• Open Access

Miscanthus × giganteus leaf senescence, decomposition and C and N inputs to soil

Authors

  • Norbert Amougou,

    1. INRA, UMR614 Fractionnement des Agro Ressources et Environnement, Reims, France
    2. Université de Reims Champagne-Ardenne, UMR614 Fractionnement des Agro Ressources et Environnement, F-51100 Reims, France
    Search for more papers by this author
  • Isabelle Bertrand,

    1. INRA, UMR614 Fractionnement des Agro Ressources et Environnement, Reims, France
    2. Université de Reims Champagne-Ardenne, UMR614 Fractionnement des Agro Ressources et Environnement, F-51100 Reims, France
    Search for more papers by this author
  • Stephane Cadoux,

    1. INRA, US1158 Agro-Impact, Laon, France
    Search for more papers by this author
  • Sylvie Recous

    Corresponding author
    1. INRA, UMR614 Fractionnement des Agro Ressources et Environnement, Reims, France
    2. Université de Reims Champagne-Ardenne, UMR614 Fractionnement des Agro Ressources et Environnement, F-51100 Reims, France
    Search for more papers by this author

Abstract

Energy crops are currently promoted as potential sources of alternative energy that can help mitigate the climate change caused by greenhouse gases (GHGs). The perennial crop Miscanthus × giganteus is considered promising due to its high potential for biomass production under conditions of low input. However, to assess its potential for GHG mitigation, a better quantification of the crop's contribution to soil organic matter recycling under various management systems is needed. The aim of this work was to study the effect of abscised leaves on carbon (C) and nitrogen (N) recycling in a Miscanthus plantation. The dynamics of senescent leaf fall, the rate of leaf decomposition (using a litter bag approach) and the leaf accumulation at the soil surface were tracked over two 1-year periods under field conditions in Northern France. The fallen leaves represented an average yearly input of 1.40 Mg C ha−1 and 16 kg N ha−1. The abscised leaves lost approximately 54% of their initial mass in 1 year due to decomposition; the remaining mass, accumulated as a mulch layer at the soil surface, was equivalent to 7 Mg dry matter (DM) ha−1 5 years after planting. Based on the estimated annual leaf-C recycling rate and a stabilization rate of 35% of the added C, the annual contribution of the senescent leaves to the soil C was estimated to be approximately 0.50 Mg C ha−1yr−1 or 10 Mg C ha−1 total over the 20-year lifespan of a Miscanthus crop. This finding suggested that for Miscanthus, the abscised leaves contribute more to the soil C accumulation than do the rhizomes or roots. In contrast, the recycling of the leaf N to the soil was less than for the other N fluxes, particularly for those involving the transfer of N from the tops of the plant to the rhizome.

Introduction

Dedicated energy crops offer an opportunity to substantially increase the renewable energy resources that can replace fossil fuels, and these crops are expected to play an important role in meeting the greenhouse gas (GHG) reduction targets (Claessens et al., 2010; Bessou et al., 2011). However, the management of these crops is not necessarily carbon (C) neutral, and the crop production must be sustainable, i.e., produce a large amount of biomass, while promoting C sequestration. An important factor for a positive energy balance is a low input of mineral fertilizer (Don et al., 2011), but other factors that are governed by agricultural practices are also important, particularly those affecting the storage of the soil organic matter (SOM), i.e., C and nutrients (Lal, 2010).

Litterfall constitutes a large proportion of the C and nutrient inputs to the soil in terrestrial ecosystems (Lee et al., 2006). Therefore, to maintain the SOM and to prevent erosion, it is important that plant residue (litter) be recycled: removing too much of the crop residue can rapidly damage the soil quality and reduce the productivity of the crop (Lal, 2010). In perennial ecosystems, the litter decomposes on the soil surface because there is no tillage, and it often accumulates due to an imbalance between the litterfall and decomposition rates (Martius et al., 2004).

Therefore, a substantial change in land use and/or the large-scale use of plant residues from agriculture and forestry for bioenergy production may actually promote environmental degradation (Don et al., 2011). Consequently, there is a growing interest in accounting for agricultural and environmental constraints when assessing crops as potential energy sources. The maintenance of the SOM appears to be crucial for mitigating carbon dioxide (CO2) emissions and for promoting ecosystem services, but relevant data for perennial bioenergy crops are scarce (Beringer et al., 2011; Don et al., 2011).

