This is the accepted version of the journal article: Peng, Yan; Vesterdal, Lars; Peñuelas, Josep; [et al.]. «Soil fauna effects on litter decomposition are better predicted by fauna communities within litterbags than by ambient soil fauna communities». Plant and Soil, Vol. 487 (June 2023), p. 49–59. DOI 10.1007/s11104-023-05902-1 This version is available at https://ddd.uab.cat/record/284326 under the terms of the license 1 Soil fauna effects on litter decomposition are better predicted by 2 fauna communities within litterbags than by ambient soil fauna 3 communities 4 Peng, Y., Vesterdal, L., Peñuelas, J., Peguero, G., Wu, Q., Heděnec, P., Yue, K., 5 Wu, F. 6 7 Abstract 8 Soil fauna is one of the major drivers of the decomposition of litter, but how its density 9 and diversity may affect decomposition remains largely unknown. We know even less 10 about whether faunal communities in litterbags or ambient soils are better for predicting 11 their effects on litter decomposition, because soil fauna frequently move into and out 12 of litterbags. We synthesized 5336 observations extracted from 46 publications to 13 assess the effects of soil fauna and its density and diversity on the rate of decomposition 14 (k). The results indicated that (1) the presence of soil fauna significantly increased k by 15 an average of 33.0% and that the effects were mainly controlled by litter concentrations 16 of phosphorus and magnesium, (2) the density and diversity of litterbag soil fauna 17 significantly affected k and/or the loss of mass, but ambient communities had limited 18 effects, and (3) the effects of the fauna on decomposition in litterbags were most 19 significant during the early stages (0−30%) of the loss of mass. Our results 20 quantitatively synthesized the patterns and drivers of the effects of soil fauna on litter 21 decomposition and indicated that litterbag communities were better for predicting the 22 effects of soil fauna on litter decomposition and that their effects were most significant 23 during the early stages of decomposition, which help us to better understand the role of 1 24 soil fauna in litter decomposition and the associated processes of carbon and nutrient 25 cycling. 26 27 Keywords: decomposition rate, mass loss, density, diversity, decomposition stage, 28 meta-analysis 29 30 1. Introduction 31 The decomposition of plant litter is a central component of ecosystem functioning, 32 because up to 90% of terrestrial plant production returns as litter to the pool of dead 33 organic matter (Berg and McClaugherty, 2020; Cebrian, 1999). Litter decomposition is 34 hierarchically driven by climate, litter quality, and decomposers (Aerts, 1997), and an 35 increasing number of studies have found that decomposers play a more important role 36 during litter decomposition than previously expected, especially in the early stages 37 (Bradford et al., 2016; Bradford et al., 2017; Yue et al., 2022). A previous global meta- 38 analysis found that the presence of soil fauna significantly increased litter 39 decomposition by an average of 37% across terrestrial biomes, and the effects were 40 modulated by climate and litter quality (García-Palacios et al., 2013). The effects of 41 soil fauna on litter decomposition have been assessed in experimental, theoretical, and 42 meta-analytical studies at both local and regional scales (Frouz, 2018; García-Palacios 43 et al., 2013; Peng et al., 2020), but the effects were assessed by comparing the rate of 44 decomposition (k) between litterbags allowing and excluding soil fauna in most of the 45 studies, which has limited our understanding of how faunal density and diversity may 2 46 affect decomposition. We know even less about whether the communities in litterbags 47 or ambient soils are better for evaluating the effects of soil fauna on litter decomposition. 48 Soil fauna can contribute to litter decomposition directly by fragmentation and 49 feeding and indirectly by modifying the soil environment and interactions with 50 microbial communities (Frouz, 2018). For example, previous studies have reported that 51 soil fauna can consume a particular amount of litter, accounting for as much as 40% or 52 even 94% of annual litterfall (Frouz et al., 2015; Knollenberg et al., 1985). The amount 53 of litter that soil fauna consume, however, is highly dependent on litter quality; quality 54 is a major factor controlling the effects of soil fauna on litter decomposition (Peng et 55 al., 2022a). Also, soil fauna usually graze selectively on fungal hyphae and invariably 56 ingest bacteria and fungi, maintaining microbes contributing to the decomposition in 57 an active stage of growth (Frouz, 2018) and thus indirectly contributing to 58 decomposition. These processes are likewise also greatly modulated by litter quality 59 (Suzuki et al., 2013). The effects of soil fauna on litter decomposition may thus also be 60 affected by plant functional type (PFT), mycorrhizal association, and leaf type, because 61 they are tightly linked to litter quality, with the quality of litter higher from herbaceous 62 than woody plants, plants associated with arbuscular mycorrhiza (AM) than 63 ectomycorrhiza (ECM) fungi, and broadleaved than coniferous plants (Lin et al., 2017; 64 Peng et al., 2020). How these variables may influence the effects of soil fauna on litter 65 decomposition at regional scales, however, remain elusive. 