Global decomposition experiment shows soil animal impacts on decomposition are climate-dependent
Article first published online: 28 JUN 2008
© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd
Global Change Biology
Volume 14, Issue 11, pages 2661–2677, November 2008
How to Cite
WALL, D. H., BRADFORD, M. A., ST. JOHN, M. G., TROFYMOW, J. A., BEHAN-PELLETIER, V., BIGNELL, D. E., DANGERFIELD, J. M., PARTON, W. J., RUSEK, J., VOIGT, W., WOLTERS, V., GARDEL, H. Z., AYUKE, F. O., BASHFORD, R., BELJAKOVA, O. I., BOHLEN, P. J., BRAUMAN, A., FLEMMING, S., HENSCHEL, J. R., JOHNSON, D. L., JONES, T. H., KOVAROVA, M., KRANABETTER, J. M., KUTNY, L., LIN, K.-C., MARYATI, M., MASSE, D., POKARZHEVSKII, A., RAHMAN, H., SABARÁ, M. G., SALAMON, J.-A., SWIFT, M. J., VARELA, A., VASCONCELOS, H. L., WHITE, D. and ZOU, X. (2008), Global decomposition experiment shows soil animal impacts on decomposition are climate-dependent. Global Change Biology, 14: 2661–2677. doi: 10.1111/j.1365-2486.2008.01672.x
- Issue published online: 22 OCT 2008
- Article first published online: 28 JUN 2008
- Received 28 February 2008 and accepted 18 April 2008
- climate decomposition index;
- soil biodiversity;
- soil carbon;
- soil fauna
- Top of page
- Material and methods
Climate and litter quality are primary drivers of terrestrial decomposition and, based on evidence from multisite experiments at regional and global scales, are universally factored into global decomposition models. In contrast, soil animals are considered key regulators of decomposition at local scales but their role at larger scales is unresolved. Soil animals are consequently excluded from global models of organic mineralization processes. Incomplete assessment of the roles of soil animals stems from the difficulties of manipulating invertebrate animals experimentally across large geographic gradients. This is compounded by deficient or inconsistent taxonomy. We report a global decomposition experiment to assess the importance of soil animals in C mineralization, in which a common grass litter substrate was exposed to natural decomposition in either control or reduced animal treatments across 30 sites distributed from 43°S to 68°N on six continents. Animals in the mesofaunal size range were recovered from the litter by Tullgren extraction and identified to common specifications, mostly at the ordinal level. The design of the trials enabled faunal contribution to be evaluated against abiotic parameters between sites. Soil animals increase decomposition rates in temperate and wet tropical climates, but have neutral effects where temperature or moisture constrain biological activity. Our findings highlight that faunal influences on decomposition are dependent on prevailing climatic conditions. We conclude that (1) inclusion of soil animals will improve the predictive capabilities of region- or biome-scale decomposition models, (2) soil animal influences on decomposition are important at the regional scale when attempting to predict global change scenarios, and (3) the statistical relationship between decomposition rates and climate, at the global scale, is robust against changes in soil faunal abundance and diversity.
- Top of page
- Material and methods
The annual global release of carbon to the atmosphere through decomposition of organic carbon by soil biota is approximately 50–75 Pg, nearly 10 times that of fossil fuel emissions (Schimel et al., 1996). Climate and plant litter quality (i.e. chemical composition) are considered the primary regulators of litter decomposition, explaining as much as 65–77% of the variation in decomposition rates (Moorhead et al., 1999; Gholz et al., 2000; Trofymow et al., 2002). The residual variation (ca. 25%) in global decomposition rate models remains a substantial source of error in estimates of contemporary and future global carbon dynamics (Schimel et al., 1996; Del Grosso et al., 2005).
Putatively, the unallocated error in global decomposition models has a significant biological component, for example organismal biomass, size distribution, taxonomic richness, and/or functional group composition. These components are normally considered as either the direct effects of physical conditions (through addition or removal of habitats and therefore species) or indirect effects (delivered via multiple climate effects defining plant communities and the area of distribution available to major groups of soil organisms). Consequently, biota (animals and microbes) are not explicitly considered in global decomposition models (VEMAP, 1995; Moorhead et al., 1999; Gholz et al., 2000; Lavelle et al., 2004), although conceptually considered to be key drivers of litter decomposition rates (e.g. Wardle, 1995; Lavelle, 1997; Coleman & Hendrix, 2000; Kibblewhite et al., 2008). Instead, soil biota have been evaluated (relative to nonbiotic agencies) for their contribution to aggregated ecosystem functions (or services, sensu Daily et al., 1997; Wall, 2004), including decomposition, together with the related processes of carbon sequestration and greenhouse gas emission. To refine this concept, ecosystem services have been apportioned to functional assemblages of named organisms (for example Lavelle, 1997; Swift et al., 2004; Kibblewhite et al., 2008), such that the effects of specific disturbances on the delivery of individual services can be elucidated (Wardle, 2002). The modeling, however, is at best semiquantitative and, most crucially, it is ecosystem-specific and cannot be deployed much beyond the landscape scale or made responsive to the incremental changes in physical environmental parameters inherent in climate change scenarios. To factor soil biota into future global decomposition modeling, it is necessary to assess decomposition against a biotic assemblage character on a supraregional scale and by an experimental procedure that will adequately distinguish between the abiotic and the biotic agencies responsible for the process.
