About this Journal Submit a Manuscript Table of Contents
Applied and Environmental Soil Science
Volume 2012 (2012), Article ID 506302, 14 pages
http://dx.doi.org/10.1155/2012/506302
Research Article

Temporal Variation of N Isotopic Composition of Decomposing Legume Roots and Its Implications to N Cycling Estimates in Tracer Studies in Agroforestry Systems

1Department of Forest Sciences, University of Helsinki, P.O. Box 27, 00014 Helsinki, Finland
2Bioversity International, P.O. Box 236, UPM Post office, Serdang, 43400 Selangor Darul Ehsan, Malaysia
3UR1321 ASTRO Agrosystèmes Tropicaux, INRA, 97170 Petit-Bourg, France

Received 8 May 2012; Revised 13 July 2012; Accepted 29 July 2012

Academic Editor: William R. Horwath

Copyright © 2012 Riina Jalonen and Jorge Sierra. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Below-ground residue of agroforestry trees is an important N source for associated crops. Several studies have shown that its isotopic signature (δ15N) may change after tree pruning, which makes it difficult to study below-ground N inputs from pruned trees by isotopic techniques. We studied how temporal variation of legume root residue δ15N could be explained by considering differential decomposition kinetics and 15N content of residue fractions. A mathematical model on the isotopic patterns of soil and a N recipient plant during root decomposition was developed and applied for testing assumptions about residue characteristics against two experimental datasets. Observed 15N patterns of the recipient plants could be satisfactorily simulated only when the residue was assumed to consist of at least two fractions with distinct δ15N and decomposition rates depending on their C : N ratio. Assuming δ15N of residue constant over time resulted in substantial underestimates of N derived from low-quality residue (%Ndfr) by the recipient plant when compared with experimental data. Results of this study suggest that residue fractionation can help improve estimation of %Ndfr in isotopic studies, as an alternative or complementary method to assuming or aiming at homogenous isotopic composition of N sources.

1. Introduction

Pruning of trees is a common practice in agroforestry systems. In legume-based systems the main purpose of pruning is to provide nitrogen to soil and the associated crops from green manure and below-ground residue. Timing and intensity of pruning can be varied to adjust the N inputs with the requirements of the crops, and amount and optimal timing of these inputs have been the interest of numerous studies [1, 2]. Although most studies have concentrated on N release from above-ground biomass, management of N release from roots may be much more important for crop nutrition, since up to 50–60% of total plant N of frequently pruned agroforestry trees may occur in roots [3].

Isotopic techniques are commonly applied for studying the fate of N in agroforestry systems and the mechanisms involved in its cycling. They can be especially useful for studying below-ground N cycling processes that are difficult to trace otherwise. The techniques require measurement or a reasonable estimate of the N isotopic compositions of the N sources. However, recent research suggests that estimating N cycling in intercropping systems with isotopic techniques becomes difficult after management interventions which affect root turnover. After shoot harvest or pruning of the N donor plant, isotopic ratio of the N recipient plants peaks rapidly within a few days and then decreases slowly over time. Such patterns were observed in different experimental setups and N donor species: in studies applying both 15N natural abundance and 15N enrichment methods for the legume tree Gliricidia sepium under field conditions [4, 5] and in pot culture [6], and in a study applying the 15N natural abundance method for the herbaceous legume Canavalia ensiformis in pot culture [7]. The similarity of the results from different setups suggests that the unexpected isotopic patterns of the N recipient plants represent a general phenomenon independent of N donor species and the 15N tracing technique applied.

A key reason for difficulties in analysing the fate of N from organic sources with isotopic techniques certainly is the nonuniform isotopic composition of these sources, which as such is well established. Discrimination against 15N in biochemical processes results in variation of 15N natural abundance between plant organs and their nitrogenous compounds, including soluble proteins, amino acids, and nitrate [8, 9]. Numerous 15N enrichment studies have also demonstrated that 15N labelling techniques and 15N allocation within the plant may result in different plant parts becoming very differently enriched with 15N [10, 11].

Isotopic variation between nitrogenous compounds of a plant could be further enhanced by their different decomposition rates, which are well known from residue decomposition research [12, 13] yet not frequently considered in 15N tracer studies. Together these two factors may result in considerable temporal variation in the isotopic patterns of N released from organic inputs (e.g., roots decomposing after pruning), which is subsequently reflected in isotopic composition of the N recipient pools such as soil compartments, soil microbial biomass, or associated N recipient crops. Studies applying modelling approaches suggest that isotopic composition of N released from decomposing roots of a legume tree differs from that of living roots just before pruning, both with 15N natural abundance and 15N enrichment methods [6] and that N uptake from 15N-enriched, surface-applied residues is best explained when decomposition rate constants are estimated separately for labile and stable residue fractions [14]. However, because isotopic heterogeneity is considered difficult to evaluate over time, N uptake or N transfer studies often assume that isotopic composition of an organic N source remains constant over the length of an experiment. Although homogeneity of 15N labelling has improved through methodological development (reviewed in [15]), it remains difficult to achieve over time, especially at compound level (cf. e.g. [10, 16]). Moreover, inconsistent isotopic signatures can impair N cycling estimates also in 15N natural abundance studies where isotopic compositions of N sources cannot be controlled [4, 6]. Alternative or complementary approaches to more reliably quantify N cycles in agroforestry systems are, therefore, needed.

Our objectives in this study were to (i) quantify how temporal variation in the isotopic composition of decomposing, 15N-enriched legume roots affects the estimates of the fate of recycled N and (ii) study how this variation could be accounted for by considering both the isotopic heterogeneity of residue compounds and their differential decomposition kinetics. To analyse the effects of heterogeneity of decomposing roots, we measured the isotopic patterns of a N recipient plant, which served as integrator of the biological processes involved in N cycling in the soil-plant system. A dynamic model on the isotopic composition of soil pools and the N recipient plant during residue decomposition was developed and then applied for testing the study assumptions against two experimental datasets. The objective of the simulations was not to find the actual parameter values for the specific residues and soil used in the experimental studies, but instead explore which overall patterns and general assumptions about residue characteristics were necessary to explain the observed temporal changes in isotopic composition of the N recipient plants. Options for improving experimental design and interpretation of results in 15N tracer studies and achieving more reliable estimates of the amount and fate of recycled N are then discussed.

