INTRODUCTION
Natural abundance of stable isotopic carbon (13C) in consumer body tissue is widely applied to indicate resource availability and accessibility to the consumers, as well as to reflect their preference for specific resources (Ponsard & Arditi Reference PONSARD and ARDITI2000, Scheu & Falca Reference SCHEU and FALCA2000). The measurement of 15N can be linked to relative trophic position and predator–prey relationships in food webs. In concert, δ13C and δ15N can explain food-web structure, reveal food-web dynamics, and track energy flow through food webs across spatial and temporal scales (Albers et al. Reference ALBERS, SCHAEFER and SCHEU2006, Hyodo et al. Reference HYODO, MATSUMOTO, TAKEMATSU, KAMOI, FUKUDA, NAKAGAWA and ITIOKA2010, Ponsard & Arditi Reference PONSARD and ARDITI2000, Post Reference POST2002, Scheu & Falca Reference SCHEU and FALCA2000, Schmidt et al. Reference SCHMIDT, CURRY, DYCKMANS, ROTA and SCRIMGEOUR2004, Tiunov Reference TIUNOV2007).
Spatial heterogeneity of resource and nutrient availability, which determine baseline δ13C and δ15N, affect 13C and 15N composition of food webs and food-web dynamics. Jennings et al. (Reference JENNINGS, RENONES, MORALES-NIN, POLUNIN, MORANTA and COLL1997) showed that fine-scale variation in resource availability altered δ13C and δ15N of fish communities, which reflects the flexible feeding strategy and trophic position changes in aquatic ecosystems. In terrestrial ecosystems, most soil and litter inhabitants experience restriction to available resources due to their cryptic life cycle and minute size (Ponsard & Arditi Reference PONSARD and ARDITI2000, Scheu & Falca Reference SCHEU and FALCA2000). Landscape-scale resource heterogeneity could possibly drive divergent litter arthropod communities and food-web structures. Yet, whether fine-scale nutrient variation has a measurable influence on litter invertebrates in terms of trophic structure of food webs is still unclear.
An existing landscape-scale soil phosphorus (P) gradient has been described during long-term ecological research in a lowland tropical rain forest at La Selva Biological Station of Costa Rica (CARBONO-project plots, Clark & Clark Reference CLARK and CLARK2000). This fine-scale soil P variation has been shown to affect tree species composition (Clark et al. Reference CLARK, PALMER and CLARK1999), density of total litter arthropods (McGlynn et al. Reference MCGLYNN, DUNN, WOOD, LAWRENCE and CLARK2007), and δ15N in forest-floor leaf litter and an ant species (body tissue, Aphaenogaster araneoides) (McGlynn et al. Reference MCGLYNN, CHOI, MATTINGLY, UPSHAW, POIRSON and BETZELBERGER2009). Soil P in tropical forests has been shown to affect net primary production and below-ground C cycling through regulating soil respiration, decomposition processes and N fixation and turnover rates (Cleveland et al. Reference CLEVELAND, TOWNSEND, TAYLOR, ALVAREZ-CLARE, BUSTAMANTE, CHUYONG, DOBROWSKI, GRIERSON, HARMS, HOULTON, MARKLEIN, PARTON, PORDER, REED, SIERRA, SILVER, TANNER and WIEDER2011, Kaspari et al. Reference KASPARI, GARCIA, HARMS, SANTANA, WRIGHT and YAVITT2008, Townsend et al. Reference TOWNSEND, CLEVELAND, HOULTON, ALDEN and WHITE2011). Decomposition rates could be facilitated via enhancing the interactions between microbes and the microvores by soil P fertilization (Kaspari et al. Reference KASPARI, GARCIA, HARMS, SANTANA, WRIGHT and YAVITT2008). Soil P variation at temporal and spatial scales could contribute to N limitation and affect N availability in a tropical ecosystem (Townsend et al. Reference TOWNSEND, CLEVELAND, HOULTON, ALDEN and WHITE2011). These findings raise the question of whether fine-scale soil P variation in La Selva has impacts on δ13C and δ15N of detrital food webs through regulating C and N dynamics of soil and litter resources. In this study, we first investigated δ13C and δ15N composition of major litter invertebrate groups in detrital food webs, the variation of C, N, δ13C and δ15N in soil organic matter, plant foliage and leaf litter across this soil P gradient. We then evaluated whether the variation in resource (soil and litter) C, N, δ13C and δ15N and soil P relates to the changes of δ13C and δ15N of detrital food webs. We hypothesized that higher δ13C and δ15N of detrital food webs in accordance with higher concentration of basal resource nutrients (i.e. soil/leaf litter C, N, δ13C and δ15N) would be observed in the P-enriched plots.