Almost 30 years ago, Swift et al. (1979) hypothesized that the relative contribution of soil fauna (vs. microflora) to decomposition was dependent on the climatic region, being greatest at midlatitudes and decreasing towards the poles. In contrast to the many multisite experiments on climate and litter quality (Moorhead et al., 1999; Trofymow et al., 2002), relationships between soil animals and litter decomposition have never been experimentally tested at global or regional scales. Furthermore, there are few data to judge the significance of changes in diversity of soil fauna at these scales (Swift et al., 1998), which reflects the scarcity of the required taxonomic expertize and the prohibitive effort needed to characterize biota in a sufficient number of sites and with a resolution that reflects true biodiversity. Inferences from local scale and the few cross-site experiments that manipulated animals (Heneghan et al., 1999; Gonzalez & Seastedt, 2001) are restricted by the low number of sites, differences in litter types and qualities between experiments, and the fact that due to a vast and mostly unknown biodiversity in many biomes, invertebrate analyses have been typically restricted to one to three taxonomic groups.
The Global Litter Invertebrate Decomposition Experiment (GLIDE) tests the hypothesis that soil animals significantly influence litter decomposition rates, over and above climate alone, at regional and global scales. A second hypothesis, related to the first and following Swift et al. (1979), is that the role of animals in terrestrial decomposition changes from region to region rather than having a single global character. The centerpiece of GLIDE's design was the use of a single plant litter substrate, of known quality and from a single origin, which was exposed in standard litterbags while manipulating the abundance and richness of soil fauna with a generalized suppressor (naphthalene). Decomposition was monitored in both animal-suppressed bags and untreated controls by gravimetric measurement and chemical analysis at 30 sites distributed across broad climatic regions from 43°S to 68°N (Table 1). Additionally, the extraction and characterization of associated (mainly mesofaunal) invertebrate animals was assessed. The use of the faunal-suppressant separated climate from animal effects on litter decomposition. Animal data, with complementary information on local climate and the weight and carbon content of litter at the start of the decomposition process and after exposure in the field, show that soil animals positively influence decomposition rates in the temperate and wet tropical biomes.
|Country||Site code||Latitude Longitude||Soil type||Vegetation type||Holdridge Life Zone||GLIDE climate region||Precipitation (cm)||Temperature (°C)||Annual CDI|
|Australia||TAS1||43°04′S 146°40′E||Planosol||Callidendrous fern rainforest||Cool temperate moist forest||Temperate||110.1||8.8||0.2711|
|Australia||TAS2||43°04′S 146°40′E||Acrisol||Intermediate rainforest||Cool temperate moist forest||Temperate||110.1||8.8||0.2711|
|Brazil||MAN||02°25′S 60°00′W||Ferralsol||Tropical moist forest||Tropical rain forest||Wet tropical||211.1||26.3||0.8311|
|Brazil||RUB||19°30′S 42°30′W||Ferralsol, Acrisol||Tropical rainforest||Subtropical rain forest||Wet tropical||116.8||22.7||0.4881|
|Burkina Faso||BON||11°20′N 4°15′W||Luvisol||Shrub to wooded savanna||Tropical dry forest||Dry tropical||95.7||27.1||0.4150|
|Canada||INU||68°19′N 133°32′W||Gleysol||Black spruce, paper birch forest||Subpolar dry tundra||Cold or dry||17.3||−11.3||0.0443|
|Canada||YUK||60°51′N 135°12′W||Luvisol||Pine, spruce, aspen forest||Boreal moist forest||Cold or dry||26.5||−4.2||0.0583|
|Canada||TOP||54°36′N 126°18′W||Luvisol||Pine, balsam, fir, spruce forest||Cool temperate moist forest||Cold or dry||48.8||2.1||0.1071|
|Canada||LET||50°11′N 113°54′W||Chernozem||Fescue grassland||Cool temperate steppe||Cold or dry||44.3||4.4||0.1477|
|Canada||VAN||48°38′N 123°42′W||Cambisol||Douglas fir forest||Cool temperate wet forest||Temperate||196.0||6.2||0.2287|
|Canada||ROC||49°32′N 58°49′W||Podzol||Balsam fir, white birch forest||Boreal wet forest||Temperate||115.2||1.6||0.2287|
|Canada||ONE||49°01′N 110°23′W||Chernozem||Shortgrass||Cool temperate steppe||Cold or dry||36.0||3.9||0.1080|
|China||XIS||21°41′N 101°25′E||Acrisol||Tropical seasonal rainforest||Subtropical moist forest||Wet tropical||132.2||20.5||0.5275|
|Colombia||BOG||04°37′N 74°18′W||Andosol, Cambisol||Montane cloud forest||Tropical wet forest||Dry tropical||117.5||14.7||0.4288|
|Czech Rep.||BIO||48°42′N 16°49′E||Fluvisol||Riparian oak forest||Cool temperate moist forest||Temperate||54.0||9.3||0.2462|
|Czech Rep.||SUM||49°97′N 13°45′E||Cambisol||Mountain climax spruce forest||Cool temperate moist forest||Temperate||64.8||7.6||0.2537|
