Alliance Bioversity-CIAT Research Online Accepted Manuscript Pasture diversification affects soil macrofauna and soil biophysical properties in tropical (silvo)pastoral systems The Alliance of Bioversity International and the International Center for Tropical Agriculture believes that open access contributes to its mission of reducing hunger and poverty, and improving human nutrition in the tropics through research aimed at increasing the eco-efficiency of agriculture. The Alliance is committed to creating and sharing knowledge and information openly and globally. We do this through collaborative research as well as through the open sharing of our data, tools, and publications. Citation: Vazquez, E.; Teutscherova, N.; Lojka, B.; Arango, J.; Pulleman, M. (2020) Pasture diversification affects soil macrofauna and soil biophysical properties in tropical (silvo)pastoral systems. Agriculture, Ecosystems & Environment 302:107083 10 p. ISSN: 0167-8809 Publisher’s DOI: https://doi.org/10.1016/j.agee.2020.107083 Access through CIAT Research Online: https://hdl.handle.net/10568/108869 Terms: © 2020. The Alliance has provided you with this accepted manuscript in line with Alliance’s open access policy and in accordance with the Publisher’s policy on self-archiving. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. You may re-use or share this manuscript as long as you acknowledge the authors by citing the version of the record listed above. You may not change this manuscript in any way or use it commercially. For more information, please contact Alliance Bioversity-CIAT - Library Alliancebioversityciat-Library@cgiar.org *Manuscript Click here to view linked References 1 Pasture diversification affects soil macrofauna and soil 2 biophysical properties in tropical (silvo)pastoral systems 1,2,3* 1,2,4,* 4 2 3 Eduardo Vazquez , Nikola Teutscherova , Bohdan Lojka , Jacobo Arango , Mirjam 2,5 4 Pulleman 1 5 Departamento de Producción Agraria, Escuela Técnica Superior de Ingeniería Agronómica, 6 Alimentaria y de Biosistemas, Universidad Politécnica de Madrid, Madrid, Spain 2 7 International Center for Tropical Agriculture (CIAT), Palmira, Colombia 3 8 Department of Soil Biochemistry and Soil Ecology, University of Bayreuth, Bayreuth, 9 Germany. 4 10 Department of Crop Sciences and Agroforestry, Faculty of Tropical AgriSciences, Czech 11 University of Life Sciences Prague, Prague, Czech Republic 5 12 Soil Biology Group, Wageningen University, Wageningen, The Netherlands 13 14 * Corresponding authors: Eduardo.vazquez@uni-bayreuth.de (Eduardo Vazquez); 15 teutscherova@ftz.czu.cz (Nikola Teutscherova) 16 Abstract 17 The diversification of tropical pastures with legumes (trees) for increased forage and animal 18 productivity has been advocated. Nevertheless, effects on soil quality and belowground 19 biodiversity, and the implications for sustainable intensification remain poorly documented, 20 particularly when cattle grazing is included in the study. We evaluated the impact of forage 21 system diversification with herbaceous and woody legumes on soil properties and soil 22 macrofauna communities and their spatial heterogeneity in a three-year-old field trial in 23 Cauca Valley, Colombia. Three forage-based systems were compared: (i) a conventional 24 monoculture-species grass pasture system of Brachiaria hybrid cv. Cayman (CP); (ii) a mixed 25 pasture system consisting of Brachiaria grass with the leguminous herb Canavalia 26 brasiliensis (LP); and (iii) a silvopastoral system with rows of the legume tree Leucaena 27 diversifolia planted within LP pastures (SPS). The experiment was arranged in a complete 28 randomized block design with three replicates and grazing cattle rotating across blocks. Plots 29 were grazed by three (treatments CP and LP) or four bulls (SPS) aiming to reflect the 30 expected cattle intensification in SPS systems. Physico-chemical soil properties and 31 macrofauna abundance and their spatial heterogeneity as affected by the distance from the tree 32 rows in SPS, were assessed. Herbaceous legumes positively affected the abundance and 33 diversity of soil macrofauna and soil physical properties in LP and the alleys between tree 34 rows in SPS, as compared to CP. In the SPS, the highest soil quality and macrofauna 35 abundance occurred at the edge of the tree lines, while the highest soil compaction and the 36 lowest abundance of soil macrofauna occurred in the tree rows, probably due to the behavioral 37 change of the grazing cattle in combination with the higher stocking rate in SPS. Soil 38 properties in LP and in the alleys between the tree rows of SPS were comparable despite 39 higher stocking rate in SPS. Overall, the SPS and LP systems, proved to be suitable 40 alternatives to CP allowing for sustainable intensification of pastures although careful 41 evaluation of possible trade-offs associated with increased spatial heterogeneity in SPS is 42 recommended to avoid localized soil compaction. Soil macrofauna, particularly functional 43 groups (classified by feeding habits) proved to be a sensitive soil quality indicator in response 44 to contrasting pasture systems. 45 46 Key words: Silvopastoral systems; soil heterogeneity; soil structure; rotational grazing; 47 tropical forages. 48 1. Introduction 49 The conversion of tropical forest to extensive pastures for cattle grazing is one of the main 50 causes of deforestation, land degradation, greenhouse gas emissions (GHG), depletion of 51 carbon (C) stocks and reduction of biodiversity (Lerner et al., 2017; Martínez and Zinck, 52 2004; Murgueitio et al., 2011). More than 120 million ha in Latin America are covered by 53 grasslands grazed by cattle (De Oliveira et al., 2004). In Colombia, extensively managed -1 54 pastures with low cattle density (0.6 animal ha ) due to low forage production occupy 55 approximately one third of the country (Lerner et al., 2017). In order to increase the 56 productivity and reduce the negative environmental impacts of conventional mono-culture 57 pastures (CP), sustainable intensification strategies are required to enhance the input use 58 efficiencies and biomass productivity while simultaneously contributing to environmental 59 benefits (Lerner et al., 2017; Rao et al., 2015). Pasture diversification, particularly with 60 legumes, often increases forage and animal production and enhances soil quality by 61 improving soil chemical, physical and biological properties. Diverse vegetation leads to soil 62 organic matter (SOM) accumulation, increase of cation exchange capacity and soil 63 aggregation while providing ecosystem services, e.g. carbon sequestration, nutrient cycling, 64 and soil and water conservation (Barros et al., 2003; De Deyn et al., 2011; Hoosbeek et al., 65 2016; Xavier et al., 2014). The incorporation of woody perennials into pastures (creating 66 silvopastoral systems, SPS) is gaining increased interest, due to its great potential to reduce 67 soil erosion (Young, 1997), enhance biodiversity (Varah et al., 2013), reduce GHG emissions 68 (Landholm et al., 2019) and improve resource utilization efficiency due to niche 69 complementarity between trees and understory herbs. 70 The adoption of SPS can increase the spatial heterogeneity, observed as so-called “fertility 71 islands” around trees (Avendaño-Yáñez et al., 2018; Scholes and Archer, 1997; Van Miegroet 72 et al., 2000). Woody perennials increase the input of organic matter in the form of litter fall, 73 root exudation and root turnover below their canopies, promoting nutrient and SOM 74 accumulation and stimulation of microbial activity (Avendaño-Yáñez et al., 2018; Diedhiou et 75 al., 2009; Vallejo et al., 2012; Van Miegroet et al., 2000). In addition, the presence of trees 76 (and legume trees in particular) often leads to changes in the animal behavior due to (i) 77 feeding preferences for legume tree foliage (Murgueitio et al., 2011), or (ii) improved 78 microclimate under tree canopy (Broom et al., 2013; Murgueitio et al., 2011), which could be 79 of particular interest under tropical climate (Dubeux Jr. et al., 2017; Dubeux Jr et al., 2015; 80 Pezzopane et al., 2019). Such changes of grazing pattern can further contribute to spatial 81 variability of soil chemical and physical properties due to accumulation of nutrients from 82 dung and urine and soil compaction by cattle trampling under the canopy (Dahlin et al., 2005; 83 Paciullo et al., 2010; Taboada et al., 2011), thus, possibly hindering the positive impact of 84 enhanced organic matter in the near proximity to the trees. 85 The heterogeneity of habitat created by trees and grazing animals together with often 86 observed enhanced productivity of SPS often increase the abundance and diversity of soil 87 macrofauna, which play the key role in SOM decomposition, stabilization and release of 88 nutrients (Lavelle, 1997). Soil macrofauna are sensitive to changes in land use and 89 management (Lavelle et al., 2006) and inadequate agronomic practices, including 90 unsustainable intensification originating from increasing stocking rates in low-production 91 pastures, could have detrimental impact on macrofaunal communities. Owing to their 92 sensitivity to disturbance, soil macrofauna are increasingly used as bioindicator of soil quality 93 (Lavelle et al., 2006; Rousseau et al., 2014). Soil macrofauna, and ecosystem engineers in 94 particular (earthworms, ants and termites) together with soil microbes, promote the 95 aggregation of soil particles into biogenic aggregates (Bottinelli et al., 2015; Brussaard et al., 96 2007), which have been both successfully used as indicators of land-use change in tropical 97 pastures (Velásquez et al., 2012). These structures tend to be enriched in organic C and 98 nutrients (Van Groenigen et al., 2019) and are often more stable than the aggregates formed 99 by other mechanisms (Pulleman et al., 2005; Wolters, 2000). 