A plant economics spectrum of litter decomposability


Correspondence author. Department of Forest Ecology and Management, Swedish University of Agricultural Science, 901 83 Umeå, Sweden. E-mail: gregoire.freschet@slu.se


1. Recent evidence indicates tight control of plant resource economics over interspecific trait variation amongst species, both within and across organs, referred to as ‘plant economics spectrum’ (PES). Whether and how these coordinated whole-plant economics strategies can influence the decomposition system and thereby impact on ecosystem carbon and nutrient cycling are yet an open question. More specifically, it is yet unknown whether plant functional traits have consistent afterlife effects across different plant organs.

2. To answer those questions, we conducted a common-garden decomposition experiment bringing together leaves, fine stems, coarse stems, fine roots and reproductive parts from a wide range of subarctic plant types, clades and environments. We measured all plant parts for the same (green and litter) plant economics traits and identified a whole-plant axis of carbon and nutrient economics.

3. We demonstrated that our local ‘PES’ has important afterlife effects on carbon turnover by driving coordinated decomposition rates of different organs across species. All organ decomposabilities were consistently controlled by the same structure-related traits (lignin, C and dry matter content) whilst nutrient-related traits (N, P, pH, phenols) had more variable influence, likely due to their contrasting functions across organs. Nevertheless, consistent shifts in elevation of parallel trait–decomposition relationships between organs indicate that other variables, potentially related to organ dimensions, configuration or chemical contents, codetermine litter decomposition rates.

4. Whilst the coordinated litter decomposabilities across species organs imply a coordinated impact of plant above-ground and below-ground litters on plant–soil feedbacks, the contrasting decomposabilities between plant parts suggest a major role for the relative inputs of organ litter as driver of soil properties and ecosystem biogeochemistry. These relationships, underpinning the afterlife effects of the PES on whole-plant litter decomposability, will provide comprehensive input of vegetation composition feedback to soil carbon turnover.


Plant litter decomposition, a major driver of carbon and nutrient cycling in terrestrial and freshwater ecosystems, controls the provision of fundamental ecosystem services such as soil formation, nutrient availability and atmospheric composition, with feedback to vegetation composition. One major aim of ecology is to model how functional features of vegetations differing in species composition feed back to soil carbon turnover, and thereby atmospheric chemistry and climate, in different biomes (Sitch et al. 2003; Cornwell et al. 2009). Here, we make a leap forward towards this aim by taking an explicit whole-plant functional approach to assessing litter decomposition rates, more specifically by linking the ‘plant economics spectrum’ (PES, Freschet et al. 2010a) to litter decomposability.

Whilst leaf litter decomposition rates are strongly determined by climate (Berg et al. 1993; Parton et al. 2007) and community composition of soil organisms (Lavelle et al. 2006), litter quality (‘species identity’) is their predominant driver within biomes (Cornwell et al. 2008). Structural and chemical leaf traits have ‘afterlife’ effects on litter decomposability (Cornelissen et al. 2004). Indeed, interspecific variation in traits of fresh leaves and that of leaf litter tends to be strongly correlated (e.g. Freschet et al. 2010b). Thus, lignin content (Meentemeyer 1978), physical toughness (Pérez-Harguindeguy et al. 2000), polyphenol content (Coq et al. 2010) or specific leaf area and dry matter content (DMC; dry to water-saturated weight ratio) (Garnier et al. 2004; Kazakou et al. 2006) can substantially affect leaf decomposition rates. Also, the high nutrient requirement of decomposer organisms creates nutrient-limited conditions for decomposition processes (Enríquez, Duarte & Sand-Jensen 1993). Thus, nitrogen (N), phosphorus (P) and calcium (Ca) contents ((Enríquez, Duarte & Sand-Jensen 1993; Aerts 1997) and pH (as a proxy for basic cation content and antimicrobial organic acids; Cornelissen et al. 2006) are usually significant predictors of leaf litter decomposition rates.