MATERIALS AND METHODS
Study sites
The CARBONO-Project plots are located in a lowland tropical rain forest at the La Selva Biological Station, Costa Rica (10°26′N, 84°00′W; 37–150 m asl; McDade et al. Reference MCDADE, BAWA, HESPENHEIDE and HARTSHORN1994). The region receives 4000 mm y−1 of rainfall, with an average daily temperature of 26 °C (Sanford et al. Reference SANFORD, PAABY, LUVALL, PHILLIPS, McDade, Bawa, Hespenheide and Hartshorn1994). The soils of all plots are categorized as Oxisols (Haplic Haploperox; Ferralsols in the FAO/WRB system) based on soil mineralogical analyses by Kleber et al. (Reference KLEBER, SCHWENDENMANN, VELDKAMP, ROBNER and JAHN2007). Three major edaphic conditions are the relatively fertile alluvial terraces (‘alluvial’ plots); less-fertile, older ridge-top sites (‘residual plateau’ plots); and less-fertile slope (‘residual slope’ plots; Sollins et al. Reference SOLLINS, SANCHO, MATA, SANFORD, McDade, Bawa, Hespenheide and Hartshorn1994). The deposited soils in the alluvial plots are deeply weathered clay, and are considered similar to the soils of residual plots but younger. Eighteen 0.5-ha CARBONO plots were established in a stratified-block design across soil type and topography (Clark & Clark Reference CLARK and CLARK2000). The plots span substantial landscape variation in soil P as well as in cation concentrations (e.g. potassium (K), calcium (Ca), manganese (Mn), iron (Fe) and aluminium (Al); Espeleta & Clark Reference ESPELETA and CLARK2007). The alluvial soil (0–100 cm depth) had higher C, P, K, Ca, Mn and Fe than the residual plateau soil (Figure 1 in Espeleta & Clark Reference ESPELETA and CLARK2007). The La Selva forest is dominated by a leguminous tree, Pentaclethra macroloba, distributed broadly along all edaphic gradients (Clark & Clark Reference CLARK and CLARK2000). The palm species, Welfia regia, Socratea exorrhiza and Iriartea deltoidea, comprise one quarter of the total density of stems larger than 10 cm in diameter (Clark & Clark Reference CLARK and CLARK2000, Clark et al. Reference CLARK, PALMER and CLARK1999). A map showing the spatial distribution of these 18 plots in different soil gradients can be found in Espeleta & Clark (2007; Appendix A: Ecological Archives M077–012-A1). More description about the background and history of CARBONO plots can be found in Espeleta & Clark (2007).
Figure 1. The δ13C (‰) and δ15N (‰) enrichments (± SD) of the different trophic levels in the detrital food webs from three edaphic site-types: alluvial (a), residual plateau (b) and residual slope (c), in a lowland tropical rain forest at La Selva Biological Station, Costa Rica. Symbols represent the following: ■ = green leaves; ▲ = soil; ♦ = leaf litter; △ = collembolans; × = oribatid mites; □ = pseudoscorpions; ○ = spiders.
