Litter quality and decomposability of species from a Mediterranean succession depend on leaf traits but not on nitrogen supply

Abstract

Background and Aims

The rate of plant decomposition depends on both the decomposition environment and the functional traits of the individual species (e.g. leaf and litter quality), but their relative importance in determining interspecific differences in litter decomposition remains unclear. The aims of this study were to: (a) determine if species from different successional stages grown on soils with low and high nitrogen levels produce leaf and litter traits that decompose differently under identical conditions; and (b) assess which trait of living leaves best relates to litter quality and litter decomposability

Methods

The study was conducted on 17 herbaceous species representative of three stages of a Mediterranean successional sere of Southern France. Plants were grown in monocultures in a common garden under two nitrogen levels. To elucidate how different leaf traits affected litter decomposition a microcosm experiment was conducted to determine decomposability under standard conditions. Tests were also carried out to determine how successional stage and nitrogen supply affected functional traits of living leaves and how these traits then modified litter quality and subsequent litter decomposability.

Key Results

The results demonstrated that leaf traits and litter decomposability varied according to species and successional stage. It was also demonstrated that while nitrogen addition affected leaf and litter traits, it had no effect on decomposition rates. Finally, leaf dry matter content stood out as the leaf trait best related to litter quality and litter decomposability

Conclusions

In this study, species litter decomposability was affected by some leaf and litter traits but not by soil nitrogen supply. The results demonstrated the strength of a trait-based approach to predict changes in ecosystem processes as a result of species shifts in ecosystems.

INTRODUCTION

Decomposition of dead plant material is a key component in carbon and nutrient cycling in most terrestrial ecosystems (Swift et al., 1979; Couteaux et al., 1995; Chapin et al., 2002). The multiple drivers of decomposition include the effects of decomposition environment, at both regional and microsite scales, the substrate quality of litter and the composition of the decomposer community (Swift et al., 1979), with the relative importance of these three factors varying across ecosystems (Lavelle et al., 1993; Pérez-Harguindeguy et al., 2000).

In a meta-analysis involving 818 species from 66 decomposition experiments on six continents, Cornwell et al. (2008) showed that the degree to which interspecific variations in leaf structure and composition affect their decomposition rate could be as large as the effect of global climatic variation. Several traits of green leaves related to physiological and protective features persist through senescence, and have been shown to affect litter decomposition. This is the case for, for example, the physical strength of leaves (Gallardo and Merino, 1993; Cornelissen, 1996; Cornelissen and Thompson, 1997) or their nitrogen and/or phosphorus concentrations (Cornelissen and Thompson, 1997; Cornwell et al., 2008). In their meta-analysis, Cornwell et al. (2008) used leaf mass per area (LMA; the ratio of leaf mass to its area) and leaf nitrogen concentration (LNC) as the two generic traits describing leaf structure and chemical composition, respectively. Although this is justified by the number of studies in which these traits have been studied, the case of LMA (or its inverse, specific leaf area, SLA) deserves further attention. SLA is a robust index of sclerophylly as a surrogate for more rigorous mechanical properties used in herbivory studies (Hanley et al., 2007) and depends on both leaf thickness and density (Witkowski and Lamont, 1991). Limited evidence suggests that litter decomposition actually depends mainly on the latter: in studies where leaf dry matter content (LDMC; the ratio of leaf dry mass to saturated fresh mass), used as a surrogate for leaf tissue density (see Garnier and Laurent, 1994; Shipley and Vu, 2002), was measured in addition to SLA, relationships with decomposition rates were always found to be stronger with LDMC, be it at the species (Kazakou et al., 2006) or the community (Cortez et al., 2007; Quested et al., 2007; Fortunel et al., 2009) level. Here the hypothesis that LDMC is the better predictor of species-level leaf decomposition rates is tested by explicitly relating plant leaf traits to several indices of litter quality known to affect decomposition. This was done for a set of plant species characterizing the different successional stages associated with the abandonment of Mediterranean arable fields.

A previous study in these successional fields had shown that litters produced by early successional species tend to decompose more rapidly than those produced by species from more advanced stages (Kazakou et al., 2006). This study was conducted on leaf and litters collected in species grown in the field however, which implies that differences in traits and decomposition rates may be due to differences in environmental conditions prevailing in the old fields from different successional stages. Such differences include soil nitrogen and carbon concentrations, litter accumulation and light interception (Garnier et al., 2004, 2007; Kazakou and Navas, 2004). Nitrogen was found to be a strong limiting factor for plant growth in this successional sere, as indicated by the estimation of nitrogen nutrition index [(NNI) approx. 50 %; see Garnier et al., 2007]. Numerous studies have shown that in an initially low nutrient environment, fertilization often causes an increase in nutrient concentration in plants, whether accompanied by an increase in growth (i.e. a limiting nutrient) or leading to luxury uptake (Chapin, 1980; Chapin et al., 1986). However, evidence supporting a link between increased endogenous leaf nutrient concentrations inducing faster decomposition rates is contradictory, sometimes showing no effects (Hobbie and Vitousek, 2000; Bridgham and Richardson, 2003; Güsewell and Verhoeven, 2006) or a stimulatory effect (Coulson and Butterfield, 1978; Pastor et al., 1987; Taylor et al., 1989; Aerts, 1997). Bridgham and Richardson (2003) proposed a conceptual model assuming that as plants from low nutrient environments have low carbon quality litter, decomposition rates will be unaffected by the increase in endogenous nutrient concentrations. If greater nutrient availability leads to better litter quality, for example through increased amino acid or soluble carbohydrate concentrations or decreased lignin or phenolic concentrations (Northrup et al., 1995), then there should be an increase in decay rates. Here we tested if nitrogen fertilization of species grown in monocultures would affect traits and thus whether there is any subsequent effect on decomposition rates.