Most litter turnover studies linking vegetation composition to decomposition have focused on the relation between leaf traits and leaf litter decomposability. Some evidence exists that interspecific variation in litter quality is also the predominant driver of root litter decomposition (Silver & Miya 2001), and the huge range in wood functional trait values (Chave et al. 2009) and within-site wood decomposition rates (Harmon et al. 1995; van Geffen et al. 2010) suggests that similar pattern may exist for plant stems too. Indeed, several chemical traits related to leaf decomposition, such as N, Ca and lignin concentrations, also impact root decomposition, (Silver & Miya 2001; Vivanco & Austin 2006), whereas N, P and tissue density affect stem decomposition (Chambers et al. 2000; Weedon et al. 2009). However, with only few studies available on interspecific variation in stem and root decomposability, we still do not know whether the traits underpinning decomposition rates, or their relative contributions, have the same effect across plant parts. Whilst it seems likely that the same traits will have broadly similar effects on litter decomposition of distinct plant organs, differences in the magnitude of their impact are likely. For instance, whilst litter N or P contents are major determinants of colonization–degradation by soil organisms (Cornwell et al. 2008), whether these nutrient pools are active (e.g. in enzymes for leaf photosynthesis, root adsorptive capacity) or passive (e.g. stem or root storage, recalcitrant defence compounds) will partly determine their chemical form after senescence and thereby modulate their availability to decomposers. Traits related to physical support functions, expressed in plant allometric relationships including organ sizes, may play important roles too; for instance, tree trunk diameter predicted variation in decomposition rates amongst 15 Bolivian tree species (van Geffen et al. 2010). Any such differences in organ function, as expressed in structural and physiological differences, might cause shifts in trait–decomposability relationships between organs.

Empirical evidence is growing that plant species possess integrated strategies across their organs with regards to C and nutrient economy (Freschet et al. 2010a), which are moreover robust to geographical scaling (Kerkhoff et al. 2006; Liu et al. 2010). In other words, (i) each vegetative plant organ (leaves, twigs, main stems, coarse roots, fine roots) seems to obey a fundamental trade-off between traits inferring rapid resource acquisition and traits leading to resource conservation, owing to direct and indirect mutual dependencies between these traits (Reich et al. 2003; Chave et al. 2009; Elser et al. 2010) and (ii) those traits seem to be generally coordinated across vegetative organs, likely due to plant physiological, ontogenetic and allometric constraints (Niklas & Enquist 2002; Wright et al. 2006; but see Baraloto et al. 2010 for decoupling of coarse wood and leaf traits amongst tropical trees). As over half the global variance in these traits is at local to plant community scale (Wright et al. 2004; Freschet et al. 2011a), many contrasting plant economic strategies are already found within a local flora (e.g. Freschet et al. 2010a).

Assuming that consistent traits and economic strategies drive decomposition across plant organs, decomposability is probably coordinated across organs too. This would indicate tight control of plant economics on the carbon and nutrient turnover in ecosystems.

At present, however, there is very little empirical evidence for this. Wang, Liu & Mo (2010) observed significant correlation between leaf and root decomposition of four tree species only, whilst Hobbie et al. (2010) did not find any consistent relationship for eleven tree species. Here, we present the first empirical evidence of a clear association between the PES and litter decomposability, in a comprehensive multispecies, multi-organ decomposition study covering a wide range of subarctic plant functional types and aquatic, riparian and terrestrial habitats. More precisely, we test the hypotheses that (i) the decomposability of distinct plant organs (leaves, fine stems, coarse stems, fine roots and reproductive parts) is controlled by the same structure-related (lignin, DMC, C) and chemical (N, P, phenolic contents, pH) traits, although the relative influence of these traits might shift across organs. Considering the strong trait coordination found across leaves, stems and roots in a subarctic flora (Freschet et al. 2010a), we further hypothesize that (ii) interspecific variation in litter decomposability is coordinated across plant vegetative organs. Finally, assuming the validity of the two previous hypotheses, we test that (iii) the locally operating PES is a good predictor of the decomposability of vegetative plant organs.

Materials and methods

Study Area, Species Types and Sampling

The plant species were sampled around the Abisko Research Station, North Sweden (68°21′N, 18°49′E), at low altitude (350–400 m a.s.l.), below the tree line. During the 1999–2008 decade, this area had a mean annual rainfall of 352 mm and mean January and July temperatures of −9·7 and 12·3 °C, respectively (meteorological data, Abisko Research Station). The forested area, which was the focus of this study, features strongly organic podsol soils and covers most of the landscape below 700 m a.s.l. except for occasional treeless mires, fens and bogs. The three most distinct ecosystem types within the selected forested sites were dry birch forest with ericaceous understorey, riparian birch forest with herbaceous and shrubby understorey and forested freshwater systems (ponds and streams). Seven sampling sites (c. 20-m transects) each including all three ecosystem types were used to identify the dominant vascular species (roughly 80–90% of total plant annual biomass production) of each plant community (see Cornelissen et al. 2003). These included 15 species from the dry forest, 18 from the riparian forest and seven from aquatic systems, altogether covering seven growth forms and six higher clades (see Appendix S1 in Supporting information for species list and characteristics).