2. Material and Methods

2.1. Experimental Design

Two separate datasets were used for evaluating the performance of the residue decomposition and fractionation model under different experimental designs. The first dataset was from a pot experiment where a fodder grass Dichanthium aristatum Poir C.E. Hubbard was grown alone in greenhouse, and 15N-enriched fine root and nodule residue of the legume tree Gliricidia sepium (Jacq.) Kunth ex Walp. were applied in the grass pots (residue application experiment, RA in the following). Evolution of the N isotopic signature (15N) of D. aristatum was then studied for 10 weeks. The experiment is described in the following. The second dataset was that described by Sierra et al. [6], where D. aristatum and G. sepium were grown together in pots in greenhouse (full interaction experiment, FI, in the following). Gliricidia sepium trees were labelled with 15N through foliar application and pruned 12 weeks thereafter to induce root turnover. Evolution of 15N of D. aristatum was then studied for 48 weeks. For this study we used the data for the first 24 weeks, after which the N source from the decomposing residue appeared exhausted [6]. Dichanthium aristatum shoots were harvested every 4–8 weeks during the FI experiment.

The RA experiment was conducted at the greenhouse facilities of the Antillean Research Centre of INRA, Guadeloupe, Lesser Antilles (16°12′N, 61°39′, 125 m a.s.l.). Soil and plant material originated from a cut-and-carry fodder production system of G. sepium and D. aristatum. The site was established in 1989 and managed thereafter by frequent tree pruning and grass cutting. The soil on the site is Vertisol with 80% of clay, pH of 7.8, organic carbon content 33.1 g kg−1, and organic N content 3.1 g kg−1 [6]. For detailed descriptions of the field site and the soil, see Daudin and Sierra [4] and Sierra et al. [17].

The experiment consisted of four potted G. sepium trees for providing the decomposing root and nodule residue, and eight pots of D. aristatum grass for studying the uptake of 15N mineralised from residue. Soil for the pot experiment was collected from the topsoil layer of the field site and sieved to <1 cm aggregates while removing plant roots. Mineral N content of the soil was analysed as described by Sierra et al. [17]. Gliricidia sepium trees were established from cuttings in nursery bags filled with the soil. One month after their propagation, the trees were transferred to pots of 14 L. Another two months later, D. aristatum swards were transplanted from the field site to similar pots as for the trees. Both series of pots were fertilised with 2 g of triple superphosphate and 2 g of K2SO4 at the time of planting and were irrigated daily throughout the experiment. Leaching was assumed negligible because soil inorganic N is mainly in the form of [18] and fixed on soil particles in the clayey soil [19]. Daily mean of air temperature in the greenhouse varied between 25.5 and 30.5°C, with a decreasing trend towards the end of the experiment.

The trees were labelled using foliar feeding of 99%  15N-enriched KNO3 four months after their propagation. The label was applied on tree leaves with a small paintbrush at three events in equal amounts with two day intervals, allowing time for absorption of the solution [6]. The total amount of 15N applied per tree was 30 mg. Foliar feeding was used because the study focused on N recycling from root and nodule turnover induced by pruning or shoot harvest of the N donor plant, common management methods in agroforestry and other intercropping systems. Labelling of above-ground plant parts with 15N is the most feasible 15N labelling methods to study N dynamics related to management practices in such systems. Foliar feeding also enabled comparison of the results with those of a larger study where various below-ground N transfer pathways between N donor and recipient plants were studied applying foliar feeding of 15N [10].

Application of fine root residue took place when two months had passed since grass transplanting in the pots, and three weeks since tree 15N-labelling. The grass shoots were first cut to approximately 2 cm height. The aim was to homogenise the initial situation and to minimise the dilution of assimilated 15N within grass biomass. Cutting was also consistent with grass management at the field site, where the grass is customarily cut every 40–50 days. After grass cutting, the trees were harvested for collecting fine roots (<2 mm of diameter) and nodules which were not detached from the roots. Coarse roots were not used because they decompose slowly, and roots recycled after tree pruning are mainly fine roots [20]. Twelve holes were then carefully drilled in the soil of the grass pots, and fresh fine roots and nodules mixed with a small amount of soil were applied in the holes. The residue was not mixed homogeneously within the soil in order to avoid destructing the grass. Each pot received 3.2 ± 0.6 g of residue, which corresponded to approximately half of the fine root and nodule mass of the trees. All remaining residue material was weighed, oven-dried at 70°C for 72 h, and ground to <0.2 mm for isotopic analysis and for determining the relation of residue fresh and dry weight. Coarse roots were analysed separately.

Grass shoots were sampled for 15N analysis immediately before residue application, and on days 28, 49, and 70 following it. Sampling was limited to these four events in order to avoid excessive disturbance of grass growth. At the end of the experiment on day 70, the grass was harvested and compartmented to shoot, stubble, and roots. All plant material was weighed, dried, and ground for isotopic analysis as described above. Characteristics of the labile and stable fractions were determined through model simulations as explained below. Biochemical properties of residue generally fail to predict N and C dynamics during residue decomposition [21], and the labile and stable fractions of residues are currently determined from laboratory incubations using a model to fit experimental data (e.g., [22]).

Sample N contents and isotopic ratios were determined at the Stable Isotope Facility of the University of California-Davis, US, using an element analyser interfaced to an isotope ratio mass spectrometer (Europa Integra CN; Sercon Ltd., Cheshire, UK).

The differences between the isotopic signatures (15N) of shoot subsample and shoot total biomass at the end of the experiment were tested with Student’s -test, in order to estimate the sampling error for the isotopic signatures during the experiment. Correlation between grass shoot 15N and N concentration during the experiment was calculated as described by Hamlett et al. [23], taking into account that values of the variables were obtained as repeated measures of individual plants and were linked over time. Correlations were expressed as Pearson’s correlation coefficients. Values of were interpreted as indicating statistically significant differences. Results were analysed with SAS statistical analysis software, version 9.1 (SAS Institute Inc., Cary, NC, USA).

