25% Hoagland nutrient solution consisting of 1.25mM Ca(NO3)2·4H2O, 1.25mM KNO3, 0.5mM MgSO4, 0.25mM KH2PO4, 0.2mM EDTA, 2NaFe·3H2O, 14μM H3BO3, 5μM MnSO4, 3μM ZnSO4, 0.7μM (NH4)6Mo7O24, 0.7μM CuSO4, 0.1μM CoCl2, renewed twice a week for 36d. The plants were then submitted to 8°C (day 10h) and 4°C (night 14h) for 46d for vernalization with the same nutrient solution renewed twice a week. After this period of vernalization, every plant was transferred to pots containing mixed 1/3 vermiculite and 2/3 perlite (one plant per pot) and submitted to a thermoperiod of 20°C (day) and 15°C (night). As indicated in Fig. 1, during different periods of growth [from GS 16 (rosette stage) to GS 32 (bolting stage), GS 53 (visible bud stage), or GS 70 (start of pod filling)], plants were supplied with 34SO42− (1 atom% excess) and 15NO3− (2 atom% excess) in order to obtain plants with homogeneous 34S and 15N labelling. Each day, the nutrient solution (25% Hoagland for control plants, i.e. HS treatment) was supplied automatically in an increasing volume as a function of the growth stages: 90, 120, 150, and 180ml per plant at the start of the bolting stage, the visible bud stage, the flowering stage, and the seed maturation stage, respectively. Mineral S restriction (LS treatments) corresponding to 8.7μM 34SO42− was applied at GS 32 for LS32, GS 53 for LS53, or GS 70 for LS70, until the end of the growth cycle (GS 99).
At final harvest, the number of mature seeds per plant was accurately determined at GS 99 for four replicates per treatment. Seeds were then used for the test of viability and the determination of seed composition. At each date of harvest, the different plant parts (lateral roots, taproot, leaves, stem, floral stem, pod walls, and seeds) were weighed, freeze-dried, and then ground to a fine powder for elemental and isotope analyses. Old, mature, and young leaves were collected after determination of the relative chlorophyll concentration using the non-destructive SPAD (Soil Plant Analysis Development) chlorophyll meter (Minolta, SPAD-502 model), and measurement of the leaf area using a LI-COR 300 area meter (LI-COR, Lincoln, NE, USA). At each date of harvest, the leaves characterized by a bottom position on the plant and a chlorophyll concentration <55 SPAD units were clustered in ‘old leaves’. Then, the upper leaves characterized by chlorophyll concentrations and areas >55 SPAD units and >55 cm2 were clustered in ‘mature leaves’. Finally, the younger leaves characterized by an area <55 cm2 were clustered in ‘young leaves’.
The leaf rank number was determined according to the date of leaf emergence using a labelled collar suspended on the petiole of each leaf rank after maturity. Thus, changes in the S and N concentrations in dead leaves were monitored for each nodal position, from seedling to the seed maturation stage. These leaf samples were freeze-dried, weighed for dry matter (DM) determination and then ground to a fine powder for S and N analyses.
The viability of seeds produced by plants submitted to the different S availabilities was tested by assessment of seed germination. Mature seeds obtained for each treatment were germinated on Whatman filter paper soaked with sterile water within Petri dishes (12×12cm). Fifty seeds per biological repetition (n=6 for HS and n=4 for each LS treatment) were sown on water for 7days with a cycle of 8h dark (18°C)/16h light (25°C). Three technical replicates were performed for each biological repetition. The percentage of plantlets with normal development indicated the number of viable seeds for each S treatment.
All the seed samples were scanned on a monochromator near infra red system (NIRSystem model 6500, FOSS NIRSystem Inc., Silver Spring, MD, USA) equipped with the transport module, in the reflectance mode. Intact seeds (~5g) were placed in a standard ring cup and scanned. The results were predicted from an external calibration established for oil and total glucosinolate content (CRAW, Gembloux, Belgium). Three determinations were performed for each sample. The results were given as a percentage of oil or proteins per seed DM and in μmol of total glucosinolates per seed DM.