100 Because of the extensive pasture degradation, particularly in tropical areas, the number of 101 studies focusing on sustainable intensification increases exponentially (Barros et al., 2003; 102 Laossi et al., 2008; Rousseau et al., 2014; Salton et al., 2014; Velásquez et al., 2012). 103 Nevertheless, the inclusion of grazing cattle is rarely included in the experiments (Webster et 104 al., 2019) and those that focus on impacts of forage diversification on cattle performance 105 rarely include more soil parameters than soil compaction or soil C storage (Paciullo et al., 106 2010; Varsha et al., 2019). Nevertheless, the cattle grazing preferences around the legume 107 trees could hinder the positive impact of tree incorporation to pastures respect to studies 108 without grazing, because grazing can lead to plant defoliation, degradation of soil physical 109 properties, alteration of nutrients and C cycling and affect the spatial distribution of soil 110 nutrients (Taboada et al., 2011). In addition, the impact of cattle can be even higher if the 111 stocking rate is adjusted to the higher biomass production of the SPS as it would happen on 112 farms. 113 In this study, we aimed to evaluate the effects of pasture diversification with legumes herbs 114 and/or trees on (i) soil biological, chemical and physical properties; and (ii) the heterogeneity 115 of silvopastoral system combining legume tree species (Leucaena diversifolia) with mix of 116 herbaceous grass (Brachiaria hybrid cv. Cayman) and herbaceous legume species (Canavalia 117 brasiliensis). Furthermore, We hypothesized that (i) the soil properties and macrofauna 118 distribution is heterogeneous in SPS with the highest macrofauna abundance found under the 119 tree canopy, that (ii) soil properties and macrofauna found in the allies of SPS are comparable 120 to LP (without trees) and different from CP, and that (iii) increased soil macrofauna diversity 121 correlates positively with improved soil physical properties and SOM accumulations. 122 2. Materials and methods 123 2.1. Study area 124 The study was conducted at the headquarters of the International Center for Tropical 125 Agriculture (CIAT) near Palmira (03°30’07”N 76°21’22”W), Cauca Valley, Colombia. The 126 altitude is 990 m a.s.l., and the area has a mean annual precipitation of 870 mm and a mean 127 annual temperature of 24°C. The precipitation, minimum, maximum and mean temperatures 128 during the experiment (January to April 2017) are shown in Fig. S-1. The soils of Cauca 129 Valley have been developed by pedogenesis of fluvial, eolic and lacustrine sediments, mostly 130 from Holocene period (Delgadillo-Vargas et al., 2016). The soil was characterized as a 131 Cumulic Haplustoll (Soil Taxonomy, USDA 2014) with a silty clay texture (50% clay in the 132 upper 20 cm soil layer). This area was originally covered with tropical dry forest. 133 Nevertheless, since the middle of the twentieth century, the most common agricultural use in 134 the area has been irrigated sugar cane monoculture (Delgadillo-Vargas et al., 2016). 135 2.2.Experimental layout 136 The experiment was established in 2013 to evaluate the effects of tropical forage 137 diversification through legume herbs and legume trees inclusion as compared to mono-crop 138 conventional pastures (CP) on pasture productivity and animal live weight gains. The 139 treatments studied here included (i) CP, consisting of Brachiaria (syn. Urochloa) hybrid cv. 140 Cayman (CIAT Br02-1752) monoculture; (ii) LP, as CP but mixed with the herbaceous 141 legume Canavalia brasiliensis (CIAT 17009), and (iii) silvopastoral system (SPS), as LP with 142 rows of Leucaena diversifolia (ILRI 15551) trees. The plots are laid out in a randomized 143 complete block design with three blocks. Each plot has an area of 0.3 ha (93 m x 36 m). L. 144 diversifolia was planted in double-rows (36 trees per row; 1 m distance between rows) with 145 eight such double-rows 10 m apart in each plot. This resulted in 576 trees per plot (1,720 tree -1 146 ha ). Approximately one fifth of L. diversifolia trees were allowed to grow naturally 147 (approximately 8 meters tall) to provide shade and C accumulation in aerial biomass and the 148 remaining trees were cut low to allow animals to browse on them. Animals could freely move 149 in the middle of the double rows as well as among alleys. 150 The plots were grazed by three (treatments CP and LP) or four bulls (SPS) that were rotated 151 across the three blocks. Each individual plot was further subdivided into 3 sub-sections (each 2 152 31 m long, 1.100 m ) with electrical fences. Each sub-section was grazed for 6 days with 48 153 days of rest (Fig 1). After grazing of all three sub-sections (18 days), the animals moved to 154 their corresponding treatment in the next block. 155 2.3.Sampling design 156 In line with the rotational grazing and given the strong effect of grazing on soil properties 157 under relatively stable climatic conditions, sampling was performed in each subsequent sub- 158 section one day before the initiation of grazing (corresponding to 48 days after the last grazing 159 event). Two transects were laid in each experimental plot and the distance from the sides of 160 each plot was selected by generating two random numbers for each plot. Along each transect 161 line in CP and LP, two samples were randomly collected, giving 4 samples from each plot 162 while each plot was replicated three times through the blocking. In SPS, samples were 163 collected at three fixed distances from the tree rows (Fig. 1): four samples at a distance of 5.5 164 m from the middle of the tree rows (SPS-5.5) which corresponded to the center of the alley; 165 four samples at 1.5 m from the middle of the tree rows (SPS-1.5) which corresponded with 1 166 m from the trees; and four samples between the tree double-rows (SPS-0.0). Therefore, 12 167 samples were collected from each of the three SPS plots. 168 2.4.Soil macrofauna 169 Soil macrofauna was sampled between February and April 2017 using the Tropical Soil 170 Biology and Fertility Institute (TSBF) method (Anderson and Ingram, 1993). One soil 171 monolith (25 x 25 x 10 cm) was collected at each sampling point from 0-10 cm and 10-20 cm 172 depth. Monoliths were immediately transported to the Laboratory of Soil Biology at CIAT at 173 the same location and all invertebrates were manually hand-sorted and stored in 70% ethanol 174 until identification. After the collection of all samples, soil macrofauna was identified and 175 classified into taxonomic groups: ants (Formicidae), earthworms (Oligochaeta), centipedes 176 (Chilopoda), millipedes (Diplopoda), beetles (Coleoptera), true bugs (Hemiptera), spiders 177 (Aranae), earwigs (Dermaptera), slug and snails (Gastropoda), woodlice (Isopoda), 178 cockroaches (Blattodea) and others. The abundance of each group was noted. Taxa from the 179 top soil layer were further classified according to their prevailing feeding habits: (i) 180 herbivorous (Dermaptera, Hemiptera); (ii) detritivorus (Blattodea, Diplopoda, Gastropoda, 181 Isopoda, Oligochaeta); and (iii) predators (Aranae, Chilopoda) similarly to (Cherubin et al., 182 2016). Nevertheless, due to the large functional diversity of Coleoptera, this order was further 183 classified to families, which were assigned to functional groups (Supplementary material 184 Table S-2). Similarly, we excluded Formicidae from all functional groups as high functional 185 variability prevented us to assign Formicidae into one particular group. 186 2.5.Soil aggregation 187 Samples for the classification of soil macroaggregates were collected within one meter from 188 soil macrofauna samples. Soil monoliths (10 x 10 x 10 cm) were excavated from the topsoil, 189 placed immediately into plastic boxes and transferred to the laboratory where they were stored 190 at 4°C until the manual inspection (within one week). Soil morphology was assessed visually 191 by the method described by Topoliantz et al. (2000) and modified by Velasquez et al. (2007). 192 Soil monoliths were gently broken down along natural planes of weakness on top of a 4 mm 193 sieve. Soil macroaggregates larger than 4 mm were separated into three groups according to 194 their origin: (i) biogenic macroaggregates (MAbio) – formed by soil macrofauna; (ii) 195 physicogenic macroaggregates (MAphys) – produced by physical processes, and (iii) 196 rhizosphere macroaggregates (MAroot) – produced by the action of plant roots and 197 rhizodeposition (Velasquez et al., 2007). The MAbio were formed by soil invertebrates, mainly 198 by the action of earthworms, termites, Coleopteran larvae and Diplopoda and were 199 characterized by rounded shapes and a dark color. Angular macroaggregates without clear 200 signs of biological activity were assigned as MAphys, while aggregates tightly bound to roots 201 were counted as MAroot. The root aggregates which were made by soil invertebrates were 202 classified as MAbio even when they adhered to plant roots. Aggregates between 2 and 4 mm 203 (MA2-4) and soil and aggregates passing the 2 mm sieve (MA0-2) are referred to as 204 unidentified soil fractions. The air-dried weight of all fractions was expressed as a percentage 205 of total soil dry weight. Soil macroaggregates were collected only in the upper soil layer (0-10 206 cm) where the highest activity of soil macrofauna was found. 