66 In addition to litter quality, the effects of soil fauna on litter decomposition can 67 also be regulated by other factors such as climate and soil properties, because these 3 68 factors are closely associated with soil-fauna communities, physiologies, and activities 69 (García-Palacios et al., 2013; Peng et al., 2022a). The effects of soil fauna on litter 70 decomposition can also vary among different stages of decomposition, because litter 71 quality varies substantially during decomposition (Yue et al., 2018). A recent global 72 meta-analysis found that the effects of invertebrates on litter decomposition in streams 73 was most significant during the initial stages (0−20%) of the loss of mass (Yue et al., 74 2022), and results from forests have also suggested that nematode communities 75 dominated the early stages (García-Palacios et al., 2016b). We know little about 76 whether the effects of soil fauna on litter decomposition vary among stages of 77 decomposition across multiple types of terrestrial ecosystems at reginal and global 78 scales. The effects of soil fauna on litter decomposition have also usually been assessed 79 by comparing differences in k between fine-mesh litterbags that exclude soil fauna and 80 coarse-mesh litterbags that allow the access of soil fauna, which may fail to fully 81 account for their effects (Chassain et al., 2021). Assessing how soil-fauna communities 82 themselves affect litter decomposition would therefore help us to better understand their 83 “real” effects on decomposition. Many studies have reported soil-fauna communities 84 along with the results of litter decomposition, but their relationships have not been well 85 assessed. Importantly, assessing the effects of soil-fauna communities in litterbags on 86 litter decomposition may underestimate the effects, because the fauna may frequently 87 move into and out of the litterbags, but how ambient communities of soil fauna may 88 affect litter decomposition has rarely been reported. 89 We systematically synthesized 5336 observations extracted from 46 publications 4 90 to evaluate the effects of soil fauna and its density and diversity on k, mass loss, and the 91 underlying driving factors. We also assessed the relationships between the effects of 92 soil fauna quantified by the differences between litterbags with fine and coarse mesh 93 sizes and between the densities and diversities of fauna in litterbags and ambient soil 94 and whether the effects varied among stages of decomposition. We hypothesized that 95 (i) soil fauna would greatly increase litter decomposition assessed by comparing the 96 differences between litterbags with fine and coarse mesh sizes, (ii) the effects of soil 97 fauna would mainly be controlled by litter quality but also by PFT, mycorrhizal 98 association, leaf type, litterbag mesh size, ecosystem type, and soil properties, (iii) the 99 density and diversity of fauna in both litterbags and ambient soil would strongly 100 positively affect litter decomposition, and (iv) the effects of soil fauna on litter 101 decomposition would be largest in the early stages of decomposition. 102 103 2. Materials and methods 104 2.1. Data 105 Peer-reviewed articles, book chapters, and academic dissertations addressing the effects 106 of soil fauna on litter decomposition were searched in the Web of Science and China 107 Knowledge Resource Integrated Database on 28 June 2022 using the search terms 108 (decomposition OR decay* OR processing) AND (litter OR leaf) AND (“soil animal*” 109 OR “soil invertebrate*” OR “soil fauna” OR macrofauna OR mesofauna) and their 110 equivalents in Chinese. To be included in our database, the primary studies had to 111 satisfy four criteria: (i) experiments were carried out in the field, and litter 5 112 decomposition was assessed using litterbags, (ii) litterbags that allowed and excluded 113 soil fauna were at the same microsite, (iii) results for both decomposition (k or mass 114 loss) and soil-fauna communities (density and the Shannon-Weiner, Simpson, Pielou, 115 and/or Margalef indices) in litterbags and/or ambient soils were reported for one 116 experiment, and (iv) the means, standard deviations, and sample sizes of the variables 117 were directly reported or could be estimated. 