The main objectives of the present study were therefore to: (a) test whether changes in decomposability and associated traits previously recorded in situ along the successional sere are also apparent when species were grown under identical environmental conditions under low and high nitrogen availabilities (we hypothesized that species effects would be stronger than environmental effects on litter decomposability); (b) assess whether differences in soil nitrogen might affect leaf traits and decomposition rates; and (c) assess which trait of living leaves best relates to litter quality and litter decomposability. Here, we hypothesize that LDMC relates better to the litter quality, hence to the decomposition rate, than to other traits screened for so far such as SLA or leaf nitrogen concentration.

MATERIALS AND METHODS

Study site, species and experimental design

The common garden experiment was conducted in the experimental field of the Centre d'Ecologie Fonctionnelle et Evolutive located in Montpellier, France (43°59′N, 43°51′E). Soil pH (7·82), soil total average nitrogen (1·38 g kg⁻¹) and carbon (14·54 g kg⁻¹) concentration were close to those recorded in the old-field succession studied by Garnier et al. (2004). Seventeen herbaceous species were selected as representative of plant communities from French Mediterranean old-field successions (Table 1). Three main stages were recognized based on the time since abandonment: early (0–6 years); intermediate (7–15 years); and advanced (16–45 years). Communities of intermediate succession showed the highest number of species (mean number of species = 7) and those of advanced succession the lowest (mean number of species = 2) (see table 2 in Garnier et al., 2004). Five species were chosen for the early stage and six species for the intermediate and advanced stages. These species represented from 57 to 97 % of the community total above-ground biomass (Vile et al., 2006). Among these species, four groups of three species from each successional stage were chosen to form four taxonomic groups: the order Lamiales and the families Asteraceae, Fabaceae and Poaceae (Table 1). In October 2003, four replicated monocultures (1·2 × 1·2 m plots) per species were established at two levels of nitrogen supply by transplanting seedlings or ramets (seeds or ramets of all the species were collected in the field and grown in a glasshouse before transplantation to the experimental plots; for a full description of the site, see Garnier et al., 2004). This ensured better survival and a standard plant density (100 plants m⁻²). In the fertilized treatment (N+ treatment), 25 g N m⁻² in the form of NH₄NO₃ were applied three times between January and March 2004. No fertilization was added in the N− treatment. Growth limitation by N was assessed by comparing the NNI of the two treatments. The NNI was calculated as the ratio between the actual nitrogen concentration of above-ground biomass and the critical nitrogen concentration (i.e. the concentration allowing potential growth), as proposed by Lemaire and Gastal (1997). The NNI was approx. 50 % in the N− treatment, and between 84 and 132 % in the N+ treatment (Kazakou et al., 2007). This compares with an NNI of approx. 50 % in the successional sere (Garnier et al., 2007). Hence the N− treatment was considered strongly growth limiting, whereas the N+ treatment was non-limiting for plant growth.

Table 1.

List of the species studied

Species	Successional status	Family/taxonomic group	Abbreviation
Bromus madritensis	Early	Poaceae (1)	BROMMADR
Crepis foetida	Early	Asteraceae (2)	CREPFOET
Geranium rotundifolium	Early	Geraniaceae	GERAROTU
Medicago minima	Early	Fabaceae (3)	MEDIMINI
Veronica persica	Early	Plantaginaceae (4)	VEROPERSI
Calamintha nepeta	Intermediate	Lamiaceae (4)	CALANEPE
Dactylis glomerata	Intermediate	Poaceae (1)	DACTGLOM
Daucus carota	Intermediate	Apiaceae	DAUCCARO
Picris hieracioides	Intermediate	Asteraceae (2)	PICRHIER
Tordylium maximum	Intermediate	Apiaceae	TORDMAXI
Trifolium angustifolium	Intermediate	Fabaceae (3)	TRIFANGU
Bituminaria bituminosa	Advanced	Fabaceae (3)	BITUBITU
Brachypodium phoenicoides	Advanced	Poaceae	BRACPHOE
Bromus erectus	Advanced	Poaceae (1)	BROMEREC
Inula conyza	Advanced	Asteraceae (2)	INULCONY
Rubia peregrina	Advanced	Rubiaceae	RUBIPERE
Teucrium chamaedrys	Advanced	Lamiaceae (4)	TEUCCHAM

Species are representative of three stages of a Mediterranean post-cultural succession (see Garnier et al., 2004 for details) following vineyard abandonment: early (2–6 years); intermediate (7–15 years); advanced (15–45 years). Four taxonomic groups were constructed (taxonomic relationships from Soltis et al., 2000) with one species per successional stage.