Miscanthus × giganteus (hereinafter referred to as Miscanthus), a C4 perennial rhizomatous grass, appears to be a promising energy crop because of its high productivity and longevity, even under temperate and cold conditions (Miguez et al., 2008), and its low N fertilization requirements compared to other crops (Cadoux et al., 2011). Over the last two decades, a number of studies have demonstrated the aboveground production potential of Miscanthus across a range of climates (Clifton-Brown et al., 2004; Miguez et al., 2008), but few studies report the effects of Miscanthus on belowground C storage. Nevertheless, there is general agreement that Miscanthus can be useful in promoting the accumulation of SOM in the topsoil (Dondini et al., 2009; Bessou et al., 2011) due to the recycling of the large amounts of plant biomass, such as leaves, rhizomes and roots, and due to the recalcitrant nature of this organic matter (Lygin et al., 2011). However, very few data are available on the decomposition of the different types of Miscanthus residues (e.g., Beuch et al., 2000; Amougou et al., 2011), and in particular, the contribution of the leaves to the accumulation of SOM has not yet been studied.

For energy production by combustion, a Miscanthus crop is generally harvested each year at maturity (at the end of winter) when the crop presents its highest dry matter (DM) content (Lewandowski & Heinz, 2003). An alternative energy use for Miscanthus is bioethanol production, which is most efficient when the plant lignocellulose is readily accessible to hydrolysis by enzymes. The degree of lignification in lignocelluloses is related to their maturity (Huyen et al., 2010), and, in the case of Miscanthus, the plants should be harvested earlier in the autumn when the aboveground parts are still green. However, an early harvest would occur before the senescent leaves fall to the ground, which might significantly alter the contribution of Miscanthus to the litter layer and therefore to SOM accumulation.

The purpose of this study was to investigate the importance of abscised leaves in the soil C and N cycling of a Miscanthus plantation. We studied a Miscanthus stand over three growing seasons, monitoring the patterns of the leaf fall during senescence, the rate of leaf decomposition (using in situ litterbags) and the leaf accumulation at the soil surface. From the data obtained, we assessed the contribution of the fallen leaves to the accumulation of the SOM and recycling under the Miscanthus crop, and we discuss the effects of the timing of Miscanthus harvesting on the soil C and N balances.

Materials and methods

Site characteristics and experimental design

The INRA (National Institute for Agricultural Research) experimental site is located in Northern France (Estrées-Mons, Picardie region, 49°80′N, 3°60′E). The soil is a deep silt loam (Orthic luvisol, Northern France), and the soil layer used (0–28 cm) is characterized by a pH(H2O) of 7.8 and a composition of 19.9% clay, 72% silt, 7.8% sand, 0.3% CaCO3, 1.05% organic C and 0.097% organic N (Mary et al., 1999). With a bulk density of 1.2 g cm−3, the stock of organic C represented 35 Mg C ha−1 in the 0–28 cm layer. Before the establishment of the Miscanthus crop in 2006, the field had a long history of intensive annual cropping with sugar beets, winter wheat, grain maize and rapeseed.

Miscanthus rhizomes obtained from ADAS (Agricultural Development and Advisory Service, UK) were planted in April 2006 in soil tilled to a depth of 20 cm, at a density of 15 625 plants ha−1 in a randomized block design of 12 m × 30 m. N fertilizer was not applied to these plots. The average annual temperature measurements (10.7 ± 0.5 °C) and rainfall (630 ± 80.4 mm) during the 3 years of the experimental study (Fig. 1) were representative of the 10-year period. A severe drought occurred in 2009 compared with the other years, whereas 2008 was the wettest year.

Figure 1.

Monthly rainfall and temperatures during the 3 years of the Miscanthus × giganteus growth study (2008–2010). The bars represent the mean rainfall (mm), and the black circles with a continuous line represent the mean temperature (°C) for each month.

At each winter harvest (in February), a corn harvester was used to cut the stems at approximately 20 cm above the ground surface. The yield of Miscanthus in the first year (winter 2007) was low (1.6 Mg DM ha−1), and the entire biomass harvest was, therefore, crushed and returned to the soil. Beginning in the second year (2008), the density was 14 941 plants ha−1, and the biomass yield ranged from 18.5 to 21 Mg DM ha−1 (Table 2) at harvest in the winter (Amougou et al., 2011; Strullu et al., 2011).