|Czech Rep.||KOM||48°42′N 16°49′E||Fluvisol||Riparian oak forest||Cool temperate moist forest||Temperate||54.0||9.3||0.2462|
|Czech Rep.||PAL||48°42′N 16°49′E||Rendzina||Xerothermic grass/deciduous forest||Cool temperate steppe||Temperate||54.0||9.3||0.2462|
|Germany||GIE||51°20′N 10°22′E||Luvisol||Temperate beech forest||Cool temperate moist forest||Temperate||68.3||8.1||0.2618|
|Germany||JEN||50°57′N 11°35′E||Rendzina||Xerothermic grassland||Cool temperate steppe||Temperate||58.6||8.2||0.2432|
|Kenya||KEN||0°15′S 34°50′E||Acrisol||Rainforest||Tropical wet forest||Wet tropical||185.6||20.7||0.6406|
|Malaysia||MAL||05°47′N 116°24′E||Acrisol||Tropical lower montane forest||Subtropical wet forest||Wet tropical||260.7||22.8||0.7587|
|Namibia||NAM||23°33′S 15°02′E||Gleysol, Histosol||Sparse dwarf shrub||Subtropical desert||Cold or dry||0.1||23.5||0.0264|
|Russia||KUR||51°34′N 36°05′E||Chernozem||Oak forest, meadow steppe||Cool temperate steppe||Temperate||65.2||5.5||0.2253|
|Senegal||SEN||16°25′N 15°30′W||Arenosol||Dry savanna, balanites||Tropical desert scrub||Dry tropical||26.1||27.1||0.1565|
|Taiwan||TAI||24°46′N 121°35′E||Acrisol, Cambisol||Evergreen, hardwood forest||Subtropical moist forest||Wet tropical||208.2||20.6||0.7313|
|UK||CAR||51°50′N 04°10′E||Cambisol||Oak forest||Cool temperate moist forest||Temperate||82.6||9.8||0.2961|
|UK||SIL||51°24′N 0°40′W||Cambisol||Oak, oak-birch, alder forest||Cool temperate moist forest||Temperate||74.1||9.3||0.2414|
|USA||COL||40°49′N 104°46′W||Yermosol||Shortgrass steppe||Cool temperate steppe||Cold or dry||38.5||8.7||0.1222|
|USA||FLO||27°09′N 81°12′W||Luvisol, Histosol||Native tallgrass wet prairie||Subtropical moist forest||Wet tropical||127.4||22.7||0.6060|
Material and methods
- Top of page
- Material and methods
Site management and location
Forty-one sites were initially established in 2001–02, as part of the DIVERSITAS IBOY (International Biodiversity Year) Project network, and with ILTER (International Long Term Ecological Research) collaborators. Sites were selected to achieve the fullest practical climatic and geographic range, but partially reflected the mostly volunteer participation by national science teams in IBOY. Eleven sites were dropped from analyses due to destruction of bags in situ, excessive extraneous organic matter or soil in the litterbags, or where export of specimens from countries was prohibited or impractical. The 30 remaining sites are described in Table 1.
Two thousand glass-fiber, 20 cm × 20 cm, 2 mm mesh litterbags (Harmon et al., 1999), were each filled with 10±0.5 g dried, gamma-sterilized grass hay [Agropyron cristatum (L.) Gaertn.] foliar litter that had been air-dried for 2 years at <20% rH. The litter, with all florets removed, was preprocessed through a 1.0 mm screen (to remove loose material before shipping) and then shipped from Colorado State University (CSU) in the constructed glass-fiber bags to sites. A single litter quality was selected to facilitate site-to-site comparisons and because the scale of resources required to analyze all taxa from just one litter type made multilitter field experiments unfeasible. Bags were weighed on-site to the nearest 0.001 g before being secured to the soil surface, and then progressively removed over time intervals. Of 24 bags at each site, six were placed randomly in each of four experimental blocks. Within each block, three bags were spaced along each of two paired 20 m transects, with each transect being assigned to either the control or animal-suppressed treatment. Transects ran parallel to each other and were 10 m apart. The three bags within each transect were spaced at 0, 10, and 20 m. All blocks were at least 10 m apart, and positioned randomly. This design was chosen to (a) ensure the suppression treatment did not influence the control litterbags, (b) reduce the impacts of disturbance by vertebrates, and (c) reduce the effects of spatial autocorrelation.
Naphthalene treatment: animal suppression
The animal inhibitor naphthalene was added to half of the litterbags; the other half was assigned to untreated controls. Naphthalene was applied at the start of the experiment in crystalline form (as ‘mothballs’ from a single commercial source) and again at each sampling occasion to litterbags that remained in the field for later retrieval. For each treatment renewal, two mothballs (∼33.17 g per mothball) were placed adjacent to treatment bags. Naphthalene was chosen to displace soil animals because it reportedly has less biocidal effects than other pesticides (Blair et al., 1989). An additional three ‘traveler bags’ were shipped to each site, placed in the field, immediately retrieved, reweighed, and reshipped to CSU for analysis. Traveler bags controlled for material loss due to handling (Harmon et al., 1999). See also http://www.nrel.colostate.edu/projects/glide/studydesign.html.
