1. Microbial communities drive plant litter breakdown. Litters originating from different plant species are often associated with specialized microbiomes that accelerate the breakdown of that litter, known as home-field advantage. Yet, how and how fast microbial communities specialize towards litter inputs is not known. 2. Here we study effects of repeated litter additions on soil microbial community structure and functioning. We set up a nine-month, full-factorial, reciprocal litter transplant experiment with soils and litters from six plant species (three grasses, three trees). We measured fungal and bacterial community composition, litter mass loss and home-field effects. 3. We found that repeated litter additions resulted in convergence in fungal community composition driven by litter functional group (trees versus grasses). Grasses enriched Sordariomycetes, while Tremellomycetes, Eurotiomycetes, and Leotiomycetes were favored by tree litter. Bacterial community composition, litter mass loss and home-field effects were not affected by litter incubation, but there was a relationship between fungal community composition and mass loss. 4. We conclude that repeated litter incubations can result in directional shifts in fungal community composition, while nine months of litter addition did not change bacterial community composition and the functioning and specialization of microbial communities. 5. Testing further how repeated litter inputs affect microbial functioning is essential for steering decomposer communities for optimal soil carbon and nutrient cycling.

Field sites

To test our hypotheses, we set up two controlled experiments. Soil and litter samples used in all experiments were collected from a long-term field site on the Veluwe, the Netherlands (Hannula et al., 2017; Kardol, Bezemer, & van der Putten, 2006; Veen, Keiser, van der Putten, & Wardle, 2018) situated between Ede (52°04′20″N, 5°44′12″) and Wolfheze (52°00′77″ N, 5°48′58″). We sampled soils from six independent locations within the Veluwe field site. All locations were situated on sandy soils. Mean annual temperature was around 10.7°C and mean annual precipitation approximately 840 mm (Veen et al., 2018) (Royal Netherlands Meteorological Institute (KNMI)). Each location consisted of a semi-natural grassland and a surrounding broad-leaved forest (Veen et al., 2018).

Experimental design

In the fall of 2016, we collected soils (8^th of December) and litter (3-27 October depending on timing of litter fall) from three grass species Agrotis capillaris, Festuca rubra, and Holcus lanatus and three tree species Betula pendula, Fagus sylvatica, and Quercus robur at each site. For the grass species we sampled soils within monoculture patches, for the tree species we sampled soils immediately underneath adult trees. For each plant species at each location we collected ~4 kg of soil from the top 10 cm by pooling ~6-10 individual soil cores. Soils were sieved over a 4-mm sieve. Soils were kept at 4°C until set up of the experiment. Litter, i.e., recently senesced biomass, was collected as a bulk sample from locations where the plant species were highly abundant. Litter was cut into 1-cm fragments and sterilized by gamma-irradiation (25 KGray). Using the soil and litter samples, we set up (i) a reciprocal transplant experiment where soils were incubated with all litter types during three subsequent periods of three month and (ii) we used the incubated soils in a subsequent litter decomposition experiment where soils were confronted with original litter (i.e., litter type as in the field) or with the new litter (i.e., litter type used during the incubation in experiment (i)).

1. Reciprocal litter incubation experiment

On 20 December 2016, we set up a full-factorial reciprocal litter transplant experiment with soils from the six replicated field locations. We filled microcosms with 240 g equivalent of dry weight soil. For each plant species at each location we set up seven microcosms which were incubated with 2 g of air-dried-sterilized plant litter from each plant species included in the experiment, according to a full-factorial design; and one mesocosm did not receive any litter (no-litter control). This resulted in a total of 6 replicates × 6 plant species × 7 litter treatments (i.e., 6 litter types, one no-litter control) = 252 microcosms (Fig. 1). Litter addition to the same pots was repeated after three months and after six months (Fig. 1). At each litter addition, litter and soil were gently homogenised; the amount of litter added was similar to average rates of litter fall in temperate ecosystems (Penuelas et al., 2007). At each 3-month litter incubation period we also added 1 g of litter in a nylon mesh bag (mesh size 0.9 × 1.0 mm), which was inserted in the soil, in order to calculate litter mass loss. Microcosms were incubated in the dark at 60% water holding capacity (WHC), 20°C and 80% air humidity. Microcosms were organized to a randomized block design, with each replicated site considered as a block. After each three-month litter incubation period, litter bags were harvested, cleaned and dried at 60°C to measure litter mass loss. Microcosms were weighed and watered to maintain WHC every two weeks. In addition, a soil subsample was collected at the start of the experiment and after each three-month incubation period to measure soil abiotic and biotic conditions (details under “Soil and litter measurements”). At the start of the experiment subsamples of the litter were oven-dried, to be able to correct mass loss calculations for the amount of moisture still present in air-dried litter.