118 When the same primary study examined leaf litter from different plant species, 119 ecosystem types, and/or litterbag mesh sizes, we considered each of these parameters 120 separately, because they represented different examples of the effects of soil fauna on 121 litter decomposition. Also, if the effect on decomposition was observed several times 122 during decomposition, we included all observations to assess how the effects varied 123 among the stages of decomposition (Yue et al., 2022). We also collected data for the 124 initial litter concentrations of carbon (C), nitrogen (N), phosphorus (P), calcium (Ca), 125 magnesium (Mg), lignin, and cellulose; the C:N and lignin:N ratios; the pH and 126 moisture and clay concentrations of the soil, when available, and latitude and elevation 127 to assess their impacts. Data were directly collected from the main texts, tables, or 128 appendices or were extracted from figures using Engauge Digitizer version 11.2 129 (https://github.com/markummitchell/engauge-digitizer/releases) for digitization. Most 130 primary studies did not report data for soil microbes, so we extracted data for microbial 131 biomass C (MBC) and microbial biomass N (MBN) from Distributed Active Archive 132 Center for Biogeochemical Dynamics (Xu et al., 2014) based on the coordinates of the 133 study sites. We also recorded plant functional type (PFT, herbaceous vs woody), 6 134 mycorrhizal association [AM, ECM, and dual (i.e. plants associated with both AM and 135 ECM fungi)] of the plants contributing leaf litter, type of leaf litter (broadleaf vs needle), 136 mesh sizes of litterbags allowing or excluding the access of soil fauna (0.5−2 vs >2 137 mm), and ecosystem type. We obtained a total of 5336 observations after extraction 138 (195 for kcoarse vs kfine litterbags, 643 for kcoarse vs litterbag communities, 248 for kcoarse 139 vs ambient communities, 2486 for mass loss vs communities in coarse-mesh litterbags, 140 and 1764 for mass loss in coarse-mesh litterbags vs ambient communities) from 46 141 publications (Fig. 1, Appendix 1). 142 143 2.2. Statistical analysis 144 To evaluate the effects of the fauna on decomposition by comparing litterbags with and 145 without fauna, we used the natural logarithm of the response ratio (lnRR) as the proxy 146 of effect size. The individual lnRRs were calculated as: 𝑘𝑝𝑟𝑒𝑠𝑒𝑛𝑐𝑒 147 lnRR = ln ( ) (1) 𝑘𝑎𝑏𝑠𝑒𝑛𝑐𝑒 148 where kpresence and kabsence are k in the litterbags with and without soil fauna, respectively. 149 The variance (v) associated with each lnRR was calculated as: 2 2 𝑠presence 𝑠absence 150 𝑣= 2 + 2 (2) 𝑛presence 𝑘presence 𝑛absence 𝑘absence 151 where npresence and nabsence are the sample sizes, and spresence and sabsence are the SDs of k 152 in the litterbags with and without soil fauna, respectively, and the weight for each lnRR 153 was then calculated as the inverse of its v (1/v). To calculate the weighted mean effect 154 sizes (lnRR++), we ran intercept-only linear mixed-effects models using the lme4 155 package in R (Bates et al., 2014), with the identities of the primary studies fitted as a 7 156 random-effect factor to explicitly account for the potential dependences among 157 observations that were extracted from one primary study. We ran meta-regressions by 158 fitting each regression as a fixed-effects factor in separate models to assess the impacts 159 of the moderator variables on the effect sizes of the fauna on decomposition. For aiding 160 interpretation, lnRR++ and the associated 95% confidence intervals (CIs) were back- 161 transformed using the equation (𝑒 𝑙𝑛𝑅𝑅++ − 1) × 100 (Koricheva et al., 2013). 162 To assess the effects of faunal density and diversity on k and mass loss, we used 163 linear mixed-effects models by fitting k or mass loss as a response variable, density or 164 a diversity index in the litterbags or ambient soils as a fixed-effects factor, and the 165 identities of the primary studies as a random-effects factor. The relationships identified 166 by the linear mixed-effects models may not be causal relationships or “effects” but can 167 to some extent account for how density and diversity may affect decomposition (Yue et 168 al., 2022). We thus used “effect” here for easy description and understanding. We fitted 169 their interactions with the faunal variables as a fixed-effects factor to evaluate the 170 impacts of the moderator variables. We used intervals of 10% mass loss (i.e. 0−10, 171 10−20, 20−30, …, 70−80, and 80−90%) for analysis to assess how the faunal effects 172 varied among different stages of decomposition. Finally, we used linear regression to 173 detect the relationships between the lnRR of k and faunal density and diversity. 174 175 2.3. Publication bias 176 We used Egger’s regression (Egger et al., 1997), funnel plots, and the trim-and-fill test 177 (Duval and Tweedie, 2000) to detect potential publication bias following previous 8 178 studies (Peng et al., 2022c; Yue et al., 2020). We applied both Egger’s regression and 179 the trim-and-fill test using the meta-analytic residuals consisting of sampling errors and 180 the effects of the level of the effect size-, which are equivalent to normal residuals 181 (Nakagawa and Santos, 2012). We used the R0 estimator with the trimfill function in 182 the metafor R package to conduct the trim-and-fill test (Viechtbauer, 2010). None of 183 the results from Egger’s regression, the funnel plots, or the trim-and-fill test provided 184 evidence of publication bias in our study (Fig. S1, Table S1), indicating that the primary 185 studies were representative samples of the available published studies. 