Table 2.

List of traits, abbreviations and units

	Abbreviation
Trait	Green leaves	Litter	Unit
Specific leaf area	SLAgreen	–	m2 kg−1
Leaf dry matter content	LDMCgreen	–	mg g−1
Leaf tensile strength	LTSgreen	–	MN m−2
Leaf resistance to fracture	LRFgreen	–	J m−2
Leaf nitrogen concentration	LNCgreen	LNClitter	mg g−1
Leaf phosphorus concentration	LPCgreen	LPClitter	mg g−1
Leaf carbon concentration	LCCgreen	LCClitter	mg g−1
Initial lignin concentration	–	LIGlitter	mg g−1
Initial cellulose concentration	–	CELlitter	mg g−1
Initial hemicellulose concentration	–	HEMlitter	mg g−1
Hollocellulose to lignocellulose quotient	–	HLQlitter	–
Fibre component	–	LCHlitter	mg g−1
Potential decomposability	–	Kpot	g kg−1 d−1

Indications (given as trait abbreviation) of the condition of the leaf material (green leaf or litter) when the trait is measured are given.

Collection of material

Traits of living leaves were measured on 12 replicate samples per species at each level of nitrogen during spring 2004, on the youngest fully expanded, well-lit leaves at their peak period of growth.

Leaf litter was collected at the season of maximum leaf senescence for each species (summer and autumn 2004). For each species litter was collected from the four plots of a given treatment and then mixed to form a composite sample. For some species that shed their leaves once they had senesced (e.g. Veronica persica), dead leaves that dropped after gently shaking the plants were collected. In species that retain dead leaves on the plant (e.g. Brachypodium phoenicoiedes and Dactylis glomerata) or that die completely above-ground (e.g. Bromus madritensis and Bromus erectus), leaves that were dead were cut off from the standing plant. Dead leaves were carefully cleaned, air-dried and stored in the laboratory.

Trait measurements

Thirteen leaf and litter traits were assessed in this study (see Table 2). Leaf traits were measured on each species, using standardized procedures (Cornelissen et al., 2003). Specific leaf area (SLA_green) and leaf dry matter content (LDMC_green) were calculated as the ratio between leaf area and leaf dry mass, and between leaf dry mass and saturated fresh mass, respectively. The physical strength of leaves was assessed by measuring leaf tensile strength (LTS_green) and leaf resistance to fracture (LRF_green). LTS_green was calculated as the force needed to tear a leaf divided by its width (e.g. Cornelissen and Thompson, 1997). It was measured with an apparatus comparable in design with that described by Hendry and Grime (1993). In order to calculate LTS_green, measurements of leaf width and average thickness at the point of rupture were made with a digital calliper and a linear variable displacement transducer, respectively. LRF (also called ‘force to fracture’ or ‘work to shear’), which was calculated as the mean force needed to cut a leaf or a leaf fragment at a constant angle (20 °) and speed (e.g. Wright and Cannon, 2001), was measured with a device adapted from that described by Wright and Cannon (2001). The measurement of LTS and LRF for Daucus carota was not possible due to its particular leaf shape.

Nitrogen and carbon concentrations were determined with an elemental analyser (model EA 1108; Carlo Erba Instruments, Milan, Italy). Phosphorus concentrations were measured colorimetrically with an autoanalyser (Evolution II; Alliance Instruments, Frépillon, France), using the molybdenum blue method following digestion by sulfuric acid (Grimshaw et al., 1989).

A sub-sample of the litter of each species was ground in a cyclone mill (Cyclotec Sample Mill, Tecator, Höganäs, Sweden) and scanned using a near infrared reflectance spectrophotometer (NIRS; NIRSystems 6500, Foss NIRSystems, Raamsdonksveer, The Netherlands). For these samples, lignin (LIG_litter), cellulose (CEL_litter) and hemicellulose (HEM_litter) concentrations were determined by NIRS according to the method described by Joffre et al. (1992). The predicted values were obtained with standard errors of calibration of 2·8 % for LIG_litter and 1·7 % for CEL_litter and HEM_litter. We also calculated the total fibre content of litter (LCH_litter = LIG_litter + CEL_litter + HEM_litter), and the hollocellulose: lignocellulose ratio [HLQ_litter = (CEL_litter + HEM_litter)/LCH_litter] which indicates the proportion of the less recalcitrant non-labile compounds (McClaugherty and Berg, 1987; Berg et al., 1993; Gillon et al., 1994; Cornelissen et al., 2004; Cortez et al., 2007).