All 40 species were sampled for leaves, 38 for fine stems (<3 mm diameter), 11 for fine roots (<2 mm diameter), 19 for reproductive organs (ovary plus receptacle), all six tree species for coarse stems (c. 50 mm diameter) and two species for coarse roots (c. 50 mm diameter). All these materials were sampled both fresh and as litter (except reproductive organs: only as litter). A minimum of 10 different plant individuals (up to 50 for some species) were used for each species and organ to ensure the representativeness of the pool collected. The details of coarse stem and coarse root sampling, for which several stages of decay were collected, are in Appendix S2 (Supporting information). To ensure fair comparison of fine root types in terms of function, only the finest root branch order visible to the naked eye was considered for each species. Twigs of woody species (<3 mm diameter) were considered analogous to stems of forbs and graminoids in terms of function.

For each organ, one part of the collected sample was placed in paper bags and air-dried for chemical analyses (fresh and litter material) and incubations (for litter material only) whilst the other was immediately placed in a closed plastic bag for DMC analysis. For root sampling, plant individuals were excavated and brought to the laboratory. Soil and alien material were washed off the root system before both mature, undamaged roots and darkening, turgescence-losing roots (i.e. root litter) were collected on the plant. Large mycorrhizal rhizomorphs were brushed off. For dead woody fine stems, only those that looked similar to living ones except for the absence of leaves were selected (on the plant). For all organs, parts with obvious symptoms of damage, infection or herbivore activity were avoided. Petioles and rachides were included as part of the leaf. To avoid the effects of seasonal variation, all living leaves and fine stems were collected whilst fully mature, in late July or early August 2007 (see Freschet et al. 2010a). All litters of leaves, fine stems and reproductive organs were sampled when fully senesced between mid-August and mid-October 2007 (see Freschet et al. 2010b). Fine roots were collected partly in August 2007 and partly in August 2008 owing to the labour-intensive process involved.

Plant Trait Measurements

All collected materials were measured for C, N, P and lignin contents, as well as for pH. For these analyses, air-dried subsamples were ground and subsequently oven-dried for 24 h at 60 °C. Carbon and nitrogen concentrations were measured by dry combustion on a NA 1500 elemental analyser (Carlo Erba, Rodana, Italy). Phosphorus was measured by acid digestion as referred to in Freschet et al. (2010a). Lignin concentration was determined by the extraction of non-ligneous compounds as described in Freschet et al. (2010a). For pH, 0·15 mL of each ground sample was shaken with 1·2 mL demineralized water in an Eppendorf tube for 1 h at 250 rpm. After centrifugation at 9000 g for 5 min, pH of the supernatant solution was measured (Cornelissen et al. 2006).

All materials except reproductive organs were measured for DMC [dry weight (mg) to water-saturated weight (g) ratio] following Cornelissen et al. (2003). The protocol was modified for woody fine stems and coarse stems–roots, which were submersed in water for 3 and 9 days respectively to ensure homogeneous filling of air spaces.

Leaves, fine stems and coarse stems–roots were measured for tannins and non-tannin phenols using the Folin–Ciocalteau method, as described by Waterman & Mole (1994), modified according to Makkar (2003).

Decomposition Study

All litter materials were incubated in an outdoor litter bed at Abisko Research Station, following Cornelissen et al. (2004). Our experiment was not designed to capture natural in situ decomposition rate but rather to provide standardized (‘common-garden’) environmental conditions for large interspecific, interorgan comparisons. Decomposition was measured over two full years of incubation, but for leaf litter additionally after 1 year.

Coarse stem and coarse root litter decomposition followed a new protocol allowing for reconstruction of long-term decomposition rates using a number of short-term incubations of woody debris from various stages of decay (details in Appendix S2, Supporting information and Freschet et al. 2011b in press). Briefly, for each wood species and type (stem or root), wood samples of a broad range of stages of decay were collected, split into four replicate subsamples, sealed in mesh and then incubated for 2 years in the litter bed. After retrieval, the changes in relative density (proxy for litter mass per volume) for fragments of each wood type were entered into an iterative optimization procedure which modelled the best-fit parameters of each of three possible decay functions most commonly used to describe wood decay: linear, exponential, sigmoid. For each wood type, the model providing the best fit was subsequently used for T1/2 calculation.

For all other materials, four subsamples (eight for leaves) of each air-dry litter sample were weighed and sealed into litterbags. An additional subsample was oven-dried (60 °C, 72 h) to correct air-dry mass of litterbag samples for residual humidity. The amount of material and the dimensions of the bags were adjusted to each material type to standardize litter densities and textures inside the bags. A 1-mm mesh was used for most material given the absence of large soil invertebrates in this subarctic area and 0·3 mm for very thin materials (providing access to microbes and micro-invertebrates). In a control test, no significant difference in 2-year mass loss % was found between litters sealed into 0·3 vs. 1-mm mesh for Betula pubescens and Vaccinium uliginosum (paired t-tests).

