Predators can disproportionately impact the structure and function of ecosystems relative to their biomass. These effects may be exacerbated under warming in ecosystems like the Arctic, where the number and diversity of predators are low and small shifts in community interactions can alter carbon cycle feedbacks. Here we show that warming alters the effects of wolf spiders, a dominant tundra predator, on belowground litter decomposition and nutrient dynamics. Specifically, while high densities of wolf spiders result in faster litter decomposition under ambient temperatures, they result instead in slower decomposition under warming. Higher spider densities are also associated with elevated levels of available soil nitrogen, potentially benefitting plant production. Changes in decomposition rates under increased wolf spider densities are accompanied by trends toward fewer fungivorous Collembola under ambient temperatures and more Collembola under warming, suggesting that Collembola mediate the indirect effects of wolf spiders on decomposition. The unexpected reversal of wolf spider effects on Collembola and decomposition suggests that in some cases, warming does not simply alter the strength of top-down effects but instead induces a different trophic cascade altogether. Our results indicate that climate change-induced effects on predators can cascade through other trophic levels, alter critical ecosystem functions, and potentially lead to climate feedbacks with important global implications. Moreover, given the expected increase in wolf spider densities with climate change, our findings suggest that the observed cascading effects of this common predator on detrital processes could potentially buffer concurrent changes in decomposition rates.
Data set ID:
EML revision ID:
Study system and experimental design: We explored the effects of spider density and warming on community composition, litter decomposition, litter N loss, and available soil nutrients through a fully factorial field experiment that was conducted from early June 2011 through late July 2012 in an area of moist acidic tundra (1) on the North Slope of Alaska near Toolik Field Station.
Mesocosms were circular, 1.5 meters in diameter, and enclosed with aluminum flashing that was buried 20 cm belowground and stood 20 cm above the soil surface in order to contain the belowground community and wolf spiders. Plots were distributed among five blocks and randomly assigned to one of the six spider density/warming treatments. For the warming treatment, we placed ITEX (International Tundra Experiment) open-topped passive warming chambers (OTCs) over half of the plots during the summer seasons only. The ITEX method typically increases mean air temperature within the OTCs by 1-2°C (2).
Wolf spider density treatments included: 1) low, 2) control, and 3) high density. At the beginning of each summer, we removed all possible wolf spiders from low density plots and added spiders to high density plots such that they would have approximately twice the early season average density of control plots, which was 1.1 wolf spider m-2. We continued to check and remove individuals from the removal plots throughout each summer. Visual inspection and live pitfall trapping at the end of the summer seasons verified that we successfully manipulated wolf spider densities during both summers of the experiment. In 2011, mean and SE of wolf spider densities within each plot over a series of three 24-hour live pitfall trapping bouts (July 20-22, 2011) were 0.8 ± 0.22 wolf spiders in high density plots, 0.167 ± 0.075 wolf spiders in control plots, and 0.20 ± 0.10 wolf spiders in low density plots. ANOVA and post-hoc Tukey tests for these 2011 data showed that the average number of spiders was significantly higher in high-density plots than in control (p = 0.0145) and removal plots (p = 0.021) but that catches did not differ between control and removal plots (p = 0.986). At the end of the second summer, we visually surveyed each plot to get a more complete count of the total number of wolf spiders per plot (as opposed to pitfalls, which catch a smaller subset of spiders present) and found that wolf spider densities differed significantly according to their pre-assigned treatments (ANOVA: F2,27 = 21.85, p = <0.0001); low density (removal) plots had approximately 0.3 ± 0.213 wolf spiders, control plots had 1.8 ± 0.20 wolf spiders, and high density plots had 3.3 ± 0.47 wolf spiders each.
Sampling of community composition: We sampled the belowground community by taking soil samples of the upper soil organic layer (average sample volume: 176 cm3) in each plot on two mid-season dates (June 20 and July 17, 2012). Microarthropods were removed through heat-extraction using Berlese-Tullgren funnels (BioQuip Products, Rancho Dominguez, CA, USA) into 90% ethanol over a period of five days. Groups that were present in at least 70% of our soil samples included Collembola, which were identified to order (Entomobrymorpha, Poduromorpha, Symphypleona), oribatid mites, and predatory mites. There were also some intermediate predators in the soil samples (centipedes, unidentifiable juvenile spiders, and other small spiders from the Dysderidae, Thomisidae, Linyphiidae, and Gnaphosidae families). Although none of these predator types were present in abundance, at least one intermediate predator was caught in 73% of the soil samples. These animals are likely predators to the detritivore groups and could potentially be prey to wolf spiders; due to their low numbers in the soil samples we did not differentiate among different soil-dwelling intermediate predators but rather used a measure of the total density of these predators. In all subsequent analyses of soil-dwelling microarthropods, we used the average densities from the June and July samplings.
In order to measure treatment effects on other intermediate predators that could affect detritivore densities, particularly Collembola, and/or be prey to wolf spiders (e.g., surface-active crab spiders, linyphiid spiders, juvenile spiders, and centipedes), we used live pitfall traps over several dates during the second summer season (June 13-15, 29 and July 9, 11, 20-23). Live traps were checked daily; surface-active intermediate predators and other by-catch were immediately returned to their respective experimental plots. Counts were corrected by the number of live pitfall traps within each plot and expressed as the average number of surface-active intermediate predators caught per plot over the summer.