186 187 3. Results 188 3.1. Effects of the soil fauna 189 The presence of soil fauna in the litterbags significantly increased k by an average of 190 33.0% (Fig. 2). The effects of the fauna on k were not affected by PFT, leaf type, 191 mycorrhizal association, or ecosystem type but were significantly affected by litterbag 192 mesh size, with 24.8 and 44.9% increases in k in litterbags with mesh sizes of 0.5−2 193 and >2 mm, respectively. Likewise, latitude, elevation, soil properties, and initial litter 194 quality had no impacts on the faunal effects on k except for the initial litter 195 concentrations of P and Mg, both of which positively affected the effects of the fauna 196 (Table 1). 197 198 3.2. Effects of the density and diversity of soil fauna 199 The density of the litterbag fauna significantly positively affected k (Fig. 3), with larger 9 200 effects for broadleaf litter, litter from plants associated with both AM and ECM fungi, 201 and in litterbags with mesh sizes of 0.5−2 mm (Fig. S2), but the diversity of the litterbag 202 fauna did not affect k. Neither the density nor the diversity of the ambient communities, 203 however, was significantly correlated with k, and PFT, leaf type, mycorrhizal 204 association, mesh size, and ecosystem type did not affect k (Fig. S3). The loss of litter 205 mass was affected positively by litterbag faunal density and the Margalef index but 206 negatively by the Pielou index, and their effects were significantly influenced by 207 multiple moderator variables (Fig. S4, Table 1). Only the Margalef index significantly 208 positively affected the loss of litter mass in the ambient communities, which were 209 modulated by litterbag mesh size, ecosystem type, latitude, and the initial litter 210 concentrations of C, N, and P (Fig. S5, Table S2). LnRR of k was correlated positively 211 with litterbag faunal density and the Pielou and Margalef indices but negatively with 212 the Simpson index. LnRR of k was only negatively correlated with the ambient faunal 213 density (Fig. 4). 214 215 3.3. Effect of soil fauna in different stages of litter decomposition 216 Litterbag faunal density and the Simpson and Pielou indices did not affect the loss of 217 litter mass within different stages of litter decomposition (Fig. 5). The positive effects 218 of the Shannon-Weiner index for the litterbag fauna on litter mass were only significant 219 in the 10−20% interval of mass loss, and the Margalef index significantly positively 220 affected mass loss during the early stages (0−30% mass loss) of decomposition. We 221 could not, however, assess how the decomposition stages affected mass loss for the 10 222 ambient communities due to the limited observations for the density and diversity of 223 the ambient fauna. 224 225 4. Discussion 226 4.1. Consistently positive effects of soil fauna 227 Our results indicated that the presence of soil fauna significantly increased k by an 228 average of 33%, consistent with the findings from previous meta-analyses by García- 229 Palacios et al. (2013) and Xu et al. (2020) that reported increases in k of 37 and 30.9%, 230 respectively, and consistent with our first hypothesis, indicating the robustness of our 231 data set and results. Our results also suggested that the fauna had consistent positive 232 effects on k regardless of PFT, leaf type, mycorrhizal association, litterbag mesh size, 233 and ecosystem type, although the effects were significantly larger in litterbags with 234 mesh sizes >2 mm than in those with mesh sizes of 0.5−2 mm, indicating that both 235 meso- and macrofauna significantly affected litter decomposition. Surprisingly, only 236 the initial litter concentrations of P and Ca significantly positively influenced the effects 237 of the soil fauna, and the initial litter N concentration and the lignin:N ratio that are 238 important for the rate of litter decomposition (Berg and McClaugherty, 2020) had no 239 effect. The positive effects of P concentration may have been because P is an essential 240 nutrient for soil fauna (Singh et al., 2018), and litters with higher P concentrations are 241 thus preferred by soil fauna. The positive effects of Mg on litter decomposition and the 242 effects of soil fauna on litter decomposition have been observed in several studies 243 (García-Palacios et al., 2016a; Peng et al., 2022a; Yue et al., 2018), because Mg is a 11 244 key element in the diet of soil fauna, and higher litter Mg concentrations will positively 245 affect the communities and activities of soil fauna (National Research Council, 2005) 246 and thus influence the effects of the fauna on litter decomposition. 