δ13C and δ15N of detrital food webs
To determine δ13C and δ15N patterns of the detrital food web, forest-floor leaf litter and litter invertebrates were collected from two 50-m bordering transects (10 m apart from one another) in each CARBONO plot during May–August 2008. At each transect, five 0.25-m2 quadrats of forest-floor leaf litter were collected (a total of ten leaf-litter samples for each plot). The quadrats (subsamples) were 10 m apart from one another to ensure that they share equivalent edaphic properties and the same species pool but are still functionally independent of one another. The multiple subsamples per plot were then aggregated to represent a cumulative assessment of invertebrate community for each plot.
Litter invertebrates were extracted from the collected leaf litter by using mini-Winkler bags (Bestelmeyer et al. Reference BESTELMEYER, AGOSTI, ALONSO, BRANDÃ, BROWN, DELABIE, SILVESTRE, Agosti, Majer, Alonso and Schultz2000, Fisher Reference FISHER1999). Extraordinary high alpha diversity of litter invertebrates is observed; however, specific trophic guilds have not been established at La Selva. In this study, taxonomic groups (i.e. collembolans, oribatid mites, spiders and pseudoscorpions) have been adapted to fulfil the sample requirement (a minimum of 1 mg dry animal tissue) for isotopic ratio analysis. The plots in which the invertebrate tissue samples were not sufficient for isotopic ratio analysis were not considered for further correlation analysis (collembolans, oribatid mites and spiders: N = 16; pseudoscorpions: N = 13). The litter invertebrate samples were maintained in aqueous solution and sorted into taxonomic categories (Collembola, Oribatida, Araneae and Pseudoscorpionida) using a stereomicroscope. The samples were then oven dried inside tin capsules prior to closure in preparation for stable-isotope analyses. Tin capsules were sent to the Stable Isotope Facility at University of California Davis, California, USA, for total C, N and δ13C and δ15N analysis.
The leaf-litter samples were drawn from a 10-L collection of fine leaf litter, from which arthropods were extracted. Leaf litter was oven-dried to a constant mass (at 60 °C) and weighed before and after combustion in a muffle furnace to determine ash-free weight. Subsamples of leaf litter were ground, shifted through No. 20-mesh (diameter = 0.8 mm; Wiley Mill; Thomas Scientific, Inc., Swedesboro, NJ, USA) and packed in tin capsules for stable-isotope analyses (C, N, δ13C and δ15N).
Topsoil samples (0–10 cm depth; N = 3 per plot) were collected simultaneously at each CARBONO plot along the sampling transects described earlier in 2009. Fresh foliar tissue from each of three dominant plant species (Pentaclethra macroloba, Welfia regia and Asterogyne martiana) was collected in August 2009. Foliar samples (mature leaves of large saplings) were collected from three arbitrarily selected individuals per plant species per CARBONO plot. Both soil and foliage samples were oven-dried to a constant mass (at 60 °C), then ground to < 2 mm and packed in tin capsules for stable-isotope analyses (C, N, δ13C and δ15N).
Leaf-litter chemistry
Chemistry of leaf litter other than 13C and 15N fractions was adapted from a long-term litterfall-monitoring project that had been conducted in CARBONO plots since 1997. Within each plot, fine litterfall (leaves, reproductive material and twigs < 1 cm in diameter) were collected every other week from nine 0.25-m2 standing basket traps and nine paired 0.25-m2 ground traps for the large (> 50 cm long) leaves; (D.A. Clark, pers. comm.). Samples from the litter traps in each plot were combined, sorted by litter category, and oven-dried to constant mass (65 °C). The dried leaf material (combined samples from both trap types) was then run through a Wiley Mill to pass through a No. 20-mesh screen (mesh size = 0.8 mm). Ground litter samples were dry-combusted on an elemental analyser (Carlo Erba, Model NA 2500, Milan, Italy) to determine total C and N. Samples were digested using a modified Kjeldahl protocol on a Tecator 2000 digestion System (Perstorp Analytical, Sweden). The method uses 30% hydrogen peroxide and concentrated sulphuric acid at 320 °C to hydrolyse organic P, Mg, Ca and K into inorganic form. Digestates were analysed on an Alpkem Flow Solution IV Autoanalyzer (OI Analytical, College Station, Texas, USA) in accordance with US EPA method for total P. Total Mg, Ca and K in digestates were analysed on an atomic absorption spectrometer (AAnalyst 100, PerkinElmer, Connecticut, USA). The data collected during 2005–2007 were adapted in this study. All data are reported on a dry mass basis. Nutrient ratios (e.g. C : N) are reported on a molar basis.