Table 1. Sampling dates for the two 1-year litter bag experiments. Three litter bags were removed from each experiment on each date
1st experimentYear 2009Year 2010
April 4April 30May 30June 30July 30October 20January 11March 5April 18
2nd experimentYear 2009Year 2010
October 15November 15December 15January 15February 15May 5July 26September 15October 25

Leaf sampling during the autumn leaf fall

The abscised leaf fall was monitored throughout the autumn/winter of 2008–2009 and autumn/winter of 2009–2010 when the Miscanthus stands were 3 and 4 years old, respectively. A total of three micro-plots of 2.4 m × 1.6 m, each containing six plants, were delineated for the collection of the fallen leaves. A nylon net (mesh size 1 cm × 1 cm) was placed at the soil surface of each micro-plot to avoid contact between the leaves and the soil, and mesh fences 40 cm high were installed around each micro-plot to minimize the loss of the fallen leaves due to wind. The leaves were then collected twice per month. Sub-samples of the leaves were dried at 35 °C for 1 week (to avoid altering their chemical composition during drying) and were later used in the litter bag decomposition experiment and biochemical analyses. The remaining leaves were dried at 80 °C for the DM, C and N determinations.

Quantification of the accumulation of mulch and leaf debris at the soil surface

An additional three 2.4 m × 1.6 m micro-plots were established to quantify the accumulation of the fallen leaves (hereinafter referred to as mulch) and leaf debris at the soil surface over time. The layer was mostly composed of leaf material, as the bases of the old stems were still standing. The measurements were performed three times, each time using a different micro-plot: in October 2009 before the start of leaf fall, in February 2010 and in October 2010. A metal ruler was inserted into the mulch layer at 10 random points within each micro-plot to measure the thickness of the layer. The mulch was then sampled for subsequent analysis by pooling four samples, each from an area of 20 cm × 20 cm and collected at random within each micro-plot. To account for the considerable spatial variability in the mulch depth, all of the mulch was removed from each micro-plot to quantify the dry mass of the mulch. The spatial variability of the leaf debris under the mulch and in contact with the ground was much smaller than that of the mulch. The leaf debris was then collected over an area of 80 cm × 80 cm per micro-plot, i.e., the mean area occupied by a Miscanthus plant. The C and N contents and biochemical characteristics of the mulch and leaf debris were then determined.

Litter bag decomposition technique

We evaluated the decomposition of the abscised leaves at the soil surface using a litter bag technique (Olson, 1963) during two different 1-year periods due to experimental constraints: from April 2009 to April 2010 and from October 2009 to October 2010. The abscised leaves collected during the winter of 2009 were dried at 35 °C for 1 week prior to analysis. Next, 15 g of leaves were placed in a 20 cm × 60 cm polyethylene bag (mesh size of 1 cm × 1 cm). The ratio of the quantity of the leaves to the area of the bag was calculated to mimic the density of the abscised leaves under the Miscanthus stands in autumn. The litter bags were closed with metal clips and placed between the rows of Miscanthus in three other micro-plots and pinned to the soil surface with metal pins. For each experiment, 27 litter bags were initially prepared, and three randomly chosen litter bags were removed from the ground on each of the nine sampling dates. The sampling dates for the two experiments are listed in Table 1. On each sampling date, the litter bags were collected and transported to the laboratory where the leaves were removed from the litter bags and dried at 35 °C for 1 week. After drying, the soil clinging to the leaves was gently removed with a brush, taking care to retain all of the leaf material. The leaf mass and C and N contents per litter bag were determined for each sampling date, and the mass, C and N losses were determined as functions of the initial dry mass and C and N contents.