Litter decomposability in microcosms

Litter was incubated in microcosms in the laboratory, under controlled temperature and humidity conditions. Microcosms, as simplified analogues of natural ecosystems, allow the study of litter decomposition by controlling temperature and humidity, with similar soil conditions and decomposer populations, while maintaining a sufficiently natural situation so that results of laboratory tests may be extrapolated to the field situation with confidence (Taylor and Parkinson, 1988). The microcosm type used for this experiment was described by Taylor and Parkinson (1988). Each microcosm chamber, 15 cm high, was made of a 15 cm diameter polyvinylchloride pipe, fitted with a lid and a sealed bottom. The lid could be opened to allow gas exchange and the plug at the bottom could be removed to drain excess water. A grid, 2 cm above the bottom, divided the chamber into two unequal parts: a usable space of 1·5 L capacity and a drainage compartment of 300 mL. One kilogram of soil, of known water-holding capacity, was placed on the grid. The soil was a 3 : 1 mixture of mineral soil and surface organic horizon taken from the common garden where the experiment was conducted (for further details, see Kazakou et al., 2006).

For each of the 17 species, ten litter samples (3 ± 0·01 g each) per nitrogen supply were sealed in a nylon litter bag of 1 mm mesh (Northern Mesh, Oldham, UK). Each litter sample was soaked for 24 h in 0·1 L of water, and then placed on the surface of the microcosm soil. In order to keep all the soluble nutrients in the system, the water used for soaking was poured into the microcosm. The soil was subsequently moistened up to 80 % of field capacity. The microcosms were kept in the dark at 22 °C throughout the experiment and watered once a week to maintain constant soil moisture during incubation. Two litter samples per species and nitrogen supply were removed from the microcosms at the end of 1, 2, 4, 6 and 8 weeks. Soil particles were carefully removed from the litter bags and the litter samples were weighed after drying for 48 h at 55 °C. Prior to the experiment, two litter samples of 2 g from each species were weighed, dried in an oven for 48 h at 55 °C and weighed again in order to correct the initial mass for the water content of the litter.

The percentage of oven-dried litter mass remaining is denoted %MR hereafter. To compare the decomposability of the different species, the single negative exponential model proposed by Olson (1963) was fitted to the %MR of each litter during the course of the experiment: %MR = 100 e^−Kpott, where K_pot is the potential decomposition rate constant (litter decomposability) over time t in days; and %MR is expressed as a percentage of the original mass. The K_pot rate constants was multiplied by 10³ and expressed in g kg⁻¹ d⁻¹.

Data analyses

For the leaf and litter traits 12 replicates per species (three individuals per plot) and per nitrogen supply were used (the plot effect was not significant for any of the traits measured). The effects of species nested within successional stage, successional stage, N supply and their interaction (successional stage × N supply) on the various variables were tested with a three-way analysis of variance (ANOVA). When the interaction effect was not significant, this was removed and the ANOVA analysis was repeated. If the effect of stage was significant, post hoc tests (Student–Newman–Keuls comparisons) were performed within each N supply in order to identify the variations between successional stages. Two-way ANOVA was carried out for the 12 species belonging to the four taxonomic groups, with successional stage, taxonomic group and their interaction as effects.

Bivariate correlations between green leaf, litter traits and K_pot were evaluated with Pearson product–moment correlation coefficients. In all analyses significance levels were corrected by the improved Bonferroni procedure (Simes, 1986; Sokal and Rohlf, 1995). All variables were log transformed when required. All the analyses were carried out with the Statistical Analysis System (SAS Institute, Cary, NC, USA, version 8).

RESULTS

Effects of successional stage and nitrogen addition on litter decomposability and traits

Litter decomposability differed significantly among species from different successional stages (Table 3 and Appendix 1). The litter mass remaining after 8 weeks of litter incubation in the microcosms ranged from 71 % (B. phoenicoides in the N− treatment) to 35 % (T. maximum in the N+ treatment). Mass loss of all litters fitted the single exponential decay model (0·01 < P < 0·1). The ANOVA indicated significant differences in K_pot according to species successional stage, with early and intermediate species decomposing more rapidly than species from advanced successional stages (Fig. 1A). Within taxonomic groups, Poaceae had the slowest decomposition rates while Fabaceae had the highest (Fig. 1A).

Table 3.