Freeze-dried samples were ground to a fine powder, weighed, and placed into tin analysis capsules. Both total S and N contents were determined with a continuous flow isotope mass spectrometer (IRMS, Isoprime, GV Instruments, Manchester, UK) linked to an analyser (EA3000, EuroVector, Milan, Italy). The IRMS analysis also provided the changes of the relative amount of 34S and 15N in excess in each sample derived from the tracer fed to the test plant.
The values can be calculated as:
34S amount in excess=isotope abundance in sample (A%)–isotope abundance in natural standard (4.2549%)
Similarly, 15N amount in excess was determined as follows:
15N amount in excess=isotope abundance in sample (A%)–isotope abundance in natural standard (0.3731%)
δ34S (‰) and δ15N (‰), that were experimentally measured in each sample, are indexes generally used and defined as:
where Rsample indicates the isotopic ratio (34S/32S) in the sample, and Rstandard=0.04415206 is the internationally accepted isotope standard for S corresponding to V-CDT (Vienna Canyon Diablo Troilite).
where Rsample indicates the isotopic ratio (15N/14N) in the sample and Ratm indicates the isotopic ratio in the atmosphere.
Accordingly, the value of Rsample can be estimated from δ34S and δ15N value as follows:
Long periods of labelling allow a homogenous distribution of tracers in different organs and different biochemical fractions containing S and/or N. Normalization of the amounts of absorbed 34S and 15N is carried out using the average amount of each of these isotopes found throughout the whole plant for each harvest date and treatment submitted to similar periods of labelling. After normalization, the partitioning of 34S and 15N in plants is expressed as the percentage of total 34S and 15N. The method of calculation of S flows is presented below and can be transposed to the determination of N flows. The calculations of flows of remobilized S depend on the source or sink status of each organ. For the source organs, this is characterized by a loss of 34S amount for a period Δt. Between the dates t0 and t0+Δt, the S amount remobilized (QSRsource) corresponded to:
where Q34St0=amount of 34S in the source organ at t0, Q34St0+Δt=amount of 34S in the source organ at t0+Δt, QSt0=amount of S in the source organ at t0, and Δt=period of chase, for example between GS 70 and GS 81.
For the sink organs, this is characterized by a gain of 34S amount for a period Δt. Between the dates t0 and t0+Δt, the S amount derived from remobilization (QSRsink) corresponded to:
where Q34St0=amount of 34S in the sink organ at t0, Q34St0+Δt=amount of 34S in the sink organ at t0+Δt, ΣQSRsource=total amount of S remobilized from source organs between t0 and t0+Δt, and Σ(Q34St0+Δt–Q34St0)=total amount of 34S accumulated in the sink organs between t0 and t0+Δt.
The inflow of S taken up (QSInflux) between two dates (i.e. for the period Δt) was calculated by subtracting the S derived from remobilization (QSRsink or source) between these two dates from the change in total S amount for this period (ΔQS):
The normality of the data was studied with the Ryan–Joiner test at 95%. Analysis of variance (ANOVA) and the Tukey test to compare the means were performed with MINITAB13 on Windows (Minitab Inc., State College, PA, USA). When the normality law of the data was not respected, the non-parametric test of Kruskal–Wallis was carried out and followed by Mood's median test. Statistical significance was postulated at P <0.05.
In LS32 conditions, the global seed DM was reduced at GS 99 by almost 45% [from 11.6±0.61g in control (HS) to 6.30±0.66g per plant in LS32; Table 1]. In addition, the number of viable seeds decreased greatly in response to LS32 treatment (Table 1) and corresponded to 15.3±1.6% of the total seeds produced. Compared with control, the oil and protein content was significantly decreased by the LS32 treatment. In addition, a strong decrease in glucosinolate content was observed in all LS treatments and especially in LS32 (–69±6.7%) and LS53 (–82±3.1%) (Table 1). The oil content in seeds was decreased in LS53 conditions (Table 1). In contrast to LS32, the protein content was not affected by the LS53 treatment as compared with the control. Interestingly, the oil and protein content in seeds was not significantly modified by the LS70 treatment (Table 1). LS70 treatment even had the benefit of lowering glucosinolate content (Table 1).