The litter bed for incubation consisted of rectangular wooden frames sunk into the ground, including a free-draining foundation layer of grit stones (particle sizes 10–20 mm) on top of the original soil profile. They were covered by a 20-mm layer of mixed fresh and old litter collected in September 2007 from nearby dry and wet birch forests and ponds, together representing the ecosystems sampled for litter. This way, all litter types were presumably inoculated with the microbial communities of all ecosystem types. On 23 April 2008, all litterbags were laid out flat, without overlap, and covered by a 10-mm layer of the same mixed litter. Four separate blocks hosted the four groups of replicates per litter type. The litterbags were subject to the local climatic influences and did not receive any treatment. They were harvested after 2 years of incubation (23 April 2010, and 23 April 2009 for first leaf harvest), whilst still frozen and stored at −16 °C. After defrosting, the adhering soil, soil fauna and other alien material were removed from the decomposed litter by gentle brushing and rinsing with tap water. Litter samples were then dried (60 °C, 96 h) and reweighed. For all litter materials except coarse stems and roots, single negative exponential models were used to derive decomposition constants (k) from % mass loss against time. Finally, half-lives of decomposing material [T1/2; time (year) needed to reach 50% mass loss] were derived from their respective decomposition models to obtain comparable measures of decomposability across all litter types including stem and root woody debris which required the use of linear or sigmoid models.

Data Analysis

To comply with normality assumptions of all subsequent parametric tests, T1/2 of all organs and most organ functional traits (except DMC, C, pH and C/N) were log10-transformed. The predictive value of plant functional traits for T1/2 of each organ was assessed using ordinary least squares (OLS) regressions. Coarse stems and coarse roots were considered similar in structure and function and pooled for this analysis. Once we had identified the best predictor of T1/2 across organs (i.e. lignin content), we looked within each organ for complementarity and interaction between this and the other traits as possible predictors of T1/2 using multiple linear regressions.

We compared the slopes and intercepts of trait–decomposability relationships for different vegetative organs by standardized major axis (SMA) regressions and routine procedures from the smatr freeware (Warton et al. 2006) (Fig. 1). Bivariate relationships between organs’T1/2 were described with SMA regressions (Fig. 2). We compared decomposition rates of leaves to those of fine stems and those of fine roots using paired t-tests.

Figure 1.

 Standardized major axis (SMA) regressions between each organ litter decomposition half-life (T1/2) and (a) lignin content, (b) dry matter content (DMC) and (c) nitrogen content. Symbols: inline image leaves, inline image fine stems, inline image fine roots, inline image coarse stems and inline image coarse roots; r2 is SMA regression coefficient; ***P < 0.001, **P < 0.01, *P < 0.05, †P < 0.10, (ns) nonsignificant P-value. Inset: significance of shifts in slope (c) or elevation – (a) (b) following nonsignificant differences in slope – between each separate plant organ SMA regression.

Figure 2.

 Standardized major axis regressions (r2) between litter decomposition half-lives (T1/2) of leaves, fine stems and fine roots. ***P < 0.001, **P < 0.01, *P < 0.05.

Data on the carbon and nutrient economics spectrum for each organ and the PES were taken from Freschet et al. (2010a), who performed principal component analyses (PCA) on trait data for living leaves, fine stems and fine roots of 35 of the same 40 species studied here. Separate PCAs were thus performed for each organ and for all three organs together (see Table S1, Supporting information for each PCA main axes’ respective factor loading, and the proportion (%) of variance explained by each economics spectrum axis). Following the high proportion of variance explained by each first PCA axis and the relatively homogeneous impacts of variables on each first PCA axis, first axis species scores of leaf, fine stem, fine root and whole-plant PCAs were used to quantify leaf, fine stem, fine root and whole-plant economics spectra, respectively. OLS regressions were used to test the predictive power of leaf, fine stem and fine root organ economics spectra on their respective organ T1/2 (Fig. S2, Supporting information), and of the PES on each organ T1/2 (Fig. 3).

Figure 3.

 Relationships between the principal component analyses first axis score representing the plant economics spectrum and decomposition half-lives of different organs. Symbols: coarse roots. ***P < 0.001, **P < 0.01, *P < 0.05, †P < 0.10.


Over the 2 years of incubation, leaf materials displayed an average 69% mass loss (from 32% to 99%), fine stems 39% (from 8% to 96%), fine roots 28% (from 8% to 65%) and reproductive organs 58% (from 17% to 95%).

