Litter species composition influences the performance of seedlings of grassland herbs


†Author to whom correspondence should be addressed. E-mail:


  • 1This study examines the impacts of plant litter species identity and the composition of litter mixtures on seedling recruitment in the context of land-use change (abandonment) in conservationally important southern Swedish semi-natural grasslands.
  • 2We found that plant litter had marked positive effects on the seedling recruitment of two common grassland species, and that these effects varied strongly with the species identity of the litter.
  • 3There was no consistent evidence that litters of species typical of earlier succession had a greater positive impact on recruitment than those typical of late succession.
  • 4The impact of mixtures of the five litter types examined was generally as expected based on the impacts of single-species litters and their contribution to the litter mixture, as predicted by the biomass ratio hypothesis. However, this was not the case for all litter and seedling species combinations, and some interactions were evident.
  • 5Species identity of litter is important even in multispecies litter mixtures. Changes in plant species dominance (and hence the proportions of litter of different species), as a result of shifts in land use, are likely to result in changes in seedling performance, with potential consequences for the persistence of plant populations in former semi-natural grasslands.


In most terrestrial ecosystems the majority of living plant biomass eventually becomes plant litter, rather than being consumed by herbivores (Odum 1960; McNaughton et al. 1989). Plant litter has important effects in structuring plant communities (Grime 1979; Facelli & Pickett 1991a), as well as being a major determinant of C and nutrient storage and fluxes through ecosystems (Chapin 1993; Wardle et al. 1997b). The magnitude of the (short-term) effects of plant litter on vegetation can be comparable with those of competition or predation (Xiong & Nilsson 1999), or site fertility (Sydes & Grime 1981). Plant litter therefore has the potential be an important factor mediating plant–plant interactions, and may in some cases play a more profound role in structuring communities than it does in nutrient cycles (Facelli & Pickett 1991a).

Strong evidence is accumulating to suggest that plant species or functional groups exert a fundamental control over community and ecosystem processes (Zinke 1962; Pastor & Post 1988; Hobbie 1992; Hooper & Vitousek 1997; Wardle et al. 1997b; Spehn et al. 2000). This topic has received increasing attention as anthropogenic environmental change brings about species shifts, as well as directly influencing abiotic factors and disturbance regimes. In this context, a number of studies have revealed species differences in the effects of litter on plant establishment and growth (Schlatterer & Tisdale 1969; Facelli & Pickett 1991b; Nilsson, Wardle & Dahlberg 1999; Lopez-Barrera & Gonzalez-Espinosa 2001; Quested, Callaghan & Press 2003; Bowman et al. 2004). In a meta-analysis of the effects of litter on vegetation (Xiong & Nilsson 1999), the largest proportion of variability explained was due to litter type, indicating that species or functional type differences in litter effects could be of central importance for impacts on populations and communities.

However, litter layers in natural habitats are rarely monospecific, usually consisting of a complex mosaic of different litter types resulting from both the species composition of a patch and redistribution of litter by wind, snow or water (Facelli & Pickett 1991a). During at least the earlier stages of succession, changes in species dominance, rather than complete shifts in species identities, are likely to occur. According to the mass ratio hypothesis (Grime 1998), species effects are proportional to their contribution to the community, such that emergent effects are not important. However, it is well established that litter of different species interacts during decomposition, often giving decomposition rates and nutrient losses that are substantially different from those expected based on single species values (Gustafson 1943; McTiernan, Ineson & Coward 1997; Wardle, Bonner & Nicholson 1997a; Finzi & Canham 1998; Wardle et al. 2003; Gartner & Cardon 2004). Litter species effects on plant establishment or growth may also be affected by litter-mixing interactions, although this has rarely been tested and the few studies conducted provide no clear picture. Mixtures of litters from subarctic species, widely differing in quality, did not influence plant growth in a way that differed from what is expected from single-species quality (Quested et al. 2003), whereas non-additive effects of mixtures of Empetrum nigrum and the moss Pleurozium schreberi were present in a boreal forest system (Nilsson et al. 1999). Further deviations from the biomass ratio hypothesis may arise because plant response to litter quantity is unlikely to be linear (Heady 1956; Bartolome, Stroud & Heady 1980; Carsson & Peterson 1990; Xiong & Nilsson 1999).