2.2. Model Structure

A dynamic model was developed for simulating N mineralisation from decomposing 15N-labelled residue, its subsequent uptake by an adjacent plant, and the plant’s isotopic signature (Figure 1). Residue decomposition was modelled according to the STICS residue decomposition model [24]. The STICS model considers a single fraction for the residue, which decomposes with first order kinetics at a rate determined by its C : N ratio. It has reasonably well-explained overall C and N mineralisation from plant residues in tropical environments [25], including for roots of G. sepium [7]. It can also be easily combined with crop or ecosystem models, as it only requires residue C : N ratio as input [26].

506302.fig.001
Figure 1: Model of residue decomposition and subsequent uptake of residue N by a recipient plant. Each pool is divided into 14N and 15N. Subscripts R1 and R2, B1 and B2, and H1 and H2 refer to the pools related to the two residue fractions in the residue, microbial biomass, and humified organic matter, respectively. Parameters k and are decomposition rate constants for residue and microbial biomass, respectively, Y is assimilation yield of C from residue by microbial biomass, and h is humification rate of microbial biomass. Parameter u is N uptake rate by the recipient plant, and its subscripts SR and SN refer to the uptake rates from the two soil inorganic pools (Table 2). Parameter Sf is N source factor which determines the relative N uptake rates from the two soil N pools. Parameter is mineralisation rate for soil organic N. Pools and processes indicated with bold lines are simulated according to Nicolardot et al. [24].

The model developed in this study applies a modified decomposition component with regard to STICS, in order to allow simulation of heterogeneity of residue and its isotopic composition, and the implications to soil and recipient plant in the system. Residue is divided into fractions, and 15N, N content, and C : N ratio can be specified individually for each of these. Uptake of N and its partitioning within the recipient plant was modelled according to a box model [18] which has explained well biomass accumulation and 15N in D. aristatum shoots [6, 18]. The final model consisted of seven types of N pools, namely, residue (NR), microbial biomass involved in decomposition (NB), humified organic matter (NH), soil inorganic N originating from residue (NSR), soil native inorganic N (NSN), and plant roots (NRO) and shoot (NS). All pools were divided to 14N and 15N according to their initial 15N values. Isotopic signatures of N flows were defined by 15N of the source pool.

All residue fractions and the associated microbial biomass pools were set to either deplete or contribute to the same soil inorganic N pool, depending on the C : N ratios of each residue fraction [24]. The model enables simulation of differential N uptake from the NSR and NSN pools by the recipient plant, by including a N source factor, Sf: where u, , and are total N uptake rate of soil N, and N uptake rates from NSR and NSN pools, respectively. Value of Sf in the model is fixed over time. If Sf > 0, then N is first absorbed from the NSR pool and only secondarily from NSN, if supply from NSR is insufficient. If Sf < 0, then N is first absorbed from the NSN pool.

Net mineralisation of soil native organic N () was simulated as a function of soil temperature according to the equation proposed for the same soil and site as in the experimental datasets by Sierra et al. [17]: Because NSN represents net N mineralisation from soil organic matter, it takes into account N immobilisation during mineralisation. However, further N immobilisation from NSN may occur when N demand for growth of the microbial biomass decomposing residues is greater than N supplied from NSR (Figure 1) [24]. Therefore, growth of microbial biomass decomposing residues is primarily supported by organic N in residues, then by NSR and then by NSN.

Ambient temperature and soil humidity affect mineralisation of both residue and microbial N pools [24] and soil native organic N. Mineralisation of humified organic matter originating from the residue was considered negligible as N source for the recipient plant, in comparison to the decomposing residue and soil inorganic N (cf. [27]). Approximately 95–98% of inorganic N in the soil used for the experimental studies is in the form of , possibly because some tropical grasses, including D. aristatum, release compounds which reduce the populations or activity of nitrifying bacteria [18, 28]. Negligible volatilisation of NH3 has been observed in a non-N-fertilised Vertisol similar to that used in the experimental studies [29], and frequent irrigation was assumed to further reduce concentration of NH4 in the soil. Therefore, denitrification and volatilisation were excluded from the model.

Daily uptake of soil inorganic N by the recipient plant was modelled as linear over the time span of the experiments and considering that grass shoots were regularly harvested in the FI experiment. Linear N uptake was based on the results of the FI experiment and previous observations for the N recipient grass [6, 30].

2.3. Calculations of N Uptake by the Recipient Plant

The enrichment of a sample with 15N in relation to the atmospheric standard is expressed as where the subscripts sa and atm refer to the sample and the atmospheric 15N atom-% (0.3663%), respectively. Proportion of N derived from residue (%Ndfr) per total N of the recipient plant was calculated as Subscripts P, SN, and SR refer to the recipient plant and to N derived from soil native N and residue, respectively, 0 to the initial values before 15N enrichment, and t to the point in time. Amount of N derived from residue () in the recipient plant was expressed as where denotes plant N content. Proportion of residue N uptaken by the recipient plant (%) was obtained from where NR is the total N content of the residue.

2.4. Model Parameterisation and Simulations

We assumed that the fine root residue consisted of two fractions, labile and stable. The following constraints were imposed for the labile fraction: (i)  the minimum C : N ratio was set to 3.0, and the maximum N content to 30% of total residue N. These are in the range of values measured for water-soluble N in plant roots [22, 31, 32]. (ii)  The maximum 15N was set to 600 for the RA experiment, and to 495 for the FI experiment. With the maximum N content and respective 15N of the labile residue fraction, 15N of the stable fraction would equal to 1 in each dataset.

Input parameters for total N content, C : N ratio, and 15N of the residue in the RA experiment were the values measured for fine roots and nodules. In the FI experiment the input parameters corresponded to the amount of N recycled after pruning, and total C : N ratio and 15N of roots (including nodules; Table 1). Rate constant equations were common for the two datasets (Table 2). Temperature in the simulations was daily mean temperature in the greenhouse during the experimental studies, and soil humidity was set equal to field capacity as the grass pots were irrigated daily.

tab1
Table 1: Parameters of the residue decomposition model for the residue application (RA) and full interaction (FI) experiments. Where 15N is not given, it is defined by 15N of the source pool.
tab2
Table 2: Common parameters in the residue decomposition model for the two experimental datasets.