Litterbag retrieval and processing
One control and one treatment litterbag were retrieved from each block on three sampling occasions. Time of field exposure was dependent on climate, longer where litter decomposition rates were slower due to high latitude (e.g. temperate sites) or high altitude (e.g. BOG site; Table 1). Three Canadian sites were sampled at 2 and 12 months, and another two were sampled at 2 and 4 months only (Table 2). Soil animals were extracted from bags at sites by a standardized Tullgren dry heat apparatus shipped from CSU. Animals were shipped in 70% ethanol to BioTrack® Australia Pty Ltd (Macquarie University, Sydney) for centralized taxonomic identification. Voucher specimens were stored until investigators requested return. After weighing, fauna-extracted litterbags were returned to CSU where subsamples of oven-dried, plant material were milled and analyzed for C on a LECO 1000 CHN analyzer (LECO Corp., St Joseph, MI, USA). Initial litter C concentration was 44.13%. The composition of the initial litter ‘lignin’ (Preston et al., 1997, 2006) was as follows: neutral detergent fiber or NDF extract, 67.91% (SD±0.74); acid hydrolyzable extract, 27.54% (SD±0.83); acid unhydrolyzable residue or AUR, 4.14% (SD±0.18); and ash, 0.41% (SD±0.44). ADF, acid detergent fiber, is the sum of the last three measures.
|Laboratory analyses||Sampling times (months)|
|Climatic region treatment||GLIDE site code||Min. FRC (%)||k (day−1)||kr2||Taxonomic richness||1||2||3||4||12|
|Cold or dry||INU||X||X||X|
|Cold or dry||YUK||X||X||X|
|Cold or dry||TOP||X||X||X|
|Cold or dry||LET||X||X||X|
|Cold or dry||ONE||X||X||X|
|Cold or dry||NAM||X||X||X|
|Cold or dry||COL||X||X||X||X|
Taxonomy was determined according to The Tree of Life (http://www.tolweb.org/tree/) and Systema Naturae 2000 (http://sn2000.taxonomy.nl/). BioTrack® (Harvey & Yen, 1990; Oliver et al., 2000) identified all animals, except for Czech Republic sites BIO, KOM, PAL, and SUM; Brazilian sites MAN and RUB; and Colombian site BOG (Table 1) where taxonomy was carried out by qualified experts. All adult animals extracted from litterbags were identified to 38 invertebrate groupings (Table 3) with Pauropoda and Symphyla identified to Class, mites (Acari) identified to Subclass, and other taxa identified to the Order level. The number of Orders ranged from one each in Phyla Annelida and Mollusca, two in Class Crustacea, four in Subclass Arachnida (Acari considered one ‘Order’), 11 in Class Myriapoda (Pauropoda and Symphyla considered as ‘Orders’), to 19 in Class Hexapoda.
|Pterygota||Blattodea Coleoptera Dermaptera Diptera Embioptera Hemiptera Hymenoptera Isoptera Lepidoptera Megaloptera Neuroptera Orthoptera Psocoptera Thysanoptera Trichoptera|
|Myriapoda||Diplopoda||Helminthomorpha||Spaerotheriida Chordeumatida Julida Spirobolida Spirostreptida Polydesmida|
Biota (Colwell, 2004), a relational database application, was used to manage data. Of four litterbags removed per treatment (control and suppression treatment) per sampling period at each site, two were for determination of C concentrations of the residual litter. Knowing the initial- and post-field-exposure litter masses, and the initial and retrieved litter C concentrations, the loge of the fraction remaining litter C (FRC) was regressed against the number of field days of litter exposure and k was determined from the slope as an estimate of litter decomposition rate (Harmon et al., 1999). By using FRC rather than fraction litter mass remaining per se, we could correct for incorporation of mineral material into litterbags, a procedure similar to the use of ash-free dry mass (Harmon et al., 1999).
For each of the 30 sites, the k-value for the inhibitor treatment bags was regressed against the k-value for control bags (Fig. 1), deviations from the 1 : 1 line being associated with potential treatment effects. Paired (by site) t-tests were performed using untransformed k-values and one-tailed tests, given that variance was determined to be equal and the hypothesis that animal inhibitors would only decrease decomposition rates respectively. These tests were first performed for all sites together and, second, to determine if there were region-specific effects, for sites within each climatic region.
For 18 of our 30 sites, we had complete animal abundance and taxonomic richness data (Tables 2 and 3) for each of the eight litterbags removed on each sampling occasion. The other 12 sites had missing values for some sample times. For the 18 sites with complete datasets, mean abundance and richness were determined for control and animal-suppressed treatments at each sampling period, and then mean abundance and richness across all sampling times. This gave single animal abundance and taxonomic richness values for both the control and animal-suppressed treatments for each site. Effects of the animal inhibitor on animal abundance and taxonomic richness were tested using, as for the k data, one-tailed paired t-tests (Fig. 2). Unequal variance in abundance data was equalized by natural log transformation.
