Experimental design
We measured the responses of plant community biomass, nitrogen mineralization rate, soil decomposition potential and litter decomposability, to the three treatments nitrogen (N) addition and above- and belowground insect herbivory in a fully factorial block experiment. Each treatment was replicated once within each block, giving a total of 64 experimental enclosed one by one m plots (Fig. 1). The experiment was established in the spring of 2013 in an organic agricultural field (59° 44′ 27.9″ N, 17° 41′ 02.9″ E) near Uppsala, Sweden, with sandy soil, low soil carbon content and a total N content of 0.08%. In each of the plots, we established plant communities of nine common grassland species: four grasses (Agrostis capillaris L., Dactylis glomerata L., Festuca rubra L., Lolium perenne L.), three non-leguminous forbs (Achillea millefolium L., Leucanthemum vulgare Lam., Plantago lanceolata L.), and two leguminous forbs (Lotus corniculatus L., Trifolium pratense L.). The plant communities were established in late May 2013 from seedlings from seeds (Herbiseed, Twyford, United Kingdom) that were sown in the greenhouse 6–8 weeks prior to planting. All plant communities had the same species composition and all species were planted at equal starting densities in a 1 × 1 m plot of soil. Plants were spaced 10 cm apart in a 9 × 9 grid formation. The position of individual plants within enclosures was randomized for each enclosure, i.e. the planting scheme was not the same in any two enclosures. We manually weeded all enclosures (i.e. removed all plants emerging from the soil’s seed bank) on one occasion during the month that followed planting. In addition, we removed all weeds that we encountered during the subsequent harvests (see “Plant community biomass” section).
Part of the enclosures (sized 1 × 1 × 2 m) used in the experiment at the initiation of the experiment in 2013 (a), and the plat community inside one of the enclosures in 2015 (b).
The plant communities were enclosed above ground by a mesh net cage (mesh size 0.2 × 0.4 mm, anti-aphid net 20/12, Artes Politecnica, Schio, Italy) of 2 m height, and below ground, by a sheet metal frame of 0.5 m depth with a mesh net bottom. The net had a vertical zipper on one side that allowed entry to the cage. Prior to planting, the base frame was re-filled with soil from the field to a depth of 0.5 m. By refilling the enclosures with unsterilized soil we accepted an uncontrolled level of background herbivory, from insects dwelling in or hatching from of the soil. We considered this natural background herbivory to be preferable to the unrealistic conditions of a fully sterile soil.
N additions
The N treatment corresponded to 40 kg N ha−1 year−1, which is in the upper range of current N deposition levels for Western European grasslands2, were applied at three occasions,early July 2013, late June 2014, and early July 2015. We applied the N by dissolving ammonium nitrate in 10 l that was watered in the enclosure. Unfertilized enclosures received the same amount of water. No further watering of the plots was carried out.
Herbivory
The aboveground herbivory treatment consisted of adults of the grasshopper Chorthippus albomarginatus (De Geer), added in mid-July 2013 at a density of 10 individuals per enclosure (5 females and 5 males) and allowed to reproduce in the enclosures in subsequent years. We chose the study species as it is one of the most common grasshoppers in the area, and all specimens were collected within a radius of approximately 5 km from the experimental site. For the belowground herbivory treatment, we added wireworms, a common generalist root herbivores in European grasslands, which are the larval stage of the click beetle genus Agriotes spp, in mid-July 2013 at a density of 10 individuals per cage. In July 2014, we estimated grasshopper density by counting the individuals in each of the enclosures. Statistical analysis showed that differences in emerged number of nymphs among enclosures were unrelated to N addition (F1,21 = 0.095, p = 0.76) and belowground herbivory (F1,21 = 0.24, p = 0.63). Based on the counting, we adjusted grasshopper densities with in each treatment combination. We did this systematically, by dividing the enclosures into quartiles, based on grasshopper density. We did not adjust densities from plots ending up within the two mid-quartiles (19–31 grasshoppers per enclosure), but we transferred grasshoppers from enclosures within the highest quartile to those within the lowest quartile, to give 25 individuals per enclosure. At the time of adjusting the grasshopper densities, we also added 10 extra individuals of the wireworms to each enclosure assigned to this treatment. In 2015, we made no adjustments of herbivore densities. However, we assessed grasshopper densities by visual counting, and grasshopper densities were unaffected by N (F1,21 = 3.2; p = 0.09) and belowground herbivory (F1,21 = 0.7; p = 0.4).
Plant community biomass
Aboveground plant community biomass, henceforth denoted as total shoot biomass, was measured in mid-September 2013, 2014 and 2015, by harvesting the plants. The timing of the harvests corresponded to the peak of standing biomass in the communities. Because all plants renew their aboveground biomass annually, the harvested biomass approximates the annual aboveground production. At harvest, we cut all aboveground plants at 5 cm height above the soil surface. All collected plant material was brought to the lab and oven-dried at 65 °C for 48 h. To simulate the management of a semi-natural grassland, we conducted an additional harvest in mid-June. This harvest was, however, not repeated in 2015 as the plants were left so they could go to seed for a parallel study on plant reproduction.