The model was run for a period of 70 days for the RA experiment and 168 days for the FI experiment. Three simulation steps were performed for each dataset to evaluate the assumptions about residue characteristics. The steps considered (i) one single residue fraction, that is homogenous residue, (ii) two residue fractions with equal isotopic composition, and (iii) two residue fractions with different isotopic composition. The aim was to test the hypotheses that simulating the observed trends of 15N of the N recipient plants requires more than one residue fraction and that the fractions differ in terms of 15N. Each simulation step was run with two options for the N source factor: N uptake proportionally equal to the size of each inorganic N pool (Sf = 0, (1)), and optimising the factor in the simulation. In simulation step 3 for the RA experiment, N content of the labile fraction was fixed to 29.0, the value obtained in all previous simulations, in order to limit the number of optimised parameters.

Agreement with simulations and experimental observations was evaluated using the coefficient of variation of the root mean square error [34]: and values <0.05 were defined to represent satisfactory model performance. Sensitivity of the model outputs was then studied by varying the values of input parameters and analysing the subsequent changes in the model outputs.

After finding the parameter values which allowed satisfactory simulation of the experimental observations, a fourth simulation step was performed to study the impacts of isotopic heterogeneity of decomposing residue on the N uptake estimates by the N recipient plant. Nitrogen uptake from the residue was estimated according to (5), using two options for 15N of N derived from the residue: (i) total 15N of the residue at the time of residue application, which corresponds to the assumption of a homogenous residue over time (uptake denoted %), and (ii) simulated 15N of N derived from residue (%NdfrS), which can vary over time if residue fractions are differently enriched with 15N and decompose at different rates.

The model was built using the Simile software, version 4.9 (Simulistics Ltd., Edinburgh, UK). Optimal fit was searched by using the PEST software (Model-Independent Parameter Estimation and Uncertainty Analysis, version 11.8; Watermark Numerical Computing, Australia). For detailed description of the software and the optimisation method, see the PEST manual [35].

3. Results

3.1. Experimental Data of the RA Experiment

Fine root and nodule residue of G. sepium applied in the grass pots had C : N ratio of 14.5 ± 0.6, N content of 97.1 ± 2.0 mg pot−1, N concentration of 3.1 ± 0.1%, and 15N of 180 ± 15. Shoot 15N of the N recipient grass increased rapidly at the beginning of the experiment from 4.3 to 33.8 ± 2.1 at the end of the first study period and decreased thereafter during the last two periods (Figure 2). The coefficient of variation for shoot 15N varied between 12 and 22% during the experiment. Grass shoot N concentration peaked simultaneously with shoot 15N (Figure 2) and correlated statistically significantly with it (, ; ). The values of 15N of shoot subsamples did not differ statistically significantly from 15N of total shoot biomass at the end of the experiment (23.5 ± 1.8 versus 21.7 ± 1.9, resp.). Value of 15N of coarse roots of G. sepium at the time of root harvest and residue application was 499 ± 38.

506302.fig.002
Figure 2: Observed change of 15N and N concentration in grass shoots during the RA experiment. Vertical bars indicate the standard error of mean ().
3.2. Simulations

Simulations with a single residue fraction resulted in low 15N values for the N recipient plant in both experimental datasets (Sf > 0, Figures 3(a), and 3(b)). Neither the trends nor the ranges of plant shoot 15N corresponded to the observations. When the residues were divided into two fractions with different C : N but equal 15N, a decreasing trend of shoot 15N after an initial peak was obtained for both datasets, as also observed in the experimental studies. However, simulations deviated too much from the observations for satisfactory results (Figures 3(c), and 3(d); Table 3). Best agreement with observations for both datasets was reached when 15N of the two residue fractions were allowed to differ (Sf > 0, Figures 3(e), and 3(f); Table 3). Characteristics of the residue fractions in those simulations are given in Table 3.

tab3
Table 3: Simulation steps, optimal parameter values and performance of the residue decomposition model. Values indicated in bold were obtained through model optimisation; other values were fixed in the respective simulation steps. Residue parameter values are for labile fraction, except for the final simulation step for which values are given for both labile and stable fraction. For CV (RMSE), asterisk (*) indicates statistically significant fit between observed and simulated δ15N values of the N recipient plant (<0.05).
fig3
Figure 3: Simulated change of plant 15N after input of 15N-labelled decomposing residue in the residue application (RA; a, c, e) and full interaction experiments (FI; b, d, f), when the residue is considered to consist of (a, b) a single fraction with uniform C : N and 15N, (c, d) two fractions with different C : N but uniform 15N, and (e, f) two fractions with different C : N and 15N. Simulations were conducted with proportionally equal N uptake from soil inorganic N sources (Sf = 0) and by optimising the N source factor in the simulations. Vertical bars indicate the standard error of mean for the observations ( and for the RA and FI experiments, resp.). For explanations and parameter values, see text and Table 2.

When N uptake was considered proportional to the size of each soil inorganic N pool (i.e., Sf = 0), plant shoot 15N remained low in all simulations for both experimental datasets, regardless of the number of residue fractions and their 15N values. When the N source factor was optimised in the simulations, best fit for both datasets was obtained when proportionally more N was absorbed from the inorganic N pool originating from residue, in comparison to soil native inorganic N (Sf > 0; Table 3).

According to the simulations, 10.3% of N in the N recipient plant originated from the residue at the end of the RA experiment (Figure 4(a)). This corresponded to 25 mg of N and 26% of initial residue N. Nitrogen uptake estimates calculated assuming homogenous residue (%NdfrR) differed only slightly from the simulated N uptake (%NdfrS), except for the first week after residue application (Figure 4(a)). In the FI experiment N originating from decomposing residue constituted 8.1% of N of the recipient plant 25 days after tree pruning (Figure 4(b)), corresponding to 12 mg and 3% of initial residue N. Nitrogen originating from residue rapidly diluted in the plant biomass thereafter as a result of frequent shoot harvests. Nitrogen uptake estimates calculated assuming homogenous residue were 40–51% lower than simulated N uptake during the first 10 days, and 36–39% lower during the rest of the experiment, with the difference slowly decreasing with time (Figure 4(b)).

fig4
Figure 4: Effect of 15N of residue on N uptake estimates (proportion of plant N derived from residue) in (a) the residue application experiment (RA), and (b) the full interaction experiment (FI). %NdfrR: calculated with total initial 15N of residue, and %NdfrS: calculated with simulated 15N of N released from residue (%NdfrS). Arrows indicate shoot harvesting in the FI experiment.