This study seeks to investigate the extent to which the individual effects of species’ litters on seedling performance can be combined to predict effects of litter mixtures, in the context of land-use change (abandonment) in southern Swedish grasslands. At a global scale, land-use change is one of the greatest threats to biodiversity and ecosystem functioning (Chapin et al. 2000; Sala et al. 2000). Ongoing abandonment of grazing land in southern Sweden has resulted in fragmentation of semi-natural grasslands and an associated decline in species richness (Eriksson & Ehrlén 2001; Eriksson, Cousins & Bruun 2002). Abandonment of grazing land is often characterized by an initial increase in litter accumulation (Facelli & Pickett 1991a), as well as shifts in the abundance of different plant species (Lindborg & Eriksson 2004). Litter accumulation generally has negative effects on the establishment and performance of plants in grassland habitats (Carsson & Peterson 1990; Tilman 1993; Foster & Gross 1997; Xiong & Nilsson 1999), and has been implicated in the declines of some species of high conservation value (Lennartsson & Svensson 1996; Eriksson & Ehrlén 2001). However, litter species identity and the composition of mixed litter layers may be of significance in determining the extent of such effects, with implications for the role of species in determining subsequent community development. Litter of plant species typical of later stages in succession may have more negative effects on plant establishment and performance – a number of studies have identified a link between increasing successional age and plants producing slowly decomposing litter that is low in nutrients and may be high in secondary metabolites (Wardle et al. 1997b; Garnier et al. 2004; Kazakou et al. 2006).

Using two grassland species as models for the effects of the species identity of litter and the composition of mixed litter layers (the relative quantities of litters of different species) on plant recruitment, this pot-based study aims to test the following three specific hypotheses: (1) species identity of single-species litter has substantial impacts on seedling performance; (2) litter of species increasing in later succession have a more negative impact on seedling performance compared with litter from species typical of earlier stages of succession; (3) effects of single-species litters are not substantially altered in litter mixtures. We discuss the results in the context of litter as a potential driver of changing species composition after abandonment of grasslands.


study sites and species

All litter and soil were collected from Nynäs Nature Reserve in the province of Södermanland, south-east Sweden (60°50′ N, 17°24′ E). In this area, there are several sites with semi-natural grasslands with a long history of management, grazing or mowing (Cousins 2001). These grasslands have substantial conservation importance. The sites used for collecting are former grazing pastures, abandoned between 15 and 100 years ago. Litter species were chosen on the basis of being widespread and common in the study area, forming a substantial part of the litter input at different stages after abandonment (unpublished data), and having litter with contrasting physical and chemical properties: the perennial herb Galium boreale L.; the perennial grass Dactylis glomerata L. and the nitrogen-fixing, perennial herb Trifolium medium L., all of which increase early after abandonment, and the trees Betula pendula Roth and Pinus sylvestris L., which increase in later stages of succession following abandonment.

Two perennial forbs, Leucanthemum vulgare Lam. and Ranunculus acris L., were used as target species in the experiments. Both species are common in semi-natural grasslands. As perennials, they have an ability to persist for several decades after abandonment, although in sparser populations than in the grasslands. Achenes (referred to as seeds, for simplicity) of these species were collected from grazed or recently abandoned sites in the vicinity of Nynäs Nature Reserve. The seed weights of the species are 0·44 mg (Leucanthemum) and 1·38 mg (Ranunculus). The target species and litter species are referred to henceforth by their generic names.