207 Additional soil samples were collected on the side of each monolith for the analysis of water 208 stable aggregates. Large soil aggregates were broken along natural fractures, passed through a 209 8 mm sieve and air dried. After manual removal of plant material, the aggregates were 210 separated by wet-sieving through five sieves (6.3, 4.75, 2, 0.25 and 0.125 mm) using a Yoder 211 apparatus. A 60 g subsample of air-dried aggregates was slowly re-wetted (20 min) by 212 capillary water at room temperature on the top of the largest sieve. The set of sieves was then 213 being moved up and down (3.8 cm) for a period of 20 min (30 oscillations per minute), after 214 which the content of all sieves was transferred to a pan and dried at 105°C until constant 215 weight. In this way, six fractions were obtained. Mean weight diameter (MWD) was 216 calculated using the proportion by weight and the mean diameter of each size fraction (van 217 Bavel, 1950). For the calculations, the upper limit was set at 8 mm. 218 2.6.Soil chemical properties 219 Additional samples from 0-10 cm and 10-20 cm depth were taken next to each monolith to 220 determine soil chemical properties. Soil was air-dried in the laboratory, manually crushed and 221 passed through a 2 mm sieve. Soil pH was determined in 1:2.5 water extract. The SOM 222 content was determined by loss on ignition (540°C, 24 hours) and total N content (TN) by 223 Kjeldahl digestion and steam distillation (Bremner and Mulvaney, 1982). The fraction of 224 particulate organic matter (POM) was determined according to Cambardella and Elliott -1 225 (1992). Briefly, we dispersed 10 g of soil with 30 mL of 5 g l sodium hexametaphosphate by 226 shaking overnight. The dispersed samples were passed through a 53 µm-mesh sieve and the 227 retained fraction was collected and dried at 60ºC. Finally, the POM content was determined 228 by loss on ignition. The potentially mineralizable N (PMN) content was evaluated following 229 the procedure of Waring and Bremner (1964) by anaerobic soil incubation during seven days + + 230 at 37 °C. The initial NH4 -N content was substracted from the amount of NH4 -N at the end + 231 of the incubation period. The sample NH4 -N was extracted with 1M KCl (1:10) and was 232 determined colorimetrically (UV-1203, Shimadzu, Kyoto, Japan) using the sodium salicylate 233 method (Forster 1995). 234 Soil available nutrients were extracted by Mehlich III procedure and P was determined 235 colorimetrically (Murphy and Riley, 1962). In the same extracts, other soil macronutrients 236 (Ca, Mg, K, Na) were quantified by elemental analysis by atomic absorption 237 spectrophotometry (AAnalyst 400, PerkinElmer, Wellesley, MA). 238 2.7.Soil physical properties 239 At each sampling location, one undisturbed soil core was collected at 0-5 cm depth, using 240 bevel-edged steel rings of 5 cm in diameter. The bulk density was determined using the same 241 samples by dividing the oven-dry soil weight (105 °C) by volume of the steel ring. Total soil -3 242 porosity was calculated from the BD assuming a soil particle density of 2.65 g cm 243 (Danielson and Sutherland, 1986). 244 The resistance to penetration (PR) was determined with an electronic cone penetrometer 2 245 (Eijkelkamp, Giesbeek, The Netherlands) using a cone with 2 cm base area, 60º angle and 80 246 cm of driving shaft. Measurements were performed at four points around the place where the 247 soil macrofauna samples were collected. A total of 16 measurements were taken per plot as 248 deep as was permitted by soil conditions. For the data analysis, the mean resistance value was 249 calculated for the 0-10 cm and for the 10-20 cm layers. Soil samples were collected at same 250 moment of the penetrometer measurements to determine the gravimetric soil moisture 251 content. We observed high variation in soil moisture between the sampling points because of 252 the quick weather changes during the sampling season. 253 2.8. Statistical analysis 254 The effects of distance to the tree row in the 0-10 cm soil layer was analyzed in SPSS 22 255 (IBM SPSS, Inc., Chicago, USA) using a generalized linear mixed model (GLMM) in which 256 the fixed factor was the distance to the tree row and block was considered a random factor. In 257 case of significant effects (p<0.05), a LCD post-hoc test was performed. Similarly, the 258 differences between the open grazing pasture systems (CP, LP and SPS-5.5) were analyzed 259 using a GLMM with pasture system as fixed effect and block as random effect. The variables 260 from 10-20 cm soil depth were evaluated separately in the same way and results are presented 261 in Supplementary Material (Tables S-1 and S-2) due to weak treatment effects in deeper soil 262 layer and low abundance of macrofauna (less than 10% of total abundance). The GLMM was 263 selected because of the absence of the normality and homogeneity of macrofauna data. In 264 addition, considering the high number of zeros found in the soil macrofaunal data, we selected 265 a negative binomial distribution with a log link function as extension of Poisson distribution 266 for macrofauna variables (Kamau et al., 2017). The studied chemical and physical soil 267 properties were compared selecting a normal distribution after the evaluation of data 268 normality and transformation (log10 (X + 1)) when necessary. In all the cases, the distribution 269 used reached the lowest Akaike Information Criterion (AIC) 270 Principal component analysis (PCA) was performed using the data from de 0-10 cm soil depth 271 including soil variables, abundance of macrofauna functional groups (herbivores, detritivores 272 and predators), ants and the richness (observed number) of macrofaunal taxa. The variable 273 “monovalent” and “divalent” represents the sum of exchangeable monovalent (Na and K) and 274 divalent (Ca and Mg) cations, respectively. The first two components (PC1 and PC2) were 275 extracted through Varimax orthogonal extraction and all the used variables in PCA were 276 plotted in the orthogonal space defined by PC1 and PC2. Additionally, the treatments (CP, 277 LP, SPS-5.5, SPS-1.5 and SPS-0.0) were plotted in the orthogonal space defined by PC1 and 278 PC2. 279 3. Results 280 3.1. Effect of distance from the tree rows on soil properties in the silvopastoral system 281 No significant differences in SOM content were detected among the three sampling distances 282 within the SPS systems (Table 1). However, we observed decreasing content of POM and TN 283 with distance from the L. diversifolia tree rows, the highest content being detected in SPS-0.0 284 (Table 1). The Na content in SPS-0.0 was also significantly higher than in the other two 285 sampling distances, while Ca, Mg, K, pH and PMN were comparable across distances in the 286 SPS (Table 1). 287 The distance from the tree rows strongly affected physical soil properties. Soil BD was lower 288 in SPS-1.5 than in SPS-0.0, while MWD increased with increasing distance from the tree 289 double-row (Table 1). According to the macroaggregate morphology, in SPS-5.5 and SPS-1.5 290 the dominant type was MAbio (58 and 70% of bulk soil, respectively) while in SPS-0.0 the 291 MAbio and MAphys were similarly distributed (41 and 50% of bulk soil, respectively) (Table 292 1). The highest amount of MAphys was detected in the middle of the tree double-rows (500 g -1 293 kg ) and the highest of MAroot in-between the alleys (Table 1). The resistance to penetration 294 (RP) was not affected by the distance from the tree rows in the 0-10 cm layer, although in the 295 10-20 cm layer, the RP was higher in SPS-0.0 when compared to SPS-1.5 (Table 1, Table S1, 296 Fig. S2). 297 3.2.Effect of distance from the tree rows on soil macrofauna in SPS 298 The soil macrofauna was strongly affected by the distance from the tree double-rows (Table 299 2). Within SPS, the highest abundances of most macrofauna groups were consistently -2 300 observed at 1.5 m distance from the tree rows, where a total of 5,859 individuals m were 301 found in the 0-10 cm layer. Between the tree double-rows and in the center of the alley, the -2 302 total abundance of macrofauna was much lower (1,533 individuals m and 2,935 individuals -2 303 m , respectively. The most abundant groups in SPS-1.5 were Formicidae. Oligochaeta was 304 significantly higher at 1.5 m when compared to the other two distances, while Formicidae, 305 Isopoda, and the total macrofauna richness was higher at 1.5 m than at 0.0 m (Table 2). In 306 addition, the larvae abundance was higher at 1.5 m than at 5.5 m (Table 2). 307 The highest abundance of predators was observed in SPS-1.5, and the detritivores were also 308 more abundant at SPS-1.5 when compared to SPS-0.0 (Fig. 2). No significant differences 309 were found in the abundance of herbivores between distances from the trees. 310 3.3. Comparison of grass-herb systems and SPS alleys 311 No differences in soil chemical properties were detected between CP, LP and SPS-5.5 (alley). 312 Nevertheless, considerable differences were found between CP and LP in soil structure (Table 313 1). The MAbio and MAphys followed an opposite pattern, where MAbio was higher in LP than in 314 CP while MAphys was higher in CP (Fig. 2). The MWD was higher in LP than in CP while no 315 differences were found in resistance to penetration and bulk density. The SPS-5.5 (alley) did 316 not differ from none of the two treeless systems. 317 Higher densities of macrofauna individuals were collected at 0-10 cm in LP (2,511 -2 -2 318 individuals m ) than in the CP system (1,383 individuals m ) (Table 2). In comparison, -2 319 2,935 individuals m were collected in the alley of the silvopastoral system (SPS-5.5). Low 320 densities (lower than 10% of all collected individuals) of macrofauna were found at 10-20 cm 321 (supplementary material (Table S-2)). The total soil macrofauna richness was also enhanced 322 by legume inclusion . In addition, the density of Diplopoda and others (groups including taxa 323 not assigned to other groups due to very low abundance) in LC was significantly higher than 324 in CP, while the highest Dermaptera density was found in SPS-5.5 (Table 2). 325 From all functional groups, detritivores were the most abundant group in all systems (Fig. 2). 