The model showed a short period of net N immobilisation between days 6 and 16 in the RA experiment. At the end of the simulations, all residue N would have decomposed, and 40% and 34% of it remained in soil microbial biomass and humified organic matter, respectively (Figure 5(a)). In the FI experiment the simulations indicated net N immobilisation from day 0 to day 47. Mineralisation of SOM exceeded immobilisation from day 12 onwards. At the end of the experiment all initial residue N would have decomposed, and 17% of it remained in soil microbial biomass and 76% in humified organic matter (Figure 5(b)). Values presented in Figure 5 correspond to total N content of each pool and include N immobilisation from NSN in soil microbial biomass and humified organic matter.

fig5
Figure 5: Simulated evolution of the N pools in (a) the residue application experiment (RA), and (b) the full interaction experiment (FI). Note that inorganic N from the residue (NSR) remains close to zero for most of the time in both experiments as it is quickly absorbed by plant or microbial biomass.

The sensitivity analyses for the RA experiment indicated that model output (15N of the N recipient plant) was sensitive to C : N ratio of the decomposing microbial biomass (RB), especially during the initial phase of residue decomposition, and to N uptake rate of the recipient plant (u), especially towards the end of the experiment (Figures 6(b), and 6(d)). The model output was negligibly affected by changes in the mineralisation rate constant of soil organic matter (; data not shown), decomposition rate constant of residue (k), and N source factor (Sf; Figures 6(a), and 6(c)) within the tested range of values. Relative values of k for the two residue fractions were retained the same in the simulations, that is, 15N of N released from the residue was not altered. The model was very sensitive to changes in Sf values between 0.0 and approximately 0.3 (data not shown) but was little affected by changes at higher Sf values because of the rapid decomposition and exhaustion of N from the labile residue fraction.

fig6
Figure 6: Sensitivity analysis for the residue application experiment (RA). Change in 15N of the N recipient plant with the change of (a) residue decomposition rate constant, k, (b) C : N of microbial biomass involved in residue decomposition, RB, (c) N source factor for the N recipient plant, Sf, and (d) N uptake rate by the N recipient plant. The actual values in time were varied between −25% and +25% of the values with which the model was optimised. Exact values are given for those parameters which have a fixed value in the model (c, d).

4. Discussion

4.1. Residue Characteristics

We observed a rapid initial peak and a slow subsequent decrease in the isotopic composition of the N recipient plant after incorporation of organic residues in the RA experiment, similar to that obtained in other studies applying both 15N natural abundance and 15N enrichment methods [47, 14]. The simulation results show that the isotopic patterns of the N recipient plants may be explained by the combined effects of a rapid initial release of N from a labile residue fraction, and its differential isotopic composition with regard to total residue. It is noteworthy that the temporal pattern of the 15N content of the recipient plant was not affected by the fact that in the RA experiment the tree roots were harvested for residue application 3 weeks after 15N labelling, whereas in the FI experiment tree pruning which induced root decomposition occurred 12 weeks after labelling. Moreover, the fact that similar trends in the isotopic composition of the N recipient plant were observed in studies applying 15N natural abundance [6, 7] indicates that the observed trends cannot be explained by the method of 15N labelling alone.

The results are in line with the observations of Sierra et al. [6] who showed that 15N of N released from decomposing roots may not correspond to that measured from the living roots just before pruning. By fractioning 15N-enriched residue into labile and stable components, Hadas et al. [14] managed to simulate fairly well the isotopic composition of the N recipient plant, except for its simulated initial peak soon after residue application. They measured isotopic composition of the recipient plant first time 47 days after residue application. This may have concealed the role of isotopic heterogeneity of the residue fractions, which became evident in the RA experiment involving earlier measurements. Recently, it was shown that nitrate content and isotopic fractionation during metabolic processes explain variation in 15N of nitrogenous compounds in rapeseed leaves (Brassica napus L.), where nitrate and amino acids were considerably more enriched than soluble proteins [9]. While fractionation is not of relevance in 15N labelling studies, the results demonstrate that N allocation and plant metabolic processes are capable of considerably altering 15N of nitrogenous compounds and that of residue fractions as a consequence.

The fact that satisfactory simulation of the observations for both experimental datasets in this study was reached with low C : N values of the labile fraction of 3.3 suggests that the fraction consisted of water-soluble components, for example amino acids and possibly inorganic N [31, 32]. Simulated proportion of labile N was much smaller in the FI experiment than in the RA experiment because residue characteristics in the FI experiment corresponded to those of total roots instead of fine roots only. Total N concentration of the root residue in the FI experiment also was approximately a third lower than in the RA experiment (data not shown).

Labile residue fraction appeared more enriched than the stable fraction in the RA experiment, but the opposite was true for the FI experiment in which 15N labelling of the trees took place much earlier. The 15N label in plants may initially occur as metabolically active N or be stored predominantly in coarse roots. In contrast, it can be assumed that N bound in structural components during their formation is generally not replaced over their life time. Therefore, labelling would not affect the isotopic composition of these components, if they were formed before 15N labelling, as presumably was the case for majority of the fine roots of G. sepium in the RA experiment. Relative 15N enrichment of root fractions in the RA experiment (coarse roots > labile fraction of fine roots > stable fraction of fine roots) is in line with these assumptions about the fate of applied 15N within the plant. Relative 15N enrichment of the fractions can be assumed to change over time, as metabolically active and stored 15N are converted to structural components during biomass growth. Differential 15N enrichment in the FI experiment may also partly be explained by the fact that stable fraction corresponded to both the structural components of the fine roots, and coarse roots as a whole, including stored 15N. Method of 15N labelling may importantly affect the relative 15N enrichment of residue fractions [15]. An early start of 15N labelling would help to enrich structural components of the plant more evenly, but may in turn disturb the system more through repeated interventions and need of higher 15N inputs.

4.2. Nitrogen Availability from the Residue to the Associated Plant

The observed correlation between shoot N concentration and 15N of the N recipient grass after the incorporation of 15N-labelled residue in the RA experiment indicates that grass obtained proportionally more N from decomposing residue than from soil native N pools. This was supported by the results of the modelling study, where proportionally higher N uptake of N mineralised from residue compared to soil native N was necessary for obtaining a satisfactory agreement with the experimental observations (Figure 3). Residue application appeared, therefore, to enhance grass N nutrition instead of merely substituting soil native N as N source (cf. [36]).