Table 2. Total dry mass (Mg DM ha−1) and amounts of carbon (Mg C ha−1) and nitrogen (kg N ha−1) of Miscanthus × giganteus abscised leaves and aboveground biomass. Leaves were collected on the ground throughout the autumn/winter of 2008–2009 and autumn/winter of 2009–2010, while the aboveground biomass was harvested in late winter, in February each year. The data are reported as the mean values (= 3). Values followed by the same letter for a given parameter are not significantly different (LSD test, P ≤ 0.05)
 Dry mass (Mg ha−1)LSDC (Mg ha−1)LSDN (kg ha−1)LSD
autumn/winterautumn/winterautumn/winter
2008–092009–102008–092009–102008–092009–10
Abscised leaves3.1a3.0a2.31.40a1.37a0.04917.7a13.6a13.5
Above-ground biomass18.5a21a4.69.2a10.5a3.234.2a35.5a10.9

Modelling mulch accumulation at the soil surface

To predict the rates of the decomposition of the leaves at the soil surface and their accumulation over a longer period, a simple model was constructed, parameterized and run using the programming language in the STELLA® software package (Costanza, 1987). STELLA® was used to view and analyse the equations created after defining the study variables, which were represented by icons (Fig. 2). The model was arranged as follows. The inflow (INPUT) is the leaf fall over 1 year, and these leaves represent the initial stock of litter (DM0) in year 0. Part of this stock (DM0) decomposes according to a constant decay rate (k), which is the outflow (DECAY), whereas the non-decomposed leaves provide a new stock (DMt) after the first year of decomposition. The new stock also decreases after a year at a decay constant k1, and the non-decomposed leaves feed the next year's stock. This cycle is then perpetuated in the ensuing years.

Figure 2.

Model structure: the annual leaf fall provides the INPUT. The stock of litter formed in year 0 (DM0) decomposes (DECAY) at a decay rate k (yr−1), and the accumulated litter represents the litter retention after the first year of decomposition.

The amount of leaves feeding stock 0 was based on our field results (INPUT = 3 Mg DM ha−1). The annual decay constant (k = 0.776 yr−1) was calculated from the two litter bag experiments, as follows:

display math(1)

with X0 as the initial mass of the leaves in the litterbags (15 g) and Xt as the residual mass of the leaves (6.9 g) at time t (1 year) (Olson, 1963). We assumed that k is constant over the years, with k = k1= kn. The evolution of the stock at the soil surface was calculated as follows:

display math(2)

Chemical analysis

To better understand the decomposition processes in our system, the chemical composition of the leaf residue was determined as follows. The total C and N concentrations of the mulch and leaf debris were determined using an elemental analyser (EURO EA, EuroVector, Italy). The leaf residue was also analysed for the soluble neutral detergent fibre (NDF), acid detergent fibre (ADF) and acid detergent lignin (ADL) according to Goering & Van Soest (1970). Briefly, the soluble fraction was obtained by grinding 1 g of leaf residue to a particle size of 4 mm (using a cross-beater mill SK-100; Retsch, Germany), which was followed by boiling (100 °C) the residue in deionized water for 30 min and extracting with a neutral detergent solution for 60 min to remove the cytoplasmic components and obtain the NDF fraction. This NDF fraction was then boiled (100 °C) in an acid detergent solution for 60 min to extract the hemicellulose and obtain the ADF fraction. This ADF fraction was then subjected to selective cellulose extraction with concentrated H2SO4 for 180 min. The final mass of the non-extractable fraction ADL was assumed to consist of lignin-like compounds. Ash measurements were performed by ignition at 550 °C for 4 h and were very low (< 3%) and not retained for the calculations. The leaf residue fractions were expressed in relation to the initial residue DM.

Data treatment and analysis

One-way analysis of variance (anova) was used to analyse the effects of the year of study on the leaf fall and decomposition. One-way anova was also used to verify the effects of the sampling date on the quantity and chemical characteristics of the mulch and leaf debris. When the F values were significant, post hoc comparisons were made using Fisher's least significant difference (LSD) test at a probability level of 0.05. The statistical analyses were performed using R 2.9.2 software.