Statistical analysis
All collected data were first analysed by PROC UNIVARIATE procedure in SAS statistic software (version 9.1, SAS Institute Inc., USA) to pass the normality of data. We used a one-way analysis of variance (ANOVA) method under PROC GLM procedure (SAS) to examine the difference in soil nutrients (C, N, P, C : N and C : P ratios), leaf-litter chemistry (C, N, P, C : N and C : P ratios), δ13C and δ15N composition of soil, leaf litter, plant foliage and litter invertebrates among the three edaphic types (alluvial, residual plateau and residual slope; six plots per edaphic type). Post hoc Tukey HSD test (significance level at P ≤ 0.05) was applied to test the differences among edaphic types. The average values of soil nutrient concentrations, litter chemistry and δ13C and δ15N of each invertebrate group from each CARBONO plot were calculated and used to run correlation analyses (soil nutrients and litter chemistry data: N = 18; collembolans, oribatid mites, and spiders: N = 16; pseudoscorpions: N = 13). Pearson correlation analysis (PROC CORR Pearson, SAS) was used to examine the relationships among soil nutrients, litter chemistry and δ13C and δ15N of litter invertebrates.
RESULTS
The alluvial site had the highest leaf-litter C, N and P concentrations, and the lowest C : N and C : P ratios. However, soil C and N did not significantly differ among the three edaphic site-types over the same landscape at La Selva (Table 1).
Table 1. Carbon (C), nitrogen (N), phosphorus (P), C : N and C : P ratios in surface soil (0–10 cm depth) and leaf litter in the alluvial, residual-plateau and residual-slope edaphic site-types within a 500-ha landscape of a lowland tropical rain forest at La Selva Biological Station, Costa Rica. Data (shown as mean ± SD) with different letters indicate significant difference between three edaphic site-types (Tukey HSD at P < 0.05). Soil P and C : P ratio data were adapted from Espeleta & Clark (2007).
Plant foliage was the most depleted in 13C (δ13C ranging from −37.8‰ to −30.4‰), followed by leaf litter (δ13C ranging from −30.6‰ to −28.0‰), and then by soil (δ13C ranging from −29.2‰ to −27.4‰) (Figure 1). Foliar δ13C of the understorey palm, Asterogyne martiana (−35.7‰ ± 1.1‰) was significantly lower than the foliage from the canopy-tree species Pentaclethra macroloba (−33.0‰ ± 1.0‰) and the subcanopy palm Welfia regia (−33.2‰ ± 1.4‰) (Tukey HSD, P < 0.0001). Litter invertebrates were higher in δ13C than basal resource leaf litter. The δ13C of the detritivores (collembolans and oribatid mites) was 2.9‰ and 3.3‰ higher than leaf litter, respectively; while the δ13C of pseudoscorpions and spiders were 3.7‰ and 3.6‰ higher than leaf litter, respectively (Tukey HSD, all P < 0.05; Figure 1). Among litter invertebrates, the spiders were significantly more 13C-enriched (δ13C = −26.6‰ ± 0.14‰) than the collembolans (δ13C = −27.2‰ ± 0.1‰, Tukey HSD; P < 0.05). The δ13C distribution patterns of topsoil, plant foliage, leaf litter and litter invertebrates were not significantly different among sites (Figure 1).