collection of material

Litter was collected in October and November 2003. Whole-plant (stems and leaves together) litter of Dactylis, Trifolium and Galium was collected; only leaves were collected for Betula and Pinus, as the majority of litter shed from these species consists of leaves, and woody litter has very different decomposition dynamics and spatial arrangement. Care was taken to collect freshly senesced, undecomposed litter. Litter was air-dried in the laboratory for a minimum of 5 days, and mixed well. The leaves of Dactylis and stems of Trifolium and Galium litters were cut, as accurately as possible, to ≈10-cm lengths to fit in the seed trays and enable mixtures to be formed. This did not appear to have a substantial influence on the physical structure of the litter layers in the trays compared with the field situation, as litter is often broken and compressed by wind, animals and snowfall (H.Q., personal observation). The cutting may have increased leaching or microbial degradation relative to the field situation, but this is unlikely to alter the broad differences in seedling performance between such contrasting litter types. All seeds and seed heads were removed.

Seeds were collected when ripe, during July and August 2003. Seeds from different collections and locations were mixed well, and seeds that appeared to be non-viable (small, pale coloured or flat) were removed. Seeds were stored in envelopes at room temperature until use.

Field soil (5 × 5 × 5-cm-deep samples) was collected using a trowel in mid-November 2003 from four sites in different stages of abandonment. This soil was used to introduce a range of soil organisms from the field situation into the pots. In order to homogenize this inoculum, the soil was sieved to pass through a 7-mm mesh, and larger roots and plant parts were removed to prevent vegetative regeneration of non-target plant species. The soil was mixed well and stored at 4 °C until use.

experimental setup

The experiment was initiated in early December 2003. A total of 180 seed trays (21 × 16 × 5 cm deep) were filled with a well mixed blend of seven parts nutrient-poor, peat-based commercial compost (Weibull's ‘P jord’, Weibull Trädgård AB, Hammenhög, Sweden) and two parts fine sand. Each seed tray was split into two parts with a strip of strong black plastic, and 10 ml sieved field soil was sprinkled evenly over the surface of each seed tray. Forty seeds of Leucanthemum vulgare were sown on one side of the partition and the same number of Ranunculus acris seeds were sown on the other side.

The 180 pots were assigned randomly to one of 12 treatments to give 15 replicates per treatment. Five single-species treatments consisted of 7 g of a single litter type per tray. Five ‘dominant mixtures’ consisted of a mixture of all five litter types, with the dominant species making up 50% of the total and the other four species making up an equal proportion of the remainder, to make a total of 7 g. In the even mixture, all five species were present in equal masses to make a total of 7 g. A no-litter treatment was also included. Litter was weighed to the nearest 0·05 g, and each litter mixture was mixed well before application to the soil surface. This litter application rate is equivalent to 208 g m−2, which compares with a mean total annual litter input of 197 g m−2 (SE 20 g m−2) estimated from harvests of litter from 14 sites abandoned between 5 and 60 years ago in the Nynäs area (unpublished data). After litter application, trays were watered and left to drain freely.

The seed trays were arranged in 15 blocks, in a fenced, outdoor experimental garden, at Stockholm University. Trays were placed on a layer of fine sand within wooden frames. One replicate of each treatment was placed in each block, and trays were arranged randomly within each block. A layer of fine, ≈2-cm mesh size nylon netting was placed over the top of each pot to prevent wind and animal disturbance. Trays were not watered, as the experiment aimed to observe the integrated impact of the litter treatments on physical, chemical and biological factors on seedling performance, including potential water stress.

Seedlings were counted four times, on 11 May, 2 June, 18 June and 28 June 2004. Dead seedlings were counted separately and removed. On the second and third counts, seedlings were classified into small (cotyledons only) or large (with true leaves). At the final harvest, seedlings were classified into small (two true leaves), medium (three to four true leaves) and large (more than four true leaves). Occasional seedlings of other species were removed. After the final count, all seedlings were removed together from the trays, their roots carefully washed and separated into roots and shoots. These were dried at 60 °C for 4 days and weighed.

