Instituto de Investigaciones Fisiológicas y Ecológicas Vinculadas a la Agricultura (IFEVA) and Consejo Nacional de Investigaciones Científicas y Tecnológicas (CONICET), Cátedra de Ecología, Facultad de Agronomía, Universidad de Buenos Aires, Av. San Martín 4453, Buenos Aires C1417DSE, Argentina
*Correspondence and present address. Ecology and Evolutionary Biology, 5205 McGaugh Hall, University of California, Irvine, CA 92697-2525, USA. E-mail: firstname.lastname@example.org
1A major challenge in predicting biodiversity effects on ecosystem functioning is to understand the linkages between above-ground and below-ground components in natural communities. However, incongruities in spatial and temporal scale between plant and soil processes, and confounding ecological factors, have impeded our understanding of biodiversity effects on below-ground processes, particularly in natural ecosystems with long-lived species such as forests.
2We designed an approach to isolate plant species composition effects from other ecosystem factors, in order to evaluate the effects of individual tree species and tree species mixtures on litter decomposition in a mixed old-growth forest in temperate South America. We identified ‘tree triangles’ where the intersection of plant canopies directly controlled micro-environmental and biogeochemical conditions on the forest floor. The monospecific treatment included triangles composed of three trees of a single species of Nothofagus dombeyi, N. obliqua or N. nervosa, while the mixed-species triangles consisted in the intersections of the three different Nothofagus species.
3We placed litterbags with N. dombeyi, N. obliqua or N. nervosa litter and mixed litter of the three species within all these triangles and estimated the decomposition constant (k) after 1 year of incubation. We also used a standard litter type in all triangles to independently evaluate the tree triangle effects on decomposition.
4We found that plant species affected decomposition through both direct and indirect effects. Direct effects were mediated through leaf litter quality, while indirect effects were related to unique conditions that the plant species created in the surrounding environment. Despite litter decomposition variation among triangles, standard soil biogeochemical conditions such as soil C : N ratios, microbial biomass and pH were similar among microsites.
5Most interestingly, we explicitly demonstrated that long-term effects of plant species created specific conditions that enhanced decomposition of their own litter, establishing affinity effects between single-species litter and their own microenvironment.
6Synthesis. Our results highlight that plant species identity and long-term plant–soil feedbacks are important in affecting litter decomposition in this temperate Patagonian forest. Thus, changes or losses in temperate forest above-ground biodiversity can directly impact litter decomposition both through changes in litter quality inputs and, additionally, through the loss of specific plant–soil interactions that affect below-ground processes.
Challenges of spatial and temporal scales and confounding ecological factors have hindered the identification of above-ground effects on below-ground processes in natural ecosystems. The average size of long-lived plants, decomposer organisms and their spheres of influence vary over several orders of magnitude, both temporally and spatially, making their relationship more difficult to study (Hooper et al. 2000; De Deyn & Van der Putten 2005). Moreover, the use of short-term, artificially constructed communities (Hector et al. 2000; Knops et al. 2001) or highly disturbed systems (Wardle et al. 1999) have complicated the exploration of natural long-term above- and below-ground relationships, which may require long time periods in order to be detected (Beare et al. 1995). Taken together, several experimental and ecological hurdles may have obscured our understanding of plant diversity effects on litter decomposition and highlight the need to simultaneously study above- and below-ground effects on ecosystem processes in natural pristine communities.
We designed an experimental approach to isolate the effects of single tree species and their mixture on below-ground characteristics in an intact natural ecosystem, with little variation in other ecosystem ‘state’ factors (Jenny 1980), such as soil age and origin, aspect or climate. The study site is a mixed southern beech old-growth forest in Patagonia, South America, where three different species of the genus Nothofagus co-occur. This study site has the highest tree richness that naturally occurs within this region (Veblen et al. 1996). Rather than the direct effects of distance from individual trees on soil characteristics (e.g. Zinske 1962), we sought to establish the species’ integrated sphere of influence on the adjacent soil environment. As such, we identified microsites or ‘tree triangles’ where the specific composition of the canopy of the overstorey trees varied and, consequently, the litter on the forest floor in the intersection of these three canopies also varied as a function of the overstorey trees (Fig. 1). Microsites were defined by the canopy intersection of three trees that delineated a ‘zone of influence’ of overstorey plant species on below-ground conditions and processes. We independently manipulated litter and overstorey composition and examined their individual and interactive effects on litter decomposition.