The δ15N of resources (plant foliage, topsoil, leaf litter) and litter invertebrates (except pseudoscorpions) were significantly higher in the alluvial site than in the other two sites (Figure 1, Table 2). Soil δ15N was significantly higher (6.8‰ ± 0.9‰) in the alluvial site than in the residual-plateau and residual-slope sites (5.2‰ ± 0.7‰ and 5.3‰ ± 0.6‰, respectively; PROC GLM, F = 26.8, df = 2, 51, P < 0.0001) (Table 2). Higher δ15N values were also observed in plant foliage (no species-specific difference), leaf litter and litter invertebrates collected from the alluvial site (Figure 1, Table 2). Delta 15N values ranged from 2.3‰ in plant foliage to the 8.4‰ in the pseudoscorpions (Figure 1). Three distinct guilds in the detrital food web can be identified by their δ15N values. The predators had the highest δ15N value (spiders: 7.9‰ ± 0.3‰ and pseudoscorpions: 8.4‰ ± 0.3‰), with oribatid mites in the middle (5.6‰ ± 0.3‰), and collembolans with the lowest δ15N (3.8‰ ± 0.3‰) (Figure 1). By adopting leaf litter as the δ15N baseline (3.3‰ ± 0.3‰) of detrital food webs, the δ15N signatures of the collembolans and oribatid mites were 0.4‰ and 2.2‰ higher, respectively, than leaf litter. Spiders and pseudoscorpions were 4.1‰ and 4.7‰ higher than collembolans. The δ15N distribution pattern of the detrital food webs was consistent in all three edaphic-type sites (Figure 1).
Table 2. Delta 15N enrichment in soil (0–10 cm depth), leaf litter and litter invertebrate groups from contrasting edaphic site-types within a 500-ha landscape of a lowland tropical rain forest at La Selva Biological Station, Costa Rica. Values (shown as mean ± SD) are from six plots sampled per site-type, and F values with degrees of freedom (df) are also shown in the table. Green leaves samples were the average values of three tree species collected in this study. Data with different letters indicate significant difference among three edaphic site-types (Tukey HSD at P < 0.05).
Across-plot δ15N variation of detrital food webs was significantly correlated to plot-specific soil P concentration and soil C : P ratio (Table 3). Higher δ15N in leaf litter, collembolans and oribatid mites was correlated with higher soil P concentration (Pearson correlation r = 0.68, 0.66, 0.6, respectively, all P < 0.05; Figure 2, Table 3) and negatively correlated to soil C : P (r = −0.59, −0.59, −0.52, respectively, all P < 0.05; Table 3). Leaf-litter δ15N value was also positively correlated to leaf-litter C and P concentrations, and negatively related to leaf litter C : P (Table 3). The δ15N of the collembolans and the pseudoscorpions were positively correlated to leaf-litter δ15N (r = 0.72 and 0.75, respectively, both P < 0.05; Table 3). Spider δ15N had a positive correlation with leaf-litter P concentration (r = 0.52, P < 0.05), but not soil P concentration. Foliar δ15N of Asterogyne martiana was correlated to soil P concentration (r = 0.58, P = 0.01). However, soil and leaf-litter P concentration and its C : P ratio did not affect the distance of δ15N between trophic levels (relative trophic position: Δ δ15N, Table 3). Interestingly, higher leaf-litter C and δ15N were found to relate to shorter δ15N distance between the pseudoscorpions and leaf litter (Δ δ15NP-L: r = −0.59 and −0.67, respectively, both P < 0.05; Table 3). The plot-specific annual litterfall and forest-floor leaf-litter stock during 2007–2009 did not relate the δ15N variation of the leaf litter and any litter invertebrates (data not shown, Pearson correlation analysis, all P > 0.05).
Table 3. Relationships (Pearson correlation coefficient, r) between soil and leaf-litter carbon (C), nitrogen (N), C : N ratio, δ15N, phosphorus (P) and C : P ratio, with δ15N values of leaf litter and litter invertebrate groups in a lowland tropical rain forest at La Selva Biological Station, Costa Rica. Asterisks indicate statistically significant. * = P < 0.05. ** = P < 0.01. Δ δ15N indicate the distance of δ15N signature between different trophic levels, and the subscript letters indicate the different trophic groups: C = collembolans; L = leaf litter; O = Oribatid mites; P = pseudoscorpions; and S = spiders.