Results

Quantification of abscised leaf fall

The leaf fall began in October or November, depending on the year, and ended by late February (Fig. 3a). The leaf fall dynamics differed between the 2 years, with a regular rate of leaf fall during the autumn/winter of 2008–2009 (Fig. 3a) and a heavy leaf fall from November to December of 2009. However, the total masses of the leaves collected were notably similar: 3.1 Mg ha−1 for autumn/winter 2008–2009 and 3.0 Mg ha−1 for autumn/winter 2009–2010, corresponding to 1.40 and 1.37 Mg C ha−1 respectively (Table 2). In the same plots, the aerial biomass at winter harvest was 18.5 Mg ha−1 for year 2009 and 21 Mg ha−1 for year 2010 and consisted mostly of stems, but the differences were not significant (P ≤ 0.05) (Table 2). The N and C contents and the C : N ratios of the collected leaves, measured at each harvest time, did not vary significantly (P ≤ 0.05) over the fall period for a given year (Fig. 3b–d). Furthermore, the C and N contents and the C : N ratio did not differ significantly between the 2 years (P ≤ 0.05): 454 ± 4 g C kg−1 DM, 5.6 ± 0.5 g N kg−1 DM and a C : N ratio of 83 ± 7 were found for the autumn/winter 2008–2009 period, and 457 ± 5 g C kg−1 DM, 4.5 ± 0.4 g N kg−1 DM and a C : N ratio of 100 ± 9 were found for the autumn/winter 2009–2010 period. In total, the leaves collected represented 17.7 and 13.6 kg N ha−1 for 2008/09 and 2009/10, respectively, but the difference between the 2 years was not statistically significant (Table 2).

Figure 3.

(a) Cumulative amounts of fallen senescent Miscanthus × giganteus leaves over time (Mg ha−1), (b) average nitrogen contents (g N kg−1DM), (c) average carbon contents (g C kg−1DM) and (d) average C : N ratio of the senescent leaves collected over time (n = 3). The leaves were collected over autumn/winter 2008–2009 and autumn/winter 2009–2010. The error bars on the graphs represent the standard errors.

Kinetics of the leaf decomposition at the soil surface

The breakdown of the leaves in the litter bags, as estimated by their mass loss, proceeded differently during the two decomposition periods (Fig. 4a), likely because the litter bag experiments were not begun at the same time of year. In fact, although the two experiments had the same duration (1 year) and nearly identical average temperatures (mean temperature of 10.5 °C for experiment 1 and 10.3 °C for experiment 2), the precipitation differed (592 and 703 mm for experiments 1 and 2 respectively) (Fig. 1). In the first period, we observed an initial phase between April and August of 2009 during which a slow, steady decay led to the loss of approximately 20% of the mass after 4 months. Next, we observed a plateau phase between August 2009 and January 2010, with no significant loss in the mass. The mass loss was then rapid from January to April of 2010, leading to a cumulative mass loss of 54% of the initial mass by the end of the experiment. The second experiment began in autumn instead of spring (October 2009), and the breakdown was initially rapid, but the mass remained constant between July and October of 2010 (Fig. 4a). The total leaf mass loss was similar to that observed during the first experiment, with a decay rate of 54% at the final sampling (October 2010).

Figure 4.

Changes in (a) the average dry mass and (b) average C : N ratio in decaying senescent Miscanthus × giganteus leaves in litterbags with a mesh size of 1 cm (n = 3) studied from April 2009 to April 2010 and October 2009 to October 2010. The error bars on the graphs represent the standard errors.

The changes in the C : N ratio in the remaining leaves was very similar for the 2 years of the study (Fig. 4b). The initial C : N ratio of the undecomposed leaves was 87, and the C : N ratio peaked at 96 for the first sampling time (1 month after the start of the experiment) and then rapidly decreased to 72 after 1 year of decomposition. As the residue decomposed, the N concentration of the remaining litter increased significantly, whereas the C concentration remained fairly constant (data not shown). On average, the leaf residues initially represented 1.40 Mg C ha−1 and 15 kg N ha−1 for the two periods. After a year of decomposition, the remaining residues in the litter bags were equivalent to an average of 0.54 Mg C ha−1 and 8 kg N ha−1.

Quantification and characterization of the leaf mulch

The mulch measured at the soil surface in October 2009 amounted to 4.9 Mg DM ha−1 and was approximately 20 mm thick (Table 3). This mulch was the result of the accumulation and decomposition of the biomass returned to the soil in the first year after planting and the leaf fall during the ensuing 2 years (the autumns of 2007 and 2008). During the winter of 2009–2010, the new fall of senescent leaves increased the dry mass and mulch thickness; by October 2010, before the new fall of leaves, however, the mulch dry mass and thickness had fallen to figures similar to those in October 2009 (Table 3). The C content of the mulch ranged from 1.26 to 2.13 Mg C ha−1, depending on the time of measurement, with a significant increase after the senescence in autumn/winter and a significant decrease after the spring/summer decomposition (Table 3). The N content of the mulch amounted to 25 to 36 kg N ha−1, and the C : N ratio of the mulch varied from 42 to 60. The amount of leaf debris collected under the mulch did not vary significantly over time. The averages of this fraction ranged from 2.3 to 2.8 Mg DM ha−1, 0.43 to 0.48 Mg C and 11 to 12 kg N ha−1, with a C : N ratio of 39 to 44.

