ARTICLE IN PRESS Pedobiologia 49 (2005) 529—538 www.elsevier.de/pedobi INTERNATIONAL SYMPOSIUM ON IMPACTS OF SOIL BIODIVERSITY ON BIOGEOCHEMICAL PROCESSES IN ECOSYSTEMS, TAIPEI, TAIWAN, 2004 Soil communities and plant litter decomposition as influenced by forest debris: Variation across tropical riparian and upland sites Honghua Ruana, Yiqing Lib, Xiaoming Zouc,d,Ã aFaculty of Forest Resources and the Environment, Nanjing Forestry University, Nanjing, China bDepartment of Ecology, Evolution, and Natural Resources, Rutgers University, 14 College Farm Road, New Brunswick, NJ 08901-8551, USA cXishuangbanna Tropical Botanical Garden, The Chinese Academy of Sciences, 88 Xuefu Road, Kunming, Yunnan 650223, China dInstitute for Tropical Ecosystem Studies, University of Puerto Rico, P.O. Box 23341, San Juan, PR 00931-3341, USA Received 13 November 2004; accepted 28 August 2005 KEYWORDS Summary Landscape Forest debris on ground surface can interact with soil biota and consequently change heterogeneity; ecosystem processes across heterogeneous landscape. We examined the interactions Litter arthropods; between forest debris and litter decomposition in riparian and upland sites within a Puerto Rico; tropical wet forest. Our experiment included control and debris-removal treatments. Soil microbial Debris-removal reduced leaf litter decomposition rates in both the riparian and upland biomass; sites. Debris-removal also reduced soil microbial biomass C in the upland site, but had Tabonuco forest; no effect on microbial biomass C in the riparian site. In contrast, debris-removal Tropical wet forest altered the density of selected arthropod groups in the riparian site. Litter decomposition rates correlated with both soil microbial biomass and the density of millipedes in a multiple stepwise regression model. Removal of forest debris can substantially reduce rates of leaf litter decomposition through suppressing soil activities. This influence can be further modified by landscape position. Forest debris plays an essential role in maintaining soil activities and ecosystem functioning in this tropical wet forest. & 2005 Elsevier GmbH. All rights reserved. Introduction Forest floor varies in the quantity of plant debris ÃCorresponding author at: Xishuangbanna Tropical Botanical Garden, The Chinese Academy of Sciences, 88 Xuefu Road, and this variation attributes to spatial heterogene- Kunming, Yunnan 650223, China. ity of soil biota in forest ecosystems because forest E-mail address: [email protected] (X. Zou). debris provides them with food and shelter 0031-4056/$ - see front matter & 2005 Elsevier GmbH. All rights reserved. doi:10.1016/j.pedobi.2005.08.001 ARTICLE IN PRESS 530 H. Ruan et al. (Coleman and Crossley, 1996; Wardle and Lavelle, spots linking upland with aquatic systems (McClain 1997). Soil biota plays an integral role in the et al., 2003). decomposition processes. Recently, soil fauna has We designed this study to examine the interac- been shown to play an important role in plant litter tions between forest debris and leaf litter decom- decomposition in tropical forests (Heneghan et al., position in riparian and upland sites within a 1998, 1999; Gonza´lez and Seastedt, 2001; Liu and tropical wet forest of Puerto Rico. Our overall Zou, 2002). Gonza´lez and Seastedt (2000) demon- hypothesis is that forest debris plays a central role strated that litter decomposition rate (k value) was in maintaining soil activity and ecosystem function- reduced by as much as 66% when arthropods were ing. We predicted that removal of forest debris will reduced using naphthalene. They further con- result in reduction in the rates of plant litter cluded that plant litter decomposition rates in decomposition through suppressing soil activity, tropical forests cannot be predicted in current and that the magnitude of this reduction in models developed from temperate ecosystems that decomposition rates and soil biota will be greater excluded soil biota as independent variables. in upland than in riparian sites because carbon and Plant litter can influence the activity of soil nutrient input can be supplemented through sub- microbes and fauna by providing them with a food surface water in the latter. We further predict that source and habitat (Zak et al., 1990, 1994; Ding plant litter decomposition rates will correlate with et al., 1992; Bengtsson et al., 1998). Zimmerman soil microbial biomass and litter fauna in the et al. (1995) reported that the experimental tropical wet forest. removal of litter and woody debris 3 years after hurricane increased soil nitrogen availability and litterfall by 40% compared to unmanipulated con- trol plots in a tropical wet forest of Puerto Rico. Li Materials and methods et al. (2005) suggested that litter removal in forests could reduce soil microbial activity and deplete the Study sites and experimental design light fraction of labile carbon pools. The influences of forest debris on leaf litter This study was conducted in riparian and upland decomposition may be achieved