Plant N uptake is affected by the availability of N in the proximity of roots [37]. The physicochemical properties of the very clayey soil in the experimental studies restrict solute flow [38, 39]. Therefore, it can be assumed that spots of high N concentration emerged in soil as a result of root decay, and that the N recipient plant absorbed proportionally more N from these spots. Such spots are also more porous as the surrounding soil, which may have facilitated colonisation by grass roots and subsequent uptake of N originating from residue [40, 41]. Soil in the holes where the residue in the RA experiment was applied was more loosely packed than the elsewhere in the pots, which may have simulated the effect of increased soil porosity. Small diameter of the holes, and the fact that grass roots had already effectively colonised the pots, likely contributed to the rapid capture of N from the residue. In the FI experiment roots of G. sepium and D. aristatum were mixed together in the pots, and N release from decomposing root residue could be assumed to have been spatially more homogenous than in the RA experiment. Intraspecific competition for soil N could explain why according to the simulations grass would have absorbed proportionally more N originating from residue than in the RA experiment. Grasses are generally more effective competitors for soil N than legumes [42]. Although heterogenous distribution of residue influences N uptake from decomposing roots also in natural systems [41], it was probably more pronounced in the experimental datasets because of experimental factors. The differences in N uptake between the two sources do not, however, affect the results about residue fractionation and distinctive 15N enrichment of labile and stable residue fractions discussed above (cf. Figures 3(e), and 3(f)).

According to the simulations, the N recipient plant had obtained 26% of initial residue N at the end of the RA experiment, and 3% of initial residue N during the first 25 days of the FI experiment. These were on the lower side of the values measured in previous studies, which range from 10 to 100% of total residue N taken up by the associated plants [4346]. Variation in the observations may result for example from residue type and method of application, biological or environmental conditions facilitating or restricting decomposition, capacity of the soil to sequester organic inputs, synchrony with the nutrient needs of the recipient plant, methods for estimating N uptake, and the length of the observation period. Nitrogen from below-ground residues appears to decompose initially faster than that of above-ground residues, and more of it is also taken up by subsequent crops [12, 43, 47]. Very little research, however, has focused on recycling of plant below-ground N, especially in tropical agroecosystems.

4.3. Implications of the Findings to N Cycling Estimates

Quantification of N cycling in agroforestry systems using isotope techniques crucially depends on the determination of the isotopic signature of the N sources. We used simulation methods to evaluate how the common assumption of homogenous isotopic composition and decomposition rates of organic N sources affects estimates of N recycling from decomposing residue, when the residue fractions actually differ in both parameters. Accuracy of the estimates depends on the quality of the residue, relative sizes, and 15N enrichment of its fractions. Nitrogen recycling estimates may not be largely affected by isotopic heterogeneity when only relatively high-quality residues (e.g. fine roots of legume trees) are concerned, which decompose entirely over a short time and N originating from the residue fractions become effectively mixed in the system. In contrast, substantial errors in N recycling estimates can be obtained for lower-quality residues with large, differentially 15N-enriched stable fractions, if isotopic composition of released N is assumed constant. Role of the residues in crop N nutrition in such cases may be underestimated as was the case for the FI experiment where the labile residue fraction was less enriched with 15N than the stable fraction. Depending on the timing or method of 15N labelling, N recycling might also be overestimated if labile residue fraction is more enriched with 15N than the stable fraction (cf. the RA experiment).

While homogenous 15N labelling of the N source is commonly assumed in 15N enrichment studies, it is generally difficult to achieve over time. Assimilation of N from soil and its allocation within the plant during growth, and in legumes also N2-fixation, result in dilution of the label at different rates for different organs [16, 48]. Moreover, in 15N natural abundance studies the isotopic composition of the N sources cannot be controlled [9]. Analysis of temporal isotopic variation of the N sources and reasons affecting it could provide an alternative or complementary method aiming at, or simply assuming, homogenous isotopic composition in 15N tracer studies. Measuring 15N and N content of labile and stable residue fractions separately can be a useful first step in evaluating whether isotopic heterogeneity is likely to result in too small or large estimates for N recycling from organic sources in the concerned system. Water-soluble N may be a good approximation for the labile fraction for this purpose (cf. [2, 31]).

If no information of 15N of residue fractions in 15N enrichment studies is available, caution should be exerted in estimating N recycling and uptake from the 15N values of the recipient pools in the short term. This is especially the case after discrete events in time such as pruning or green manure application, or under pronounced seasonality, if there are large inputs of residue from senescing biomass over a short-time period. Previous studies indicate that these considerations are important also for studies using the 15N natural abundance method [4, 6, 7, 9], and the results explained here reveal that the role of residue fractionation in explaining isotopic patterns in 15N natural abundance studies merits research.

Acknowledgments

The authors thank Dr. Pekka Nygren for his valuable comments on the paper, and Saint-Ange Sophie for his skillful technical assistance during the experiment. They also thank two anonymous reviewers for their valuable comments. Contribution of R. Jalonen was financed by the Academy of Finland (Grants 111796 and 129166). The authors declare no conflict of interests.