Here we report results from a study that simultaneously compares the direct effects of leaf litter composition and indirect effects of living plant species on the decomposition microenvironment in a natural temperate forest ecosystem. Specifically, we tested the following hypotheses: (i) plant species composition effects on litter decomposition are seen more strongly through changes in litter quality (litter effects) rather than through changes in the soil environment (microsite effects); (ii) plant species produce long-term effects on below-ground properties that create specific interactions that promote decomposition of its own litter (affinity effects); and (iii) Plant species diversity has additive effects on decomposition either through litter mixtures or plant mixture effects on the soil environment (diversity effects).
The study was conducted in a Nothofagus mixed forest, in Parque Nacional Lanín, 30 km west from San Martín de los Andes (40°08′ S, 71°30′ W), Argentina. Average annual precipitation in this area is 2300 mm and mean monthly temperatures range from 3 °C in winter (July) to almost 15 °C in summer (January). Soils are Inceptisols, suborder Distrandepts, derived from postglacial volcanic ashes (Ferrer & Irrisarri 1990). Understorey vegetation is dense and monospecific, almost completely dominated by Chusquea culeou (caña colihue). We followed the US Forest Inventory program protocol (Bechtold & Zarnoch 1999) to estimate total basal area and d.b.h. for each tree species at our study forest site. The average tree age in the site was 108 years.
We identified four types of microsites or ‘tree triangles’ within the forest matrix defined by the canopy intersection of three trees in an area of approximately 4 m2 (Fig. 1). These tree triangles delineate a ‘zone of influence’ of overstorey plant species on the soil environment. The specific composition of the overstorey trees differed among triangles: (i) microsites delimited by three trees of Nothofagus dombeyi (Mirb) Blume, (ii) microsites delimited by three trees of Nothofagus nervosa (Phil) Krasser, (iii) microsites delimited by three trees of Nothofagus obliqua (Birb) Blume and (iv) microsites delimited by a single tree of N. dombeyi, a single tree of N. nervosa and a single tree of N. obliqua. Five replicates of each microsite type were identified; the microsites were distributed throughout the 6-ha study site. We evaluated the effectiveness of our selection criteria for microsites by sampling the forest floor detritus, harvesting a rectangular area of 100 cm2 in the centre of the microsite, and separating the litter by category into the different constituent overstorey species and the understorey C. culeou.
We collected freshly senesced leaves of all Nothofagus using nets suspended from the soil surface in the study site. We prepared litterbags (2 mm fibreglass mesh) with 2 g of air-dried leaf litter of N. dombeyi, N. nervosa, N. obliqua or an equivalent mixture of the three species (mixed litter). We also prepared litterbags with litter of Populus nigra (2 g), a non-native tree species which did not grow in the study area and whose litter was collected in an adjacent black poplar plantation. Litterbags were 15 cm × 15 cm, except for N. dombeyi whose leaves were smaller than the other species, so we used bags of 10 cm × 10 cm, creating the same litter incubation conditions among litter types. Litterbags were placed in the centre of the plot on bare soil in all microsites (4 litterbags × 4 microsites × 5 replicates), where all detritus and understorey vegetation had been removed. Litter accumulation on the litterbags was prevented by hand cleaning during most of the year to avoid confounding effects of overstorey litter on litterbags. In autumn, we covered the area occupied by litterbags with a mesh tent that enabled water and snow to pass but avoided litter accumulation on litterbags. Litterbags were placed in the field in January 2003 and collected at 43, 122, 268 and 366 days and analysed for changes in organic matter over time. Ash-free dry mass was determined for all samples to correct for soil contamination from the field (Harmon et al. 1999). Mass loss over time was approximated using a single exponential decay model using ln (Mt/Mo) = –kt, where Mo is the initial ash-free dry mass, Mt is the ash-free dry mass at time t and k is the decomposition constant (Swift et al. 1979). Linear regressions were performed setting the intercept to zero. In the few cases when samples did not fit a significant regression, values were considered to be outliers and were replaced by a weighted mean as a function of litter type and microsite, following the missing value procedure (Steel & Torrie 1980).