326 The highest herbivores density was found in SPS-5.5 and the lowest detritivores density in 327 CP. Finally, the predators density was higher in LC than in CP while the herbivores were 328 similar in the three systems. 329 3.4.Relationships among variables 330 The first two components in PCA accounted for a 38.6% of the total variance (Fig. 3). On the 331 PC1 (explaining 19.60% of total variance) SOM, TN, POM, PMN, and monovalent cations 332 (the sum of available Na and K) loaded on the positive side, while MWD loaded on negative 333 side (Table S-3). The macrofauna functional groups (detritivors, predators and herbivores), 334 richness and ants abundance loaded on the PC2 (19.04% of total variance). The aggregate 335 morphological fraction MAbio and BD loaded on the positive and negative side of PC2, 336 respectively. When the sampling points were plotted in the orthogonal space defined by PC1 337 and PC2, the highest differences were found between SPS-1.5, SPS-0.0 and CP. The samples 338 from SPS-1.5 were plotted on the right side of PC1 and the upper side of PC2 (high 339 macrofauna abundance) and positive side of the PC1 (high SOM content), while the SPS-0.0 340 samples were located on the positive side of PC1 but on the negative side of PC2. The CP 341 treatment was located in the negative side of PC1 and PC2 due to low macrofauna abundance, 342 low SOM and high BD. Finally, LP and SPS-5.5 were overlapping both located in the 343 negative side of PC1 and slightly positive side of the PC2. 344 345 4. Discussion 346 4.1.Differences in carrying capacity due to pasture diversification 347 Many studies have confirmed strong relationship between vegetation and soil biota (Bardgett 348 et al., 2005; Wardle et al., 2004), but causal effects of plant biomass and species composition 349 on soil macrofauna remain poorly understood. Although the quantification of plant biomass 350 was not the specific objective of the present study, enhanced plant productivity due to pasture 351 diversification and/or legume inclusion permitted higher stocking rates and higher animal 352 productivity in terms of meat production in the same experiment as described in previous 353 studies (Durango et al., 2017; Enciso et al., 2019). Since the establishment of the experiment, 354 the carrying capacity of pastures was adapted to pasture productivity and cattle stocking rates 355 were increased to four bulls in SPS (when compared to three bulls in CP and LP). Although 356 this increase of pasture carrying capacity could potentially hinder the positive effects of 357 pasture diversification and/or the presence of leguminous herbs (and hence, higher biomass 358 productivity) on soil properties, it resembles real situation on farm where stocking rates are 359 continuously being adjusted to pasture productivity. 360 4.2.Spatial heterogeneity of soil properties and macrofauna in the silvopastoral 361 system 362 Two main drivers of soil heterogeneity can be expected with the incorporation of legume trees 363 into pastures: (i) the increased input of resources from the litterfall, root exudation and 364 turnover forming “fertility islands” around trees (Avendaño-Yáñez et al., 2018; Scholes and 365 Archer, 1997; Van Miegroet et al., 2000); and (ii) alterations of animal behavior leading to 366 gradients in soil compaction and nutrient accumulation (Taboada et al., 2011), which were 367 confirmed by the differences in soil properties and biodiversity parameters within SPS in the 368 present study. 369 The so-called “fertility islands” with high soil C and nutrients contents around trees have been 370 observed and described by several authors (Avendaño-Yáñez et al., 2018; Van Miegroet et al., 371 2000). Beside its N-fixing capacity, L. diversifolia can influence nutrient availability also by 372 the transfer of nutrients from deeper zones (Rowe et al., 1998) to aboveground plant parts and 373 their deposition on soil surface followed by decomposition by soil macroinvertebrates and soil 374 microbes. Both sampling points located near the trees (SPS-1.5 and SPS-0.0) differed 375 considerably from other treatments and sampling points in several soil chemical properties as 376 observed in PCA. The higher richness of soil macrofauna taxa and the abundance of 377 detritivores, predators and herbivores near at SPS-1.5 translated into lower soil BD the 378 content of MAphys confirming the key role of soil macrofauna in soil structure formation via 379 bioturbation. 380 Furthermore, shade-providing trees can have direct impact on soil moisture content and soil 381 temperature, which are among the key abiotic variables affecting soil biological activity and, 382 therefore, the release of nutrients from SOM (Breshears, 2006). We observed decreasing 383 content of POM with increasing distance from the L. diversifolia tree rows, indicating a 384 positive influence of tree litter inputs on soil labile C content. The POM consists mainly of 385 partly decomposed organic tissues from above and belowground plant litter which has been 386 fragmented and partially decomposed by soil organisms (Lavallee et al., 2020). The POM 387 fraction has been found to be a valuable early indicator of soil fertility and land use changes 388 due to its low protection to decomposition and faster accumulation rates in comparison with 389 mineral-associated C and total SOM (Lajtha et al., 2014; Lavallee et al., 2020). The faster 390 accumulation rate of POM (when compared to SOM) likely explains the observed differences 391 in POM in early stages of the experiment but only a slight (not significant) trend of increasing 392 SOM. For this reason, SOM fractions such as POM or permanganate oxidizable C are more 393 useful indicators of pasture management effects on soil organic matter dynamics (Webster et 394 al., 2019). The increase of SOM content in the proximity of L. diversifolia tree rows could be 395 expected in the long-term as a result of high-quality litter from the legumes which may lead 396 to an accumulation of C in a mineral-associated form (Cotrufo et al., 2013). 397 In the SPS, soil properties and biomass production are also strongly affected by livestock 398 activity and movement (Taboada et al., 2011), which likely influenced soil BD and 399 biodiversity of macroinvertebrates. The introduction of palatable and protein-rich legume 400 trees in Brachiaria pastures may influence animal behavior due to the changes in diet 401 (preference of the tree fodder over grass) and due to the animal preference for shade (Broom 402 et al., 2013; Dubeux Jr. et al., 2017; Murgueitio et al., 2011) (see photo in Fig. S1). Lower 403 radiant thermal load in agroforestry systems compared to open grasslands, has been linked to 404 increased animal comfort (Pezzopane et al. (2019). If livestock spends more time in double 405 rows of L. diversifolia, higher input of animal urine and dung and accumulation of nutrients 406 can be expected in those areas (Dahlin et al.,2005). A change in behavior of cattle (see photo 407 in Fig. 1), together with the N-fixing capacity of legume tree, could explain the higher TN 408 content between the double-rows of L. diversifolia than in the middle of the alley and at 1.5 m 409 distance from the trees. The higher Na content between the tree double-rows also suggests a 410 concentration of animal deposition in the understory (Haynes and Willisms, 1992; Taboada et 411 al., 2011). Animal-derived nutrient depositions could increase the abundance of soil 412 burrowing macrofauna and lead to enhanced soil physical properties (Herrick and Lal, 1995). 413 Nevertheless, in the present study, we observed deterioration of soil physical properties under 414 the trees, probably due to cattle trampling causing soil compaction (Paciullo et al., 2010) 415 and/or reduction of grass cover (and grass fine roots) in the shaded area (visual observation) 416 due to light or water competition in the understory (Breshears, 2006; Podwojewski et al., 417 2014), which was further aggravated by higher cattle stocking rate in SPS. 418 Localized soil compaction was also confirmed by higher resistance to penetration in the 10-20 419 cm layer at SPS-0.0 and the observed frequent waterlogging after rainfall events indicating 420 poor water drainage (Fig. S3). It is plausible that the resistance to penetration was strongly 421 affected by soil moisture content, which differ between studied treatments and could thus 422 hinder the differences in soil compaction (see Fig. S2). The differences in soil resistance to 423 penetration should be evaluated by repetitive measurements to disentangle the influence of 424 soil compaction from changed in soil moisture. 425 Lower MWD of soil aggregates and higher proportion of MAphys was detected in the vicinity 426 to the tree-rows (SPS-0.0). Reduced MWD of water-stable aggregates could indicate the 427 disruption of larger aggregates during the drier months by trampling (Taboada et al., 2011; 428 Warren et al., 1986). Nevertheless, this reduction could be also linked to lower soil 429 macrofauna and root density of grasses between the tree-double rows (Fonte et al., 2012; 430 Podwojewski et al., 2014) and is likely related to the reduction of the abundance of MAbio 431 which tend to have higher stability than MAphys (Jouquet et al., 2009, 2008) as confirmed by 432 PCA in the present study. In addition, the PCA revealed a negative relation between 433 monovalent cations and MWD. The higher urine deposition in the double rows of L. 434 diversifolia likely increased the Na content which could promote a dispersion of soil 435 aggregates (Guo et al., 2019). Hence, the combined effect of dispersive monovalent cations 436 (Na), higher intensity of trampling and lower abundance of soil macrofauna and grass fine 437 roots, could have a stronger (negative) impact on soil aggregation when compared to (not 438 significant) effect of SOM in the proximity to the tree rows. Furthermore, the methodology of 439 MWD evaluation including the pre-analysis treatment can have a strong impact on the results. 