To test whether the impacts of the animal suppressor on litter decomposition rates were correlated with treatment effects on animal abundance and/or taxonomic richness, we used multiple linear regression and regression tree approaches (Crawley, 2002). The primary reason for conducting these analyses was to evaluate whether the expected statistical relationship between decomposition rate and climate was altered by changes in the abundance and/or diversity of fauna. In both regression analyses, decomposition rate (k) was used as the response variable, natural log-transformed to meet assumptions of homogeneity of variance. There were three explanatory variables: mean animal taxonomic richness, mean animal abundance, and the function ‘climate decomposition index’ (CDI; Table 1). The influence of climate on decomposition is usually represented in global decomposition models as a function of temperature and water availability (Liski et al., 2003; Parton et al., 2007). For analysis of our data, we used the function CDI, a widely recognized index for predicting climate effects on litter decomposition (Moorhead et al., 1999; Gholz et al., 2000) and an integral part of the global carbon model CENTURY (Parton, 1996). An annual CDI value for each site was calculated from monthly values for precipitation and temperature (Tables 1 and A1) (Parton et al., 2007). The temperature function uses average monthly maximum and minimum air temperatures and has been validated on an extensive global soil respiration dataset (Del Grosso et al., 2005). The water stress term is calculated as a function of the ratio of monthly rainfall to potential evapotranspiration rate (Parton et al., 2007). For GLIDE, average CDI values were calculated for the period that corresponded to the maximum sampling time achieved (e.g. 12 months for temperate sites) (Table 1). Monthly values of maximum and minimum air temperature and precipitation data from weather stations close to sites were used (see Appendix) because there were incomplete data at all sites for these years.
A total of 35 values were used for regression analyses, from the subset of 18 sites for which complete animal data were available. These comprised two sets of animal variables (richness, abundance), one from each treatment (control, animal-suppressed). Through model checking, one point (Kenya, animal-suppressed litterbag) was omitted because it consistently violated assumptions of homogeneity of variance irrespective of the transformations used.
For regressions using CDI, richness, and abundance values, the simplest possible model that included all three explanatory variables was fitted first. Both k and CDI were plotted as natural log-transformed variables. Next, the curvature was tested by fitting the quadratic terms for the variables; a square-root transformation of the abundance was found necessary to correct for curvature.
- Top of page
- Material and methods
Soil animal diversity
The 2 mm litterbag mesh prevented the access of soil animals greater than 2 mm in diameter, excluding larger animals. Some macrofauna (e.g. spiders, termites, beetles, and other insects) could gain access as immatures, but of the 80 606 individuals sorted to 38 higher taxonomic groupings (i.e. Class, Subclass, Order), the majority were soil mesofauna (defined as having maximum body widths ranging from 100 μm to 2 mm as adults; Swift et al., 1979) (Table 3).
Latitudinal effects of animal suppressant naphthalene
The k-values from the 30 sites (Table 1) were used to test whether the naphthalene reduced litter decomposition rates. For each site, the k-value for the animal-suppressed treatment bags was plotted against the k-value for control bags (Fig. 1), deviations from the 1 : 1 line being associated with potential treatment effects. In some climate regions, the inhibitor had a negative effect on k.
Across all 30 sites, treatment with naphthalene did not affect litter decomposition rates (k-value, Fig. 1a). When the same analyses were performed for each of our four climatic regions (Table 1), decomposition rates were significantly reduced by the animal suppressant in temperate and wet tropical regions (temperate, t1,12=3.6, P<0.01; wet tropical, t1,6=2.1, P<0.05), but not in cold or dry regions, or the tropics as a whole (P>0.05 for all three datasets) (Fig. 1a and b). To corroborate that these negative and neutral effects on decomposition rates were associated with animal responses to the suppressant, we evaluated whether naphthalene decreased the abundance and/or mean taxonomic richness (Table 3) of animals extracted from litterbags. When the suppressant was present, there were significant reductions in both abundance (t1,17=1.8, P<0.05) and richness (t1,17=2.8, P<0.01). Log-transformed abundance and taxonomic richness were found to be correlated but weakly, using Pearson's product-moment statistic (R2=0.376, t34=4.5, P<0.01, n=18 sites). Treatment values (means±1 SE) for each site were calculated as the mean of all three sample periods (Fig. 2a and b).
Animal effects on the relationship between climate and litter decomposition
Analysis of covariance was used to test whether the global CDI–decomposition relationship was influenced by the presence of naphthalene. The interaction between the CDI and suppressant terms was not significant (P>0.05), suggesting that the global relationship between climate and litter decomposition rates is robust to experimental reductions in soil animal abundance and taxonomic richness. The animal-suppressed treatment was then substituted for mean animal abundance and mean animal richness data. These variables could be used along with CDI annual values (Tables 1 and A1) as continuous explanatory variables, providing 35 independent data points for regression analyses (see ‘Material and methods’). That is, the subset of 18 sites (Table 2) for which complete animal abundance and taxonomic richness data were available (Table 3) was used for the regression. This confounds the animal and climate effects and therefore requires cautious interpretation (Loreau et al., 2002; Wardle, 2002); this limitation is more fully evaluated in ‘Discussion’. The minimal adequate model (loge[k-value]=0.1271[taxonomic richness]+0.6299 × loge[CDI]−5.755) obtained using step-wise multiple linear regression retained CDI and taxonomic richness as explanatory variables and accounted for 77% of the variation in decomposition rates [P<0.001 (F2,32=53; n=35)] (Fig. 3).