through altering sites within a tabonuco forest at El Verde Field community structure and activities of soil biota as Station (181200N, 651490W) in the Luquillo Experi- well as soil chemical and physical properties. These mental Forest (LEF) of Puerto Rico. The vegetation influences can be further modified by topography of was classified as a subtropical wet forest. The land where ridge and slope areas are often most forest is dominated by tabonuco trees (Dacryodes exposed to hurricane damages and valleys are excelsa). Elevation of the tabonuco forest ranges relatively protected (Zimmerman et al., 1996). In from 300 to 600 m above sea level. Mean monthly addition, soil chemical and physical conditions may temperature varies from 20.8 to 24.4 1C with a differ between ridges and valleys. In a tropical wet mean annual precipitation of 3456 mm. forest of Puerto Rico, ridge and slope areas are On September 21 of 1998, a category 2 hurri- regarded as uplands that are typically dominated cane, Georges, struck the LEF with a maximum by an Oxisol, with low pH and cation concentration wind speed of 185 km hÀ1. We started our experi- and with high carbon content, whereas valleys ment 7 months after the passage of Georges. are represented by riparian soils with higher Freshly senesced leaves were collected in April of pH values and cation contents and with lower 1999. Leaf litterbags were constructed and placed carbon levels (Scatena and Lugo, 1995; Soil Survey in the field in May. We established a total of eight Staff, 1995). Compared to upland zones, riparian blocks with two plots (2 Â 2 m in size) in each block soils are often water-saturated, and periodically along a slope from ridge top to stream bank in the flooded. These differences in soil chemical tabonuco forest, of which four blocks were located and physical properties between riparian and in riparian sites near a stream and four in upland upland sites may result in variations in soil activity sites that include both ridge and slope areas (Ruan (Badejo and Van Straalen, 1993; Willig et al., 1996). et al., 2004). Two treatments were randomly For example, Ettema et al. (1999) reported that assigned to the two plots within each block: control soil microbial C and respiration in riparian sites and debris-removal. For the debris-removal plots, were significantly higher than in upland sites. existing debris (averaged 450 g mÀ2) from both pre- Riparian soils are recognized to provide ecosystem hurricane and post-hurricane input was removed on services disproportionate to their relatively small April 17, 1999. The removal of pre-hurricane debris land area (Scatena, 1990; Frangi and Lugo, 1991; represents the extreme bare soil conditions in Fennessy and Cronk, 1997) and as ecological hot heterogeneous landscape in the forest. To prevent ARTICLE IN PRESS Soil biota as influenced by forest debris 531 Table 1. Species composition and weight (g) of air- filled with a mixture of air-dried and newly dried leaf litter in litterbags that represent the main tree senesced fresh leaf litter collected from the species composition of natural litterfall in April 1999 riparian and upland areas (Zou et al., 1995). This from both the riparian and upland sites within a tropical fresh leaf litter represented the main tree species wet forest of Puerto Rico. composition in natural litterfall in April (Table 1). Species Riparian Upland Tree species composition of litterfall was deter- 2 (g) (g) mined using leaffall collected from the 4 m tent above the debris-removal plots. Leaves of the five Tabebuia heterophlla (DC.) Britton 0.5 most common species from each of the riparian and Prestoea montana (R. Grah.) 0.4 upland sites, representing 81–85% of the total fresh Nichols. leaf-fall mass, were placed in litterbags in propor- Byrsonima spicata (Cav.) HBK 0.4 tion to their biomass in litterfall (Table 1). Leaves Inga laurina (Sw.) Willd.ex L. 0.2 Sapium laurocerasus Desf. 0.2 of the remaining species were randomly selected to Buchenavia capitata (Vahl) Eichl. 0.6 obtain a total of 5 g air-dried leaf material (4.6 g Dacryodes excelsa Vahl 0.4 mean oven-dried mass) in each litterbag. Initial Homalium recemosum Jacq. 0.4 leaf litter chemistry for C, N, and P contents Guarea guidonia (L.) Sleumer 0.3 was 51.3%, 1.57% and 0.57 mg gÀ1, respectively Sloanea berteriana Choisy 0.2 (Gonza´lez and Zou, 1999; Liu and Zou, 2002). Other species 0.6 0.8 The decomposition experiment was initiated in Total 2.3 2.7 May 1999 and continued until May 2000. Litterbags The common mixed-species litterbags were used for a decom- were placed onto forest floor inside each plot. Two position experiment in both riparian and upland sites. randomly selected litterbags were collected after 0, 60, 120, 180, 240, 300, 360 days in the field from each plot with a total of eight litterbags from the post-hurricane debris input that remained in the riparian or upland sites at each collection date for canopy and the subsequent new litterfall, a tent of each of the control and debris-removal treatment.