Fungal and bacterial biomass were estimated using epiflourescent microscopy techniques (3) from 5-g soil samples that were collected from the upper organic layer at the conclusion of the experiment in late July. Fungal samples were stained with calcoflour fluorescent brightener (see 4) and read between 334-365 nm wavelengths. Bacterial samples were stained with 5-(4,6 dichlorotriazin-2-yl) aminoflouorescein (DTAF) and read at a 490 nm wavelength. We estimated active fungal biomass as 10% of the total fungal biomass (see 5).
Measures of decomposition and nutrient availability: We measured decomposition rates and litter N loss in the experimental plots using litter bags that were filled with standing dead leaves of the dominant plant, Eriophorum vaginatum, that were collected from an area adjacent to our experimental plots in June 2010. Litter was dried at 40° C for 48 hours, mixed, and sub-sampled for litter bag preparation. Litter bags were 8×8 cm with 3 mm mesh size on the top and bottom to allow access by most arthropods and each bag contained 1.5 g of litter. Subsamples of the initial litter mixture were ground and processed for total nitrogen (N) content using a CE Elantech Flash EA 1112 Elemental Analyzer (CE Elantech, Inc., Lakewood, NJ, USA) at Duke University.
We deployed a pair of litter bags in each plot during mid-June 2011 by placing one litter bag on the soil surface and by burying the other in the lower litter layer under the moss surface (ca. 5-10 cm). We collected both litter bags in August 2012 (14 months) after the experimental start date and manually removed accumulated soil, ingrown moss and roots, and microarthropods from the decomposed leaves before drying them at 40° C for 72 hours. Litter bag contents were weighed to determine proportional mass loss from the initial litter, and subsamples were ground and analyzed for %N as described above.
We used ion exchange membranes (Plant Root Simulator (PRS)TM probes; Western Ag Innovations Inc., Saskatoon, Canada) to measure treatment effects on available soil N, P, and K. Three cation- and three anion-exchange resin membranes were inserted at equal spacing within each plot in June 2012 and incubated three weeks prior to collection. The initial probes were replaced with a second set of probes that were similarly incubated for an additional three weeks. Probes were analyzed by Western Ag Innovations Inc. (Saskatoon, Canada) for NO3-, NH4+, total N, P, and K; nutrient supply rates are expressed as the total over these two periods.
Experimental warming can reduce soil moisture, which can affect litter decomposition (e.g., 6, 7), microarthropod abundances (e.g., 8), and microarthropod effects on decomposition (9). To account for this, we took measures of soil moisture at three locations within each plot at the beginning, middle, and end of the 2012 summer season using a HydroSense portable soil moisture probe (Campbell Scientific, Logan, Utah, USA).
1. Bliss LC & Matveyeva NN (1992) Circumpolar arctic vegetation. Arctic Ecosystems in a Changing Climate: An Ecophysiological Perspective, ed Chapin FS III JR, Reynolds JF, Shaver GR and Svoboda J (Academic Press, San Diego), pp 59–89.
2. Marion GM, Henry, G. H. R., Mølgaard, P., Oechel, W.C., Jones, M. H., and Vourlitis, G. (1993) Passive techniques for manipulating field temperatures in tundra ecosystems. in Passive techniques for manipulating field temperatures in tundra ecosystems (Hanover, NH).
3. Bloem J (1995) Fluorescent staining of microbes for total direct counts. Molecular microbial ecology manual, eds Akkermans ADL, Van Elsas JD, & De Bruijn F (Springer Netherlands), pp 367-378.
4. Frey SD, Elliott ET, & Paustian K (1999) Bacterial and fungal abundance and biomass in conventional and no-tillage agroecosystems along two climatic gradients. Soil Biology and Biochemistry 31(4):573-585.
5. Ingham ER & Klein DA (1984) Soil fungi: Relationships between hyphal activity and staining with fluorescein diacetate. Soil Biology and Biochemistry 16(3):273-278.
6. Makkonen M, et al. (2012) Highly consistent effects of plant litter identity and functional traits on decomposition across a latitudinal gradient. Ecology letters 15(9):1033-1041.
7. Aerts R (2006) The freezer defrosting: global warming and litter decomposition rates in cold biomes. Journal of Ecology 94(4):713-724.
8. Hodkinson ID, et al. (1998) Global Change and Arctic Ecosystems: Conclusions and Predictions from Experiments with Terrestrial Invertebrates on Spitsbergen. Arctic and Alpine Research 30(3):306-313.
9. Wall DH, et al. (2008) Global decomposition experiment shows soil animal impacts on decomposition are climate-dependent. Global Change Biology 14(11):2661-2677.
This research was partially funded under NSF Graduate Research Prgogram grant number 1106401.
Version 2: Added all the data tables to the data set. July 2018
Version 3: Added Amanda's ORCID and funding source. Aug 2018
Full Metadata and data files (either comma delimited (csv) or Excel) - Environmental Data Initiative repository.
Use of the data requires acceptance of the data use policy --> Arctic LTER Data Use Policy