Exotic-induced changes to soil have been proposed as a limiting factor in native restoration success, with short-term greenhouse studies demonstrating that native perennial grasses perform worse in soil conditioned by exotic annuals . Exotic annual grass roots concentrate in the top 30 cm of soil while native perennial roots extend over 1 m and so invasion reduces nutrient cycling, soil organic matter and water holding capacity in deeper soils . The exotic and native grasses have also been found to cultivate different soil food webs and microbial communities , which may have many direct and indirect effects on plant performance . Further, exotic annual seedlings outcompete native grass seedlings for soil moisture and light and so exotic vs native competition needs to be included to properly address the importance of PSFs in California grasslands. In this experiment, mixes of native perennial or exotic annual grasses conditioned soil for 11 years before a 2-year feedback phase where the native and exotic mixes were seeded and measured for multiple traits. The native and exotic mixes were seeded alone to test for general feedbacks, as well as together to address the role of competitor identity in the feedback. Both the conditioning and feedback phases of the experiment experienced extreme variability in precipitation. As with most PSF experiments, our experimental design focuses on the netfeedback effect of all plant-induced changes to the soil, and we can only make inferences about which mechanisms may drive these feedbacks. We hypothesize that 1) the exotic and native grasses differ in their effects on the physical and chemical properties of the soil and cultivate distinct microbial communities. We expect these differences to be particularly strong in the sub-surface soil, commerical grow racks where native roots are abundant, and exotic roots are sparse.
We also hypothesize that 2) the soil changes will lead to feedbacks in plant performance, and whether the feedback is positive or negative will vary across life stages. 2a) One possible outcome could be that both species groups perform best on their own soil, mediated by changes in carbon and nitrogen cycling, indicating that feedbacks are playing a role in the exotic invasion . Alternatively: 2b) despite the chemical changes to soil discussed above, plant performance for both native and exotic species will be dominated by negative plant-soil feedbacks due to the build-up of localized pathogens . Further, we hypothesize 3) that competition that occurs from growing the natives and exotics together will eclipse the effects of any feedbacks, as exotic annuals generally outcompete native seedlings .Measures in the conditioning phase address our first hypothesis, that native vs. exotic grassland communities differ in their effects on soil water holding capacity, soil organic matter, total C and N content, soil nitrogen cycling , and microbial community composition . Soil cores were collected in August 2018, after 11 growing seasons of soil conditioning. Soil samples were taken at four depths to capture effects due to differences in rooting depths between the exotic annuals and native perennials. Soil was sieved within 36 hours and stored at 4℃ from the time of collection until analysis. Time sensitive analysis of N cycling extractions and incubations were performed within 48 hours. Subsamples were stored at -20°C for microbial DNA sequencing for 26 months. Soil chemical and physical properties Water holding capacity was measured by saturating soil and determining the % soil moisture the soil retained after draining under gravity for 24 hours at 100% humidity . Soil organic matter was determined by combustion in a muffle furnace at 550°C for 4 hours . Total carbon and nitrogen content of dried, ground soil was analyzed with an elemental analyzer interfaced with a mass spectrophotometer by the UC Davis Stable Isotope Facility.