440 Under tropical conditions where strong rainfall events are common, rapid rewetting of soil 441 aggregates could better mimic the slacking conditions likely occurring on the field when 442 compared to slow re-wetting of aggregates by capillarity. However, in soils with high 443 aggregate stability, such as the Vertisol under study, the different pre-treatments lead to a 444 comparable MWD (Le Bissonnais, 1996). We can thus consider our results reliable. 445 Soil macrofauna abundance (and diversity) was hypothesized to be higher close to the tree 446 lines due to higher organic matter input under the tree canopy. Nevertheless, the highest 447 abundance and richness was found at 1.5 m distance from the center of the tree lines. This is 448 likely explained by colonization of this area of both shade-thriving species (under the tree 449 canopy) and species more commonly found on the open grasslands (alley) which together 450 enhance the biodiversity as described by the “edge effects theory” (Harris, 1988). This theory 451 is also confirmed by the PCA analysis (Fig. 3) where SPS-1.5 m is positively defined by 452 macrofaunal taxa richness and abundance as well as by MAbio. The higher soil macrofauna in 453 SPS-1.5 than in SPS-0.0, (mainly ants in the present study) can explain the lower BD at 1.5 m 454 than at in the middle of the double-tree row. The spatial arrangements of trees within SPS can 455 thus influence the overall impact of land-use change on soil properties and soil macrofauna 456 diversity through effects on the behavior of grazing animals and resources diversity and 457 allocation. 458 4.3. Comparison of open pasture systems and SPS alleys 459 While no differences in soil chemical properties were observed among open pasture systems, 460 considerable differences in soil structure were found between pastures with legumes (LP and 461 SPS-5.5) and grass only pastures (CP) in soil structure. While soil under LP had higher MWD 462 of water stable aggregates and content of MAbio, CP soil constituted predominantly of MAphys, 463 indicating higher biological activity (and hence, formation of biostructures) in diversified 464 systems and negative impact of animal trampling on soil macrofauna. 465 High heterogeneity in tree-based systems makes comparison between land-use types 466 challenging. While trees have clearly the strongest impact on soil and understory vegetation in 467 the near vicinity, the open areas outside the tree canopies often cover proportionally larger 468 areas. The overlap in soil properties between SPS-5.5m (alley) and LP in the PCA indicate 469 that soil properties and macrofauna at 5.5 m distance from the center of the tree lines (alley) 470 are not affected by the presence of the trees in the SPS system, nor by the higher cattle 471 stocking rate. 472 Laossi et al. (2008) found a positive relation between biomass production of herbaceous 473 legumes and soil macrofauna abundance, which indicates the importance of this key plant 474 functional group. In the present study, the inclusion of legumes in open pasture systems 475 positively affected the taxonomic richness and total abundance of soil macrofauna and the 476 abundance of beetle larvae, all being lower in CP when compared to LP and SPS-5.5, even in 477 relatively young systems. The higher abundance of herbivorous macrofauna in legume- 478 containing pastures can be linked either with enhanced biomass production or with 479 diversification of substrate and/or habitat. Similarly, the increase in LP and SPS-5.5 of 480 detritivores (represented mainly by cockroaches, millipedes, woodlice and earthworms) which 481 participate in fragmentation of organic matter and nutrient recycling within the systems, are 482 likely linked with higher organic matter input and quality after legume inclusion (Velásquez 483 et al., 2012). 484 Despite generally low impact of forage diversification on soil macrofauna (Laossi et al., 2008; 485 Wardle et al., 2006), in the present study, soil macrofauna resulted to be a suitable predictor 486 of system productivity even at the early stage of development. Nevertheless, the effect of 487 forage diversification (legume inclusion) cannot be distinguished from effects of enhanced 488 biomass production (described by Enciso et al., 2019), which corresponds to legume-bearing 489 treatments. 490 4.4.Implications for sustainable intensification 491 Even though no significant differences were observed in the majority of soil chemical and 492 physical properties between SPS-5.5 (alleys) and CP or LP, increases in soil parameters and 493 biodiversity with decreasing distance from tree rows within SPS, suggest a positive effect of 494 tree inclusion at the system level. While increased biomass productivity and higher stocking 495 rates in SPS may underlay the idea of sustainable intensification, the prevention of possible 496 localized soil structure damage should remain the priority. This suggests a negative impact of 497 grazing intensification, a finding that could not have been detected in the studies without 498 cattle. The majority of studies considered the effects of pasture diversity on soil parameters 499 ignoring the grazing dimension, which can lead to biased conclusions on sustainable 500 intensification. To the best of our knowledge, our study is one of very few studies that 501 allowed to evaluate the effect of SPS adoption on soil properties including cattle grazing, and 502 especially when cattle grazing is adapted to the carrying capacity of the system (based on 503 available forage biomass) as modified by pasture system design resembling real farm 504 conditions. We recommend that in future studies of silvopastoral systems, like ours, more 505 detailed soil sampling schemes are used in order to grasp the high soil spatial variability, 506 including the areas more vulnerable to soil degradation. 507 5. Conclusions 508 This study confirms a clear positive impact of diversification of grazed pastures, via 509 incorporation of legume herbs and/or trees, on key soil properties and the abundance and 510 diversity of soil macroinvertebrates. The incorporation of legume trees grown in double-rows 511 resulted in a spatial heterogeneity of SPS with high fertility zones on the edge of tree canopy 512 (1.5 m from the trees, distinguished by enhanced SOM and nutrient content) and zones with 513 elevated compaction risk (between the double-rows, characterized by high content of MAphys 514 and high BD). The taxonomic abundance and functional diversity of soil macrofauna were 515 found to be responsive to both (i) legume herbs/trees inclusion (particularly under the edge of 516 the tree canopy) and (ii) cattle trampling causing localized soil compaction within the 517 silvopastoral system. 518 Our results clearly indicate the great opportunities for improvement of soil properties and soil 519 biodiversity in grass-legume systems and in silvopatoral systems in particular, while allowing 520 for increased cattle stocking rates under SPS thus providing a promising strategy for 521 sustainable intensification of pastoral systems. Nevertheless, the increased stocking rates and 522 changing grazing patterns in SPS need to be carefully accounted for to avoid possible 523 negative impacts on soil quality and other effects such as GHG emissions. Therefore, trade- 524 offs between positive (biodiversity, soil properties etc.) and negative effects (possible soil 525 compaction resulting from locally high stocking rates) need to be carefully evaluated. 526 Acknowledgements 527 Special thanks belong to Yamileth Chagüeza and Enna Díaz at CIAT for the assistance 528 with soil macrofauna extraction. We also thank Yamileth Chagüeza for beetle identification to 529 family level. The authors are also grateful to Mauricio Sotelo for the establishment and 530 maintenance of the trial and assistance during the experiment and to César Botero for the 531 assistance with soil physical properties analysis, and Joana Frazão for advice on the statistical 532 approach. This work was implemented as part of the CGIAR Research Program (CRP) on 533 Climate Change, Agriculture and Food Security (CCAFS), and the Livestock CRP which are 534 carried out with support from CGIAR Fund Donors and through bilateral funding agreements. 535 For details, please visit https://ccafs.cgiar.org/donors. We also acknowledge the financial 536 assistance of BBSRC grant Advancing sustainable forage-based livestock production systems 537 in Colombia (CoForLife) (BB/S01893X/1) and GROW Colombia from the UK Research and 538 Innovation (UKRI) Global Challenges Research Fund (GCRF) (BB/P028098/1). Eduardo 539 Vázquez thanks the Spanish Ministry of Education for his FPU fellowship. Nikola 540 Teutscherova thanks Cátedra Rafael Dal-Re/TRAGSA. Financial support was also obtained 541 from the Internal Grant Agency of Czech University of Life Sciences Prague (no. 20185004, 542 and no. 20195005). 543 544 References 545 Anderson, J.M., Ingram, J.S.I., 1993. Tropical Soil Biology and Fertility: A handbook of 546 methods. Trop. Soil Biol. Fertil. A Handb. methods 2 Ed., 88–91. 547 https://doi.org/10.1017/S0014479700018354 548 Avendaño-Yáñez, M. de la L., López-Ortiz, S., Perroni, Y., Pérez-Elizalde, S., 2018. 549 Leguminous trees from tropical dry forest generate fertility islands in pastures. Arid L. 550 Res. Manag. https://doi.org/10.1080/15324982.2017.1377782 551 Bardgett, R.D., Bowman, W.D., Kaufmann, R., Schmidt, S.K., 2005. A temporal approach to 552 linking aboveground and belowground ecology. Trends Ecol. Evol. 553 https://doi.org/10.1016/j.tree.2005.08.005 554 Barros, E., Neves, A., Blanchart, E., Fernandes, E.C.M., Wandelli, E., Lavelle, P., 2003. 555 Development of the soil macrofauna community under silvopastoral and agrosilvicultural 556 systems in Amazonia. Pedobiologia (Jena). 