Litter concentrations of C and N were measured on three replicate subsamples of the bulk collection per species. Analyses were carried out on a CHONS microanalyser (Carlo Erba 1500, Carlo Erba, Milan, Italy), which uses chromatography in the gaseous phase to measure the C and N contents of solid-phase samples. Two replicate measurements were made for each of these samples, and the average value taken. Decomposition rates of litter were assessed in five replicate litter bags per species, incubated for 7 months (November to May 2004) in an outdoor litter-bed experiment (sensu Cornelissen 1996), which was carried out in the same fenced garden (unpublished data). Specific leaf area (SLA) and leaf dry matter content were measured on 10–25 fresh leaves from separate individuals collected from plots in Nynäs at different stages of abandonment. Methods follow the standard methods in Cornelissen et al. (2003).

data analysis

The shoot weight ratio (total shoot weight per tray and species divided by total seedling mass per tray and species), and average mass per seedling (total mass of all seedlings of one species in a tray divided by the number of seedlings of that species), were calculated. Shoot weight ratio was log-transformed before analysis. Seedling numbers were analysed as the total number of seedlings per species and harvest date, and were square-root transformed prior to analysis. Effects of litter treatments were tested with a general linear model (GLM) with treatment as a fixed effect and block as a random effect, followed by Tukey tests (P < 0·05). Seedling number data were analysed separately for each date.

The impact of litter mixing was analysed in two ways. First, GLMs were used with species, diversity (single species or dominant mixture), and block as factors. The no-litter and even-mixture treatments were excluded from these analyses. Second, expected values of response variables were calculated as the sum of the single species values from each block, divided by the proportion of that litter type involved in the mixture (Nilsson et al. 1999). To test for non-additive litter-mixing effects, calculated expected values were compared with the observed values from the same block using paired sample t-tests. Power calculations were performed to calculate the smallest difference detectable between observed and expected values (observed minus expected values divided by expected values) and n required to detect a specific difference assuming α = 0·05 and β = 0·2 (Crawley 2002). These calculations were carried out with estimates of the mean and variance of each mixed litter treatment separately.


single-species litters

The litter chemistry, decomposability and SLA of the five litter types examined are summarized in Table 1. As expected, in general the litter of ‘late-successional’ species has lower N concentrations, higher C : N ratios, slower decomposition rates and lower specific leaf areas than species typical of earlier succession.

Table 1.  Characteristics of litter, and specific leaf area (SLA) of fresh leaves, for the five species used in the experiment
SpeciesLitter nitrogen concentration (%) (n = 3)Litter C : N ratio (n = 3)Decomposability (% mass loss) (n = 5)Fresh leaf SLA (m2 kg−1) (n = 10)
  1. Decomposability is percentage mass loss after 7 months in an outdoor litter bed. Means and SE.

Dactylis glomerata0·91 (0·006) 51·2 (0·43)21·0 (2·30)24·8
Galium boreale1·25 (0·009) 39·0 (0·30)31·4 (2·26)27·3
Trifolium medium2·25 (0·040) 22·4 (0·33)31·4 (2·17)21·6
Betula pendula0·74 (0·009) 72·4 (1·14)20·3 (0·50)15·5
Pinus sylvestris0·41 (0·012)131·2 (4·80)15·1 (0·74) 5·0

Litter treatments had significant impacts on seedling performance (Table 2) in terms of the total mass of seedlings per tray (Fig. 1a,b), due to differences in both mean mass per seedling (Fig. 1c,d) and the number of seedlings (Figs 2a and 3a). The presence of any litter significantly and substantially increased the number of seedlings (with the exception of the final counting date for Leucanthemum, Figs 2a and 3a), the average mass per seedling and the total seedling mass (Fig. 1a–d). The presence of litter had a stronger positive effect on seedlings of Ranunculus than on Leucanthemum.

Table 2.  Results of general linear models for the response variables total mass of seedlings per tray, average seedling mass per tray, and average shoot weight ratio per tray, with litter treatment as a fixed effect and block as a random effect
VariabledfTotal mass of seedlingsMass per seedlingShoot weight ratio
  1. For all treatments n = 15.

Block14 6·51<0·001 3·87<0·0016·58<0·001
Treatment1114·91<0·00112·25<0·0013·13  0·001
Block14 3·63<0·001 4·25<0·0011·40  0·159
Figure 1.