Figure 2. Relationships between soil phosphorus (P) concentration and δ15N (‰) in leaf litter, collembolans and oribatid mites in a lowland tropical rain forest at La Selva Biological Station, Costa Rica. Symbols represent the following: ♦ = leaf litter (dashed line); ○ = collembolans (dotted line); × = oribatid mites (solid line). Each data point indicates the average value of soil P concentration and δ15N of each invertebrate group from each CARBONO plot.
DISCUSSION
Overall, δ13C and δ15N distribution patterns of the detrital food webs were relatively consistent across the landscape-scale variation of soil P in this lowland tropical rain forest. The δ15N signature spanned over 8‰, which suggests at least three trophic levels in the detrital food webs (assuming 2.3‰–3.4‰ per trophic level; DeNiro & Epstein Reference DENIRO and EPSTEIN1981, McCutchan et al. Reference MCCUTCHAN, LEWIS, KENDALL and MCGRATH2003, Minagawa & Wada Reference MINAGAWA and WADA1984). The significant enrichment in 15N of the spiders and the pseudoscorpions (4.1‰ and 4.7‰, respectively, higher than the collembolans) indicated the occurrence of intraguild predation, cannibalism among the spider and pseudoscorpion populations, or a mixed diet of other 15N-enriched secondary decomposers (i.e. earthworms) (DeNiro & Epstein Reference DENIRO and EPSTEIN1978, McCutchan et al. Reference MCCUTCHAN, LEWIS, KENDALL and MCGRATH2003, McNabb et al. Reference MCNABB, HALAJ and WISE2001, Post Reference POST2002, Tiunov Reference TIUNOV2007). However, it could also be a residual effect of a wide range of life stages we analysed (Ponsard & Arditi Reference PONSARD and ARDITI2000, Scheu & Falca Reference SCHEU and FALCA2000).
Higher δ13C and δ15N signatures in soil and leaf litter than in the foliage samples can be attributed to the accumulation of heavy isotopic 13C in the decayed leaf litter during the litter decomposition process. Microbial activities often primarily utilize 12C-compound carbohydrates and release 13C-depleted CO2 (Hyodo et al. Reference HYODO, MATSUMOTO, TAKEMATSU, KAMOI, FUKUDA, NAKAGAWA and ITIOKA2010, Melillo et al. Reference MELILLO, ABER, LINKINS, RICCA, FRY and NADELHOFFER1989, Nadelhoffer & Fry Reference NADELHOFFER and FRY1988). Fluxes of relatively 13C-enriched dissolved organic carbon from leaf litter and/or sorption of humic fraction within soils can result in the heavier 13C enrichment in the soils (Cleveland et al. Reference CLEVELAND, NEFF, TOWNSEND and HOOD2004). The collembolans and oribatid mites (the detritivores) were 2.9‰ and 3.1‰ higher in δ13C, respectively, than their basal resource leaf litter (Figure 1), which falls at the higher end of the 1–3‰ range suggested by DeNiro & Epstein (1978). Higher δ13C in the detritivores could result from their preferential assimilation of 13C-enriched compounds from decayed/humified debris (such as cellulose and starch; Pollierer et al. Reference POLLIERER, LANGEL, SCHEU and MARAUN2009), and consumption of 13C-enriched fungal hyphae as a significant portion of their diet (Beare et al. Reference BEARE, PARMELEE, HENDRIX, CHENG, COLEMAN and CROSSLEY1992, Hobbie et al. Reference HOBBIE, MACKO and SHUGART1999, Hyodo et al. Reference HYODO, MATSUMOTO, TAKEMATSU, KAMOI, FUKUDA, NAKAGAWA and ITIOKA2010, Pollierer et al. Reference POLLIERER, LANGEL, KORNER, MARAUN and SCHEU2007, Ruf et al. Reference RUF, KUZYAKOV and LOPATOVSKAYA2006). The preferential utilization of fungal hyphae (often relatively 15N-enriched) may also account for the higher δ15N values in oribatid mites than that in the collembolans and leaf litter (Schneider et al. Reference SCHNEIDER, MIGGE, NORTON, SCHEU, LANGEL, REINEKING and MARAUN2004).