47, 273–280. https://doi.org/10.1078/0031- 557 4056-00190 558 Bottinelli, N., Jouquet, P., Capowiez, Y., Podwojewski, P., Grimaldi, M., Peng, X., 2015. 559 Why is the influence of soil macrofauna on soil structure only considered by soil 560 ecologists? Soil Tillage Res. https://doi.org/10.1016/j.still.2014.01.007 561 Bremner, J.M., Mulvaney, C.S., 1982. Nitrogen-Total, in: Page, A.L., Miller, R, H., D.R., K. 562 (Eds.), Methods of Soil Analysis. Part 2. Chemical and Microbiological Properties 563 (Agronomy Series n 9) ASA. SSSA Madison, Wisconsin, pp. 595–624. 564 Breshears, D.D., 2006. The grassland-forest continuum: Trends in ecosystem properties for 565 woody plant mosaics? Front. Ecol. Environ. https://doi.org/10.1890/1540- 566 9295(2006)004[0096:TGCTIE]2.0.CO;2 567 Broom, D.M., Galindo, F.A., Murgueitio, E., 2013. Sustainable, efficient livestock production 568 with high biodiversity and good welfare for animals. Proc. R. Soc. B Biol. Sci. 280, 569 20132025–20132025. https://doi.org/10.1098/rspb.2013.2025 570 Brussaard, L., Pulleman, M.M., Ouédraogo, É., Mando, A., Six, J., 2007. Soil fauna and soil 571 function in the fabric of the food web. Pedobiologia (Jena). 50, 447–462. 572 https://doi.org/10.1016/j.pedobi.2006.10.007 573 Cambardella, C.A., Elliott, E.T., 1992. Particulate Soil Organic-Matter Changes across a 574 Grassland Cultivation Sequence. Soil Sci. Soc. Am. J. 56, 777. 575 https://doi.org/10.2136/sssaj1992.03615995005600030017x 576 Cherubin, M.R., Franco, A.L.C., Cerri, C.E.P., Karlen, D.L., Pavinato, P.S., Rodrigues, M., 577 Davies, C.A., Cerri, C.C., 2016. Phosphorus pools responses to land-use change for 578 sugarcane expansion in weathered Brazilian soils. Geoderma. 579 https://doi.org/10.1016/j.geoderma.2015.11.017 580 Cotrufo, M.F., Wallenstein, M.D., Boot, C.M., Denef, K., Paul, E., 2013. The Microbial 581 Efficiency-Matrix Stabilization (MEMS) framework integrates plant litter decomposition 582 with soil organic matter stabilization: Do labile plant inputs form stable soil organic 583 matter? Glob. Chang. Biol. https://doi.org/10.1111/gcb.12113 584 Dahlin, A.S., Emanuelsson, U., McAdam, J.H., 2005. Nutrient management in low input 585 grazing-based systems of meat production. Soil Use Manag. 586 https://doi.org/10.1079/SUM2005299 587 Danielson, R.E., Sutherland, P.L., 1986. Porosity, in: Klute, A. (Ed.), Methods of Soil 588 Analysis: Part 1—Physical and Mineralogical Methods. Soil Science Society of 589 America, American Society of Agronomy, pp. 443–461. 590 https://doi.org/10.2136/sssabookser5.1.2ed.c18 591 De Deyn, G.B., Shiel, R.S., Ostle, N.J., Mcnamara, N.P., Oakley, S., Young, I., Freeman, C., 592 Fenner, N., Quirk, H., Bardgett, R.D., 2011. Additional carbon sequestration benefits of 593 grassland diversity restoration. J. Appl. Ecol. https://doi.org/10.1111/j.1365- 594 2664.2010.01925.x 595 De Oliveira, O.C., De Oliveira, I.P., Alves, B.J.R., Urquiaga, S., Boddey, R.M., 2004. 596 Chemical and biological indicators of decline/degradation of Brachiaria pastures in the 597 Brazilian Cerrado. Agric. Ecosyst. Environ. https://doi.org/10.1016/j.agee.2003.12.004 598 Delgadillo-Vargas, O., Garcia-Ruiz, R., Forero-Álvarez, J., 2016. Fertilising techniques and 599 nutrient balances in the agriculture industrialization transition: The case of sugarcane in 600 the Cauca river valley (Colombia), 1943-2010. Agric. Ecosyst. Environ. 601 https://doi.org/10.1016/j.agee.2015.11.003 602 Diedhiou, S., Dossa, E.L., Badiane, A.N., Diedhiou, I., Sène, M., Dick, R.P., 2009. 603 Decomposition and spatial microbial heterogeneity associated with native shrubs in soils 604 of agroecosystems in semi-arid Senegal. Pedobiologia (Jena). 605 https://doi.org/10.1016/j.pedobi.2008.11.002 606 Dubeux Jr., J.C.B., Muir, J.P., Apolinário, V.X. de O., Nair, P.K.R., Lira, M. de A., 607 Sollenberger, L.E., 2017. Tree legumes: An underexploited resource in warm-climate 608 silvopastures. Rev. Bras. Zootec. https://doi.org/10.1590/S1806-92902017000800010 609 Dubeux Jr, J.C., Muir, J.P., Nair, P.R., Sollenberger, L.E., Silva, H.M., de Mello, A.C., 2015. 610 The advantages and challenges of integrating tree legumes into pastoral systems. Forages 611 Warm Clim. 612 Durango, S., Gaviria, X., González, R., Sotelo Cabrera, M.E., Gutierrez, J.F., Chirinda, N., 613 Arango, J., Barahona Rosales, R., 2017. Climate change mitigation initiatives in beef 614 production systems in tropical countries. 615 Enciso, K., Sotelo, M., Peters, M., Burkart, S., 2019. The inclusion of Leucaena diversifolia 616 in a Colombian beef cattle production system: An economic perspective. Trop. 617 Grasslands-Forrajes Trop. https://doi.org/10.17138/tgft(7)359-369 618 Fonte, S.J., Quintero, D.C., Velásquez, E., Lavelle, P., 2012. Interactive effects of plants and 619 earthworms on the physical stabilization of soil organic matter in aggregates. Plant Soil 620 359, 205–214. https://doi.org/10.1007/s11104-012-1199-2 621 Forster JC, 1995. Soil Nitrogen, in: Alef K, N.P. (Ed.), Methods in Applied Soil Microbiology 622 and Biochemistry. pp. 79–87. 623 Guo, Z., Zhang, J., Fan, J., Yang, X., Yi, Y., Han, X., Wang, D., Zhu, P., Peng, X., 2019. 624 Does animal manure application improve soil aggregation? Insights from nine long-term 625 fertilization experiments. Sci. Total Environ. 626 https://doi.org/10.1016/j.scitotenv.2019.01.051 627 Harris, L.D., 1988. Edge effects and conservation of biotic diversity. Conserv. Biol. 628 https://doi.org/10.1111/j.1523-1739.1988.tb00196.x 629 Haynes, R.J., Willisms, P.H., 1992. Changes in soil solution composition and pH in urine- 630 affected areas of pasture. J. Soil Sci. 43, 323–334. https://doi.org/10.1111/j.1365- 631 2389.1992.tb00140.x 632 Herrick, J.E., Lal, R., 1995. Soil Physical Property Changes during Dung Decomposition in a 633 Tropical Pasture. Soil Sci. Soc. Am. J. 634 https://doi.org/10.2136/sssaj1995.03615995005900030040x 635 Hoosbeek, M.R., Remme, R.P., Rusch, G.M., 2016. Trees enhance soil carbon sequestration 636 and nutrient cycling in a silvopastoral system in south-western Nicaragua. Agrofor. Syst. 637 1–11. https://doi.org/10.1007/s10457-016-0049-2 638 Jouquet, P., Bottinelli, N., Podwojewski, P., Hallaire, V., Tran Duc, T., 2008. Chemical and 639 physical properties of earthworm casts as compared to bulk soil under a range of 640 different land-use systems in Vietnam. Geoderma. 641 https://doi.org/10.1016/j.geoderma.2008.05.030 642 Jouquet, P., Zangerle, A., Rumpel, C., Brunet, D., Bottinelli, N., Tran Duc, T., 2009. 643 Relevance and limitations of biogenic and physicogenic classification: A comparison of 644 approaches for differentiating the origin of soil aggregates. Eur. J. Soil Sci. 645 https://doi.org/10.1111/j.1365-2389.2009.01168.x 646 Kamau, S., Barrios, E., Karanja, N.K., Ayuke, F.O., Lehmann, J., 2017. Soil macrofauna 647 abundance under dominant tree species increases along a soil degradation gradient. Soil 648 Biol. Biochem. 112, 35–46. https://doi.org/10.1016/j.soilbio.2017.04.016 649 Lajtha, K., Townsend, K.L., Kramer, M.G., Swanston, C., Bowden, R.D., Nadelhoffer, K., 650 2014. Changes to particulate versus mineral-associated soil carbon after 50 years of litter 651 manipulation in forest and prairie experimental ecosystems. Biogeochemistry. 652 https://doi.org/10.1007/s10533-014-9970-5 653 Landholm, D.M., Pradhan, P., Wegmann, P., Sánchez, M.A.R., Salazar, J.C.S., Kropp, J.P., 654 2019. Reducing deforestation and improving livestock productivity: greenhouse gas 655 mitigation potential of silvopastoral systems in Caquetá. Environ. Res. Lett. 656 https://doi.org/10.1088/1748-9326/ab3db6 657 Laossi, K., Barot, S., Carvalho, D., Desjardins, T., Lavelle, P., Martins, M., Mitja, D., 658 Rendeiro, A.C., Rousseau, G., Sarrazin, M., Velasquez, E., Grimaldi, M., 2008. Effects 659 of plant diversity on plant biomass production and soil macrofauna in Amazonian 660 pastures. Pedobiologia (Jena). 51, 397–407. https://doi.org/10.1016/j.pedobi.2007.11.001 661 Lavallee, J.M., Soong, J.L., Cotrufo, M.F., 2020. Conceptualizing soil organic matter into 662 particulate and mineral-associated forms to address global change in the 21st century. 663 Glob. Chang. Biol. https://doi.org/10.1111/gcb.14859 664 Lavelle, P., 1997. Faunal Activities and Soil Processes: Adaptive Strategies That Determine 665 Ecosystem Function. Adv. Ecol. Res. 27, 93–132. https://doi.org/10.1016/S0065- 666 2504(08)60007-0 667 Lavelle, P., Decaëns, T., Aubert, M., Barot, S., Blouin, M., Bureau, F., Margerie, P., Mora, P., 668 Rossi, J.-P., 2006. Soil Invertebrates and Ecosystem Services. Eur. J. Soil Biol. 42. 669 https://doi.org/10.1016/j.ejsobi.2006.10.002 670 Le Bissonnais, Y., 1996. Aggregate stability and assessment of soil crustability and 671 erodibility: I. Theory and methodology. Eur. J. Soil Sci. 47, 425–437. 672 https://doi.org/10.1111/ejss.4_12311 673 Lerner, A.M., Zuluaga, A.F., Chará, J., Etter, A., Searchinger, T., 2017. Sustainable Cattle 674 Ranching in Practice: Moving from Theory to Planning in Colombia’s Livestock Sector. 675 Environ. Manage. 60, 176–184. https://doi.org/10.1007/s00267-017-0902-8 676 Martínez, L.J., Zinck, J.A., 2004. Temporal variation of soil compaction and deterioration of 677 soil quality in pasture areas of Colombian Amazonia. Soil Tillage Res. 678 https://doi.org/10.1016/j.still.2002.12.001 679 Murgueitio, E., Calle, Z., Uribe, F., Calle, A., Solorio, B., 2011. Native trees and shrubs for 680 the productive rehabilitation of tropical cattle ranching lands. For. Ecol. Manage. 261, 681 1654–1663. https://doi.org/10.1016/j.foreco.2010.09.027 682 Murphy, J., Riley, J., 1962. A modified single solution method for the determinatio of 683 phosphate in natural waters. Anal. Chem. ACTA 27, 31–36. 