Initial litter quality was assessed for leaf litter of three Nothofagus species and for P. nigra. Total carbon (%C), nitrogen (%N) content were determined by dry combustion with a Carlo-Erba® NA 2500 elemental analyser (Haake Buchler Instruments, Inc., Saddle Brook, NJ) at the Pontificia Universidad Católica de Chile. Soluble compounds, hemicellulose and lignin concentration were determined by successive extractions with acid detergent reactions (Van Soest 1963). Total soluble polyphenols were extracted with a 50% methanol solution and analysed by the Folin-Ciocalteu method (Anderson & Ingram 1993; Constantinides & Fownes 1994; Palm & Rowland 1997). Leaf litter tensile strength was measured as an index of litter physical quality (Hendry & Grime 1993; Pérez-Harguindeguy et al. 2000). Soil characteristics in each microsite were assessed with soil cores taken to a depth of 10 cm in each microsite. We determined soil gravimetric water content and soil inorganic nitrogen (NH4-N and NO3-N) concentration at five different time points during the experiment (March, May, June and October 2003 and January 2004), total carbon (%C) and nitrogen (%N) in March 2003 and February 2004, pH in May 2003, and soil bulk density in January 2004. Inorganic N was evaluated using 2 N KCl extracts and measured colorimetrically with an Alpkem Flow IV® autoanalyser (O-I Analytical, College Station, TX) at the University of Buenos Aires. Total soil C and N were analysed for leaf litter samples and the average of both sampling dates is reported. pH was measured in a 1 : 1 mixture with distilled water. Soil microbial biomass carbon of soil samples that had been taken in January 2003 was measured using a modified chloroform fumigation–extraction technique (Paul et al. 1999) with a conversion factor of 0.45 (Kc). Soil nitrogen mineralization was estimated in the field with the buried bag technique (Stark 2000). An initial sample was used to determine soil inorganic N content (as described above) and a second sample was incubated in a polyethylene bag of 40 µm until the following sampling date. Net ammonification and net nitrification were estimated as the difference in soil ammonium concentration and soil nitrate between the initial and the incubated sample, respectively. Annual rates were calculated as the sum of daily values obtained for each sampling period along the year.
Tree species diversity effects on decomposition constants (k) were analysed with a two-way anova, with litter and microsite as main factors. Microsite effects for P. nigra litter were analysed by one-way anova. Affinity effects were tested by a two-way anova, with individual litter species (N. dombeyi, N. nervosa and N. obliqua) and location (in situ vs. ex situ) as the main factors. In situ location corresponded to individual litter species decomposing in their own monospecific microsite (3 species × 5 replicates). Ex situ decomposition corresponded to individual litter species decomposing in monospecific microsites of different species (3 species × 2 microsites × 5 replicates). Product–moment correlations were performed between species decomposition constants and initial litter quality parameters. An α of 0.05 was used in all the statistical analyses.
Non-additive effects of tree species mixture on decomposition were analysed for (i) mixed litter and for (ii) mixed tree species present in a microsite. In the first case, we compared the observed organic matter remaining of mixed litter against the predicted values calculated as the average of organic matter remaining of species decomposing alone for each plot at each sampling date, using a one-way anova. We obtained predicted values of decomposition constant (k) for each plot (5 plots × 4 microsites) regressing the log of the fraction of predicted organic matter remaining against time. Two-way anova was performed to compare observed and predicted values in each of the microsites studied. In the second case, we compared average organic matter remaining and k values obtained in monospecific microsites with the averaged organic matter remaining and k values obtained in the plurispecific microsite for all litter types using a one-way anova. All post hoc comparisons were performed with a Tukey test and an α of 0.05 was used for all the statistical analysis.