2. Litter decomposition feedback experiment

At the end of the reciprocal litter transplant experiment, i.e., after nine months of incubation, we collected a soil subsample from each microcosm from experiment (i) to set up a litter decomposition experiment. Each soil sample was split into two subsamples, used to set up two new microcosms using 50 ml falcon tubes: one microcosm was incubated with the litter type from the plant species where the soil originated from in the field, the other microcosm was incubated with the litter type that the soil had been incubated with during the reciprocal transplant experiment (Fig. 1). Each microcosm received the equivalent of 0.50 g of dry soil and 0.50 g of dry litter (Keiser et al., 2011). This resulted in 252 soil subsamples × 2 litter types (i.e., the historical field litter type and the current incubation litter type; for the no-litter control samples we only incubated with the historical field litter) = 468 microcosms. We used small amounts of soil in this experiment in order to inoculate the soil microbiome, while minimizing effects of soil physical and chemical conditions on litter breakdown (Keiser et al., 2011). Microcosms were incubated in the dark at 20°C, 60% water holding capacity and 80% air humidity for three months and then freeze-dried to determine litter mass loss.

Soil and litter measurements

At the start of the experiment we measured initial soil and litter chemical properties from all soil and litter types. A soil subsample was dried at 105°C for 24 hours to determine soil moisture content. Soil organic matter content was determined by loss-on-ignition in a muffle furnace (550°C, 4 hr). We determined pH in fresh soil samples with a Mettler Toledo pH meter after shaking the equivalent of 10 g dry weight soil in 25 ml of demi-water for 2 hr at 250 RPM. Inorganic nitrogen content (N-NO_x and N-NH₄⁺) were determined with an autoanalyzer (Quaatro, Seal Analytical, Norderstedt, Germany) after shaking the dry weight equivalent of 10 g soil in 50 ml 1M KCl (2 hr, 250 RPM). Soil inorganic nitrogen content was determined again after 9 months of litter incubation, i.e., at the end of experiment (i). A soil subsample was dried at 40°C and ground and used to determine total soil C and N content with an element analyser (Flash 2000, Thermo Fisher Scientific, Bremen, Germany). Soil P availability was measured as P-Olsen and measured with an Auto­Analyzer (Quaatro, Seal Analytical, Norderstedt, Germany) (Olsen, 1954). Litter C and N content was determined with an element analyser (Flash 2000, Thermo Fisher Scientific, Bremen, Germany). Litter P content was determined by digestion with a 2.5% potassium persulfate solution. The obtained extract was measured colorimetrically with an Auto­Analyzer (Quaatro, Seal Analytical, Norderstedt, Germany) (Murphy & Riley, 1962). We determined lignin content using methanol–chloroform extractions and hydrolysis (Rowland & Roberts, 1994).

Data analysis

Before the analysis of the reciprocal litter incubation experiment (i) we standardized litter mass loss values to 90-day periods, in order to correct for differences in litter incubation length (range between 90-94 days). We then used a general linear mixed model to determine how soil source, litter type and experimental period affected litter mass loss. Site (1|site), experimental period and mesocosm (period|mesocosm) were used a random factors to control for the experimental set up (sites as replicated blocks in field and greenhouse) and for repeated measures, respectively. We tested the effect of soil source, litter type and experimental period on home-field advantage effects (expressed as the percentage of additional decomposition at home; ADH) using a general linear mixed model with site and experimental period as random factors (period|site). For the first experimental period, we tested how home-field effects differed between plant functional groups using a general linear model with transplant type (i.e., transplants between two grass species, two tree species or a grass and a tree species) as a fixed factor and site (1|site) as a random factor. Data from the litter feedback experiment (ii) were analysed from two different perspectives. First, we tested how litter incubation history (from experiment i) affected the mass loss and HFA of the original litter type, allowing us to analyse whether microbial lost affinity and thus HFA for the original litter type. We used general linear mixed models with mass loss and HFA as respective response variables, litter incubation history and litter type as fixed factors and site (1|site) as a random factor. Second, we tested how field history affected the mass loss of the incubation litter type, allowing us to analyse whether microbial communities developed affinity and thus HFA for the incubation litter type. We used general linear mixed models with mass loss and HFA as respective response variables, field history and litter type as fixed factors and site (1|site) as a random factor. We were not able to perform one full model for this experiment, because as a result of logistic constraints we did not include all full-factorial reciprocal transplants in this experiment (see Fig. 1 for set up). For all analysis we used post hoc Tukey HSD tests to test which treatments differed from each other. We explored for a normal distribution of residuals using QQ-plots and a Shapiro-Wilk test and homogeneity of variances using a Levene’s test. All data were analysed in R version 3.6 (Team, 2013) using the lme4 (Bates & Maechler, 2009) and lmerTest package (Kuznetsova et al., 2013) package.

For this publication we have uploaded:

1. a file containing the data

2. a READ ME file supporting the data file