Relationships between Plant Organ Litter Traits and Litter Decomposability

All structure-related traits (DMC, C, lignin content) were good predictors of the interspecific variation in decomposability of any given organ (except DMC vs. fine root decomposability; Table 1). In contrast, the goodness-of-fit of relationships between several chemical traits (pH, N, non-tannin phenols) or structural to chemical trait ratios (lignin/N, C/N) and decomposability varied widely amongst organs. Only leaves displayed significant relationships between N and T1/2 and pH scaled with leaf, fine stem and coarse stem–root T1/2 but not with fine root or reproductive organ T1/2. Tannin and P contents showed no clear trends with T1/2 within any organ.

Table 1.   Relationships between functional traits of litter and living organs and ‘decomposition half-life’ (T1/2) across species for each organ
 Leaves (n = 40)Fine stems (n = 38)Fine roots (n = 11)Coarse stems and roots (n = 8)Reproductive parts (n = 19)
R P-value R P-value R P-value R P-value R P-value
  1. OLS regression coefficients (R) are displayed in bold when relationships are significant (P < 0.05). For multiple regressions, significance is displayed for each parameter except interactions. na is non-available data. n is the number of species.

Litter traits
 Lignin (%) 0.73 <0.001 0.90 <0.001 0.91 <0.001 0.85 0.007 0.73 <0.001
 DMC (mg g−1) 0.75 <0.001 0.88 <0.0010.400.22 0.89 0.003nana
 Carbon (%) 0.55 <0.001 0.88 <0.001 0.63 0.036 0.80 0.017 0.85 <0.001
 Nitrogen (%) −0.42 0.0070.040.83−0.150.66−0.360.390.080.75
 Phosphorus (%)−0.030.84−0.010.97−0.340.31−0.400.33−0.320.18
 C/N 0.47 0.0020.120.460.270.430.480.230.270.26
 Lignin/N 0.75 <0.001 0.74 <0.001 0.72 0.0130.620.1 0.52 0.022
 Lignin + N 0.80 <0.001; 0.025 0.92 <0.001; 0.027 0.92 0.49; 0.270.870.25; 0.64 0.86 <0.001; 0.004
 Lignin + DMC 0.81 0.007; 0.022 0.94 <0.001; <0.001 0.94 0.024; 0.34 0.93 0.55; 0.73nana
 Lignin + N + DMC 0.87 0.20; 0.07; 0.85 0.95 0.046; 0.59; 0.45 0.96 0.80; 0.67; 0.80 0.95 0.54; 0.19; 0.07nana
Living organ traits
 Lignin (%) 0.66 <0.001 0.87 <0.001 0.82 <0.001 0.86 0.006nana
 DMC (mg g−1) 0.86 <0.001 0.94 <0.0010.340.3−0.280.5nana
 Carbon (%) 0.63 <0.001 0.90 <0.001 0.73 0.0100.370.37nana
 Nitrogen (%)0.450.0030.390.0150.440.17−0.670.07nana
 Phosphorus (%)−0.160.330.410.0100.150.66−0.790.06nana
 C/N 0.55 <0.001 0.51 0.001−0.140.68 0.78 0.022nana
 Lignin/N 0.71 <0.001 0.82 <0.001 0.73 0.011 0.88 0.004nana
 Tannins (mg g−1)−0.470.51nana
 Non-tannin phenols (mg g−1)0.270.09 0.52 <0.001nana−0.790.07nana

Lignin consistently explained >53% of the variance in organ T1/2 and was generally the best single trait predictor of decomposability, followed by DMC (>56%; except for fine roots) and C content (>30%; Table 1). Combining lignin and DMC in multiple regressions against T1/2 added substantial explanatory power to lignin predictions alone for leaves (+13% explained variance; P < 0·05) and fine stems (+8%; P < 0·001) but not for fine roots and coarse stems or roots. Litter chemical traits did not significantly increase predictive power of lignin with respect to T1/2 variance, except N content for leaves (+15% explained variance; P < 0·05), fine stems (+5%; P < 0·05) and reproductive parts (+21%; P < 0·01). Combining lignin, DMC and N significantly explained over 76% of the variance in leaf T1/2 (P < 0·001) and 90% of stem T1/2 variance (P < 0·001). With respect to organ decomposabilities, the information provided by structural and nutritional traits was thus partially redundant or complementary depending on the organ.