The SHI (i.e. the S amount in seeds expressed as a percentage of the total S amount in plants at GS 99) corresponded to 26±1.3% of total S in control plants and was similar in LS32 conditions, whereas it reached 45±1.8% and 55±1.7% in LS53 and LS70 conditions, respectively (Table 1). The highest SHI was thus obtained in LS70 conditions and was 2-fold higher than in control, suggesting a better targeting of S mobilization to seeds in response to this treatment. The N Harvest Index (NHI, i.e. the N amount in seeds expressed as a percentage of the total N amount in plants at GS 99) was 35±3.9% in LS32 conditions whereas it reached 49.1±2.1% in control (Table 1). The other LS treatments did not affect the seed N amount and NHI in comparison with control.
The production of DM of mature seeds was used to calculate the S or N utilization efficiency (SUtE and NUtE, expressed as seed DM produced per unit of S or N accumulated in vegetative shoots; Table 2). The highest values of SUtE were obtained in LS53 and LS70 conditions and reached 461±24mg and 379±24mg of mature seed DM per mg of S in shoots, respectively (Table 2). Compared with control, the NUtE was significantly increased only in LS53 conditions (with 48±5.6mg versus 32±2.1mg of mature seed DM per mg of N in shoots in HS).
Using double 34S and 15N long-term labelling it was possible to estimate for the chase period the distribution of S reserves in plants. Figure 2 illustrates the 34S partitioning from GS 32 to GS 81 in response to the different sulphate availabilities. The analysis of 34S partitioning as a function of growth stages allows a determination of sink–source relationships for S at the whole-plant level.
In HS32 conditions, all leaves (young, mature, old, and dead leaves) contained the largest proportion of the total 34S from GS 32 to GS 81. As a consequence, the foliar 34S remobilization efficiency (SREleaf, corresponding to the loss of 34S in leaves between two growth stages, expressed in a percentage of total 34S labelling) was only 26±0.8% between GS 32 and GS 81 (Fig. 2A). At GS 70, the weak proportion of 34S remobilized from leaves of HS32 plants was transiently allocated towards the stems and roots. After GS 70, the proportion of 34S in roots remained stable (15±1.8%) and high amounts of 34S were lost in dead leaves at GS 81 (54±6.0%; Fig. 2A).
Compared with HS32, the proportion of 34S allocated to stems, floral stems, and pod walls from GS 32 (bolting stage) to GS 81 (seed colouring) was increased by LS32 treatment (Fig. 2A). Interestingly, in response to LS32 treatment, roots became a transient major sink organ (until GS 70) before becoming a source for S (from GS 70 to GS 81) (Fig. 2A). From the beginning of the chase period, 34S stored in the mature and old leaves of LS32 plants was mobilized earlier than in HS32 conditions, and this 34S re-allocation was to the benefit of roots and floral stem. The residual 34S in dead leaves was strongly decreased at GS 81, from 54±6.0% in HS32 to 24±0.3% of total 34S in LS32. Nevertheless, a remobilization from all leaves to other plant parts did not take place between GS 70 and GS 81 in LS32 conditions. Indeed, the total proportion of 34S in all leaves remained stable between GS 70 and GS 81 (28±0.3%; Fig. 2A).
In HS53, the 34S partitioning from GS 32 to GS 53 illustrates the allocation associated with the S uptake before GS 53. After GS 53 (start of the chase period; Fig. 2B), the 34S partitioning illustrates the pattern of remobilization of the S previously acquired in the plant. The SREleaf from GS 53 to GS 81 (42±1.8% for HS53) is higher than the SREleaf obtained between GS 32 and GS 81 (26±0.8% for HS32) (Fig. 2A, ,B).B). The mobilization of 34S in leaves was associated with an increasing sink status, first of the floral stems (at GS 70) and secondly of pod walls and seeds (at GS 81). Roots and stems did not act as source or sink organs for 34S between GS 53 and GS 81 in HS53 conditions (Fig. 2B). In response to LS53 conditions, compared with HS53, the 34S was allocated particularly to stems at GS 70 (but only transiently), while the 34S accumulated in leaves decreased. In contrast to the LS32 conditions, the LS53 treatment did not provoke transient redistribution of 34S towards roots. In comparison with HS53, the final 34S partitioning in LS53 conditions was characterized by the highest redistribution of 34S in seeds (corresponding to 46±2.0% in LS53 versus 27±1.7% of total 34S in HS53 at GS 81), whereas a better remobilization of 34S reserves was noticed in stems and leaves (Fig. 2B). Consequently, at GS 81, dead leaves of LS53 contained 20.8±0.5% of total 34S versus 31.1±0.7% in HS53.