Means of leaf and litter traits and decomposition decay at each successional stage and nitrogen supply

	N−			N+
Traits	Early	Intermediate	Advanced	Early	Intermediate	Advanced
Green leaves
SLAgreen (m2 kg−1)	26·6 (7·5)	20·8 (3·0)	13·8 (2·8)	29·0 (9·5)	21·9 (2·3)	15·9 (3·4)
LDMCgreen (mg g−1)	190 (41·3)	195 (50·3)	258 (60·8)	187 (56·4)	180 (51·3)	240 (65·2)
LTSgreen (MN m−2)	181 (320)	187 (243)	365 (541)	96 (145)	182 (243)	287 (423)
LRFgreen (J m−2)	787 (635)	816 (651)	1455 (1644)	731 (455)	854 (779)	1166 (1366)
LNCgreen (mg g−1)	24·4 (7·7)	26·3 (7·2)	21·1 (3·6)	35·7 (6·1)	39·3 (6·0)	35·4 (8·6)
LPCgreen (mg g−1)	2·8 (0·4)	3·9 (1·2)	2·3 (0·4)	2·8 (0·7)	3·2 (0·8)	2·4 (0·5)
LCCgreen (mg g−1)	409 (8·97)	410 (26·5)	415 (15·6)	420 (13·4)	426 (14·6)	427 (18·6)
Litter
LNClitter (mg g−1)	12·6 (4·1)	10·1 (6·3)	7·7 (2·8)	21·0 (5·4)	14·5 (4·3)	10·7 (4·2)
LPClitter (mg g−1)	2·1 (1·0)	2·4 (1·2)	0·9 (0·4)	1·6 (0·6)	1·2 (0·7)	0·7 (0·3)
LCClitter (mg g−1)	384 (10·5)	362 (30·0)	372 (47·0)	389 (13·7)	381 (21·2)	383 (43·5)
LIG litter (mg g−1)	167 (50·6)	15·3 (73·6)	149 (59·2)	158 (44·8)	165 (91·9)	134 (57)
CEL litter (mg g−1)	198 (65)	157 (84·5)	165 (70·9)	153 (62·8)	161 (45·1)	120 (30·5)
HEM litter (mg g−1)	138 (5·05)	128 (51·0)	134·4 (64·9)	163 (33·4)	179 (91·4)	156 (82·5)
LCH litter (mg g−1)	503 (13·3)	438 (185)	485 (73)	548 (94)	434 (109)	553 (136)
HLQ litter	0·66 (0·08)	0·65 (0·13)	0·68 (0·15)	0·65 (0·10)	0·65 (0·09)	0·68 (0·13)
Kpot (g kg−1 d−1)	22·7 (7·6)	23·4 (8·4)	17·4 (10·8)	23·4 (8·1)	27·1 (9·6)	16·6 (8·2)

Standard errors are given in parentheses (n = 5 for the early stage and n = 6 for the intermediate and advanced stage).

Fig. 1.

Means and s.e. of (A) decomposition rate K_pot, (B) leaf dry mass content (LDMC_green), (C) leaf tensile strength (LTS_green) and (D) resistance to fracture (LRF_green) for species from different successional stages (E, early; I, intermediate; A, advanced succession) at each nitrogen supply (N− and N+ treatments, as indicated). In the first part of each histogram, the mean and s.e. for all species combined (n = 17 for each N supply) are indicated. Results of post hoc tests (Student–Newman–Keuls comparisons) with successional stage as factor are given with Greek letters (F- and P-values are given in Table 3). Means and s.e. in the 12 species from each taxonomic group are shown in the second part of the histograms (separated by the dotted line). Results of post hoc tests with taxonomic group as factor are given with Roman letters (F- and P-values are given in Appendix 2).

Leaf and litter traits varied according to species successional stage. SLA_green decreased in advanced successional species. Species from advanced successional stages also had higher LDMC_green, LTS_green and LRF_green than species from earlier stages in both N supply treatments (Table 3 and Fig. 1B–D). Poaceae showed higher values of structural traits (LDMC_green, LTS_green and LRF_green) than species from other taxonomic groups (Table 3 and Fig. 1B–D). Species from the intermediate stage had higher LNC_green and LPC_green values than species typical of other stages (Table 3), with Fabaceae species showing the highest LNC_green values and Asteraceae showing the highest LPC_green values (Appendix 2). Regarding litter traits, species from early successional stages had higher LNC_litter values and species from early and intermediate stages had higher LPC_litter values (Table 3). LCC (green and litter), LIG_litter, CEL_litter, HEM_litter, non-labile compounds and HLQ_litter did not vary with species successional status (Table 3). Poaceae species had higher HEM_litter and LCH_litter values than species from other taxonomic groups (Appendix 2).

N addition had no effect on the litter decomposition rate but affected leaf and litter traits (Table 3). With increased N supply, SLA_green, LNC_green and LNC_litter increased (9, 33 and 36 %, respectively), LDMC_green, LTS_green and LPC_litter decreased (6·5, 6 and 28 %, respectively) whereas all other traits were not affected by N supply (Table 3).