Table 3. Thickness (mm), dry matter (Mg ha−1), amounts of carbon (Mg C ha−1) and nitrogen (kg N ha−1), and C : N ratios of mulch and of leaf debris underneath the mulch. Mulch and leaf debris were collected in October 2009, February 2010 and October 2010. For the thickness measurements, the mean values per plot (n = 10) are shown. For the other parameters, the means of three replicates (n = 3) are shown. Values followed by the same letter for a given parameter are not significantly different (LSD test, P ≤ 0.05)
 Thickness (mm)Dry matter (Mg ha−1)C (Mg ha−1)N (kg ha−1)C : N
MulchOctober 0922.23b4.9a1.30b31a42b
February 1042.31a5.8a2.13a36a60a
October 1022.08b4.6ab1.49b25ac60a
Leaf debrisOctober 09nd2.3c0.48c11b44b
February 10nd2.6c0.47c12bc39b
October 10nd2.8bc0.43c11b39b
LSD0.61.90.6213.5 9.6

The Van Soest analysis of the senescent leaves, mulch and leaf debris showed significant variations (P ≤ 0.05) in the relative amounts of soluble, hemicellulose, cellulose and lignin fractions among the three types of litter (Table 4). The leaf debris contained the highest proportions of cellulose and lignin, whereas the recently abscised leaves contained the highest level of soluble C compounds and hemicelluloses.

Table 4. Chemical characterization of senescent leaves, mulch and leaf debris. Mulch and leaf debris were collected in October 2009, February 2010 and October 2010, and senescent leaves were collected in early October 2009. The data are reported as the means (n = 3). Values followed by the same letter for a given parameter are not significantly different (LSD test, P ≤ 0.05)
 (% dry mass)
SolubleHemicelluloseCelluloseLignin
Senescent leavesOctober 200823a34a36abc5d
October 200922a35a36abc6d
MulchOctober 200922a29b35bc14b
February 201020b34a35bc10c
October 201023a31b34c13b
Leaf debrisOctober 200919b25c39a17a
February 201019b26c37ab18a
October 201019b24c38ab18a
 LSD 2.0 2.2 3.3 2.3

Simulation of the accumulation of leaf residues on the ground

In the simulation, the leaves gradually accumulated up to a total DM equivalent of 6.9 Mg ha−1 after 4 years of leaf fall, and the mass then remained constant until year 10 (the duration of the simulation), reflecting an equilibrium between the input and decomposition of the material (Fig. 5). Experimentally, the amounts of DM at the soil surface (mulch + leaf debris) measured after 4 and 5 years of Miscanthus growth (Fig. 5) were similar and represented 7.2 Mg ha−1 and 7.4 Mg ha−1, respectively, which are similar to the simulated values.

Figure 5.

Simulated amounts (white diamonds with a dashed line) of leaves accumulated on the ground over 10 years and the experimental measurements (black diamonds) obtained at 4 and 5 years after planting. For the experimental measurements (n = 3), the error bars represent the standard errors.

Discussion

Abscised leaf-C and N

The main objective of this study was to characterize the dynamics of Miscanthus senescent leaf fall and the amounts of leaf C and N recycled annually to evaluate the impact of Miscanthus harvesting strategies on the accumulation of SOM. The amounts of fallen leaves were markedly similar in the 2 years, equivalent to 3 Mg DM ha−1, i.e., 1.40 Mg C ha−1 and 16 kg N ha−1 and represented 15% of the total aboveground biomass (stems + leaves), which was constant over the 2 years of the study. However, compared to a similar range of Miscanthus total aboveground biomass production, our data for the biomass of the leaves lie at the lower end of the range reported in the literature, i.e., 3–7 Mg DM ha−1 yr−1 returned annually to the ground before the maturity stage (Beuch et al., 2000; Kahle et al., 2001). We attribute the discrepancy between our measurements and the cited literature to the fact that the Miscanthus litterfall consisted of 61% senescent leaves and 39% shoot fragments in the previous studies, whereas our study focused on the input of senescent leaves only.