Trait–decomposability regressions typically revealed similar slopes between organs, with consistent shifts in elevations (intercepts). For instance, at a given lignin content, decomposition rates ranked as leaves >fine stems >fine roots >coarse stems and roots (Fig. 1a). However, the change in slope in decomposition rate per unit change in lignin content was similar across organs. Very similar patterns of common slope and significant elevation shifts were found for DMC (Fig. 1b), P, C, pH, tannins, non-tannin phenols and lignin/N (data not shown). In contrast, N (Fig. 1c) and C/N displayed some dissimilarities between slopes. For instance, the N–T1/2 relationship of coarse stems had flatter slope than those of any other vegetative organ (Fig. 1c).

Strong Covariation of Decomposability of Different Vegetative Plant Organs

Interspecific variations in leaf, fine stem and fine root T1/2 were strongly coordinated (Fig. 2). Coarse stem and reproductive organ T1/2 were not significantly coordinated with that of any other organs (data not shown).

The position of species relative to the 1 : 1 axes of each regression (Fig. 2) illustrates that leaves decomposed generally faster than fine stems (P < 0·001), which themselves decomposed marginally slower than fine roots (P = 0·08). Visual inspection of those relationships indicated that the coordinated pattern of T1/2 across organs was partly owing to relationships within but also between phylogenetic, plant type or environmental groups (Fig. S1, Supporting information). Nevertheless, we observed no clear phylogenetic, plant type or environmental group clustering away from the general regression lines, for any organ. In other words, despite a few outliers, no plant type or clade seemed to consistently offset either the slope or intercept of organ decomposability relationships. Similarly, plants from aquatic, riparian and terrestrial environments fitted consistently in the pattern of interspecific decomposability covariation between vegetative organs.

Consistent with the above results, leaf and fine stem economics spectra were good predictors of interspecific variation in their respective organ decomposability (R2 of 0·57 and 0·76, respectively, P < 0·001 in both cases; Fig. S2, Supporting information). In contrast, the fine root economics spectrum did not significantly predict fine root decomposability (R2 = 0·02; P = 0·70; Fig. S2, Supporting information). Nevertheless, the PES was a highly significant predictor of the decomposabilities of each of the vegetative organs (Fig. 3). Thus, following a gradient from the most nutrient acquisitive to the most nutrient conservative species, the decomposition half-lives of leaves (R2 = 0·53), fine stems (R2 = 0·64), fine roots (R2 = 0·49) and coarse stems (R2 = 0·86) decreased exponentially (P < 0·05 in all cases; Fig. 3).


This study revealed a remarkable interspecific coordination in decomposability between the main organs of plants resulting in a ‘PES of litter decomposability’, concomitant with consistent shifts in intercept of trait–decomposability relations between organs. Below, we synthesize these findings and their ecological implications.

The Same Structure-Related Traits Determine Interspecific Variation in Decomposability for Different Plant Organs

Our results show that the relative investment of plants in dense, long-lived, mechanically reinforced structures (reflected by traits such as DMC, C and lignin content) is what predominantly drives the interspecific variation in decomposability of every plant organ in our subarctic flora. In contrast, nutrient and phenolic contents were typically weaker predictors, with diverging predictive powers across organs. Nevertheless, structural (lignin, DMC) and chemical (N) traits together were better predictors than structural traits alone for several high-turnover organs (leaves, fine stems and reproductive parts). Our first hypothesis is thus not unequivocally supported by our findings.

In line with our findings, the few studies that investigated the role of structural traits on non-leaf litter decomposition also attributed a major role for lignin (e.g. Taylor et al. 1991; Vivanco & Austin 2006; Weedon et al. 2009). However, whilst Taylor et al. (1991) suggested a disruption in lignin–decomposition relationship for materials of high (>28%) lignin content, we found a linear relationship encompassing plant material with more than 40% lignin (Fig. 1). The strong impact of structural features on litter decomposabilities is supported by the complementarity of lignin and DMC in predicting litter decomposition rates, especially for leaf and fine stem litter (Table 1). As DMC represents the proportion of dense to light tissues, it differentiates for instance between the soft, nutrient-rich mesophyll of leaves and the tough supportive and conductive tissues of leaf veins (Garnier & Laurent 1994; Kazakou et al. 2006). For stems, which generally consist mostly of supportive and conductive tissues, this distinction appeared almost as relevant as for leaves. This confirms the value of DMC as a predictive plant trait for litter decomposability (Garnier et al. 2004). The poor relation with fine root decomposability is inconclusive in our study, given the low number of species represented. The predictive power of C content for decomposability is particularly good for highly lignified tissues such as fine and coarse stems where it correlates strongly with lignin and DMC. However, C is also abundant in nonstructural compounds and can therefore only be a crude indicator of litter structural composition.