Relationships between litter decomposability, leaf and litter traits

There were clear links between litter decomposability and leaf structural traits. LDMC_green, for instance, was most closely related to litter decomposability: species with low LDMC_green tend to decompose more rapidly than species with high LDMC_green (Table 4 and Fig. 2A). A negative relationship was found between LTS_green, LRF_green and K_pot (Table 4 and Fig. 2C, D). The relationship between K_pot and the two traits related to leaf physical strength (LTS_green and LRF_green) is mainly due to the leaves of Poaceae species which have higher values than all the other species. LPC_green was positively correlated to litter decomposability under the N+ treatment (Table 4, Fig. 2B). The non-significant relationship between K_pot and LPC_green in the N− treatment is due to D. glomerata: when this species is removed from the analysis, the relationship becomes significant (r_N− = 0·54, P = 0·03). This relationship could be attributed to the negative correlation between LPC_green and LDMC_green (Table 3). No significant relationship was found between K_pot and SLA_green, LNC and LCC of green leaves and litter.

Table 4.

Relationships (Pearson's correlation coefficients) between measured green living traits and initial litter traits (two levels of N supply are considered together) (n = 34 for each correlation)

	Kpot	SLAgreen	LDMCgreen	LTSgreen	LRFgreen	LNCgreen	LPCgreen	LCCgreen	LNClitter	LPClitter	LCClitter	LIGlitter	CELlitter	HEMlitter	LCHlitter
SLAgreen	ns	–	–	–	–	–	–	–	–	–	–	–	–	–	–
LDMCgreen	−0·79***	ns	–	–	–	–	–	–	–	–	–	–	–	–	–
LTSgreen	−0·66***	ns	0·69***	–	–	–	–	–	–	–	–	–	–	–	–
LRFgreen	−0·62***	ns	0·64***	0·92***	–	–	–	–	–	–	–	–	–	–	–
LNCgreen	ns	ns	ns	ns	ns	–	–	–	–	–	–	–	–	–	–
LPCgreen	0·57**	ns	−0·51*	ns	ns	ns	–	–	–	–	–	–	–	–	–
LCCgreen	ns	ns	0·4*	ns	ns	ns	ns	–	–	–	–	–	–	–	–
LNClitter	ns	0·37*	ns	−0·42*	–0·35	0·34*	ns	ns	–	–	–	–	–	–	–
LPClitter	ns	0·45*	ns	ns	ns	−0·42*	0·7***	ns	ns	–	–	–	–	–	–
LCClitter	ns	ns	0·38*	ns	ns	−0·45*	ns	0·86***	ns	ns	–	–	–	–	–
LIGlitter	ns	ns	ns	–0·36	−0·39*	ns	ns	0·44*	ns	ns	ns	–	–	–	–
CELlitter	ns	ns	0·38*	0·57**	0·64***	ns	ns	ns	ns	ns	ns	ns	–	–	–
HEMlitter	−0·41**	ns	0·47**	0·71***	0·75***	ns	ns	ns	ns	ns	ns	ns	0·93***	–	–
LCHlitter	–0·35	ns	0·39*	0·45*	0·49**	ns	ns	ns	ns	ns	ns	0·43*	0·95***	0·87***	–
HLQlitter	ns	ns	ns	0·68***	0·76***	ns	ns	ns	ns	ns	ns	−0·56**	0·70***	0·77***	0·47**

*P < 0·05; **P < 0·01; ***P < 0·001 (significance levels corrected by the improved Bonferroni procedure. Marginally significant results (0·05 < P < 0·10) are shown in italics.

Fig. 2.

Relationships between potential decomposition decay (K_pot) and (A) leaf dry matter content (LDMC_green), (B) leaf phosphorus content (LPC_green), (C) leaf tensile strength (LTS_green), (D) resistance to fracture of green leaves (LRF_green), (E) initial litter hemicellulose (HEM_litter) and (F) initial litter non-labile compounds (LCH_litter). Each point represent each species (n = 17 for each N supply); N− and N+ treatments as indicated. Pearson's correlation coefficients between all variables (r_total) and with each of the fertilization treatments are given (r_N− and r_N+).

Among the chemical traits of litter, HEM_litter and LCH_litter (marginally) were most closely negatively related to the decay decomposition rate (Table 4 and Fig. 2E, F). These relationships were conserved at both N levels. Leaf structural traits (LDMC_green, LTS_green and LRF_green) were closely correlated all together (Table 4).

Traits of living leaves are related well to litter quality. Strong positive correlations were found between LDMC_green and HEM_litter or LCH_litter (Fig. 3A, B). The same positive correlation was found between LTS_green and LRF_green, and HEM_litter and LCH_litter (Fig. 3C–F) with the Poaceae species having a determinant role in the relationship between LTS_green and litter traits.

Fig. 3.

Relationships between initial litter chemistry (hemicelluloses and non-labile compounds LCH) and (A, B) leaf dry matter content (LDMC_green), (C, D) tensile strength and (E, F) resistance to fracture of green leaves; N− and N+ treatments as indicated. Pearson's correlation coefficients between all variables (r_total) and with each of the fertilization treatment are given (r_N− and r_N+).