A large amount of 34S was allocated to stems before GS 70 in HS70 conditions. The decline in 34S from leaves (with an SREleaf of 15±1.1% from GS 70 to GS 81) was associated with the decrease of 34S in the roots and stems and this 34S was re-allocated towards seeds, which reached 30±3.0% at GS 81 (HS70, Fig. 2C). As compared with HS70, the decrease in 34S in leaves was more important in LS70 conditions as indicated by the value of SREleaf (–35±0.5%) observed between GS 70 and GS 81. Indeed, the dead leaves corresponded finally to 14±0.5% in LS70 versus 27±0.8% of total 34S in HS70 at GS 81 (Fig. 2C). The 34S in seeds finally reached 45±3.0% of total 34S in LS70.
The changes in 15N partitioning from GS 32 to GS 81 in response to the different S treatments are given in Fig. 3.
In HS conditions, while the proportions and dynamics of 15N were similar to 34S in roots (Figs 2, ,3),3), leaves contained a large proportion of 15N which, in contrast to 34S, greatly decreased before GS 81. The foliar N remobilization efficiency (NREleaf, corresponding to the loss of 15N in leaves between two growth stages, expressed as a percentage of total 15N labelling) reached, on average, 71±2.9% between GS 32 and GS 81 (Fig. 3A). After GS 70, leaves were the main source of 15N for seed filling and the final proportion of 15N in seeds reached 47±5.2% in HS32 (Fig. 3A).
In response to LS32 treatment, the proportion of 15N transiently and strongly increased in roots until GS 70 before becoming a source for 15N from GS 70 to GS 81 (Fig. 3A). At GS 70, the 15N in roots of plants submitted to LS32 conditions reached 31±3.7% of total 15N versus only 18±0.8% in HS32. Finally, the 15N found in seeds at GS 81 in response to the LS32 treatment reached 34±7.3% of the total 15N, and the 15N in roots remained high (Fig. 3A).
In contrast to LS32, the LS53 and LS70 treatments did not significantly alter the partitioning of 15N in comparison with the respective controls, HS53 and HS70 (Fig. 3B, ,C).C). In LS53 and LS70 conditions, the 15N reallocated to seeds corresponded to more than half of the total 15N (Fig. 3B, C). Globally, all the leaves constitute the main source of 15N for seed 15N filling (Fig. 3B, ,CC).
In contrast to LS32 and LS53, the LS70 treatment consisting of a restriction of sulphate supply since GS 70 (i.e. start of pod filling) did not alter seed yield and quality (Table 1). In addition, LS70 treatment did not affect the NUtE (Table 2) and 15N partitioning (Fig. 3), and led to the most efficient seed production with high SHI (Table 1) and SUtE (Table 2). In response to this treatment, the SREleaf was more than doubled in comparison with control (35±0.5% versus 15±1.1% in HS70 between GS 70 and GS 81) (Fig. 2C). In these circumstances, the allocation of S and N taken up from the soil and the endogenous S and N remobilizations during the reproductive phase of oilseed rape development were examined on the basis of 34S and 15N enrichment (see Materials and methods for details) and illustrated for HS70 (Figs 4A, ,5A)5A) and LS70 conditions (Figs 4B, ,5B5B).