To our knowledge, this is the first study to demonstrate a positive correlation between δ15N signature in soil, leaf litter and body tissue of the collembolans and oribatid mites and soil P fertility, an essential nutrient other than C and N. In a review by Cleveland et al. (Reference CLEVELAND, TOWNSEND, TAYLOR, ALVAREZ-CLARE, BUSTAMANTE, CHUYONG, DOBROWSKI, GRIERSON, HARMS, HOULTON, MARKLEIN, PARTON, PORDER, REED, SIERRA, SILVER, TANNER and WIEDER2011), soil P as a limiting nutrient affects net primary production and below-ground C cycling via regulating N fixation and turnover rates, soil respiration and decomposition processes of the tropical forests. Resource C and N (i.e. N concentration and C : N ratio) have been known to affect δ15N values of animal consumers at higher trophic levels via trophic cascade (Adams & Sterner Reference ADAMS and STERNER2000, Haubert et al. Reference HAUBERT, LANGEL, SCHEU and RUESS2005, Jennings et al. Reference JENNINGS, RENONES, MORALES-NIN, POLUNIN, MORANTA and COLL1997, Robbins et al. Reference ROBBINS, FELICETTI and SPONDEIMER2005, Thomas & Cahoon Reference THOMAS and CAHOON1993, Webb et al. Reference WEBB, HEDGEES and SIMPSON1998). Thus, higher P concentration could possibly cause stronger δ15N signature in the collembolans and the oribatid mites by facilitation of higher N concentration along with enriched15N in the leaf litter from the alluvial plots. No correlations of leaf litter (or soil) N to its own δ15N and to detritivore δ15N eliminates the possibility of higher detritivore δ15N driven by N-rich resources. However, we did find collembolan and pseudoscorpion δ15N had a positive correlation with leaf-litter δ15N. The increase of soil P fertility has been shown to relate to higher foliage δ15N signature (and lower plant foliage N : P ratio) in nutrient-limited tropical forest ecosystems (Clarkson et al. Reference CLARKSON, SCHIPPER, MOYERSOEN and SILVESTER2005, Hidaka & Kitayama Reference HIDAKA and KITAYAMA2011, McKee et al. Reference MCKEE, FELLER, POPP and WANEK2002). Clarkson et al. (Reference CLARKSON, SCHIPPER, MOYERSOEN and SILVESTER2005) proposed that soil P enhances plant N demand and diminishes plant discrimination between 15N and 14N isotopes to cause the heavier foliar 15N. In this study, Asterogyne martiana foliar δ15N and leaf-litter δ15N were both observed in the P-enriched alluvial plots. Higher δ15N values in the collembolans and oribatid mites might be derived from their consumption on 15N-enriched leaf litter at the P-enriched plots. Strong δ15N signals in collembolans further cascaded down to the spiders and/or pseudoscorpions. The collembolans and oribatid mites are easily affected by environmental stoichiometry; often seeking resources to meet their minimum N and P requirements (Davidson et al. Reference DAVIDSON, COOK, SNELLING and CHUA2003, Sterner & Elser Reference STERNER and ELSER2002). Such sensitivity to nutrient limitation may explain why collembolans and oribatid mites responded strongly to a smaller magnitude of soil P variability than spider and pseudoscorpions at higher trophic levels. However, stronger detritivore δ15N derived as a result of 15N-enriched decayed/humified debris or their preferential consumptions of fungal hyphae colonized on leaf litter (or both) need further investigation.