DISCUSSION

Leaf traits, litter quality, litter decomposability and successional stage

The first objective of the present study was to test whether successional changes in litter decomposability and associated traits were also apparent when species were grown under identical environmental conditions. Our common garden experiment demonstrated that species typical of later successional stages had tougher leaves (higher LDMC_green, LTS_green and LRF_green), producing litter that decomposed more slowly than species from earlier successional stages. The patterns observed here were consistent with studies conducted on individuals growing in the field (Kazakou et al., 2006) and also on ‘average’ community litters from the same old-field succession (Garnier et al., 2004; Cortez et al., 2007), and in successional seres from other sites in Europe (Quested et al., 2007; Fortunel et al., 2009). These studies demonstrated that litter from early successional stages or regularly disturbed sites tended to decompose more rapidly than litter from later stages or more stable sites. When we compared the K_pot of species grown in the common garden experiment (present study) with those grown in situ (Kazakou et al., 2006), a strong positive relationship was found (Pearson's r = 0·91; P < 0·001 for seven species common in both studies). Combining these results with the finding of the present study, we can advocate that the differences in litter decay observed during succession could be attributed to the differences in species composition rather than to changes in environmental conditions. Moreover, when a common litter was incubated for decomposition in the three successional stages no significant effect of the successional stage was found (data not shown). Based on these conclusions, we suggest that the patterns of litter decay rate observed along our successional gradient are robust enough to scale up from species to community level, in spite of potential mixing effects in the plurispecific litters when decomposition is assessed at the community level (see Hättenschwiler et al., 2005).

Does nitrogen supply affect leaf traits and litter decomposability?

Despite the recognition that N availability is an important factor controlling decomposition (Swift et al., 1979), the relative effects of exogenous (N in the environment, external to the decaying material) vs. endogenous N (N concentration of the litter itself) on decomposition remain unclear (Bridgham and Richardson, 2003). Several studies have reported significantly faster decay rates in response to increased external N availability (Hunt et al., 1988; Hobbie, 2000). Many others have reported either no significant change (Pastor et al., 1987; Hunt et al., 1988; Prescott, 1995; Bryant et al., 1998; Carreiro et al., 2000; Hobbie and Vitousek, 2000) or a decrease in decay rates (Magill and Aber, 1998; Carreiro et al., 2000).

In the present study, we tested whether species grown on nutrient-rich soils produced leaves and litters that decomposed faster than species grown on nutrient-poor soils. Our results demonstrated that N addition did result in changes of some leaf and litter traits. However, litter decomposability was not affected by these changes in leaf and litter traits. Even if some structural and chemical traits were modified by nitrogen addition, the responses of leaf traits to increased N supply were not strong enough to affect litter decomposability.

Our study therefore provides clear evidence that greater N availability does not result in more rapid litter decomposition for the studied species. These patterns are in agreement with the results of Knorr et al. (2005), Berg et al. (1993), Edmonds (1980) and Prescott (1995) who found that even if litter from fertilized trees contained twice as much N as litter from control trees they decomposed at the same rate.

Relationships between leaf traits and litter decomposability

Our second aim was to assess which trait of living leaves best relates to litter quality and litter decomposability. LDMC_green, LTS_green and LRF_green were negatively correlated with litter K_pot (Fig. 2A, C, D). Comparable results were found in species from Argentina and Great Britain for LTS_green (Cornelissen et al., 1999; Pérez-Harguindeguy et al., 2000) and in the same set of species measured in situ for the LDMC_green (Kazakou et al., 2006). Our results support the hypothesis of ‘afterlife’ effects of functional traits of living leaves on decomposability (Cornelissen and Thompson, 1997; Wardle et al., 1998; Pérez-Harguindeguy et al., 2000; Cornelissen et al., 2004) for plants grown under identical environmental conditions. Relationships between K_pot and leaf structural traits (LDMC_green, LTS_green and LRF_green) were found under both N supplies, which means that these patterns are independent of variations in soil N availability. In addition, when data from the present study are combined with experiments conducted when species were grown in situ and decomposed in microcosms (Kazakou et al., 2006) or litter beds (N. Pèrez-Harguindeguy et al., UNC-CONICET, Córdoba, Argentina, unpubl. res.), a single relationship between K_pot and LDMC is observed (Fig. 4). We may therefore conclude that the relationships between LDMC_green and litter decomposability are independent from species responses to environmental factors.

Fig. 4.

Relationships between potential decomposition decay (K_pot) and leaf dry matter content (LDMC_green). N− and N+ treatments (this study, r_N− = −0·81, P < 0·001, r_N+ = −0·78, P < 0·001); data from Kazakou et al. (2006, r_Kaz. = −0·75, P < 0·001), and data from N. Pérez-Harguindeguy et al. (UNC-CONICET, Córdoba, Argentina, unpubl. res.; r_P-H. = −0·74, P = 0·04*): r_total = −0·76***.