References

  1. K. E. Giller, Nitrogen Fixation in Tropical Cropping Systems, CABI, Wallingford, UK, 2001.
  2. P. L. Mafongoya, K. E. Giller, and C. A. Palm, “Decomposition and nitrogen release patterns of tree prunings and litter,” Agroforestry Systems, vol. 38, no. 1–3, pp. 77–97, 1998. View at Scopus
  3. N. Sanginga, B. Vanlauwe, and S. K. A. Danso, “Management of biological N2 fixation in alley cropping systems: estimation and contribution to N balance,” Plant and Soil, vol. 174, no. 1-2, pp. 119–141, 1995. View at Publisher · View at Google Scholar · View at Scopus
  4. D. Daudin and J. Sierra, “Spatial and temporal variation of below-ground N transfer from a leguminous tree to an associated grass in an agroforestry system,” Agriculture, Ecosystems and Environment, vol. 126, no. 3-4, pp. 275–280, 2008. View at Publisher · View at Google Scholar · View at Scopus
  5. J. Sierra and D. Daudin, “Limited 15N transfer from stem-labeled leguminous trees to associated grass in an agroforestry system,” European Journal of Agronomy, vol. 32, no. 3, pp. 240–242, 2010. View at Publisher · View at Google Scholar · View at Scopus
  6. J. Sierra, D. Daudin, A. M. Domenach, P. Nygren, and L. Desfontaines, “Nitrogen transfer from a legume tree to the associated grass estimated by the isotopic signature of tree root exudates: a comparison of the 15N leaf feeding and natural 15N abundance methods,” European Journal of Agronomy, vol. 27, no. 2–4, pp. 178–186, 2007. View at Publisher · View at Google Scholar · View at Scopus
  7. J. Sierra and N. Motisi, “Shift in C and N humification during legume litter decomposition in an acid tropical ferralsol,” Soil Research, vol. 50, no. 5, pp. 380–389, 2012. View at Publisher · View at Google Scholar
  8. P. Högberg, “Tansley review no. 95 natural abundance in soil-plant systems,” New Phytologist, vol. 137, no. 2, pp. 179–203, 1997. View at Publisher · View at Google Scholar · View at Scopus
  9. P. P. G. Gauthier, M. Lamothe, A. Mahé et al., “Metabolic origin of δ15N values in nitrogenous compounds from Brassica napus L. leaves,” Plant, Cell and Environment. In press. View at Publisher · View at Google Scholar · View at Scopus
  10. R. Jalonen, P. Nygren, and J. Sierra, “Transfer of nitrogen from a tropical legume tree to an associated fodder grass via root exudation and common mycelial networks,” Plant, Cell and Environment, vol. 32, no. 10, pp. 1366–1376, 2009. View at Publisher · View at Google Scholar · View at Scopus
  11. W. D. F. Khan, M. B. Peoples, and D. F. Herridge, “Quantifying below-ground nitrogen of legumes: 1. Optimising procedures for 15N shoot-labelling,” Plant and Soil, vol. 245, no. 2, pp. 327–334, 2002. View at Publisher · View at Google Scholar · View at Scopus
  12. E. C. Adair, W. J. Parton, S. J. Del Grosso et al., “Simple three-pool model accurately describes patterns of long-term litter decomposition in diverse climates,” Global Change Biology, vol. 14, no. 11, pp. 2636–2660, 2008. View at Publisher · View at Google Scholar · View at Scopus
  13. T. M. Henriksen and T. A. Breland, “Evaluation of criteria for describing crop residue degradability in a model of carbon and nitrogen turnover in soil,” Soil Biology and Biochemistry, vol. 31, no. 8, pp. 1135–1149, 1999. View at Publisher · View at Google Scholar · View at Scopus
  14. A. Hadas, T. A. Doane, A. W. Kramer, C. Van Kessel, and W. R. Horwath, “Modelling the turnover of 15N-labelled fertilizer and cover crop in soil and its recovery by maize,” European Journal of Soil Science, vol. 53, no. 4, pp. 541–552, 2002. View at Publisher · View at Google Scholar · View at Scopus
  15. F. Wichern, E. Eberhardt, J. Mayer, R. G. Joergensen, and T. Müller, “Nitrogen rhizodeposition in agricultural crops: methods, estimates and future prospects,” Soil Biology and Biochemistry, vol. 40, no. 1, pp. 30–48, 2008. View at Publisher · View at Google Scholar · View at Scopus
  16. D. F. Khan, M. B. Peoples, P. M. Chalk, and D. F. Herridge, “Quantifying below-ground nitrogen of legumes. 2. A comparison of 15N and non isotopic methods,” Plant and Soil, vol. 239, no. 2, pp. 277–289, 2002. View at Publisher · View at Google Scholar · View at Scopus
  17. J. Sierra, M. Dulormne, and L. Desfontaines, “Soil nitrogen as affected by Gliricidia sepium in a silvopastoral system in Guadeloupe, French Antilles,” Agroforestry Systems, vol. 54, no. 2, pp. 87–97, 2002. View at Publisher · View at Google Scholar · View at Scopus
  18. J. Sierra and P. Nygren, “Transfer of N fixed by a legume tree to the associated grass in a tropical silvopastoral system,” Soil Biology and Biochemistry, vol. 38, no. 7, pp. 1893–1903, 2006. View at Publisher · View at Google Scholar · View at Scopus
  19. E. A. Paul and F. E. Clark, Soil Microbiology and Biochemistry, Academic Press, San Diego, Calif, USA, 2nd edition, 1996.
  20. G. Schroth and W. Zech, “Above- and below-ground biomass dynamics in a sole cropping and an alley cropping system with Gliricidia sepium in the semi-deciduous rainforest zone of West Africa,” Agroforestry Systems, vol. 31, no. 2, pp. 181–198, 1995. View at Scopus
  21. S. Abiven, S. Recous, V. Reyes, and R. Oliver, “Mineralisation of C and N from root, stem and leaf residues in soil and role of their biochemical quality,” Biology and Fertility of Soils, vol. 42, no. 2, pp. 119–128, 2005. View at Publisher · View at Google Scholar · View at Scopus
  22. L. S. Jensen, T. Salo, F. Palmason et al., “Influence of biochemical quality on C and N mineralisation from a broad variety of plant materials in soil,” Plant and Soil, vol. 273, no. 1-2, pp. 307–326, 2005. View at Publisher · View at Google Scholar · View at Scopus
  23. A. Hamlett, L. Ryan, and R. Wolfinger, “On the use of PROC MIXED to estimate correlation in the presence of repeated measures,” in Proceedings of the SAS SUGI Conference, Montréal, Canada, May 2004.
  24. B. Nicolardot, S. Recous, and B. Mary, “Simulation of C and N mineralisation during crop residue decomposition: a simple dynamic model based on the C:N ratio of the residues,” Plant and Soil, vol. 228, no. 1, pp. 83–103, 2001. View at Publisher · View at Google Scholar · View at Scopus