The ‘tree triangle’ approach created a natural gradient of tree species diversity within the forest mosaic, represented by microsites with different overstorey composition and a single dominant understorey bamboo (Fig. 1). The three Nothofagus species were well represented in the total basal area of the study site (33 m2/ha, Table 1). There were no other tree species present in the study site. Litter chemistry and morphology significantly differed among Nothofagus species in a number of characteristics including total nitrogen and carbon content; lignin, soluble compounds, hemicellulose and soluble polyphenols concentrations, and tensile strength (Table 1). Species composition of the forest floor litter in each type of tree triangle demonstrated significant differences in the proportion of tree species’ litter contributing to the detritus (P < 0.05, Table 2). A large percentage (~70%) of the total detritus in the single species’ microsites came from the target species, similar proportions of detritus from the three Nothofagus species were found in three species microsites, and a similar contribution of the dominant understorey species (C. culeou) was detected in all microsites (Table 2).
Table 1. Tree species characteristics in the study site and their initial litter quality (mean ±1 SE). Different letters indicate significant differences among species from post hoc comparisons (P < 0.05)
Table 2. Forest floor detritus and soil characteristics of microsites (mean ± 1 SE). Different letters indicate significant differences among microsites (P < 0.05)
Leaf litter detritus
Chusquea culeou leaf litter (%)
12.7 ± 8.3
13.6 ± 2.3
16.7 ± 2.6
19.3 ± 2.6
N. dombeyi leaf litter (%)
74.7a ± 9.2
0.2b ± 0.1
1.7b ± 0.7
35.2c ± 9.2
N. nervosa leaf litter (%)
1.8a ± 1.8
80.4b ± 4.1
0.0a ± 0
31.9c ± 4.2
N. obliqua leaf litter (%)
10.8a ± 4.9
5.9a ± 1.8
81.6b ± 3.0
13.6a ± 7.5
Gravimetric water content (%)
53 ± 3
55 ± 3
56 ± 2
52 ± 4
5.7 ± 0.2
5.1 ± 0.1
5.4 ± 0.2
5.3 ± 0.2
N-NH4 (µg/g dry soil)
116 ± 18
95 ± 13
96 ± 9
115 ± 17
N-NO3 (µg/g dry soil)
0.68 ± 0.18
0.59 ± 0.37
1.00 ± 0.38
0.98 ± 0.68
(µg/g dry soil)
1.13 ± 0.49
2.66 ± 0.55
0.72 ± 0.10
2.83 ± 0.96
13.2 ± 2.2
8.9 ± 1.4
11.1 ± 1.4
13.5 ± 3.0
0.76 ± 0.12
0.50 ± 0.08
0.73 ± 0.06
0.72 ± 0.12
17.1 ± 0.5
17.9 ± 0.8
14.9 ± 0.9
17.6 ± 1.0
Bulk density (g/cm3)
0.43 ± 0.07
0.46 ± 0.05
0.48 ± 0.03
0.56 ± 0.04
Soil microbial biomass carbon (mg C/g dry soil)
2.72 ± 1.60
2.81 ± 0.18
5.13 ± 1.12
4.39 ± 1.06
Soil N mineralization
Annual net ammonification (µg N-NH4/g dry soil × year)
94.2 ± 39.8
163.4 ± 75.9
5.5 ± 57.9
–19.9 ± 105.9
Annual net nitrification (µg N-NO3/g dry soil × year)
171.5 ± 53.1
135.2 ± 27.6
144.8 ± 20.8
133.1 ± 33.2
litter effects on decomposition
Above-ground plant diversity demonstrated significant effects on leaf litter decomposition through litter effects, microsite effects and their interaction. Litter types showed significant differences in organic mass remaining within the tree triangle microsites over time (see Supplementary Figure S1 in Supplementary Material). Similarly, decomposition constants significantly differed among litter types within microsites (Fig. 2), demonstrating direct litter quality effects based on different plant species composition. Decomposition constants of Nothofagus species showed a positive association with soluble polyphenol content (r = 0.81, P < 0.001) and total soluble compounds (r = 0.70, P = 0.003) and a negative association with hemicelluose concentration (r = –0.80, P < 0.001), total carbon content (r = –0.53, P = 0.035) and lignin concentration (r = –0.52, P = 0.037). For example, N. nervosa litter, with the lowest soluble polyphenol content and the highest hemicellulose concentration, decomposed more slowly than the rest of the Nothofagus species across all microsites (Fig. 2).