247 248 4.2. Effects of the density and diversity of soil fauna 249 Our results indicated that the density of the litterbag soil fauna positively affected both 250 k and mass loss. The density and diversity of the ambient fauna were not correlated 251 with decomposition, and the Margalef index was correlated with mass loss, inconsistent 252 with our third hypothesis. Positive effects of faunal density on decomposition have also 253 been reported across global stream ecosystems (Yue et al., 2022), indicating a common 254 pattern of the effects of invertebrates on litter decomposition across terrestrial and 255 aquatic ecosystems and supporting the positive effects quantified by comparing 256 differences between fine- and coarse-mesh litterbags. The negative effects of the Pielou 257 index and the positive effects of the Margalef index on the effect sizes of soil fauna in 258 our study suggest that faunal diversity would facilitate the effects on decomposition, 259 but the evenness of different species would decrease their effects. A more 260 taxonomically diverse soil fauna has more opportunities for niche differentiation of the 261 efficiency in the use of energy and nutrients from litter and of food preferences and 262 interactions with microbes that facilitate decomposition (Barbier et al., 2008; 263 Hättenschwiler et al., 2005). Likewise, the effects of the density of litterbag fauna on k 264 in our study were also positively affected by the litter concentrations of P and Mg, and 265 litter mass was more sensitive to litter quality, latitude, elevation, and soil properties 12 266 (Table S3). The effects of the diversity of the soil fauna on k and mass loss were also 267 affected by litter quality, latitude, elevation, and soil properties but varied among the 268 indices. 269 Our results also suggested that the ambient communities had limited effects on 270 decomposition, indicating that the common use of soil-fauna communities in litterbags 271 for predicting litter decomposition is meaningful. The limited effects of the ambient 272 communities on decomposition may be attributed to the relative stability of the density 273 and diversity of the ambient communities compared with the litterbag communities 274 (Wang, 2016), so ambient communities cannot well describe variations in litterbag litter 275 decomposition. Not all ambient soil fauna would move into litterbags, so the 276 contributions of soil fauna to litter decomposition were thus those in litterbags, which 277 was supported by our result that faunal lnRR for k was negatively correlated with the 278 density of the ambient soil fauna. 279 280 4.3. Larger effects of the diversity of soil fauna in the early stages of decomposition 281 Our results indicated that the effects of the density of the litterbag soil fauna on the loss 282 of litter mass were not significant when divided into different stages of decomposition, 283 but the Shannon-Weiner and Margalef indices had significantly positive effects during 284 the 0−30% intervals of mass loss, partly supporting our fourth hypothesis. Significantly 285 positive effects of soil fauna on decomposition during the early stages have been 286 observed in previous studies in both terrestrial and aquatic ecosystems (García-Palacios 287 et al., 2016b; Liu et al., 2019; Yue et al., 2022), where the soil fauna preferred high- 13 288 quality litter (Peng et al., 2022b; Yang et al., 2020), which decreases during 289 decomposition with the release of liable components (Berg, 2014). Another potential 290 mechanism may be the early regulation of litter decomposition by microbes (García- 291 Palacios et al., 2016b), and the higher colonization and larger effects of microbes in the 292 early stages of decomposition can render litter more accessible to invertebrates (Jinggut 293 and Yule, 2015), thereby increasing the effects of the soil fauna. Another study, however, 294 found that the net contribution of soil fauna to litter decomposition increased as 295 conditions for microbial decomposition became increasingly adverse, particularly when 296 the concentrations of N and other nutrients in the litter substrate and the surrounding 297 environment decreased (Peguero et al., 2019). These divergent findings may be because 298 some taxonomic groups such as Oribatida that are not sensitive to litter quality (Bluhm 299 et al., 2019) contribute more to decomposition. We did not assess the effects of ambient 300 soil fauna on litter decomposition at different stages, however, due to the limited 301 observations for ambient communities. 302 303 4.4. Uncertainty analysis 304 Several uncertainties and knowledge gaps attributed to limitations in experimental 305 designs and the lack of data points from major regions of the globe still need more 306 attention, despite the overall patterns and drivers of the effects of soil fauna on litter 307 decomposition reported in our study. Firstly, the studies assessing the effects of soil 308 fauna generally compared the differences in litter decomposition among fine- and 309 coarse-mesh litterbags and did not report soil-fauna communities, which prevented a 14 310 robust assessment of the effects of density and diversity on decomposition. Secondly, 311 the dynamics of both ambient communities and litter decomposition were seldom 312 reported, and the nonsignificant effects of the density and diversity of ambient soil 313 fauna on litter decomposition may be attributed to the limited number of data points. 