Belowground plant community biomass, henceforth denoted as total root biomass, was assessed in September 2015 at the end of the 3-year experiment. We collected five soil cores from each enclosure, to a depth of 15 cm, using a cylindrical soil corer (ϕ10 cm). We pooled the five cores into one composite sample (volume of 5.9 dm3). The samples were kept refrigerated at 4 °C for 3 months before sieving, first with a 5 mm mesh and then a 2 mm mesh sieve. Prior to the second sieving the samples were left to dry for 3 days to facilitate the separation of soil and roots. Finally, the root samples were oven-dried at 65 °C for 48 h, and thereafter weighed.
Decomposition potential of the soil
To assess decomposition potential of the soil under the different herbivory and nitrogen treatments, we used the tea bag method that produces standardized decomposition rate estimates26. We used two types of tea, Lipton Rooibos and Lipton Green Tea (Unilever Belgium, Brussels, Belgium). In in mid-June 2015, we buried (depth of c. 8 cm) two bags of rooibos tea (recalcitrant) and two bags of green tea (easily degradable) in the soil of each enclosure. All tea bags were weighed before being buried. After 90 days, we dug up, dried at 70 °C for 48 h, and re-weighed the bags. The mass loss of the tea bag was used as an estimate of decomposition potential of the soil.
Decomposition of plant litter
To assess treatment effects on the decomposition of plant litter produced in the enclosures, we used dried plant material from the September 2014 harvest. The decomposition was measured using plant litter harvested from the same enclosure. The material was stored under cool, dry conditions from the harvest in September 2014 until May 2015, when litter for P. lanceolata, T. pratense, and D. glomerata was extracted for litterbag construction. We used these species as they based on the 2014 biomass harvest were dominant species, and as they represent three functional groups, namely grasses (D. glomerata), non-leguminous forbs (P. lanceolata) and leguminous forbs (T. pratense).
The litterbags were manufactured using polyamide mesh (Sefar Nitex 03—50/37, Sintab, Oxie, Sweden) with a pore size of 50 µm, which allows entry of bacteria, fungi and certain microfauna only. Approximately 1 g of dry litter of P. lanceolata, T. pratense, or D. glomerata was enclosed in a right triangular bag with 14 cm sides. After sealing and weighing we placed the bags (one for each plant species), on the soil surface in the enclosure from which the litter originated. We put out the litterbags in mid-June 2015, and collected them after 90 days, when we after dried at 70 °C for 48 h and weighed them. Litter mass loss over the experimental period was then used as an estimate of plant litter decomposability.
Nitrogen mineralization rate
We assessed the amount of inorganic N mineralized over the growing season in the enclosures with the buried bag technique26,27. A soil sample of about 300 g was taken from each enclosure in mid-June 2015 by extracting six evenly spaced cores (ϕ25 mm, 10 cm deep) and mixing them into one composite sample. We spitted the composite sample into two samples and put them in polyethylene bags. One bag was buried c. 8 cm below the soil surface in the middle of the enclosure, and the other was brought back to the lab for storage in a freezer (− 20 °C) until analysis. After 90 days, we recollected the buried bags, brought them to the lab and stored them in a freezer until further analyses. These samples were later analyzed for inorganic N content (g/kg of NO3/NO2 and NH4 respectively) using 2 M KCl extraction (Agrilab AB, Uppsala, Sweden). We estimate soil net N mineralization produced over the 90-day period by was subtracting the amount of nitrate and ammonium in the control bags from that in the buried bags.
Statistical analyses
We used linear mixed effects models28 to test how primary production, soil decomposition potential, N mineralization, and litter decomposability responded to N fertilization and above- and belowground herbivory.
Total shoot biomass was analyzed as a dependent variable of the fixed factors N and above- and belowground herbivory, including all possible interactions between the three factors. Year was included as a fixed factor, to account for variation among the harvests in the different years. Enclosure was nested within block in the random structure of the model. To account for autocorrelation of repeated measures within enclosures, we added a first-order autoregressive correlation structure to the model. Total root biomass was analysed with N and above- and belowground herbivory as fixed factors, with all possible interactions among the factors, and block included as a random factor. We also tested for treatment effects on the ratio between root and shoot production in 2015. Finally, we explored whether herbivory generated effects on aboveground shoot biomass were dependent on differences in soil decomposability (mass loss of read tea) and nitrate production by including these variables as covariates in the analyses.
In the analysis of mass loss of the red and green tea, we first calculated the mean mass loss for each tea type (i.e. the mean for the two bags of each type) in each enclosure. The mean mass losses of each tea type and the litter mass losses of D. glomerata, P. lanceolata and T. pratense were then analyzed in linear mixed effects models with block as random factor. In each model, N and above- and belowground herbivory were included as fixed factors, including all possible interactions between the three. For T. pratense, two enclosures were excluded as there was no T. pratense litter from 2014 to use (i.e. the species had gone extinct in these enclosures). A third enclosure was excluded from the analysis, as its amount of T. pratense litter from 2014 was only 0.1 g and skewed the analysis. All analyses were performed in R version 3.2.3 (2015).
Source: Ecology - nature.com