Given that CDI alone explained 70% of the variation in measured decomposition rates, as shown in other multisite global decomposition experiments, the validity of designating fauna as an explanatory variable was further tested using a regression tree approach (results not shown). This confirmed that animal abundance per se offered little explanatory power, but that taxonomic richness of soil fauna, in addition to climate, was correlated with decomposition rates across broad climatic regions.
- Top of page
- Material and methods
GLIDE provides the first high-resolution taxonomic database relating soil fauna to decomposition on a global scale. It improves upon pairwise comparisons of temperate vs. tropical regions (Heneghan et al., 1999; Gonzalez & Seastedt, 2001) to include a gradient of climatic conditions from 30 sites globally, and improves taxon breadth from three or less to 38 taxonomic groups. The findings of this experiment indicate that soil mesofaunal assemblages, dominated by arthropods, influence litter decay across broad climatic regions, namely temperate and wet tropical, as predicted by Swift et al. (1979). If this applies to all soil invertebrates, the original conceptual model of three primary drivers of litter decomposition (climate, litter quality, and biota; Swift et al., 1979) is validated.
Investigations using biogeographical comparisons to explore linkages between ecosystem processes and organism diversity are typically frustrated by the positive correlation between diversity and the product of temperature and water availability (Coûteaux et al., 1995; De Deyn & Van der Putten, 2005), though Maraun et al. (2007) have offered oribatid mites, one of the most abundant and species-rich groups of soil mesofauna, as an exception to this rule. In our study, we overcame this limitation by experimentally reducing animal richness and abundance at each site independently of climate, an approach already used effectively in smaller scale, cross-site experiments (Heneghan et al., 1999; Gonzalez & Seastedt, 2001). By using an analysis of covariance (as opposed to regression approaches) in this initial analysis, we ensured that our statistical tests did not confound the animal and climate effects (Loreau et al., 2002; Wardle, 2002). One potential limitation of the approach, however, is that naphthalene may negatively affect other components of soil assemblages. Under field conditions, however, naphthalene tends to have negligible effects on soil bacterial and fungal growth (Blair et al., 1989) and is relatively persistent. Using this approach, we demonstrate that soil fauna positively influence decomposition rates in the temperate and wet tropical biomes and are neutral in regions where biological activity is more constrained by temperature and/or moisture, which was hypothesized by Swift et al. (1979) but untested until now.
Our regression analyses, where the categorical factor naphthalene was substituted for the continuous variables of abundance and richness, requires cautious interpretation because of the confounding influence of climate once this statistical approach is used. That is, in contrast to the categorical factor naphthalene, abundance and richness may be correlated with temperature and moisture (i.e. climate), and so relationships between abundance and richness with decomposition rates may be artefacts of this correlation. It is, however, noteworthy that taxonomic richness of soil fauna, and not abundance, was positively related to litter decomposition rates. In contrast, experimental evidence from microcosm and local-scale field experiments suggests that it is animal abundance or biomass, and not richness, which operate as the driver (Cole et al., 2004). Our experiment did not allow for the determination of animal biomass per se, so we cannot discount the possibility that the effect of taxonomic richness was driven by biomass. The key difference between our experiment and previous investigations into relationships between soil animal diversity and litter decomposition rates, aside from geographic scale, is that we identified the entire extracted animal assemblage in our litterbags to Class, Subclass or Order, rather than identifying only a few select groups to species (Table 3). The high species richness of soil animals is assumed to be associated with high functional redundancy (Andrén & Balandreau, 1999), whereas animal richness at the Order level or higher is concluded to be associated with higher functional dissimilarity (Naeem & Wright, 2003; Heemsbergen et al., 2004; St. John et al., 2006); that is, the greater the value of taxonomic richness at this level, the more functionally diverse the animal assemblage. However, this concept is difficult to test in the field because no single soil or litter decomposition study has had the resources to analyze all species of soil fauna present at one time, owing to the expertize required for the accurate identification of each group and the high percentage (∼85%) of soil fauna yet to be described (Eggleton & Bignell, 1995; Lawton et al., 1998; St. John et al., 2006). The present study, utilizing the BioTrack® facilities, takes advantage of recent developments in semiautomated methods of rapid biodiversity assessment (Oliver et al., 2000) and suggests that functional richness of soil fauna may be an important driver of decomposition across large geographic regions. Further work will be required to test this conclusion definitively. We recommend that additional studies should address fewer representative sites, but with greater replication per site, higher taxonomic resolution, biomass determination, and the use of robust litterbags excluding mesofauna and macrofauna as an additional suppression treatment augmenting naphthalene. The use of multiple litter species of differing qualities should also be considered, given that faunal effects may be quality dependent. A more recalcitrant litter than used in our study, decomposing for a longer time, would likely involve a different biodiversity and abundance and perhaps show other effects on decomposition rates. However, the effort required for this and the other improvements to the field procedure is very large.