Attributes of Ranunculus (left) and Leucanthemum (right) seedlings after 7 weeks’ growth with different litter types (SS) and mixtures (DM). (a,b) Total seedling mass per pot; (c,d) average mass per seedling; (e,f) shoot weight ratio. Means and SE are given, n = 15. Results of general linear models are given in Table 2; bars not sharing a common letter differ significantly, Tukey's test (P < 0·05).

Figure 2.

Number of Ranunculus seedlings per tray on each of four counting dates. (a) Single-species litter and no-litter treatments; (b–f) single-species litter, dominant mixture of that species and even litter mixture. Means and SE are given, n = 15. Letters indicate significant differences within each counting date, Tukey's test (P < 0·05).

Figure 3.

Number of Leucanthmum seedlings per tray on each of four counting dates. (a) Single-species litter and no-litter treatments; (b–f) single-species litter, dominant mixture of that species and even litter mixture. Means and SE are given, n = 15. Letters indicate significant differences within each counting date, Tukey's test (P < 0·05).

As hypothesized, litter of different types had strongly different impacts on seedling performance. In particular, Dactylis litter had a strong significant positive effect on both the number of Ranunculus seedlings (from the second counting date onwards; Fig. 2a), and the mean mass per seedling (Fig. 1c), resulting in the total seedling mass being 2–3·6 times greater than with any of the other single-species litters (Fig. 1a). There were significantly more Leucanthemum seedlings present with Dactylis litter for the first three counting dates (Fig. 3a), but by the final harvest neither seedling number nor average seedling mass differed significantly between the single-species treatments. The total seedling mass in the Dactylis litter treatment differed significantly only from the Galium and Betula litter treatments (Fig. 1b).

There were few significant differences in seedling performance between the other litter types. Ranunculus seedling number was significantly greater with Trifolium litter than with Pinus litter at the second counting date (Fig. 2a). Both average mass per seedling and total mass tended to be consistently lower in Pinus and Galium litter, and greater in Trifolium litter (Fig. 3a,c).

effects of litter mixing

The GLM using litter species identity and litter composition (mixed or monospecific) as factors revealed that litter identity explained a substantially greater proportion of the variability in seedling performance than either litter composition or the interaction between these factors (Table 3). These models also revealed a significant interaction between litter species and litter diversity for the total mass of Ranunculus seedlings, indicating that litter mixing had different effects depending on the identity of the dominant litter species. It is not possible to be specific, as no significant differences between single-species litter treatments and dominant mixtures of the same species in Ranunculus were detected (Table 2; Fig. 1a), probably due to the less powerful nature of the Tukey test. Similarly, the shoot weight ratio of both Leucanthemum and Ranunculus revealed a significant interaction between species identity and composition (Table 3), but no significant differences were detected between single-species litter treatments and dominant mixtures of the same species (Fig. 1e,f).

Table 3.  Results of general linear models for the response variables total mass of seedlings per tray, average seedling mass per tray, and average shoot weight ratio per tray, with litter species identity and composition (single species or mixed litter) as fixed effects and block as a random effect
VariabledfTotal mass of seedlingsMass per seedlingShoot weight ratio
  1. For all treatments n = 15.

Species 414·15<0·00113·88<0·0015·20  0·001
Composition 1 6·71  0·011 3·82  0·0530·08  0·771
Species × composition 4 2·53  0·044 1·81  0·1323·04  0·020
Block14 5·40<0·001 3·85<0·0015·45<0·001
Species 4 9·76<0·001 6·76<0·0014·51  0·002
Composition 1 0·09  0·766 0·00  0·9641·02  0·315
Species × composition 4 0·57  0·687 0·83  0·5083·35  0·012
Block14 3·87<0·001 5·50<0·0011·87  0·035