The positive relationship between LPC_green and K_pot in the N+ treatment, (Fig. 2B, r_N+ = 0·89, P < 0·001) can result from a higher P demand under high N supply as plants may have suffered from P limitation: P resorption efficiency as well as the N to P ratio were higher in the N+ treatment (see Kazakou et al., 2007). Phosphorus control over decomposition has been reported in regions where P availability is low compared with N availability (Aerts, 1997; Hoorens et al., 2003; Vivanco and Austin, 2006).

Concerning the relationships between traits of living leaves and litter traits, LDMC_green was positively related to HEM_litter which was the trait of litter best correlated to K_pot (Fig. 3A). Strong correlations were also found between leaf structure-related traits and the LIG_litter, CEL_litter, HEM_litter and non-labile compounds. Support for this interpretation is available from independent studies (Choong et al., 1992; Wright and Illius, 1995; Cornelissen et al., 1999), confirming that high concentrations of lignin and other carbon-rich compounds, particularly when invested in fibres, strengthen living leaves considerably.

In conclusion, among the traits quantified here, LDMC_green stood out as being best related to litter quality and thus K_pot. LDMC_green reflects the amount of mesophyll vs. structural compounds in a leaf (Garnier and Laurent, 1994; van Arendonk and Poorter, 1994). A high LDMC_green corresponds to a high proportion of vascular tissues and sclerenchyma (dense tissues) (Dijkstra and Lambers, 1989; Garnier and Laurent, 1994). In terms of chemical composition, this corresponds to leaves rich in (hemi)cellulose, insoluble sugars and lignin (Poorter and Bergkotte, 1992). In our study, high values of LDMC_green (dense leaf tissues) in leaf result in initial litter rich in HEM_litter with a slow decomposition rate. Moreover, LDMC_green is an easily measurable trait, much less variable and the best single variable for locating plant species on a resource acquisition–conservation axis (Wilson et al., 1999): species with high LDMC_green tend to conserve resources more efficiently in resource-poor environments and have lower growth rates than species with low LDMC_green (Poorter and Bergotte, 1992; Ryser and Aeschlimann, 1999). Additionally, the strong connection between LDMC_green and litter decomposition was also confirmed in studies measuring litter decomposition at the community level (Garnier et al., 2004; Cortez et al., 2007; Quested et al., 2007; Quetier et al., 2007; Fortunel et al., 2009).

Conclusions

Our results provide evidence that leaf traits and litter decomposability vary according to species successional stage. Nitrogen fertilization affected some leaf and litter traits but these changes were not translated into changes in their decomposability. Leaf dry matter content appears as the single trait of green leaves best related to decomposability, closely related to initial litter quality. This demonstrates the strength of a trait-based approach to predict changes in ecosystem processes as a result of species shifts in ecosystems.

APPENDIX 1

Results of three-way ANOVA (F-values and probabilities) on the effects of species, successional stage and N supply for the 17 species on each measured variable

Parameter	Species (d.f. = 14)	Successional stage (d.f. = 2)	N supply (d.f. = 1)
SLAgreen	23·9***	212***abc	12·5**
LDMCgreen	48·7***	117***bba	11·6**
LTSgreen	33·3***	17·2***bba	3·48ns
LRFgreen	37·6***	17·1***bba	1·76ns
LNCgreen	1·86ns	2·16ns	47·27***
LPCgreen	3·02*	14·9**bac	1·24ns
LCCgreen	4·05**	1·21ns	12·5**
LNClitter	3·12*	15·5**abc	19·6**
LPClitter	2·21ns	10·7**aab	10·6**
LCClitter	10·5***	3·36ns	7·11*
LIGlitter	1·13ns	0·99ns	0·75ns
CELlitter	1·66ns	0·70ns	0·16ns
HEMlitter	2·20ns	1·38ns	0·43ns
LCHlitter	1·06ns	0·35ns	0·29ns
HLQlitter	3·08*	1·83ns	0·88ns
Kpot	11·0***	25·4***aab	0·05ns

Superscript letters indicate the results of post hoc tests (Student–Newman–Keuls comparisons) with successional stage as factor (results were identical for each nitrogen supply).

* P < 0·05; ** P < 0·01; *** P < 0·001; ns, not significant.

APPENDIX 2

Results of two-way ANOVA (F values and probabilities) on the effects of successional stage, taxonomic groups and their interaction for the 12 species (Table 1)

Variable	Successional stage	Taxonomic group	Suc. Stage × Phylo. group
SLAgreen	173***	63·2***	20·1***
LDMCgreen	77·7***	179***	40·5***
LTSgreen	ns	131***	ns
LRFgreen	10·7***	180***	18·7***
LNCgreen	ns	6·80**	ns
LPCgreen	17·7***	8·55***	6·47***
LCCgreen	3·79*	ns	ns
LNClitter	9·66**	8·81***	2·94*
LPClitter	8·35**	3·92*	2·24*
LCClitter	ns	15·9***	16·72***
LIGlitter	ns	ns	ns
HEMlitter	ns	7·15*	ns
CELlitter	ns	10·1**	ns
Kpot	6·58*	19·2***	3·55*