It is unlikely that potential nontarget effects of the suppressant on microbial activity explain the results in the field, because inhibitor effects were restricted to specific climatic regions. If nontarget effects had occurred, then litter decomposition should have been retarded in all climatic regions because the climate driver in decomposition models (CDI) is a surrogate for bacterial and fungal activity per se. Therefore, soil mesofauna, likely through their functional or taxonomic richness, are an important driver of litter decomposition rates in some climatic regions. When these regions are viewed at a global scale under current climatic conditions (Fig. 4), they account for 27.2% of the Earth's land area.
These results indicate that explicit inclusion of soil animals in decomposition models may reduce the unexplained variation in relationships between litter decomposition and climate across regional scales. We found evidence that inclusion of soil animals (specifically, their diversity) in models of global-scale decomposition rates may provide modest improvements in predictability [from 70% to 77% variance explained over abiotic factors (CDI) alone] but the underlying mechanism for this remains unclear and warrants further study. Notably, however, our finding that animals have important influences on decomposition in certain climatic regions has potential implications for global carbon dynamics under changing climate, and may help explain regional variability in soil respiration (Davidson et al., 2006). For example, regions predicted to have warmer and wetter climates may see positive feedbacks to soil communities, resulting in accelerated decomposition rates and release of carbon. In higher latitude systems, this feedback could substantially increase respiration of C to the atmosphere given the greater proportions of relatively undecomposed, organic carbon that has accumulated due to historically cold and dry conditions (Lal, 2005). Alternatively, global changes such as land use conversion (Swift et al., 1998; Kaiser, 2004) may alter soil animal diversity and/or biomass with consequences for decomposition rates in warm, wet climatic regions. Thus, the specific inclusion of regional impacts of soil animals on decomposition is necessary to adequately model the impacts of global change scenarios on decomposition and atmospheric C dynamics. Additionally, because litter quality may modify animal effects at a single site, future global experiments will need to test simultaneously all three primary drivers of decomposition.
- Top of page
- Material and methods
We thank H. A. Mooney, D. A. Wardle, and R. D. Bardgett, and the following for their assistance and advice throughout the project: G. Adams, B. J. Adams, A. I. Breymeyer, E. J. Broos, V. K. Brown, M. Chauvat, D. C. Coleman, R. K. Colwell, N. M. DeCrappeo, A. N. Gillison, J. R. Gosz, J. Green, D. Gustine, M. E. Harmon, S. W. James, M. Lane, A. N. Parsons, K. L. Prior, S. D. Salvo, P. Smilauer, G. W. Yeates, X. Yang, and A. S. Zaitsev. GLIDE research was a project of the DIVERSITAS International Biodiversity Observation Year and supported by funding to D. H. W. from the Winslow Foundation, the National Science Foundation DEB 980637 and 0344834, and the Soil Science Society of America Outreach Program. Funding to W. J. P. from DOE 4000060456 contributed to publication cost.
- Top of page
- Material and methods
- 1999) Biodiversity and soil functioning – from black box to can of worms? Applied Soil Ecology, 13, 105–108. , (
- 1989) Effects of naphthalene on microbial activity and nitrogen pools in soil-litter microcosms. Soil Biology Biochemistry, 21, 507–510. , , (
- 2004) Effect of density and species richness of soil mesofauna on nutrient mineralisation and plant growth. Biology and Fertility of Soils, 39, 337–343. , , , (
- 2000) Invertebrates as Webmasters in Ecosystems. CABI Publishing, Wallingford. , (
- 2004) Biota 2: The Biodiversity Database Manager. Sinauer Associates Inc., Sunderland, MA. (
- 1995) Litter decomposition, climate and litter quality. Trends in Ecology and Evolution, 10, 63–66. , , (
- 2002) Statistical Computing: An Introduction to Data Analysis Using S-plus. Wiley, Chichester, UK. (
- 1997) Ecosystem services supplied by soil. In: Nature's Services: Societal Dependence on Natural Ecosystems (ed. DailyGC), pp. 113–132. Island Press, Washington, DC. , , (
- 2006) On the variability of respiration in terrestrial ecosystems: moving beyond Q10. Global Change Biology, 12, 154–164. , , (
- 2005) Linking aboveground and belowground diversity. Trends in Ecology and Evolution, 20, 625–633. , (
- 2005) Modeling soil CO2 emissions from ecosystems. Biogeochemistry, 73, 71–91. , , , , , , (
- 1995) Monitoring the response of tropical insects to changes in the environment: troubles with termites. In: Insects in a Changing Environment (eds HarringtonR, StorkN), pp. 473–497. Academic Press, London. , (
- 2000) Long-term dynamics of pine and hardwood litter in contrasting environments: toward a global model of decomposition. Global Change Biology, 6, 751–765. , , , , (
- 2001) Soil fauna and plant litter decomposition in tropical and subalpine forests. Ecology, 82, 955–964. , (