microsite effects on decomposition
Plant species composition also exerted a significant control on litter decomposition mediated through changes in the plants’ sphere of influence in the surrounding soil environment. Microsite effects significantly affected decomposition of leaf litter from native Nothofagus species (Fig. 2) and also from an independent standard litter substrate, P. nigra (Fig. 3). Populus nigra litter showed significant differences in organic mass remaining among microsites over time. Accordingly, P. nigra decomposition varied significantly in the different tree triangles, with significantly faster decomposition in N. obliqua microsites than in N. dombeyi microsites (0.54 vs. 0.41 year−1 respectively, P < 0.05, Fig. 3). Despite differences in litter decomposition among microsites, standard soil biogeochemical conditions such as soil C : N ratios, microbial biomass and pH were similar among microsites (Table 2). In addition, there was no indication of microclimatic variations among microsites, as we did not find differences in soil water content (Table 2), soil temperature or intercepted solar radiation among tree triangles (data not shown). Annual soil nitrogen transformations, although variable, did not show clear differences among microsites (Table 2).
affinity and diversity effects on decomposition
A closer analysis of litter type × microsite interaction reveals that individual tree species produced effects on below-ground properties that enhanced decomposition of their own litter. We evaluated whether litter decomposition from individual tree species was enhanced underneath that same tree species, comparing in situ location (decomposition observed when litter and microsite were from the same species) with ex situ location (decomposition of a monospecific litter types averaged across the other monospecific microsites). Organic matter loss in the litter was significantly higher in situ than ex situ incubations for the last two sampling dates for all single litter types (Fig. 4a). Similarly, leaf litter decomposition constants were significantly higher when litter was placed in its microsite of origin than in other microsites, independent of which litter species was examined (Fig. 4b). For example, N. obliqua litter decomposed significantly faster in a N. obliqua microsite (0.38 year−1) than it did in the other monospecific microsites (0.30 year−1). In contrast, trees species litter mixtures resulted in additive effects on decomposition. Organic matter loss for mixed litter was the same as predicted by the average of the single species decomposing alone (see Supplementary Figure S1), showing non-synergistic or non-antagonistic effects of litter mixtures. Likewise, observed decomposition constants for the mixed litter were similar to the prediction based on the average decomposition of the component species in the mixture at all the studied microsites (see Supplementary Figure S2a). In the same way, decomposition constants in the plurispecific microsite did not differ from the average decomposition in monospecific microsites (see Supplementary Figure S2b).
Here we show that long-term above- and below-ground interactions exerted important and unique controls on litter decomposition in temperate forests and represent an example of how plant–soil feedbacks can regulate ecosystem processes (Wardle et al. 2004). We demonstrated that simultaneous direct (through litter, Fig. 2) and indirect (through microsite, Fig. 3) effects of plant species can affect litter decomposition on the forest floor. Finally, the interaction of microsite and litter type demonstrated that individual tree species produced effects on below-ground properties that enhanced decomposition of their own litter, with plant–soil interactions creating a home-field advantage for the species’ own litter (Figs. 2, 4).
litter effects on decomposition
As hypothesized, it appears that the direct control of litter quality on decomposition in this forest is stronger than the indirect effects of the soil environment. Although several studies in Patagonian forests have demonstrated that Nothofagus spp. can modulate their nitrogen use efficiency in response to changes in soil fertility (Diehl et al. 2003) and consequently affect soil N mineralization (Satti et al. 2003; Bertiller et al. 2006), our results did not demonstrate a relationship between initial nutrient concentrations of the litter and rates of decomposition independent of microsite. However, correlations with initial litter quality suggest that litter breakdown is more closely related to carbon quality of the litter rather than nutrient concentrations, C : N ratios or lignin : N ratios, which have been observed in other forest ecosystems (Aber & Melillo 1982; Melillo et al. 1982; Berg 2000). In particular, initial soluble polyphenols and total soluble compounds were positively correlated with mass loss. Total polyphenols involve a great variety of compounds, from soluble tannins and condensed tannins to non-tannin polyphenols that have variable reactivity and functionality (Palm & Rowland 1997). While the role of phenolic compounds on carbon and nutrient cycling is still unclear, they generally demonstrate negative effects on decomposition (Harborne 1997). At the same time, there is evidence that certain types of phenolic compounds, particularly low molecular weight polyphenols, can stimulate microbial activity, thus accelerating litter decomposition (Hättenschwiler & Vitousek 2000). Hemicellulose and lignin concentrations, on the other hand, were negatively correlated with decomposition. The complexity of these structural polysaccharides of plant cell walls may retard the access of microorganisms to more easily degradable soluble litter compounds (Swift et al. 1979). Taken together, these results suggest a strong carbon chemistry control determining the differences in decomposition rates among these species with little direct control of nitrogen concentrations on the observed variation in decomposition.