For HS70 plants, little of the S taken up was allocated to the roots; the main sinks for S taken up were floral stems, pod walls, and seeds, with an equivalent allocation of S to pod walls and seeds (Fig. 4A). Leaves were the major source organ for remobilized S (60±2.2% of the total S remobilized from GS 70 to GS 81; Fig. 4A) while stems, floral stems, and roots contributed poorly to the supply of endogenous S to other tissues in control plants. The restriction of S availability (LS70 treatment) greatly reduced total S uptake to a level that was insignificant, whereas 67±2.2mg of S were taken up in HS70 conditions (Fig. 4B). Compared with HS70, LS70 conditions also changed the source–sink relationships for endogenous S (Fig. 4B). The LS70 treatment increased the SRE (i.e. the proportion of the total S amount remobilized into the plant which was recycled towards seeds) with a redistribution of 65±1.2% of S remobilized to seeds versus 44±0.9% in HS70 (Fig. 4A, ,B).B). Compared with HS70, the highest mobilization of S for seed filling observed in LS70 conditions would be related to a lower loss of S by dead leaves (which was ~2-fold less in LS70 than in HS70 conditions; Fig. 4). Finally, the S amount quantified in seeds reached 41±2.3mg in LS70 thanks to remobilization from vegetative plant parts. The source status of roots was significantly lower in LS70 than in control (7±0.1% in LS70 versus 13±0.3% of total endogenous S recycled in HS70 plants, Fig. 4A, B).
The restriction of S availability applied at GS 70 did not significantly reduce the total N uptake between GS 70 and GS 81 (with an average of 297±10mg of N taken up) and did not drastically change the N partitioning within the different plant tissues (Fig. 5A, ,B).B). The N remobilization efficiency (NRE) to seeds in LS70 conditions reached 77% and was not significantly different from the control (Fig. 5A, ,B).B). Finally, about half of the total N in seeds at GS 81 was derived from mobilization in both treatments. It appeared that leaves represented the major source organ for N, to the main benefit of the seeds, and to a lesser extent to the pod walls. The residual N lost by dead leaves (14±2.1mg of N) was unchanged by the treatment restricting S availability. Compared with control, the N remobilization from roots towards reproductive tissues was reduced in LS70 (5.2±0.1% in LS70 versus 9.6±0.3% of total endogenous N recycled in HS70 plants; Fig. 4A, ,B).B). Whatever the treatment, and as observed for S, roots therefore contributed poorly to the supply of endogenous N to other plant tissues.
Since the residual DM of each leaf rank was not affected by LS treatments (data not shown), the S and N concentration in dead leaves in response to the treatments was examined in relation to their nodal positions (Fig. 6A, ,B).B). The average of residual S in leaf ranks below nodal position #13 was 0.67±0.03% of DM while the residual S concentration was >0.8% of DM in upper leaf ranks (ranging from 0.88±0.05% of DM in leaf rank #14 to 1.78±0.22% of DM in leaf rank #16). These upper leaves corresponded to the smallest leaves (with a leaf area <6cm2, data not shown) that appeared at the visible bud stage. As expected, in response to mineral S restriction treatments, the residual S concentration in leaves was significantly reduced. The S concentration in dead leaves of LS32 plants was significantly affected from node #5 while this decrease happened in leaves above node #7 for LS53 and above node #9 for LS70 plants (Fig. 6A). Minimal values of residual S concentration in dead leaves (comprised between 0.1% and 0.2% of DM) were observed in response to the three LS treatments, particularly in leaves above leaf #11 (emerged at GS 32). These minimal foliar S concentrations were observed earlier for the LS32 treatment (from node #7).
In comparison with HS, and with the exception of leaf rank #14 (with a concentration of N significantly reduced in response to the LS32 treatment; Fig. 6B), the residual N concentration in leaves was not affected by sulphate restriction treatments. Residual N gradually increased from basal to upper leaves and was globally below 1% of DM (Fig. 6B). While the residual S concentration in the control was higher than the residual N concentration in leaves emerged before the 11th rank (emerged at GS 32), it is interesting that cross-talk between S and N concentrations (corresponding to an N/S ratio of 1) was observed in lower leaf ranks in response to LS treatments (Fig. 6). The residual N concentration was higher than the S concentration for leaf ranks ≥#5 for LS32, for leaf ranks ≥#6 for LS53, and for leaf ranks ≥#9 for LS70 (Fig. 6).