314 Thirdly, the data points in our compiled database were mainly from China, with several 315 others distributed in America, Europe, and Australia, and the lack of data points in 316 major biomes from America, Europe, Australia, and Africa greatly limited our ability 317 to identify a common pattern of the effects of soil fauna on litter decomposition. We 318 therefore recommend that the communities in both litterbags and ambient soils should 319 be considered and reported in studies assessing the effects of soil fauna on litter 320 decomposition, especially in critical regions that have been rarely studied. Studies 321 should also report information such as initial litter quality and soil properties that are 322 closely associated with the decomposition of litter. 323 324 5. Conclusions 325 Our meta-analysis synthesizing published data clearly demonstrated that soil fauna 326 consistently had positive effects on litter decomposition regardless of PFT, leaf type, 327 mycorrhizal association, litterbag mesh size, and ecosystem type, increasing k by an 328 average of 33.0%, and that these effects were dominantly controlled by the initial litter 329 concentrations of P and Mg. Our results also suggested that the density and diversity of 330 soil fauna in litterbags significantly affected k and/or mass loss, but ambient 331 communities had limited effects on decomposition. The effects of the diversity of the 15 332 litterbag soil fauna on the loss of litter mass were most significant in the early stages 333 (0−30% mass loss) of decomposition. Our meta-analysis both quantitatively 334 summarized the patterns of the effects of soil fauna on litter decomposition at a regional 335 scale and indicated that communities in litterbags were better than ambient 336 communities for predicting the effects on decomposition. These results will help us to 337 better understand the role of soil fauna in the decomposition of litter and the associated 338 processes of C and nutrient cycles in terrestrial ecosystems. 339 340 341 Data availability 342 Raw data used in the study will be deposited in figshare (https://figshare.com) if this 343 manuscript is accepted. 344 345 References 346 Aerts, R., 1997. Climate, leaf litter chemistry and leaf litter decomposition in terrestrial ecosystems: 347 a triangular relationship. Oikos 79(3), 439-449. 348 Barbier, S., Gosselin, F., Balandier, P., 2008. Influence of tree species on understory vegetation 349 diversity and mechanisms involved—A critical review for temperate and boreal forests. 350 Forest Ecology and Management 254(1), 1-15. 351 Bates, D., Mächler, M., Bolker, B., Walker, S., 2014. Fitting linear mixed-effects models using lme4. 352 Journal of Statistical Software 67(1), 1-48. 353 Berg, B., 2014. Decomposition patterns for foliar litter – A theory for influencing factors. Soil 354 Biology and Biochemistry 78, 222-232. 355 Berg, B., McClaugherty, C., 2020. Plant litter: decomposition, humus formation, carbon 356 sequestration. 4th edition ed. Springer Nature, Cham, Switzerland. 357 Bluhm, C., Butenschoen, O., Maraun, M., Scheu, S., 2019. Effects of root and leaf litter identity and 358 diversity on oribatid mite abundance, species richness and community composition. Plos 359 One 14(7), e0219166. 360 Bradford, M.A., Berg, B., Maynard, D.S., Wieder, W.R., Wood, S.A., 2016. Understanding the 361 dominant controls on litter decomposition. Journal of Ecology 104(1), 229-238. 16 362 Bradford, M.A., Veen, G., Bonis, A., Bradford, E.M., Classen, A.T., Cornelissen, J.H.C., Crowther, T., 363 De Long, J.R., Freschet, G.T., Kardol, P., 2017. A test of the hierarchical model of litter 364 decomposition. Nature Ecology and Evolution 1(12), 1836-1845. 365 Cebrian, J., 1999. Patterns in the fate of production in plant communities. The American Naturalist 366 154(4), 449-468. 367 Chassain, J., Vieublé Gonod, L., Chenu, C., Joimel, S., 2021. Role of different size classes of 368 organisms in cropped soils: What do litterbag experiments tell us? A meta-analysis. Soil 369 Biology and Biochemistry 162, 108394. 370 Duval, S., Tweedie, R., 2000. Trim and fill: a simple funnel‐plot–based method of testing and 371 adjusting for publication bias in meta‐analysis. Biometrics 56(2), 455-463. 372 Egger, M., Smith, G.D., Schneider, M., Minder, C., 1997. Bias in meta-analysis detected by a simple, 373 graphical test. British Medical Journal 315(7109), 629-634. 374 Frouz, J., 2018. Effects of soil macro- and mesofauna on litter decomposition and soil organic 375 matter stabilization. Geoderma 332, 161-172. 376 Frouz, J., Jedlička, P., Šimáčková, H., Lhotáková, Z., 2015. The life cycle, population dynamics, and 377 contribution to litter decomposition of Penthetria holosericea (Diptera: Bibionidae) in an 378 alder forest. European Journal of Soil Biology 71, 21-27. 379 García-Palacios, P., Maestre, F.T., Kattge, J., Wall, D.H., 2013. Climate and litter quality differently 380 modulate the effects of soil fauna on litter decomposition across biomes. Ecology Letters 381 16(8), 1045-1053. 382 García-Palacios, P., McKie, B.G., Handa, I.T., Frainer, A., Hättenschwiler, S., 2016a. The importance 383 of litter traits and decomposers for litter decomposition: a comparison of aquatic and 384 terrestrial ecosystems within and across biomes. Functional Ecology 30(5), 819-829. 385 García-Palacios, P., Shaw, E.A., Wall, D.H., Hättenschwiler, S., 2016b. Temporal dynamics of biotic 386 and abiotic drivers of litter decomposition. Ecology Letters 19(5), 554-563. 