Whilst Wardle et al. (1998) observed a consistent effect of organ N on leaf, stem and root decomposition of 20 herbaceous species, we found inconsistent effects of N content on decomposition, depending on the organ considered. Whilst leaf nitrogen content influenced leaf decomposition, this relation was not apparent in any other organs. To explain these contrasting findings, we hypothesize that initial N content substantially affects litter decomposition only if it is higher than that of the surrounding litter substrate. In low N litters, decomposers rely largely on exogenous N sources (Parton et al. 2007). When N availability is higher in the surrounding substrate than in the litter under study, such as non-leaf litter incubated in the abundant young and old, thus less N-immobilizing leaf-dominated litter mixture matrix in our litter bed, decomposers of the non-leaf litter may have relatively steady access to exogenous N sources. Additionally, the relationship between N and decomposition could reach a threshold at high litter N content where microbial growth is no longer N-limited (Taylor et al. 1991). The N–decomposition relationship may thus hold only for intermediate values of litter N, explaining the weak impact of N on leaf decomposability and its nonsignificant impact on all other plant organs. If true, this implies strong interactive effects between different litter types, also according to their position in the litter layer, and could potentially explain part of the inconsistent effect of initial N content on leaf litter decay across studies (Hobbie 2005). However, in contrast with this explanation, some evidence also exists that interspecific rankings of litter decomposition rates across multiple species of contrasting N contents are remarkably robust to the contrasting nutritional status of different litter bed matrices (Cornelissen et al. 1999; Quested et al. 2003; J. H. C. Cornelissen and H. M. Quested, unpublished data). Further testing is thus needed in this respect. Another potential explanation for the inconsistent impact of N on litter decomposability relates to litter secondary chemistry. The wide range of plant types used in our study, with highly variable secondary chemistry, implies highly species-dependent complexation reactions between N and phenolic compounds which are likely to obscure N–decomposability relationships. This is however poorly supported by the largely nonsignificant impact of phenolics on litter decay rates. As put forward by Hobbie (2005), it is also possible that commonly observed relationship between litter N and decomposability may not stem from straightforward N limitation but at least partly through indirect correlation with other traits, as observed in this study between N and DMC for leaves but no other organs.

The differential influences of litter pH and the amount of low–molecular weight phenolic compounds on the decomposability of different plant organs suggest that these traits may have different meanings across organs with respect to the decomposition process. Indeed, whilst pH appears to be a good proxy of basic cation content and/or antimicrobial organic acids (Cornelissen et al. 2006) in leaf, fine stem and wood litters, pH variation in fine roots does not seem to be associated with the same properties in the subarctic flora. Similarly, whilst a large proportion of low–molecular weight phenolic compounds appear to have antimicrobial properties in fine stems, they seem to predominantly serve different functions in leaves (see also Aerts, van Logtestijn & Karlsson 2006).

The consistency of trait–decomposability relationships across organs is confirmed by the similarities in regression slopes between organs (except for N and C/N), which supports our first hypothesis. At the same time, major differences between organs in terms of trait–decomposability regression elevation (intercept) reflect substantial intrinsic differences between organs. For instance, at a given lignin content (e.g. 20%), leaves decomposed 1·6 times faster than fine stems, 2·1 times faster than fine roots and 3·1 times faster than coarse stems. These elevation shifts suggest the predominant impact of additive effects of several covarying traits, e.g. lignin, DMC, N, rather than interactions between traits, which would imply slope heterogeneity. However, these shifts may also stem, at least partly, from the changing ecological or physiological function of some traits across organs. For instance, potential differences in the configuration of lignin deposition between organs would lead to more or less recalcitrant structures for decomposers (see Vanholme et al. 2010). Contrasting proportions of different forms of organic N may also affect the decomposability of different organs. Organs key to uptake and assimilation, such as leaves and fine roots, hold more easily degradable forms of organic N (particularly photosynthetic enzymes or transmembrane proteins) than organs with predominant support and transport functions, in which most N might be in structural or defence-related compounds (e.g. microtubules, phloem proteins; Hayashi et al. 2000). Large differences between organ structure and function are also likely to explain shifts in organ decomposability. For instance, the typically flat structure of leaves should provide a relatively larger surface for microbial attacks and selective feeding by soil fauna compared to cylindrical organs such as stems. Relative access to decomposers, as reflected by the area/volume ratio of plant parts, may also explain why stem diameter, which is inversely proportional to the area/volume ratio, is an important factor for stem decomposition (Harmon et al. 1995; van Geffen et al. 2010), potentially explaining part of the shift in decomposability between fine and coarse stems and roots. Besides, the surface properties of each organ differ strongly. The barrier created by the bark of coarse stem and root litter likely provides better protection against decomposers than the lower side of leaves, with relatively thin cuticles and high densities of stomata providing microbial access, or absorptive fine root tissues. This is illustrated by the often ‘empty bark’ of highly lignified or suberized phloem of highly decomposed roots or stems compared to the vein skeletons of decomposed leaves.