This double 34S and 15N labelling experiment (Fig. 1), undertaken in control conditions, was designed to follow the course of remobilization of endogenous S and N in oilseed rape with particular attention to leaves that correspond to the main source organs for S and N (Figs 2, ,3).3). Except for LS70, the SHI values obtained were noticeably lower than those observed for the NHI (Table 1), indicating that S is remobilized to seeds less efficiently than N (Sexton et al., 1998). The results obtained for yield and quality of seeds reveal that the mineral S availability between GS 32 and GS 70 would be a determinant for seed filling processes and seed quality. In response to the LS32 treatment, the NHI and NUtE were reduced and the seed composition was affected (Table 1). Fismes et al. (2000) have shown using field-grown oilseed rape that S deficiency can reduce NUtE and protein level in seeds. The present results indicate that an S fertilization regime with the ability to satisfy the growth needs of oilseed rape until GS 53 is required to maintain a sufficient NUtE and protein level in seeds. In response to the LS treatment consisting of a restriction of sulphate supply from GS 70 (LS70), oilseed rape was able to optimize its SUtE (Table 2) in order to produce high quality seeds (Table 1). The LS70 treatment led to the highest SRE to seeds, with a redistribution of 65±1.2% of remobilized S towards seeds, in contrast to the 4 ±0.9% observed in HS70 (Fig. 4A, ,BB).
The enhanced remobilization of endogenous S towards the seeds observed in response to the LS53 or LS70 treatments was not associated with noticeable modifications of the source–sink relationships for N (Figs 3B, ,C,C, ,5).5). This shows that the interaction of the two nutrients is strongly affected by development. Thus, the altered seed yield and quality in response to the LS32 treatment would be partially attributed to significant modifications in N dynamics. After GS 53, it appears that oilseed rape can optimize the mobilization of endogenous S to seeds in response to S restriction, independently of the N distribution.
About half of the N content in reproductive tissues of oilseed rape was derived from N mobilization (Fig. 5) occurring mostly in leaves and stems. The roots did not significantly contribute to endogenous S remobilization. The lack of S remobilization from roots suggests sequestration of sulphate and/or the presence of a high proportion of organic S reserves that were difficult to mobilize. In HS plants at GS 81, 35±2.4% of the total S in roots was in the sulphate form (data not shown).
Hoefgen and Nikiforova (2008) suggested enhanced lateral root formation thanks to activation of auxin-inducible genes as a possible adaptation to prospect for available S in soil in the case of sulphate deficiency. In response to the drastic restriction of mineral S which started from a younger stage (LS32 treatment), there was a transient and strong increase of S and N demands in roots (increase of sink status until GS 70) (Figs 2A, ,3A).3A). Compared with control, this temporary sink status of roots in response to severe S restriction (LS32) in oilseed rape was also associated with an accumulation of N amount (but not of S) and a higher DM production in roots (data not shown).
These results underline the importance of S mineral availability before flowering and emphasize the level of S reserves in vegetative aerial tissues. It appears that oilseed rape will more efficiently mobilize previously acquired S and N towards the seeds if S is supplied in adequate amount to support growth of the plant up to the beginning of seed formation. After GS 70, S taken up later (in HS70) was more likely to be disproportionally allocated to the pods (and seeds) (Fig. 4), and was not necessary for the maintenance of seed yield and quality (Table 1).
The importance of leaves for N storage and mobilization to seeds has been well established (Noquet et al., 2004; Malagoli et al., 2005a) and was verified in the present experiment (Figs 3, 5, 6). In contrast, the contribution of leaves to S storage and subsequent S distribution to sustain seed formation and filling remains unclear in oilseed rape (Hawkesford and De Kok, 2006). While Sunarpi and Anderson (1997) reported that soybean leaves contribute little to seed S filling, the present work underlined that leaves of oilseed rape would be crucial for their role as a major source organ for S in response to S restriction (Figs 2, ,3).3). More specifically, if S limitation occurred at GS 70, leaves may improve their SRE in order to cover the demand for S for seed growth (Fig. 4). Interestingly, despite an enhanced remobilization of foliar S reserves (Fig. 6), the lifespan of the leaves emerging during the whole of the growth cycle was unaltered by the LS treatments (data not shown). Besides, the total N amount in dead leaves was not significantly different between HS and LS treatments (Figs 5, ,6B),6B), suggesting that LS conditions improved the S mobilization in leaves independently of N [higher SREleaf (Fig. 2) versus unaltered NREleaf (Fig. 3)]. To sustain the S demand for growth under S limitation, a strong SO42– mobilization in leaves was already reported at the rosette stage without any acceleration of leaf senescence (Dubousset et al., 2009).