387 Hättenschwiler, S., Tiunov, A.V., Scheu, S., 2005. Biodiversity and litter decomposition in terrestrial 388 ecosystems. Annual Review of Ecology, Evolution, and Systematics 36, 191-218. 389 Jinggut, T., Yule, C.M., 2015. Leaf-litter breakdown in streams of East Malaysia (Borneo) along an 390 altitudinal gradient: initial nitrogen content of litter limits shredder feeding. Freshwater 391 Science 34(2), 691-701. 392 Knollenberg, W.G., Merritt, R.W., Lawson, D.L., 1985. Consumption of leaf litter by Lumbricus 393 terrestris (Oligochaeta) on a Michigan woodland floodplain. The American Midland 394 Naturalist 113(1), 1-6. 395 Koricheva, J., Gurevitch, J., Mengersen, K., 2013. Handbook of meta-analysis in ecology and 396 evolution. Princeton University Press, Princeton and Oxford. 397 Lin, G., McCormack, M.L., Ma, C., Guo, D., 2017. Similar below-ground carbon cycling dynamics 398 but contrasting modes of nitrogen cycling between arbuscular mycorrhizal and 399 ectomycorrhizal forests. New Phytologist 213(3), 1440-1451. 400 Liu, Y., Wang, L., He, R., Chen, Y., Xu, Z., Tan, B., Zhang, L., Xiao, J., Zhu, P., Chen, L., 2019. Higher 401 soil fauna abundance accelerates litter carbon release across an alpine forest-tundra 402 ecotone. Scientific Reports 9(1), 10561. 403 Nakagawa, S., Santos, E.S., 2012. Methodological issues and advances in biological meta-analysis. 404 Evolutionary Ecology 26(5), 1253-1274. 405 National Research Council, 2005. Mineral Tolerance of Animals. The National Academies Press, 17 406 Washington D.C. 407 Peguero, G., Sardans, J., Asensio, D., Fernández-Martínez, M., Gargallo-Garriga, A., Grau, O., Llusià, 408 J., Margalef, O., Márquez, L., Ogaya, R., 2019. Nutrient scarcity strengthens soil fauna 409 control over leaf litter decomposition in tropical rainforests. Proceedings of the Royal 410 Society B 286(1910), 20191300. 411 Peng, Y., Holmstrup, M., Schmidt, I.K., Bachega, L.R., Schelfhout, S., Zheng, H., Heděnec, P., Yue, K., 412 Vesterdal, L., 2022a. Tree species identity is the predominant modulator of the effects of 413 soil fauna on leaf litter decomposition. Forest Ecology and Management 520, 120396. 414 Peng, Y., Holmstrup, M., Schmidt, I.K., De Schrijver, A., Schelfhout, S., Heděnec, P., Zheng, H., 415 Bachega, L.R., Yue, K., Vesterdal, L., 2022b. Litter quality, mycorrhizal association, and soil 416 properties regulate effects of tree species on the soil fauna community. Geoderma 407, 417 115570. 418 Peng, Y., Peñuelas, J., Vesterdal, L., Yue, K., Peguero, G., Fornara, D.A., Heděnec, P., Steffens, C., 419 Wu, F., 2022c. Responses of soil fauna communities to the individual and combined effects 420 of multiple global change factors. Ecology Letters 25, 1961-1973. 421 Peng, Y., Schmidt, I.K., Zheng, H., Heděnec, P., Bachega, L.R., Yue, K., Wu, F., Vesterdal, L., 2020. 422 Tree species effects on topsoil carbon stock and concentration are mediated by tree 423 species type, mycorrhizal association, and N-fixing ability at the global scale. Forest 424 Ecology and Management 478, 118510. 425 Singh, J., Hundal, J., Sharma, A., Singh, U., Sethi, A., Singh, P., 2018. Phosphorus nutrition in dairy 426 animals: A review. International Journal of Current Microbiology and Applied Sciences 7(4), 427 3518-3530. 428 Suzuki, Y., Grayston, S.J., Prescott, C.E., 2013. Effects of leaf litter consumption by millipedes 429 (Harpaphe haydeniana) on subsequent decomposition depends on litter type. Soil Biology 430 and Biochemistry 57, 116-123. 431 Viechtbauer, W., 2010. Conducting meta-analyses in R with the metafor package. Journal of 432 statistical software 36(3), 1-48. 433 Wang, Z., 2016. Litter decomposition and the effects of soil fauna on litter decomposition and 434 nutrient release in the coniferous forest of the Changbai Mountains, China, Norheast 435 Normal University, Changchun, China, 144 pp. 436 Xu, X., Sun, Y., Sun, J., Cao, P., Wang, Y., Chen, H.Y., Wang, W., Ruan, H., 2020. Cellulose dominantly 437 affects soil fauna in the decomposition of forest litter: A meta-analysis. Geoderma 378, 438 114620. 439 Xu, X., Thornton, P., Potapov, P., 2014. A aompilation of global soil microbial biomass carbon, 440 nitrogen, and phosphorus data. ORNL DAAC, Oak Ridge, Tennessee. 441 Yang, Y., Wu, Q., Yang, W., Wu, F., Zhang, L., Xu, Z., Liu, Y., Tan, B., Li, H., Zhou, W., 2020. 442 Temperature and soil nutrients drive the spatial distributions of soil macroinvertebrates 443 on the eastern Tibetan Plateau. Ecosphere 11(3), e03075. 444 Yue, K., De Frenne, P., Van Meerbeek, K., Ferreira, V., Fornara, D.A., Wu, Q., Ni, X., Peng, Y., Wang, 445 D., Heděnec, P., Yang, Y., Wu, F., Peñuelas, J., 2022. Litter quality and stream 446 physicochemical properties drive global invertebrate effects on instream litter 447 decomposition. Biological Reviews 97, 1-16. 448 Yue, K., García‐Palacios, P., Parsons, S.A., Yang, W., Peng, Y., Tan, B., Huang, C., Wu, F., 2018. 449 Assessing the temporal dynamics of aquatic and terrestrial litter decomposition in an 18 450 alpine forest. Functional Ecology 32(10), 2464-2475. 451 Yue, K., Jarvie, S., Senior, A.M., Van Meerbeek, K., Peng, Y., Ni, X., Wu, F., Svenning, J.-C., 2020. 