- 1999) Measuring decomposition, nutrient turnover, and stores in plant litter. In: Standard Soil Methods for Long-Term Ecological Research (eds RobertsonGP, ColemanDC, BledsoeCS, SollinsP), pp. 202–240. Oxford University Press, New York. , , (
- 1990) Worms to Wasps: An Illustrated Guide to Australia's Terrestrial Invertebrates. Oxford University Press, Melbourne. , (
- 2004) Biodiversity effects on soil processes explained by interspecific functional dissimilarity. Science, 306, 1019–1020. , , , , , (
- 1999) Soil microarthropod contributions to decomposition dynamics: tropical–temperate comparisons of a single substrate. Ecology, 80, 1873–1882. , , , , (
- 2004) Wounding Earth's fragile skin. Science, 304, 1616–1618. (
- 2008) Soil health in agricultural systems. Philosophical Transactions of the Royal Society Series B, 363, 685–701. , , (
- 2005) Forest soils and carbon sequestration. Forest Ecological Management, 220, 242–258. (
- 1997) Faunal activities and soil processes: adaptive strategies that determine ecosystem function. Advances in Ecological Research, 27, 93–132. (
- 2004) Connecting soil and sediment biodiversity: the role of scale and implications for management. In: Sustaining Biodiversity and Ecosystem Services in Soils and Sediments (ed. WallDH), pp. 193–224. Island Press, Washington. , , et al. (
- 1998) Biodiversity inventories, indicator taxa and effects of habitat modification in tropical forest. Nature, 391, 72–76. , , et al. (
- 1990) Possible changes in natural vegetation patterns due to global warming. IIASA Working Paper WP-90-008. International Institute for Applied Systems Analysis, Laxenburg, Austria. (
- 2003) Climatic effects on litter decomposition from arctic tundra to tropical rainforest. Global Change Biology, 9, 575–584. , , , (
- 2002) Biodiversity and Ecosystem Functioning: Synthesis and Perspectives. Oxford University Press, Oxford. , , (
- 2007) Awesome or ordinary? Global diversity patterns of oribatid mites. Ecography, 30, 209–216. , , (
- 1999) Climate and litter quality controls on decomposition: an analysis of modeling approaches. Global Biogeochemical Cycling, 13, 575–589. , , , , (
- 2003) Disentangling biodiversity effects on ecosystem functioning: deriving solutions to a seemingly insurmountable problem. Ecology Letters, 6, 567–579. , (
- 2000) Virtual biodiversity assessment systems. BioScience, 50, 441–450. , , , , , (
- 1996) The CENTURY model. In: Evaluation of Soil Organic Matter Models Using Existing Long-term Datasets (eds PowlsonDS, SmithP, SmithJU), pp. 283–293. Springer, Berlin. (
- 2007) Global-scale similarities in nitrogen release patterns during long-term decomposition. Science, 315, 361–364. , , et al. (
- 2006) Decomposition, δ13C, and the “lignin paradox”. Canadian Journal of Soil Science, 86, 235–245. , , (
- 1997) 13C nuclear magnetic resonance spectroscopy with cross-polarization and magic-angle spinning investigation of the proximate-analysis fractions used to assess litter quality in decomposition studies. Canadian Journal of Botany, 75, 1601–1613. , , , (
- 1996) CO2 and the carbon cycle. In: Climate Change 1995 (eds HoughtonJ, FilhoLM, CallanderBA, HarrisN, KattenbergA, MaskellK), pp. 76–86. Cambridge University Press, Cambridge. , , , , , , (
- 2006) Does plant species co-occurrence influence soil mite diversity? Ecology, 87, 625–633. , , (
- 1998) Global change, soil biodiversity, and nitrogen cycling in terrestrial ecosystems: three case studies. Global Change Biology, 4, 729–743. , , et al. (
- 1979) Decomposition in Terrestrial Ecosystems. Blackwell Scientific, Oxford. , , (
- 2004) Biodiversity and ecosystem services in agricultural landscapes – are we asking the right questions? Agriculture, Ecosystems & Environment, 104, 113–134. , , (
- 2002) Rates of litter decomposition over 6 years in Canadian forests: influence of litter quality and climate. Canadian Journal of Forestry Research, 32, 789–804. , , et al. (
- VEMAP (1995) Vegetation/ecosystem modeling and analysis project: comparing biogeography and biogeochemistry models in a continental-scale study of terrestrial ecosystem responses to climate-change and CO2 doubling. Global Biogeochemical Cycling, 9, 407–437.
- WallDH, (ed) (2004) Sustaining Biodiversity and Ecosystem Services in Soils and Sediments. Island Press, Washington, DC.
- 1995) Impacts of disturbance on detritus food webs in agro-ecosystems of contrasting tillage and weed management practices. Advances in Ecological Research, 26, 105–185. (
- 2002) Communities and Ecosystems: Linking the Aboveground and Belowground Components. Princeton University Press, Princeton. (
- Top of page
- Material and methods
Monthly values for mean precipitation (Precip, cm), maximum (Tmax) and minimum (Tmin) temperatures ( °C), and monthly climate decomposition index (CDI) calculation for each site. Italicised values indicate actual months in the field (see Table 2) for sites where litterbags were in the field for <12 months. For these sites only the italicised values were used to calculate the mean CDI. See Table 1 for explanation of site codes.
|TAS 1 & 2||Precip||8.04||6.30||4.71||8.28||8.17||9.55||12.18||12.22||9.62||11.78||8.94||10.32|