microsite effects on decomposition
The direct significant effects of microsite on litter decomposition of both native litter species and a common substrate (Fig. 4) suggest that the ‘tree triangles’ significantly affected decomposition independent of changes in litter quality of the decomposing material. Surprisingly, we did not detect differences in soil fertility or other standard soil biogeochemical characteristics which could explain the variation in decomposition (Table 2), or differences in microclimatic conditions that have been suggested to explain differences in decomposition in monospecific tree plantations (Hobbie et al. 2006). The lack of a straightforward biogeochemical explanation for these results suggests that biotic factors may play a critical role in these microsite effects, independent of changes in coarse-scale biogeochemical pools and turnover. Different live plant species can sustain a diversity of microbial and fauna communities in the soil and the rhizosphere (Waldrop & Firestone 2004), which in turn could directly and indirectly affect the activity of the decomposer community in different microsites (Grayston et al. 1998; Marschner et al. 2001; Porazinska et al. 2003; Waldrop & Firestone 2004). The demonstrated importance of micro-, macro- and mesofauna in affecting mass loss in a range of terrestrial ecosystems (González & Seastedt 2001; Bradford et al. 2002b; Hättenschwiler & Gasser 2005) suggests that variation in soil biota distribution related to different plant species may explain the microsite effects on litter decomposition in this ecosystem. While the effect of microsite in the litter layer was clear, these effects did not translate to changes in soil organic matter dynamics, suggesting that the principal microsite effect is occurring in the litter layer in the first stages of decomposition.
affinity relationships between litter and microsite
Although the idea of specificity between plant and soil biota affecting litter decomposition is appealing, it has been difficult to detect in natural ecosystems. It is important to note that the observed affinity effects for decomposition in this temperate forest are substantially different from synergistic effects of litter quality and soil fertility. High-quality litter may decompose faster in a fertile site due to interactions between soil resource availability and litter resource availability, independent of its origin (e.g. priming effect), but true affinity effects occur only when a lower quality litter decomposes faster in its own low fertility environment, demonstrated here in the case of all three Nothofagus species (Fig. 4).
Some studies have provided indirect evidence that decomposer communities may be specialized to a characteristic litter type of a given ecosystem. For example, in a regional-scale study by Gholz et al. (2000), litter from broadleaved trees decomposed faster than pine litter in a broadleaved forest, but in much smaller magnitude, pine leaf litter decomposed faster in coniferous forest than in broadleaved forest. In partial support of the affinity effect, Hunt (1988) showed that lodgepole pine decomposed faster in a forest than in adjacent grasslands, but this trend did not hold for grasses, which decomposed more slowly in grasslands than in the forest. These results suggest that specificity at the level of life form could occur. However, large differences between ecosystems including soil origin, climate and age could have important effects beyond that of the direct effect of the in situ ecosystem. As such, specificity effects have not been reported within a given ecosystem, where microsite variations are mainly due to the plant species composition. The only explicit test of affinity effects in a single ecosystem, to our knowledge, was conducted by Ayres et al. (2006), who incubated litter from different tree species with soil inoculums prepared from soil beneath a stand of each species. This study did not demonstrate specificity effects; the authors suggested that the disruption of soil structure, the exclusion of soil fauna and the relatively short duration of the experiment did not allow for the detection of these relationships.