The non-additive effects of litter mixing on seedling number can be assessed by comparing the results from the single-species litter treatments, dominant litter mixtures and even litter mixture results (Figs 2 and 3b–f). In the absence of substantial non-additive interactions, the dominant mixture values will lie part way between the even and single-species mixture values. However, as the impacts of single-species litter were rather similar (with the exception of Dactylis), and within-treatment variability was rather high, we have limited power to detect such patterns statistically. Despite this, at the second counting date there were significantly fewer Ranunculus seedlings in the Pinus single-species treatment than in either the pine-dominant mixture or the even mixture, which had nearly identical seedling numbers. A similar but non-significant trend was evident for Galium. Interestingly, in the case of Leucanthmum there were greater numbers of seedlings in the dominant Trifolium mixture than in the even mixture or single-species treatments, but only significantly so for the first counting date. Despite this evidence for non-additive effects of litter mixing, calculated expected values did not differ significantly from observed values for any of the response variables analysed, in either species (analyses not shown). To determine whether the absence of significant differences was likely to be due to lack of statistical power rather than the genuine absence of interactions, we performed power calculations. These revealed that the detectable deviation from calculated expected values (δ), averaged across the litter mixture treatments, was 46% for Ranunculus and 32% for Leucanthemum total seedling mass; 27 and 30% for mass per seedling; and 12 and 13% for shoot weight ratio. The δ value for seedling numbers, at the four counting dates, was 151, 78, 41 and 30% for Ranunculus and 66, 24, 21 and 20% for Leucanthemum. To detect a 20% deviation in expected values of total seedling mass, 68 replicates of Ranunculus and 35 of Leucanthemum would be required. Similarly, for mass per seedling 34 and 33 replicates would be required, and for seedling numbers at the final count 28 and 15 replicates would be needed for Ranunculus and Leucanthemum, respectively.


litter species identity

In answer to our first question, the species identity of the litter in the single-litter treatments has a clear and substantial impact on the seedling performance and recruitment of both Ranunculus and Leucanthemum. Litter of Dactylis, a species that increases early in succession, had a strong positive effect on seedling performance of both target species in comparison with other species’ litter.

In answer to our second question, based on the five species examined in this study, there is no unequivocal evidence to support the hypothesis that species typical of later succession have consistently more negative effects on seedlings of our target species. Aside from the dramatic positive effects of Dactylis litter, there was some evidence that Trifolium litter influenced seedling performance consistently more positively than Pinus litter. However, Betula and Galium litters did not differ significantly from other litter types in their effects on seedling performance. Species typical of later successional stages generally have lower leaf nutrient concentrations, lower SLA and slower litter decomposition rates than earlier successional species (Wardle et al. 1997b; Garnier et al. 2004; Kazakou et al. 2006). However, in the current study these differences are not clearly linked to impacts on seedling performance, in contrast to a hypothesis suggested by Sydes & Grime (1981).

It has been proposed that traits, or ‘functional markers’, can be used to link species shifts and their effects on community and ecosystem properties and processes (Lavorel & Garnier 2002). In contrast to studies linking traits such as SLA to decomposition (Cornelissen et al. 1999; Pérez-Harguindeguy et al. 2000; Cornelissen et al. 2005), we did not find a link between the litter characteristics measured and the impact of species on seedling performance. The net effect of litter on plants is likely to be due to a balance of a variety of chemical, physical and biological factors, and interactions between them (Facelli & Pickett 1991a). For example, litter can influence the light, moisture and temperature conditions experienced by plants (Heady 1956; Sydes & Grime 1981; Facelli & Pickett 1991a, 1991b; Foster & Gross 1997); release nutrients, directly influencing plant growth (Brearly, Press & Scholes 2003; Quested et al. 2003); release labile C or secondary metabolites including phenolics, potentially inhibiting plant growth (Schmidt, Michelsen & Jonasson 1997a, 1997b; Hättenschwiler & Vitousek 2000; Bowman et al. 2004); act as a mechanical barrier (Sydes & Grime 1981); and alter interactions with other organisms (Sydes & Grime 1981; Facelli & Pickett 1991a; Facelli 1994). It might thus be expected that, to predict litter effects on seedling performance, a combination of traits reflecting the physical (e.g. light interception, temperature insulation) and chemical (e.g. nutrient-release rates) properties of the litter would be needed, for example leaf area, dry matter content, content of nutrients and phenolic compounds. The relative importance of these traits is likely to change according to the ecological factors that are most limiting to seedling performance, so the generality of any proposed trait-based scheme must be tested under a variety of contrasting conditions. Much of the variation in the current study appears to be due to structural differences between the litters produced by different growth forms (e.g. needles, grasses and forbs), which suggests that functional groupings may explain at least a proportion of the variation in litter effects on seedlings. However, to draw general conclusions about potential links between functional traits or groups and litter species effects on seedling performance, more extensive screening of litter from large numbers of species would be necessary (Cornelissen 1996).