The N content of the fallen leaves did not vary significantly over time. This consistency can be attributed to the physiological state of senescence reached before leaf fall. The N content of the senescent leaves was much lower than that of the Miscanthus leaves during the growing phase, and the drop in N content is due to the remobilization of N from the aboveground parts of the plants to the rhizomes before the onset of winter (Strullu et al., 2011). Under our experimental conditions, the autumn remobilization was equivalent to approximately 66% of the leaf N content measured at flowering (data not shown). This result is consistent with those of Himken et al. (1997) and Neukirchen et al. (1999) who showed that 70% of the N contained in Miscanthus leaves was remobilized to the underground parts before the complete senescence of the plant. Therefore, the amount of N recycled to the soil through leaf fall was small in comparison with the other N pools that may supply the crop with N, i.e., the N stored in the rhizome and roots and the soil N (Kahle et al., 2001; Strullu et al., 2011).

Leaf and mulch decomposition dynamics

To our knowledge, only this study and that of Yamane & Sato, 1975 have attempted to quantify the decomposition of Miscanthus leaves under field conditions. In each of our two decomposition experiments, approximately 54% of the leaf DM disappeared from the litterbags after 1 year. Studying the leaf decomposition of Miscanthus sinensis, one of the parent species of Miscanthus × giganteus, Yamane & Sato, 1975 showed a 40% DM loss after 1 year, which is equivalent to an annual decay constant = 0.511 yr−1, as calculated from their data using equation (1). We assume that the different decomposition kinetics observed in the 2 years of our experiment were due to the different times at which decomposition was begun each year and were influenced by the climatic conditions. For example, in this study, the drought that occurred between June 2009 and September 2009 probably explains the plateau observed during the first year of decomposition. Indeed, the mulch moisture is an important factor for leaf decomposition at the soil surface, which is in agreement with the conclusions of Coppens et al. (2007) who studied the effect of rainfall on the decomposition of residue mulch.

The chemical characteristics of the litter are another important factor explaining the rate of decomposition. The Miscanthus lignocellulosic biomass, and the leaves in particular, has a high lignin content relative to other plants, particularly other grasses, and has been shown to be recalcitrant to enzyme hydrolysis (Dresboll & Magid, 2006; Huyen et al., 2010). Under controlled conditions, Amougou et al. (2011) showed that the lignin content of Miscanthus residue negatively affected their decomposition rate, which confirms the general findings for a large range of residues (e.g., Trinsoutrot et al., 2000). In addition, these abscised leaves have low N contents, and, as they decomposed at the soil surface in the field, their decomposition was probably slowed by the lack of available mineral N. Low N availability can alter the kinetics of C mineralization in high C : N residues due to the shortage of available N for assimilation by heterotrophic decomposers (Recous et al., 1995; Keeler et al., 2009). Indeed, studies of senescent leaf decomposition in other perennial crops (e.g., vines and perennial legumes) and various types of broadleaf forest trees (beech, poplar) have reported decay rates similar to those observed after 1 year of decomposition in this study using the litterbag technique, which is considered to reflect fairly well the natural conditions of residue decomposition on the ground (Semwal et al., 2003; Duchesne et al., 2010; Nikolaidou et al., 2010).

The Miscanthus stand in our study had a high annual leaf input and a slow decomposition rate, resulting in the accumulation of partially decomposed leaves on the ground. After 3 years of Miscanthus growth, the mulch formed a 20 mm thick biomass layer at the soil surface, and a considerable pool of C accumulated. The variations in mass, thickness and biochemical composition over 1 year were consistent with the seasonal cycle of senescent leaf inputs and decomposition. However, the net change in DM quantity between the two autumn samplings was negligible, indicating an equilibrium between the amount of biomass decomposed and the input of new biomass to the existing mulch layer. Interestingly, we observed that the DM of the leaf debris in direct contact with the ground under the mulch layer was constant, with features (C : N ratio and biochemical composition) indicative of a more advanced degree of decomposition than the overlying mulch. This observation is consistent with the conceptual representation of several mulch models that describe a layer of mulch consisting of non-decomposed residue “fuelling” an underlying layer formed of matter undergoing decomposition and in direct contact with the ground (Probert et al., 1998; Findeling et al., 2007). Of course, this view is a simplification, and a gradient of decomposition exists from the top of the mulch down to the soil surface; however, this gradient within the mulch layer was not characterized in this study.