What could explain affinity effects between above-ground plant species and below-ground processes? Above- and below-ground components of terrestrial ecosystems are interdependent since plants provide a source of carbon and nutrients for soil biota and soil biota, in turn, decompose dead plant material and release nutrients that regulate plant growth. The essential nature of this relationship suggests that long-term interactions between plant species and soil biota could promote the presence of a decomposer community that was more efficient in the breakdown of litter produced in situ, creating unique conditions for ‘home field advantage’ decomposition. Although this mechanism postulates strong bottom-up control of plant species composition on below-ground communities, a top-down feedback might additionally exist through soil fauna preferences for certain microsite conditions and food resources provided by different plant species (Hättenschwiler & Bretscher 2001; Hättenschwiler & Gasser 2005). It appears that our ‘tree triangle’ approach allowed for an appropriate spatial scale to examine plant species’ effects on litter decomposition, as the trees’ sphere of influence and the studied process were appropriately scaled in this natural old-growth ecosystem.
plant species diversity effects on decomposition
In contrast to the observed specificity effects on litter decomposition, plant species diversity demonstrated additive effects as a function of litter and microsite (see Supplementary Figure S2). In accordance with our hypothesis, in each microsite, litter mixtures decomposed at the same rate as the average of the component species decomposing alone (see Supplementary Figure S2). Litter mixture experiments have shown variable responses, from completely additive effects, as in this study, to strong antagonistic and synergistic effects (Gartner & Cardon 2004). Although the mechanisms behind these responses are still not clear, some recent evidence suggests that specific interactions between soil fauna and litter species could explain such variety in litter mixture effects on mass loss (Hättenschwiler & Gasser 2005). In addition, living plant species richness also showed no detectable effect on decomposition (Figs. 2, 3). This result is in accordance with the lack of response of litter decomposition to a increasing number of living plant species in artificially constructed grassland ecosystems (Hector et al. 2000; Knops et al. 2001; Knops et al. 2007). The lack of a direct relationship between above-ground diversity and diversity of soil biota (Wardle 2006) complicates the linkages that occur between above- and below-ground communities. In concordance with a number of other studies, it appears that the relationships between above-ground diversity and below-ground processes are more complex than simple numerical relationships between plant species and decomposition.
Understanding what controls carbon turnover in terrestrial ecosystems is critical in order to evaluate the consequences of global environmental changes (Cao & Woodward 1998). While traditional models of litter decomposition suggest that litter chemistry and climatic conditions are the principal controls of carbon loss in terrestrial ecosystems at regional scales (Meentemeyer 1978; Vitousek et al. 1994; Austin & Vitousek 2000; Gholz et al. 2000), there is increasing recognition that the relative importance of these controls varies across ecosystem type (Aerts 2006; Austin & Vivanco 2006). Our results highlight that in addition to these controls, long-term plant–soil feedbacks are important in affecting litter decomposition at the ecosystem scale in this temperate forest. Thus, changes or losses in temperate forest biodiversity both above- and below-ground can directly impact litter decomposition through changes in litter input quality and, additionally, through the loss of specific plant–soil interactions that affect below-ground processes. These relations should be taken into account in decisions attempting to conserve natural forest biodiversity (Simberloff 1999), and emphasize the need for future studies evaluating biodiversity effects on ecosystem processes in natural intact ecosystems.
We thank A. González Arzac, P. I. Araujo, V. A. Marchesini, M. Gonzalez-Polo, A. Fernández Souto, F. Biganzoli, and J. Vrsalovic for field assistance. A. González Arzac, A. Grasso, E. Galli, L. Gherardi, and P. Rojas Machado provided laboratory assistance. C. Pérez from Pontificia Universidad Católica de Chile analysed carbon and nitrogen content of litter and soil samples. P. Flombaum and O. E. Sala provided important inputs on various aspects of this project. We thank the College of Agronomy of the University of Buenos Aires and IFEVA-CONICET for logistic support in Buenos Aires, Parque Nacional Lanín for permission to conduct research within the park, and Brown University for support in the final stage of this manuscript. Financial support came from the Inter-American Institute for Global Change Research (CRN-012), Fundación Antorchas of Argentina, the National Science Foundation of the United States, and ANPCyT and UBACyT of Argentina. LV was supported by graduate fellowships from the University of Buenos Aires, and Fundación YPF of Argentina.