Detrital food webs, in terms of trophic structure and community composition, are a key component in regulating below-ground nutrient dynamics and biogeochemical cycles (e.g. mineralization and decomposition processes), particularly in the tropics (González & Seastedt Reference GONZÁLEZ and SEASTEDT2001, Heneghan et al. Reference HENEGHAN, COLEMAN, ZOU, CROSSLEY and HAINES1999, Powers et al. Reference POWERS, MONTGOMERY, ADAIR, BREARLEY, DEWALT, CASTANHO, CHAVE, DEINERT, GANZHORN, GILBERT, GONZÁLEZ-ITURBE, BUNYAVEJCHEWIN, GRAU, HARMS, HIREMATH, IRIARTE-VIVAR, MANZANE, OLIVERIRA, POORTER, RAMANAMANJATO, SALK, VARELA, WEIBLEN and LERDAU2009). Unfortunately there is still limited understanding of relationships between the microbes and detrital invertebrates, and their interactions with soil nutrient dynamics and biogeochemical processes in the tropics (Hättenschwiler & Jørgensen Reference HÄTTENSCHWILER and JØRGENSEN2010). Ecological roles of detrital invertebrates are often omitted and probably considered negligible while addressing plant–soil nutrient dynamics and cycling at ecosystem scales (Cleveland et al. Reference CLEVELAND, TOWNSEND, TAYLOR, ALVAREZ-CLARE, BUSTAMANTE, CHUYONG, DOBROWSKI, GRIERSON, HARMS, HOULTON, MARKLEIN, PARTON, PORDER, REED, SIERRA, SILVER, TANNER and WIEDER2011). In addition to the positive soil P-detritivore δ15N correlation found in this study, fine-scale soil P variation over the same landscape also affected below-ground soil CO2 efflux (Schwendenmann et al. Reference SCHWENDENMANN, VELDKAMP, BRENES, O'BRIEN and MACKENSEN2003), plant root dynamics (Espeleta & Clark Reference ESPELETA and CLARK2007), density of total litter invertebrates (McGlynn et al. Reference MCGLYNN, DUNN, WOOD, LAWRENCE and CLARK2007), and δ15N signature of leaf litter and a gypsy ant, Aphaenogaster araneoides (McGlynn et al. Reference MCGLYNN, CHOI, MATTINGLY, UPSHAW, POIRSON and BETZELBERGER2009). This study addresses the relevance of fine-scale soil P variation in regulation of nutrient dynamics through its interrelations with leaf litter and arthropod components of detrital food webs. Interrelationships between soil nutrients, plants and components of detrital food webs (such as the microbes, detritivores and predators) are important for better understanding of soil nutrient dynamics and food-web structures, especially in the tropics. As for many broader-scale comparative studies, interpretation and comparison of results are often complicated due to methodological differences, contrasting soil types, differences in species composition and divergent climate and precipitation regimes between sites (Cleveland et al. Reference CLEVELAND, TOWNSEND, TAYLOR, ALVAREZ-CLARE, BUSTAMANTE, CHUYONG, DOBROWSKI, GRIERSON, HARMS, HOULTON, MARKLEIN, PARTON, PORDER, REED, SIERRA, SILVER, TANNER and WIEDER2011, Powers et al. Reference POWERS, MONTGOMERY, ADAIR, BREARLEY, DEWALT, CASTANHO, CHAVE, DEINERT, GANZHORN, GILBERT, GONZÁLEZ-ITURBE, BUNYAVEJCHEWIN, GRAU, HARMS, HIREMATH, IRIARTE-VIVAR, MANZANE, OLIVERIRA, POORTER, RAMANAMANJATO, SALK, VARELA, WEIBLEN and LERDAU2009). Thus, field research at the fine-scale provides an opportunity to test specific hypotheses, to identify mechanisms, as well as to obtain in-depth understandings of interactions between soil, plants and the detrital food web.
ACKNOWLEDGEMENTS
We thank Danilo Brenes and Danilo Villegas for helping field sampling of leaf litter, green leaves and soil; Justin Betzelberger and Gilbert Lam for conducting litter arthropod sampling. This study and student participation was supported by the US National Science Foundation (NSF/EAR0421178; NSF/OISE0854259; NSF/HRD0802628). Support for the litter collections, the maintenance of the CARBONO plots, was provided by the US National Science Foundation (NSF/LTREB0841872). We thank three anonymous reviewers for the useful comments on the earlier draft of this paper.