While the leaves have been shown to be the primary donors of N for mobilization to seeds (Noquet et al., 2004; Malagoli et al., 2005a), their importance as a major source organ for S has been demonstrated as well. In LS70 conditions, leaves supplied the seed with up to 75±3.7% of the mobilized S during reproductive development. The present study showed that leaves of control plants had high S concentrations (0.67±0.03% of DM for nodes 1–13) when they abscised, indicating that a significant proportion of leaf S was not mobilized before abscission (as verified in Fig. 2). In the absence of deficiency of sulphate, the high proportion of residual S of dead leaves characterized in controlled conditions (Fig. 6) is in accordance with the potential sequestration of S in leaves (in sulphate form) suggested by previous studies (Blake-Kalff et al., 1998; Hawkesford, 2000; Matula and Pechovà, 2002). Under restricted sulphate availability, the residual concentration of S in dead leaves seemed clearly to reflect the balance between supply and demand of S for growth and seed filling. The present experiment suggests that the conjunction of a residual S concentration of 0.1–0.2% of DM with a value of the N/S ratio ≥1 in dead leaves (corresponding to leaves emerged before the bolting stage) could be used as indicators of S deficiency leading to alteration in seed quality (Fig. 6). Nevertheless, the N/S ratio in leaves depends on S and N availability, which leads to difficulties in using this ratio as an accurate diagnosis of the plant S status (Blake-Kalff et al., 2002).
Analysis of the effects of sulphate limitations applied at different growth stages on S and N partitioning reveals disruption between S and N distribution patterns in oilseed rape in response to this nutrient deficiency. By using stable isotopes (34S, 15N) as a tracer system (Monaghan et al., 1999), the determination of 34S/15N partitioning and S/N flows allowed characterization of the contribution of each organ to seed S/N filling. The data obtained in the present work confirm that S is relatively immobile in plants in control (HS) conditions, as the proportion of S redistributed from leaf tissue was considerably smaller than that of N. Under recommended levels of S fertilization, the loss of S through leaf fall from Brassica napus L. cv. Capitol in field conditions can reach 22±0.7kg S ha−1 (LD, unpublished results). This also indicates that S is not recycled during leaf senescence if oilseed rape is grown under optimal S nutrition. In response to LS70 treatments, the highest S remobilization (SRE) to seeds was associated with a high foliar mobilization of S (leaves supplied the seed with ~75±3.7% of the SRE between GS 70 and GS 81). The minimal values of S concentration, comprised between 0.1% and 0.2% of DM in dead leaves, would indicate an enhanced remobilization in response to mineral S restriction. Results observed in response to LS70 treatment indicate the possibility of increasing the SRE (without altering seed quality) by limiting the practice of fertilization. Considering that the presence of glucosinolates in seeds restricts the use of meal in animal feed (Zhao et al., 1994), a limited fertilization with sulphate after GS 70 did not alter the seed quality for meal. To adapt S inputs, the residual S concentration of leaves that emerged before GS 32 may serve as an indicator of a sufficient S reserve status for reproductive growth if it is >0.5% of DM. These results should be taken into account for the development of field diagnosis tests to determine whether plants are deficient in mineral S.
This work was supported by the French National Research Agency (programme ANR-COSMOS no. ANR-05-JC05-51097) and by a PhD grant to LD from the French Ministry of Research. The authors thank Mrs Nathalie Nesi and Véronique Gautier for their help in seed composition analysis, Ms Marie-Paule Bataillé and Mr Raphaël Ségura for their technical help in S and N analyses, Mr Raphaël Ségura, Ms Anne-Françoise Ameline, and Ms Alexandra Girondé for their valuable help in the greenhouse experiment, and Mrs Michelle Coustenoble, Ms Josiane Pichon, Anne-Sophie Desfeux, Josette Bonnefoy, Virginie Séguin, Julie Levallois, Marie-Paule Bataillé, Bénédicte De Loynes d'Estrées, and Mr Jérôme Dubousset for their casual assistance in the harvest process.