Considering that the major influence of climatic factors on litter decay rates is through indirect effects via plant and litter quality (Zhang et al. 2008), the trait–decomposability relationships revealed here are likely to hold across climatic areas (e.g. Cornelissen et al. 1999). The relatively wide range of plant types and trait values (except perhaps for secondary chemistry) covered in our study also suggest that these relationships might generally hold for other floras, albeit with potential changes in the relative impact of each trait, e.g. with larger impact of secondary compounds on decomposability in the tropics (Coq et al. 2010). Besides, as litter decomposability is largely the result of interactions between litter quality and the decomposer community, the relative importance of litter traits might also vary slightly with the perception of quality by distinct decomposer communities (Strickland et al. 2009).

Coordinated Variation of Plant Organ Decomposabilities: The ‘Afterlife’ of the Plant Economics Spectrum

In support of our second hypothesis, we demonstrate for the first time that across a large number of plant functional types, contrasting clades and both aquatic and terrestrial species, plant decomposability is coordinated across its vegetative organs. This striking coordination could be a direct consequence of, simultaneously, the consistent role of functional traits on decomposition across organs identified here and the consistent covariation of those traits between vegetative organs (Kerkhoff et al. 2006; Freschet et al. 2010a). The absence of a relationship between leaf and fine root decomposabilities of 11 woody deciduous and evergreen species (Hobbie et al. 2010) suggests that the positive coordination reported in our study might be partly driven by plant functional type or clade differences. Visual inspection of our data appears to partly support this (Fig. S1, Supporting information), although no formal test could be performed with the low number of species per plant functional type or clade.

In line with the idea that the leaf economics spectrum drives leaf decomposability (Grime et al. 1997; Santiago 2007), we found here that not only leaf but also fine stem economics drive their respective organ decomposabilities. Thus, within above-ground organs, the numerous trade-offs between economic traits (Reich, Walters & Ellsworth 1997; Roumet, Urcelay & Díaz 2006) also constrain parallel covariation in decomposability. This was not found for fine roots. However, in support for our third hypothesis, we show here that the PES, reflecting an integrated plant strategy (Freschet et al. 2010a), controls the decomposability of each of its vegetative organs. This is consistent with the correlations found by Wardle et al. (1998) between herbaceous species’ growth rates and leaf, stem and root decomposabilities. Thus, at the whole-plant level, physiological and allometric interactions between organs (e.g. Wright et al. 2006; Maire et al. 2009) lead to coordinated trait variation across organs (Kerkhoff et al. 2006; Freschet et al. 2010a) which in turn drive coordinated organ decomposabilities.

At the ecosystem level, such coupling between plant strategy and organ decomposability yields considerable implications for plant control over soil processes. Litter decomposition drives biogeochemistry including nutrient turnover and immobilization, soil formation and atmospheric composition. Soil properties, in turn, have been shown to have strong feedback effects on plant performance and thus plant selection (Aerts 1999; Wardle et al. 2004). The coordinated litter decomposabilities across species organs imply a coordinated impact of plant above-ground and below-ground litters on the direction and magnitude of this potential feedback loop. Whether the impact of this whole-plant decomposability coordination is substantial with regards to soil properties is nevertheless uncertain. Considering the substantial differences found here between organ decomposabilities, both the absolute and the relative fluxes of litter inputs from each organ could indeed be an even stronger control on soil properties and ecosystem biogeochemistry. In freshwater systems, however, where detritus displacements generally occur, it is less likely that such plant–sediment feedback be of much influence. The relationships underpinning the afterlife effects of the PES on whole-plant litter decomposability will provide comprehensive input of vegetation composition feedback to soil carbon turnover, which is urgently needed in next-generation global models linking carbon dynamics to climate.


We are grateful to the Abisko Scientific Research Station (ANS) for providing facilities and ideal working conditions, Jurgen van Hal for his most welcome help with the final litter harvest, Richard van Logtestijn for assisting with the litterbag making, James Weedon for his precious help with wood decay models and Emilie Kichenin, Maxime Belondrade and Damien Lamouroux for their invaluable input during the experiment setup. We would also like to thank David Whitehead and several anonymous reviewers for their constructive comments on this manuscript. G.T.F. was supported by EU Marie Curie training network MULTIARC contract MEST-CT-2005-021143; R.A. by EU ATANS grant Fp6 506004; and J.H.C.C. by grants 047.017.010 and 047.018.003 of the Netherlands Organisation for Scientific Research (NWO).