The dramatic positive effect of litter of any type on seedling performance observed in the current study suggests that amelioration of the physical environment, and/or protection from herbivores, are the dominant factors under the conditions employed here. As far as we can tell from the available evidence, nutritional mechanisms appeared to play a smaller role. Assuming that plants invest abundant resources to enhance the acquisition of scarcer ones, increased allocation to roots can be interpreted as an indication of nutrient limitation (Bloom, Chapin & Mooney 1985; Ericsson 1995), although root–shoot allocation patterns also change as a plant grows (Gedroc, McConnaughay & Coleman 1996). The few seedlings that survived in the no-litter treatment had similar or greater allocation to shoots than those in the litter treatments (Fig. 1). The trend towards enhanced seedling performance with Trifolium litter relative to other litter types may, however, be a nutritional effect, given the markedly higher N content of Trifolium litter (Table 1).

We suggest that the absence of litter may have resulted in higher temperatures at the soil surface early in the season, which may have stimulated earlier germination, leaving seedlings vulnerable to late frosts. Dactylis litter formed a thicker and less compact litter layer than the other species, which suggests that it gave more effective protection against soil desiccation and/or extremes of temperature, without acting as a mechanical barrier or extinguishing as much light as, for example, birch litter. In support of this, Xiong & Nilsson (1999) found that grass litter had weaker negative effects on germination and establishment than other litter types, and Facelli & Pickett (1991b) found that litter of the grass Setaria extinguished less light than that of Quercus. Both Galium and Trifolium have a high proportion of stems in their litter, and the leaves fell off the stems early in the decomposition process, leaving a thin, compact litter layer that is unlikely to have a substantial insulating effect. Direct measurements of the impact of litter treatments on, for example, temperature, moisture, light and soil nutrient availability would allow us to test these hypotheses about the mechanistic basis for the observed results.

effects of mixing litters

In answer to our third question, we observed few, and rather small, statistically significant interactions between litters in mixtures, although there was some evidence for effects of mixed litters that are not simply related to those of single-species litters. The observed seedling performance did not differ significantly from expected values calculated on the basis of single-species litter. Further, the performance of target-species seedlings was statistically indistinguishable in single-species litters and where the same litter species formed only 50% of the litter layer. However, power calculations reveal that we could detect differences from expected values of around 25–40% for most response variables. This compares with typical deviations from expected values in decomposition studies of ≈10–25%, up to 65% in some cases (Gartner & Cardon 2004), indicating that we could only expect to detect substantial interactions with this type of analysis.

There was, however, evidence for effects of mixed litters that are not simply related to those of single-species litters. The significant interaction between species identity and composition for Ranunculus total seedling mass indicates that litter mixing had different impacts depending on the identity of the litter species. Although it is not possible to be specific, due to the absence of significant differences between single-species litters and their dominant mixtures, this result is consistent with the significant, transient effect on Ranunculus seedling number of mixing Pinus litter (Fig. 2e), and a similar, although non-significant trend for Galium litter (Fig. 2d). Fewer seedlings were present in the Pinus or Galium litter monoculture than in either the dominant or even mixtures, indicating that the impacts of these litters are not proportional to the mass of litter present. In the case of Pinus, this may be because its litter forms a compact layer, offering minimal protection from drought (Lopez-Barrera & Gonzalez-Espinosa 2001) or temperature extremes, whereas the presence of 50 or 80% litter of other species is enough to form a substantial litter layer.