452 Changes in plant diversity and its relationship with productivity in response to nitrogen 453 addition, warming and increased rainfall. Oikos 129(7), 939-952. 454 19 455 Table 1 Impacts of latitude, elevation, initial litter quality, and soil property on the size of the effect (lnRR) of soil fauna on the rate of litter decomposition 456 (k, per year) and on the effects litterbag soil-fauna density and the Shannon-Weiner, Simpson, Pielou index, and Margalef indices assessed using linear 457 mixed-effects models. Slope, the p value, and the number (n) of observations are reported. Bold indicates significant effects. *, p < 0.05; **, p < 0.01; ***, 458 p < 0.001. Moderator variable lnRR Litterbag soil-fauna community Density Shannon Simpson Pielou Margalef Estimate n Estimate n Estimate n Estimate n Estimate n Estimate n Latitude (°) -0.009 195 -0.003 125 0.007 178 0.038** 106 -0.012 124 0.017* 110 Elevation (m) -0.001 187 -0.001 125 0.001 170 -0.001 98 0.001 116 -0.001 102 Initial litter quality C (%) -0.057 177 -0.019* 111 0.003 148 0.051* 88 0.019 110 -0.052*** 92 N (%) 0.168 177 0.009 111 -0.378*** 156 -0.233*** 92 -1.283*** 110 0.350*** 92 P (%) 0.285** 154 0.111*** 100 0.586 137 -1.737** 74 -1.008 92 0.213 74 C:N ratio -0.022 177 -0.002 111 0.006* 148 0.010 88 0.027*** 92 -0.011 92 Lignin (%) 0.108 132 0.002 78 0.005 113 -0.041** 70 0.021 80 -0.012* 74 Cellulose (%) 0.078 124 0.001 76 0.018* 109 0.001 0.965 0.001 78 -0.001 68 Lignin:N ratio 0.144 127 0.002 78 0.002 113 -0.029** 70 0.005 80 -0.007 73 Ca (%) 0.065 87 0.033 38 -0.018 62 -0.274 35 -0.497*** 78 0.105 29 Mg (%) 0.212* 99 0.045*** 38 -0.569 70 -1.746*** 41 -1.364* 45 1.203** 35 Soil property pH -0.039 193 -0.013 125 0.15*** 168 0.122 106 -0.075 122 0.075 108 Moisture (%) 0.273 193 0.001 125 -0.053*** 168 -0.078 106 -0.109* 122 0.025 108 Clay (%) 0.052 193 0.001 125 -0.036* 168 -0.035 106 -0.138** 122 0.026 108 MBC (g C/m2) -0.005 189 -0.001 121 0.006* 168 0.009 106 0.003 122 0.002 108 MBN (g N/m2) -0.041 189 -0.003 121 0.069** 168 0.195 106 0.219** 122 0.033 108 MBC:MBN ratio 0.025 189 0.003 121 0.013 168 0.069 106 -0.040 122 0.041* 108 459 C, carbon; N, nitrogen; P, phosphorus; MBC, microbial biomass carbon; MBN, microbial biomass nitrogen; MBP, microbial biomass phosphorus. 460 20 461 462 Figure 1 Map of the distribution of the study sites in our study. the size of shapes 463 indicate sample size,. 464 21 Total (195) *** PFTns Herbaceous (73) ** Woody (118) *** Leaf typens Brodleaf (94) *** Needle (38) *** Mycorrhizal associationns AM (62) *** ECM (61) ** Dual (21) ** Mesh size*** 0.5–2 mm (106) *** > 2 mm (89) *** Ecosystemns Forest (124) *** Grassland (70) ** Tundra (1) -30 0 30 60 90 465 Changes in percentage (%) 466 Figure 2 Effects of soil fauna on litter decomposition assessed by litterbag mesh sizes. 467 22 (a) Decomposition rate (b) Mass loss Density (125) *** (712) *** Shannon (178) (680) Simpson (106) (308) Pielou (124) (402) *** Margalef (110) (384) *** Litterbag Litterbag Density (72) (474) Shannon (50) (334) Simpson (46) (286) Pielou (46) (286) Margalef (34) (384) *** Ambient Ambient -1.6 -0.8 0.0 0.8 -24 -12 0 12 24 468 Effects of soil fauna density and diversity 469 Figure 3 Effects of the density and diversity of the litterbag and ambient soil fauna on 470 the rate of litter decomposition (a) and on mass loss (b). Values are the estimated slopes 471 and 95% confidence intervals of the linear mixed-effects models. Brown and blue 472 indicate significantly positive and negative effects, respectively, and gray indicates 473 nonsignificant effects. Shannon, Shannon-Weiner index; Simpson, Simpson index; 474 Pielou, Pielou index; Margalef, Margalef index; ***, p < 0.001. 475 23 476 477 Figure 4 Relationships between the effect sizes (lnRR) of the soil fauna on the rate of 478 litter decomposition and the densities and diversities of the litterbag and ambient soil 479 fauna using pairwise data points. Fitted linear regression lines and 95% confidence 480 intervals are shown. Shannon, Shannon-Weiner index; Simpson, Simpson index; Pielou, 481 Pielou index; Margalef, Margalef index. 482 24 Density Shannon Simpson Pielou Margalef (a) (b) (c) (d) (e) Litter decomposition stage (mass loss interval) 0-10 (100) (62) (27) (36) (97) *** 10-20 (124) (124) * (72) (95) (77) *** 20-30 (103) (83) (37) (61) (42) * 30-40 (92) (78) (25) (42) (30) 40-50 (99) (115) (54) (67) (53) 50-60 (92) (120) (61) (65) (49) 60-70 (77) (79) (26) (29) (29) 70-80 (20) (16) (5) (6) (6) 80-90 (5) (3) p = 0.195 p = 0.009 p = 0.096 p = 0.338 p < 0.001 -0.5 0.0 0.5 -4 0 4 -4 0 4 -10 0 10 -4 0 4 483 Effects of soil fauna density and diversity on litter mass loss 484 Figure 5 Effects of the density and diversity of soil fauna on the loss of litter mass at 485 different stages of decomposition represented by 0−10% intervals of mass loss. Values 486 are the estimated slopes and 95% confidence intervals of the linear mixed-effects 487 models. Brown indicates a significantly positive effect, and gray indicates 488 nonsignificant effects. Shannon, Shannon-Weiner index; Simpson, Simpson index; 489 Pielou, Pielou index; Margalef, Margalef index; *, p < 0.05; ***, p < 0.001. 490 25