The mulch accumulation model simulated the accumulation of leaves over time, using parameters fitted with the data obtained in our experiment. Decomposition is a complex process that does not always follow the basic exponential decay function, but this function often provides an excellent approximation (Berg & McClaugherty, 2008). We found a good agreement with the field measurements and the model predictions that the mass of residual mulch would be fairly stable over the medium term. The modelling assumption that the mulch layer has a constant mean decay rate over 10 years is probably not completely accurate, particularly during the first 3 years of Miscanthus growth when the mulch layer was young and, therefore, probably decomposed faster. However, the quantification of this C pool should encompass a longer duration than 2 years to obtain a more accurate approximation of the mulch decay, and the stabilization of the amounts of mulch C accumulated under Miscanthus stands over longer time periods should be confirmed under field conditions.

C and N inputs to the soil

In a previous study, Amougou (2011) showed that under controlled conditions optimal for decomposition (ground residue mixed with soil and non-limiting N availability), the apparent C mineralization of Miscanthus leaves peaked at approximately 65% of the added C after 700 days of decomposition, which agrees with most of the studies of crop residue decomposition (e.g., Trinsoutrot et al., 2000; Jensen et al., 2005). The unmineralized portion of the C (approximately 35%) is, therefore, considered to be “stabilized” (either as the recalcitrant plant fraction or as microbial metabolites). These data allow us to estimate the annual input of leaf-C to the soil as follows: based on a recycling rate of 1.4 Mg of leaf-C per hectare per year at equilibrium and assuming that 35% of the added C is stabilized over time, the annual contribution of senescent leaves to the soil C would be approximately 0.50 Mg C ha−1 yr−1, i.e., 10 Mg C ha−1over the 20-year lifespan of a Miscanthus crop. This annual input of C (0.50 Mg ha−1 yr−1) represents 1.4% of the initial soil C of the (0–28) cm soil layer in our experiment. For a similar range of Miscanthus biomass production, Don et al. (2011) calculated an annual total C input to the soil of 0.66 Mg C ha−1 yr−1, originating from fallen leaves and from the turnover of root and rhizome C, whereas Beuch et al. (2000) calculated 0.7 Mg C ha−1 yr−1 derived from senescent leaves using the humus balance method. Our data show similar results and suggest that abscised leaves make an important contribution to the accumulation of the soil organic carbon under Miscanthus. This conclusion is plausible because the root-C pool of Miscanthus is much lower (Monti & Zatta, 2009; Amougou et al., 2011) and its annual turnover rate is unknown. By comparison, organic C storage in the Miscanthus rhizome is higher (8–10 Mg C ha−1), but its annual contribution to the SOC is difficult to quantify and mostly unknown (Neukirchen et al., 1999; Christian et al., 2009). At the end of the life of the crop and assuming a similar ratio between the mineralization and stabilization for the rhizome and leaf-C (Amougou et al., 2011), the destruction of plant rhizomes would lead to a net contribution to the soil of approximately 4 Mg C ha−1. The contributions of the different parts of the Miscanthus plant to its C storage potential must therefore be taken into account when assessing the GHG balance of the crop.

As a result, early harvest of Miscanthus when the aboveground parts are still green, which aims to increase the amount of lignocellulosic biomass collected and reduce the recalcitrance of this biomass to hydrolysis by enzymes (Huyen et al., 2010), would deprive the soil of the return of the leaves and an annual significant input of organic C to the soil and deprive the crop of a significant remobilization of N (Strullu et al., 2011). Therefore, this scenario does not seem desirable from an environmental point of view, considering that the substitution of fossil C with the C from plants must be undertaken, while preserving the ecosystem functions of the soil, particularly for preserving the beneficial physical, chemical and biological qualities by maintaining the soil organic matter content (Lal & Pimentel, 2009).

Acknowledgements

This study was funded by INRA, the Region Champagne-Ardenne, which provided a doctoral grant to N. Amougou, and the Region Picardie (MISQUAL project AAP07-52). The authors thank Professor D. Moorhead for fruitful discussions regarding this study, JM. Machet, M. Preudhomme and E. Fourdinier (INRA Agro-Impact) for field experiment management and for providing the plant material and F. Millon and O. Delfosse for their technical assistance.

Ancillary