Although it must be borne in mind that the ability of the current study to resolve small interactions between litters in mixtures is limited, we conclude that the biomass ratio hypothesis is partially confirmed: for the majority of mixtures studied here, the effects of mixed litter layers can be predicted, at least roughly, as a ‘sum of the parts’, dependent on the contribution of a species (by mass) to the litter layer, and its behaviour in monoculture. Litter-mixing interactions and non-linear effects of litter quantity on seedling performance do not appear to play a large role within the range of species and litter quantity used here. There are few other studies that examine the effect of litter-mixing interactions on plant performance, and these concern only mixtures of two species’ litters (Nilsson et al. 1999; Quested et al. 2003) – the current study is thus unique in this respect. As far as is possible to tell from the available evidence, substantial non-additive effects of litter mixtures on plant performance appear to be rare.

implications for abandoned grasslands

The results of this pot-based study enable only tentative conclusions about likely litter effects in the field, as a number of factors, including the absence of vegetation, the soil type and the age and spatial structure of the litter layer, differ between the two situations. The direction and magnitude of litter effects on plants are linked to the life stage of the target plants, the quantity of litter accumulated, and the dominant biotic and abiotic factors involved (Facelli & Pickett 1991a; Nash Suding & Goldberg 1999). Litter accumulation is usually associated with negative effects on seedling establishment and population performance in mesic habitats (Carsson & Peterson 1990; Tilman 1993; Xiong & Nilsson 1999; Jakobsson & Eriksson 2000; Eriksson & Ehrlén 2001). However, in the context of the current study, it is clear that plant litter does not have solely negative effects on the seedling performance of common grassland species. In our study sites, there are areas of thin soil over bedrock where drought or exposure to temperature extremes are likely to be recurring ecological factors. The accumulation of litter that occurs as a result of abandonment may well have a positive impact on seedling establishment under these conditions (Facelli & Pickett 1991a), although field investigations would be needed to confirm this.

The quantity of litter present has an impact on the magnitude and direction of impacts on seedling performance, with, in general, increasingly negative effects with greater quantities of litter (Carsson & Peterson 1990; Xiong & Nilsson 1999). The quantities of litter applied here are closely matched with annual litter inputs in recently abandoned sites (unpublished data), and are within the range where positive effects are likely to dominate (100–300 g m−2; Carsson & Peterson 1990). The total litter accumulation (standing dead and litter) increases from ≈175 g m−2 in managed sites to ≈380 g m−2 in sites abandoned 15–60 years before this study (unpublished data). Further studies are needed to assess the role of changing litter quantities in these abandoned grasslands – this study emphasizes the importance of taking into account the species composition of that litter accumulation. The impacts of litter are also likely to vary with the life stage of the target species, with facilitative effects dominating during establishment (Nash Suding & Goldberg 1999). This is often a crucial stage in plant population performance (Harper 1977), and the observed positive effects of the moderate quantities of litter examined here may have important implications for plant populations and hence biodiversity in these grasslands. However, it is important to complete the current study with investigations of the impact of litter composition on subsequent life stages, and of target species with differing traits.

This study indicates that litter cannot be treated as a uniform material – litter species identity matters, and these differences persist even in mixed litter layers. Changes in plant species dominance (and hence the proportions of species’ litter present), as a result of shifts in land use, are likely to result in changes in seedling performance, with potential consequences for the persistence of plant populations in former semi-natural grasslands. At small spatial scales, variation in litter layer species composition is likely to create significant heterogeneity in conditions for seedlings. The next step is to investigate the role of multispecies litter mixtures on species in different life stages and their interactions in complete communities in the field, where both vegetation and litter can influence seedling performance.


We are very grateful to Irja Hedman, Malte Ehrlén, Jerker Eriksson and Åsa Eriksson for first-class field and harvesting help, and to the EU VISTA project (EVK2-2002-00168) for financial support.