684 https://doi.org/10.1016/S0003-2670(00)88444-5 685 Paciullo, D.S.C., Castro, C.R.T. de, Gomide, C.A. de M., Fernandes, P.B., Rocha, W.S.D. da, 686 Müller, M.D., Rossiello, R.O.P., 2010. Soil bulk density and biomass partitioning of 687 Brachiaria decumbens in a silvopastoral system. Sci. Agric. 688 https://doi.org/10.1590/s0103-90162010000500014 689 Pezzopane, J.R.M., Nicodemo, M.L.F., Bosi, C., Garcia, A.R., Lulu, J., 2019. Animal thermal 690 comfort indexes in silvopastoral systems with different tree arrangements. J. Therm. 691 Biol. 79, 103–111. https://doi.org/10.1016/J.JTHERBIO.2018.12.015 692 Podwojewski, P., Grellier, S., Mthimkhulu, S., Titshall, L., 2014. How Tree Encroachment 693 and Soil Properties Affect Soil Aggregate Stability in an Eroded Grassland in South 694 Africa. Soil Sci. Soc. Am. J. https://doi.org/10.2136/sssaj2013.12.0511 695 Pulleman, M.., Six, J., Uyl, A., Marinissen, J.C.Y., Jongmans, A.G., 2005. Earthworms and 696 management affect organic matter incorporation and microaggregate formation in 697 agricultural soils. Appl. Soil Ecol. 29, 1–15. https://doi.org/10.1016/j.apsoil.2004.10.003 698 Rao, I., Peters, M., Castro, A., Schultze-Kraft, R., White, D., Fisher, M., Miles, J., Lascano, 699 C., Blümmel, M., Bungenstab, D., Tapasco, J., Hyman, G., Bolliger, A., Paul, B., Van 700 Der Hoek, R., Maass, B., Tiemann, T., Cuchillo, M., Douxchamps, S., Villanueva, C., 701 Rincón, A., Ayarza, M., Rosenstock, T., Subbarao, G., Arango, J., Cardoso, J., 702 Worthington, M., Chirinda, N., Notenbaert, A., Jenet, A., Schmidt, A., Vivas, N., Lefroy, 703 R., Fahrney, K., Guimarães, E., Tohme, J., Cook, S., Herrero, M., Chacón, M., 704 Searchinger, T., Rudel, T., 2015. LivestockPlus - The sustainable intensification of 705 forage-based agricultural systems to improve livelihoods and ecosystem services in the 706 tropics. Trop. Grasslands-Forrajes Trop. https://doi.org/10.17138/TGFT(3)59-82 707 Rousseau, L., Fonte, S.J., Tellez, O., van der Hoek, R., Lavelle, P., 2014. Soil macrofauna as 708 indicators of soil quality and land use impacts in smallholder agroecosystems of ... Ecol. 709 Indic. 27, 71–82. https://doi.org/10.1016/j.ecolind.2012.11.020 710 Rowe, E.C., Hairiah, K., Giller, K.E., Van Noordwijk, M., Cadisch, G., 1998. Testing the 711 safety-net role of hedgerow tree roots by N-15 placement at different soil depths. 712 Agrofor. Syst. https://doi.org/10.1023/A:1022123020738 713 Salton, J.C., Mercante, F.M., Tomazi, M., Zanatta, J.A., Concenço, G., Silva, W.M., Retore, 714 M., 2014. Integrated crop-livestock system in tropical Brazil: Toward a sustainable 715 production system. Agric. Ecosyst. Environ. https://doi.org/10.1016/j.agee.2013.09.023 716 Scholes, R.J., Archer, S.R., 1997. Tree-grass interactions in Savannas. Annu. Rev. Ecol. Syst. 717 https://doi.org/10.1146/annurev.ecolsys.28.1.517 718 Taboada, M.A., Rubio, G., Chaneton, E.J., Hatfield, J.L., Sauer, T.J., 2011. Grazing Impacts 719 on Soil Physical, Chemical, and Ecological Properties in Forage Production Systems, in: 720 Soil Management: Building a Stable Base for Agriculture. Soil Science Society of 721 America, pp. 301–320. https://doi.org/10.2136/2011.soilmanagement.c20 722 Topoliantz, S., Ponge, J.F., Viaux, P., 2000. Earthworm and enchytraeid activity under 723 different arable farming systems, as exemplified by biogenic structures. Plant Soil 225, 724 39–51. https://doi.org/10.1023/A:1026537632468 725 Vallejo, V.E., Arbeli, Z., Terán, W., Lorenz, N., Dick, R.P., Roldan, F., 2012. Effect of land 726 management and Prosopis juliflora (Sw.) DC trees on soil microbial community and 727 enzymatic activities in intensive silvopastoral systems of Colombia. Agric. Ecosyst. 728 Environ. 150, 139–148. https://doi.org/10.1016/j.agee.2012.01.022 729 van Bavel, C.H.M., 1950. Mean Weight-Diameter of Soil Aggregates as a Statistical Index of 730 Aggregation1. Soil Sci. Soc. Am. J. 731 https://doi.org/10.2136/sssaj1950.036159950014000c0005x 732 Van Groenigen, J.W., Van Groenigen, K.J., Koopmans, G.F., Stokkermans, L., Vos, H.M.J., 733 Lubbers, I.M., 2019. How fertile are earthworm casts? A meta-analysis. Geoderma 338, 734 525–535. https://doi.org/10.1016/J.GEODERMA.2018.11.001 735 Van Miegroet, H., Hysell, M.T., Johnson, A.D., 2000. Soil microclimate and chemistry of 736 spruce-fir tree islands in Northern Utah. Soil Sci. Soc. Am. J. 737 Varah, A., Jones, H., Smith, J., Potts, S.G., 2013. Enhanced biodiversity and pollination in 738 UK agroforestry systems. J. Sci. Food Agric. https://doi.org/10.1002/jsfa.6148 739 Varsha, K.M., Raj, A.K., Kurien, E.K., Bastin, B., Kunhamu, T.K., Pradeep, K.P., 2019. High 740 density silvopasture systems for quality forage production and carbon sequestration in 741 humid tropics of Southern India. Agrofor. Syst. https://doi.org/10.1007/s10457-016- 742 0059-0 743 Velásquez, E., Fonte, S.J., Barot, S., Grimaldi, M., Desjardins, T., Lavelle, P., 2012. Soil 744 macrofauna-mediated impacts of plant species composition on soil functioning in 745 Amazonian pastures. Appl. Soil Ecol. 56, 43–50. 746 https://doi.org/10.1016/j.apsoil.2012.01.008 747 Velasquez, E., Pelosi, C., Brunet, D., Grimaldi, M., Martins, M., Rendeiro, A.C., Barrios, E., 748 Lavelle, P., 2007. This ped is my ped: Visual separation and near infrared spectra allow 749 determination of the origins of soil macroaggregates. Pedobiologia (Jena). 51, 75–87. 750 https://doi.org/10.1016/j.pedobi.2007.01.002 751 Wardle, D.A., Bardgett, R.D., Klironomos, J.N., Setälä, H., van der Putten, W.H., Wall, D.H., 752 2004. Ecological linkages between aboveground and belowground biota. Science 304, 753 1629–33. https://doi.org/10.1126/science.1094875 754 Wardle, D.A., Yeates, G.W., Barker, G.M., Bonner, K.I., 2006. The influence of plant litter 755 diversity on decomposer abundance and diversity. Soil Biol. Biochem. 756 https://doi.org/10.1016/j.soilbio.2005.09.003 757 Waring, S.A., Bremner, J.M., 1964. Ammonium Production in Soil under Waterlogged 758 Conditions as an Index of Nitrogen Availability. Nature 201, 951–952. 759 https://doi.org/10.1038/201951a0 760 Warren, S.D., Thurow, T.L., Blackburn, W.H., Garza, N.E., 1986. The Influence of Livestock 761 Trampling under Intensive Rotation Grazing on Soil Hydrologic Characteristics. J. 762 Range Manag. https://doi.org/10.2307/3898755 763 Webster, E., Gaudin, A.C.M., Pulleman, M., Siles, P., Fonte, S.J., 2019. Improved Pastures 764 Support Early Indicators of Soil Restoration in Low-input Agroecosystems of Nicaragua. 765 Environ. Manage. https://doi.org/10.1007/s00267-019-01181-8 766 Wolters, V., 2000. Invertebrate control of soil organic matter stability 1–19. 767 Xavier, D.F., da Silva Lédo, F.J., de Campos Paciullo, D.S., Urquiaga, S., Alves, B.J.R., 768 Boddey, R.M., 2014. Nitrogen cycling in a Brachiaria-based silvopastoral system in the 769 Atlantic forest region of Minas Gerais, Brazil. Nutr. Cycl. Agroecosystems 99, 45–62. 770 https://doi.org/10.1007/s10705-014-9617-x 771 Young, A., 1997. Agroforestry for soil management. Second edition, Agroforestry for soil 772 management. Second edition. 773 Tables Click here to download Tables: _Table 1.docx Table 1 Soil properties of the three different forage-based systems and distances from the tree rows in the silvopastoral system (0-10 cm soil layer). Mean values followed by standard errors in parenthesis (n=12) 1 Pasture system CP LP SPS-5.5 SPS-1.5 SPS-0.0 Soil variable -1 SOM (g kg ) 45.8 (0.92) 46.9 (0.56) 47.3 (0.65) 49.4 (1.36) 48.7 (0.95) -1 POM (g kg ) 5.52 (0.32) 6.16 (0.23) 5.91 (0.36) B 6.79 (0.65) AB 7.23 (0.59) A -1 TN (g kg ) 1.16 (0.06) 1.21 (0.04) 1.17 (0.04) B 1.22 (0.06) B 1.37 (0.06) A -1 PMN (mg N kg ) 29.1 (1.71) 32.8 (2.46) 35.2 (2.82) 39.5 (3.27) 34.1 (3.73) -1 P (mg kg ) 21.76 (4.49) 24.29 (6.41) 18.06 (3.97) 21.87 (3.45) 15.82 (3.02) -1 Ca (mg kg ) 2,268 (39) 2,331 (54) 2,237 (37) 2,262 (46) 2,247 (49) -1 Mg (mg kg ) 859 (8.6) 888 (25.2) 869 (8.3) 871 (8.4) 871 (10.7) -1 K (mg kg ) 626 (130) 767 (159.7) 864 (236.4) 888 (134.6) 947 (134.8) -1 Na (mg kg ) 45.9 (3.52) 46.2 (2.18) 46.7 (2.35) B 41.6 (2.02) B 54.00 (1.60) A pH 7.40 (0.07) 7.51 (0.10) 7.61 (0.08) 7.49 (0.08) 7.47 (0.09) -3 BD (g cm ) 1.55 (0.02) 1.52 (0.02) 1.51 (0.02) AB 1.46 (0.02) B 1.52 (0.02) A PR (kPa) 849 (69.0) 831 (49) 877 (58) 863 (81) 917 (87) MWD (mm) 5.56 (0.18) b 5.99 (0.09) a 5.91(0.15) abA 5.30 (0.18) B 4.19 (0.26) C -1 MAbio (g kg ) 479 (42.0) b 654 (38) a 577 (34) abB 698 (28) A 405 (52) C -1 MAphys (g kg ) 349 (48.0) a 183 (48) b 259 (46) abB 178 (28) B 500 (63) A -1 MAroot (g kg ) 87 (27.0) 62 (17) 79 (20) A 21 (4) B 5 (2) C SOM, soil organic matter; POM, particulate organic matter; TN, total nitrogen; PMN, potentially mineralizable nitrogen; BD, bulk density; PR, penetrometer resistance; MWD, mean weight diameter of water-table aggregates; MAbio, biogenic macroaggregates; MAphys, physicogenic macroaggregates; MAroot, root-associated macroaggregates 1 pasture type: CP, conventional pasture of Brachiaria hybrid cv. Cayman monoculture; LP, Brachiaria pasture improved by incorporation of Canavalia brasiliensis; SPS, silvopastoral system with Brachiaria, Canavalia brasiliensis and Leucaena diversifolia Different lowercase letters indicate significant difference (p<0.05) between CP, LP and SPS-5.5. Different upper case letters indicate difference (p<0.05) between the distances from L. diversifolia tree row within SPS. Tables Click here to download Tables: _Table 2.docx Table 2 Soil macrofauna density in three different pastures systems and distances from the tree rows in the silvopastoral system (0-10 cm soil layer). Mean values followed by standard errors (n-=12) 1 Pasture system CP LP SPS-5.5 SPS-1.5 SPS-0.0 Macrofauna group -2 individuals m Formicidae (ants) 1,028 (627) 1,707 (629) 2,115 (805) AB 4,240 (1194) A 741 (272) B Oligochaeta (earthworms) 65 (30) 60 (19) 117 (40) B 277 (132) A 87 (28) B Chilopoda (centipedes) 39 (19) 72 (25) 49 (15) 160 (45) 32 (14) Diplopoda (millipedes) 60 (18) b 137 (36) a 