  25. L. Raphael, J. Sierra, S. Recous, H. Ozier-Lafontaine, and L. Desfontaines, “Soil turnover of crop residues from the banana (Musa AAA cv. Petite-Naine) mother plant and simultaneous uptake by the daughter plant of released nitrogen,” European Journal of Agronomy, vol. 38, no. 1, pp. 117–123, 2012. View at Publisher · View at Google Scholar · View at Scopus
  26. N. Brisson, C. Gary, E. Justes et al., “An overview of the crop model STICS,” European Journal of Agronomy, vol. 18, no. 3-4, pp. 309–332, 2003. View at Publisher · View at Google Scholar · View at Scopus
  27. N. Brisson, B. Mary, D. Ripoche et al., “STICS: a generic model for the simulation of crops and their water and nitrogen balances. I. Theory and parameterization applied to wheat and corn,” Agronomie, vol. 18, no. 5-6, pp. 311–346, 1998. View at Scopus
  28. T. Ishikawa, G. V. Subbarao, O. Ito, and K. Okada, “Suppression of nitrification and nitrous oxide emission by the tropical grass Brachiaria humidicola,” Plant and Soil, vol. 255, no. 1, pp. 413–419, 2003. View at Publisher · View at Google Scholar · View at Scopus
  29. N. Courtaillac, R. Baran, R. Oliver, H. Casabianca, and F. Ganry, “Efficiency of nitrogen fertilizer in the sugarcane-vertical system in Guadeloupe according to growth and ratoon age of the cane,” Nutrient Cycling in Agroecosystems, vol. 52, no. 1, pp. 9–17, 1998. View at Publisher · View at Google Scholar · View at Scopus
  30. P. Cruz, “Effect of shade on the carbon and nitrogen allocation in a perennial tropical grass, Dichanthium aristatum,” Journal of Experimental Botany, vol. 48, no. 306, pp. 15–24, 1997. View at Scopus
  31. G. D. Bending, M. K. Turner, and I. G. Burns, “Fate of nitrogen from crop residues as affected by biochemical quality and the microbial biomass,” Soil Biology and Biochemistry, vol. 30, no. 14, pp. 2055–2065, 1998. View at Publisher · View at Google Scholar · View at Scopus
  32. B. Chaves, S. De Neve, G. Hofman, P. Boeckx, and O. Van Cleemput, “Nitrogen mineralization of vegetable root residues and green manures as related to their (bio)chemical composition,” European Journal of Agronomy, vol. 21, no. 2, pp. 161–170, 2004. View at Publisher · View at Google Scholar · View at Scopus
  33. J. Sierra and P. Nygren, “Role of root inputs from a dinitrogen-fixing tree in soil carbon and nitrogen sequestration in a tropical agroforestry system,” Australian Journal of Soil Research, vol. 43, no. 5, pp. 667–675, 2005. View at Publisher · View at Google Scholar · View at Scopus
  34. D. G. Mayer and D. G. Butler, “Statistical validation,” Ecological Modelling, vol. 68, no. 1-2, pp. 21–32, 1993. View at Scopus
  35. PEST, Model-Independent Parameter Estimation. User Manual, Watermark Numerical Computing, Brisbane, Australia, 2004.
  36. D. S. Jenkinson, R. H. Fox, and J. H. Rayner, “Interactions between fertilizer nitrogen and soil nitrogen—the so-called “priming” effect,” Journal of Soil Science, vol. 36, no. 3, pp. 425–444, 1985.
  37. D. Loqué and N. Von Wirén, “Regulatory levels for the transport of ammonium in plant roots,” Journal of Experimental Botany, vol. 55, no. 401, pp. 1293–1305, 2004. View at Publisher · View at Google Scholar · View at Scopus
  38. Y. M. Cabidoche and P. Guillaume, “A casting method for the three-dimensionql analysis of the intraprism structural pores in vertisols,” European Journal of Soil Science, vol. 49, no. 2, pp. 187–196, 1998. View at Publisher · View at Google Scholar · View at Scopus
  39. R. Jalonen, P. Nygren, and J. Sierra, “Root exudates of a legume tree as a nitrogen source for a tropical fodder grass,” Nutrient Cycling in Agroecosystems, vol. 85, no. 2, pp. 203–213, 2009. View at Publisher · View at Google Scholar · View at Scopus
  40. W. Jingguo and L. R. Bakken, “Competition for nitrogen during decomposition of plant residues in soil: effect of spatial placement of N-rich and N-poor plant residues,” Soil Biology and Biochemistry, vol. 29, no. 2, pp. 153–162, 1997. View at Publisher · View at Google Scholar · View at Scopus
  41. P. Hinsinger, A. G. Bengough, D. Vetterlein, and I. M. Young, “Rhizosphere: biophysics, biogeochemistry and ecological relevance,” Plant and Soil, vol. 321, no. 1-2, pp. 117–152, 2009. View at Publisher · View at Google Scholar · View at Scopus
  42. S. F. Ledgard and K. W. Steele, “Biological nitrogen fixation in mixed legume/grass pastures,” Plant and Soil, vol. 141, no. 1-2, pp. 137–153, 1992. View at Publisher · View at Google Scholar · View at Scopus
  43. R. Chintu and A. R. Zaharah, “Nitrogen uptake of maize (Zea mays. L) from isotope-labeled biomass of Paraserianthes falcataria grown under controlled conditions,” Agroforestry Systems, vol. 57, no. 2, pp. 101–107, 2003. View at Publisher · View at Google Scholar · View at Scopus
  44. O. B. Hesterman, M. P. Russelle, C. C. Sheaffer, and G. H. Heichel, “Nitrogen utilization from fertilizer and legume residues in legume-corn rotations,” Agronomy Journal, vol. 79, no. 4, pp. 726–731, 1987.
  45. E. S. Jensen, “Availability of nitrogen in 15N-labelled mature pea residues to subsequent crops in the field,” Soil Biology and Biochemistry, vol. 26, no. 4, pp. 465–472, 1994. View at Scopus
  46. C. A. Palm, “Contribution of agroforestry trees to nutrient requirements of intercropped plants,” Agroforestry Systems, vol. 30, no. 1-2, pp. 105–124, 1995. View at Publisher · View at Google Scholar · View at Scopus
  47. I. K. Thomsen, J. M. Oades, and M. Amato, “Turnover of 15N in undisturbed root systems and plants materials added to three soils,” Soil Biology and Biochemistry, vol. 28, no. 10-11, pp. 1333–1339, 1996. View at Publisher · View at Google Scholar · View at Scopus
  48. A. M. McNeill, C. Zhu, and I. R. P. Fillery, “Use of in situ15N-labelling to estimate the total below-ground nitrogen of pasture legumes in intact soil-plant systems,” Australian Journal of Agriculture Research, vol. 48, no. 3, pp. 295–304, 1997. View at Scopus