125 (43) ab 331 (104) 157 (102) Coleoptera (beetles) 43 (15) 92 (22) 117 (38) 181 (34) 155 (21) Hemiptera (true bugs) 5 (3) 16 (10) 8 (5) 21 (6) 7 (3) Aranea (spiders) 11 (5) 61 (18) 25 (9) 53 (19) 15 (5) Dermaptera (earwigs) 21 (9) b 55 (15) b 125 (30) a 155 (36) 72 (17) Gastropoda (molluscs) 9 (6) 16 (11) 4 (3) 16 (11) 5 (5) Isopoda (woodlice) 31 (12) 69 (20) 59 (18) AB 108 (30) A 17 (5) B Blattodea (cockroaches) 3 (2) 11 (5) 23 (7) 21 (6) 12 (7) Others 7 (4) b 112 (70) a 40 (14) ab 55 (14) 17 (7) Larvae 61 (14) b 103 (30) a 127 (2) a B 240 (50) A 216 (39) AB Total abundance 1,383 (657) b 2,511 (685) a 2,935 (834) aAB 5,859 (1308) A 1,533 (137) B Richness (S) 7.0 (0.4) b 8.8 (0.7) a 9.4 (0.6) aAB 10.8(0.2) A 8.7 (0.5) B 1 pasture type: CP, conventional pasture of Brachiaria cv. Cayman monoculture; LP, Brachiaria pasture improved by incorporation of Canavalia brasiliensis; SPS, silvopastoral system with Brachiaria, Canavalia brasiliensis and Leucaena diversifolia Different lower-case letters indicate significant difference (p<0.05) between CP, LP and SPS-5.5. Different upper-case letters indicate difference (p<0.05) between the distances from L. diversifolia tree row within SPS. Figure Click here to download Figure: Fig 1 Sampling design capture.docx Fig. 1. Lay-out of the field experiment with different forage-based systems. Figure Click here to download Figure: Fig 1 Sampling desing PDF.pdf 36 m * Conventional pastures (CP) with Brachiaria hybrid cv. Cayman Legume pastures (LP) with Brachiaria and Canavalia brasiliensis Double-row of Leucaena diversifolia Sampling point * Photo courtesy of Belisario Hincapié Carvajal Block 1 Block 2 Block 3 93 m 93 m 93 m 31 m Figure Click here to download Figure: Fig 2 Functional groups.docx Fig. 2. Functional groups of soil macrofauna in the three (silvo)pastoral system and with different distance from the tree lines in SPS. CP, conventional pastures of Brachiaria hybrid cv. Cayman; LP, legume-improved pasture of Brachiaria with Canavalia brasiliensis; SPS, silvopastoral system of LP with Leucaena diversifolia tree rows. Mean (SEM). Different lowercase letters indicate differences (p<0.05) between CP, LP and SPS-5.5; different uppercase letter indicate differences (p<0.05) between distances from the tree double-row. Herbivores, predators, detritivors and others are grouped according to prevailing feeding habits. Figure Click here to download Figure: Fig 3 PCA cations.docx 1 Predators Herbivores Richness Detritivores Formicidae MAbio pH PMN MWD MonovalePnOt M SOM 0 TN Resist Divalent P MAroot MAphys BD -1 -1 0 1 PC1: 19.60% (SOM, PMN, POM, N, -MWD, Monovalent, P, MAphys) 2 CP LP SPS-5.5 1 SPS-1.5 SPS-0.0 0 -1 -2 -1 0 1 2 PC1: 19.60% (SOM, PMN, POM, N, -MWD, Monovalent, P, MAphys) Fig. 3. PCA loading plots of the collected samples based on SOM, soil organic matter; TN, total nitrogen; POM, particulate organic matter; PMN, potentially mineralizable nitrogen; MWD, mean weight diameter of soil aggregates; Monovalent, the sum of available K and Na; P, available (Mehlich 3) phosphorus; MAphys, soil macroaggregates formed by physical forces; Divalent, the sum of available Ca and Mg; Richness, the richness of soil macrofauna taxa; MAbio, soil macroaggregates formed by the activity of soil macrofauna; BD, soil bulk density; MAroot, soil macroaggregates formed by plant roots; Resist, resistance to penetration. The highest PC1 scores had SOM (0.858), TN (0.757), POM (0.737) and PMN (0.717) while the highest PC2 scores had Predators (0.803), Richness (0.775) and Herbivores (0.768). Scores of all PC1 and PC2 can be found in Table S-3 (Supplementary material). CP, conventional monoculture pasture (grass only); LP, pasture improved by the incorporation of legume; SPS-5.5, SPS-1.5 and SPS- PC2: 19.04% PC2: 19.04% (Predators, Richness, Herviores, Detrivores, (Predators, Richness, Herviores, Detrivores, Formicidae, MAbio, -BD) Formicidae, MAbio, -BD) 0.0, silvopastoral system at distance of 5.5 m, 1.5 m and 0.0 m from the tree rows, respectively. Means (n=12) with standard errors of the means. Supplementary Material for publication online only Click here to download Supplementary Material for publication online only: Supplementary material.pdf 1 Table S-1 2 Soil properties of the three different forage-based systems and distances from the tree rows in the silvopastoral 3 system (10-20 cm soil layer). Mean values followed by standard errors in parenthesis (n=12) Pasture system 1 CP LP 5.5 m 1.5 m 0.0 m Soil variable SOM (g kg-1) 41.4 (0.53) 42.2 (0.54) 42.2 (0.55) 42.7 (0.60) 42.0 (0.53) POM (g kg-1) 2.91 (0.10) b 3.42 (0.17) a 3.11 (0.16) ab 3.44 (0.25) 3.06 (0.17) TN (g kg-1) 1.00 (0.03) 1.08 (0.03) 1.05 (0.02) 1.06 (0.05) 1.04 (0.04) PMN (mgN kg-1) 21.6 (2.26) b 28.3 (1.83) a 24.6 (1.97) ab 28.2 (2.28) 26.1 (2.16) P (mg kg-1) 10.9 (2.44) 14.4 (4.15) 8.58 (1.34) 8.69 (1.52) 7.33 (1.13) Ca (mg kg-1) 2,353 (44) 2,442 (52) 2,340 (46) 2,325 (47) 2,384 (47) Mg (mg kg-1) 875 (10.4) 897 (24.9) 874 (10.4) B 886 (9.62) AB 906 (8.82) A K (mg kg-1) 514 (103) 696 (136) 568 (109) 519 (92) 505 (173) Na (mg kg-1) 51.8 (1.92) 49.5 (2.34) 46.6 (2.61) B 46.5 (2.03) B 55.4 (2.08) A pH 7.31 (0.08) 7.47 (0.09) 7.34 (0.09) 7.33 (0.08) 7.34 (0.10) PR (kPa) 1,217 (108) 1,242 (110) 1,240 (85) AB 1,191 (79) B 1,386 (138) A 4 SOM, soil organic matter; POM, particulate organic matter; TN, total nitrogen; PMN, potentially mineralizable 5 nitrogen; PR, penetrometer resistance. 6 1 pasture type: CP, conventional pasture of Brachiaria hybrid cv. Cayman monoculture; LP, Brachiaria pasture 7 improved by incorporation of Canavalia brasiliensis; SPS, silvopastoral system with Brachiaria, Canavalia 8 brasiliensis and Leucaena diversifolia 9 Different lowercase letters indicate significant difference (p<0.05) between CP, LP and SPS-5.5. Different 10 uppercase letters indicate difference (p<0.05) between the distances from L. diversifolia tree row within SPS. 11 12 Table S-2 13 Soil macrofauna density in three different pastures systems and distances from the tree rows in the silvopastoral 14 system (10-20 cm soil layer). Mean values followed by standard errors (n-=12) Pasture system 1 B BC 5.5 m 1.5 m 0.0 m Macrofauna group individuals m-2 Formicidae (ants) 244 (171) 100 (57) 128 (83) 11 (5) 27 (7) Oligochaeta (earthworms) 13 (6) 33 (18) 5 (9) 35 (9) 21 (12) Chilopoda (centipedes) 0 (0) 5 (3) 0 (0)B 9 (4)A 3 (2)B Diplopoda (millipedes) 12 (5) 37 (17) 32 (20) 91 (52) 139 (102) Coleoptera (beetles) 11 (5) 4 (2) 3 (3) 12 (6) 9 (4) Hemiptera (true bugs) 0 0 0 0 0 Aranea (spiders) 0 (0) 3 (2) 0 (0)B 1 (1)B 8 (4)A Dermaptera (earwigs) 1 (1) 0 (0) 1(1) 3 (2) 1 (1) Gastropoda (molluscs) 0 (0) 3 (3) 1 (1) 0 (0) 0 (0) Isopoda (woodlice) 4 (4) 0 (0) 3 (2) 0 (0) 1 (1) Blattodea (cockroaches) 0 (0) 0 (0) 0 (0) 0 (0) 3 (2) Others 1 (1) 16 (16) 3 (2) 8 (3) 7 (4) Larvae 31 (9) 4 (3) 7 (3) 15 (5) 8 (4) Total abundance 317 (173) 205 (83) 183 (86) 184 (55) 227 (116) Richness (S) 2.92 (0.40) 2.50 (0.62) 1.92 (0.51) B 3.58 (0.53) A 3.33 (0.72) A 15 1 pasture type: CP, conventional pasture of Brachiaria cv. Cayman monoculture; LP, Brachiaria pasture improved 16 by incorporation of Canavalia brasiliensis; SPS, silvopastoral system with Brachiaria, Canavalia brasiliensis and 17 Leucaena diversifolia 18 Different lowercase letters indicate significant difference (p<0.05) between CP, LP and SPS-5.5. Different upper 19 case letters indicate difference (p<0.05) between the distances from L. diversifolia tree row within SPS. 20 21 22 Table S-3 23 Principle component loadings for measured soil parameters Rotated component Matrix Soil parameter PC1 PC2 20.67% 19.55% SOM 0.824 -0.200 PMN 0.704 -0.045 POM 0.704 -0.233 TN 0.663 -0.322 Cations 0.444 -0.073 MWD -0.429 0.401 BD -0.404 -0.257 MAroot -0.392 0.118 P 0.339 -0.308 Resist -0.231 -0.009 Richness 0.424 0.726 Pred 0.326 0.723 Formicidae 0.022 0.688 Detritiv 0.539 0.643 Herb 0.243 0.615 pH -0.049 0.534 MAbio -0.213 0.476 MAphys 0.306 -0.342 24 SOM, soil organic matter; PMN, potentially mineralizable nitrogen; POM, particulate organic matter; N, total 25 nitrogen; Cations, the sum of available base cations (Ca, Mg, K, Na); MWD, mean weight diameter of soil 26 aggregates; BD, bulk density; MAroot, soil macroaggregates formed by plant roots; P, available (Mehlich 3) 27 phosphorus; Resist, resistance to penetration; Richnes, the richness of soil macrofauna taxa; Pred, predators, 28 Formicidae, ants; Detritiv, detritivores; Herb, herbivores; MAbio, soil macroaggregates formed by the activity of 29 soil macrofauna; MAphys, soil macroaggregates formed by physical forces. 30 31 40 140 35 120 30 100 25 80 20 60 15 Total precipitation (mm) 10 Maximum temperature (°C) 40 Minimum temperature (°C) 5 20 0 0 32 0-Jan 15-Jan 30-Jan 14-Feb 29-Feb 15-Mar 30-Mar 14-Apr 29-Apr 14-May29-May 13-Jun 28-Jun 33 Fig. S-1. Maximum and minimum temperatures and precipitation between January 2017 and June 2017 at the 34 experimental site of CIAT-Palmira, Colombia. 35 Temperature (°C) Precipitation (mm day-1) Penetrometer Resistance (kPa) Penetrometer Resistance (kPa) - Block 1 0 500 1000 1500 2000 0 500 1000 1500 2000 0 0 A B 5 5 10 10 15 15 20 20 25 25 B 30 30 B BC BC 35 BCL 5.5 35 BCL 5.5 BCL 1.5 40 40 BCL 1.5 BCL 0.0 BCL 0.0 45 45 Penetrometer Resistance (kPa) - Block 2 Penetrometer Resistance (kPa) - Block 3 0 500 1000 1500 2000 0 500 1000 1500 2000 0 0 C D 5 5 10 10 15 15 20 20 25 25 30 B 30 B BC BC 35 BCL 5.5 35 BCL 5.5 40 BCL 1.5 40 BCL 1.5 BCL 0.0 BCL 0.0 45 45 Fig. S-2. Penetrometer resistance of forage-based plots with B, conventional pasture of Brachiaria cv. Cayman monoculture; BC, Brachiaria pasture improved by incorporation of Canavalia brasiliensis; BCL, silvopastoral system with Brachiaria, Canavalia brasiliensis and Leucaena diversifolia (A), block 1 (B), block 2 (C) and block 3 (D). Bars indicate standard error of the mean (n=12). Soil depth (cm) Soil depth (cm)