To grind the soil, air-dried samples were placed in a scintillation vial with grinding bars and then set on a roller mill for four days. To determine net mineralization and nitrification rates, we measured inorganic N concentrations in 5g of firesh soil as well as in 5 g of soil that was incubated for a week in the dark at room temperature. These soils were extracted with 25 mL of 2 M KCl , shaken on a mechanical shaker for an hour, filtered using pre-leached Whatman No. 1 filter paper, and stored at -20℃ until analysis on the spectrophotometer. From these extracts we quantified nitrate and ammonium concentrations following methods developed by Forster . To determine gravimetric soil moisture, within 24 hours of soil sampling, soil was dried at 105°C until reaching constant mass. Microbial DNA extraction, amplification, and sequencing DNA was extracted from the 88 soil samples using the Power SoilTM kit . The 16S rRNA and ITS2 region were amplified in a two-step PCR procedure, with the final pooled library quantified via qPCR and sequenced using 300-bp paired-end method with an Illumina MiSeq instrument in the Genome Center DNA Technologies Core, University of California, Davis. DNA extractions and library preparation were performed by the UC Davis Host Microbe Systems Biology Core Facility. Detailed experimental procedures and primers are found in Supplementary Methods and Tables S1.1 and S1.2.The sequencing data for both bacteria and fungi were analyzed as Amplicon sequence variants using the “dada2” package following the dada2 pipeline workflows . The SILVA and UNITE databases were used to taxonomically classify the bacteria and fungi sequences. Data were further processed using the “phyloseq” package for downstream analysis and raw sequence reads were normalized using the “metagenomeSeq” package , as rarefying reads has statistical concerns . Further ASV processing details are found in Supplementary Methods. To assess functional differences in the fungal community, functional guilds were assigned to the already taxonomically classified fungal ASV dataset using the FUNGuild database . Only ASVs with guild assignations of ‘probable’ or ‘highly probable’ were used for analysis on functional guilds, representing 43% of the taxonomically classified dataset. We simplified the guilds to the following: arbuscular mycorrhizae, ectomycorrhiza, plant pathogen, endophyte, and saprotroph, as well as plant pathogen – saprotroph, endophyte – saprotroph, and plant pathogen – endophyte – saprotroph which exhibit traits of multiple guilds. We excluded non-plant pathotrophs, orchid and ericoid mycorrhizae, and plant pathogen – endophytes due to low presence.To address our second hypothesis and determine whether exotic and native communities were structured by negative or positive feedbacks, we compared both native and exotic performance on soils that were cultivated by either native or exotic plants. Each group was grown alone and in competition with the other group to determine whether competition influences the strength of the feedback. Seeding treatments were applied to a subset of the Phase 1 plots, resulting in 8 native-soil and 8 exotic-soil plots. Phase 1 plots not used in Phase 2 were excluded due to the 2018 percent cover levels not meeting the original plot selection criteria . The 8 exotic-soil plots included 7 of the original 11 Phase 1 plots with another exotic-soil plot that met the selection criteria for Phase 1 but whose soil was not sampled.
To initiate Phase 2, all above ground vegetation and litter were removed from the plots shortly after the soil sampling of Phase 1. In December of 2018, after rains induced germination of the seedbank, the seedlings were killed with RoundUp ProMax . After waiting 10 days for complete herbicide disintegration in the soil, and directly preceding another germinating rain, vertical grow weed the original 1.5-m x1.5-m plot was split into three subplots in a design that gave equal area and perimeter to each subplot. The locations of seedling treatments were fixed , and each subplot was seeded with the dominant grasses found in the community treatments used in the conditioning phase: native community, exotic community, and native + exotic mix . In the 2nd growing season, two native-conditioned, native community subplots were excluded due to ground squirrel or flooding disturbance. We assessed a variety of responses encompassing the life span of the plants for two growing seasons to test how soil conditioning influences plant performance. Learning which life stages are most affected by plant-soil feedbacks will help determine the overall strength of the feedback and the ultimate impact on plant fitness . While the entire plot was seeded, only the inner 1.2-m x 1.2-m square was used for measurements, resulting in each subplot having an area of 0.48 m2 .Height, which provides a useful proxy for growth and avoids destructive sampling, was measured throughout both growing seasons to assess how soil conditioning affects timing of growth. In each subplot, eight individuals of each species were haphazardly chosen and measured. Each season, natives were measured at three monthly time points once they could be identified from the exotic seedlings. The exotics were measured only once each season, as they reach full height by the time they are identifiable by seed heads. Above ground biomass of the exotic seeded subplots was taken twice each growing season because phenology varies, and no one time point can capture peak biomass for all species. The native subplot was assessed only in the late season to minimize destructive sampling of the young perennials. Samples were clipped within a haphazardly tossed 10-cm diameter ring, oven dried at 50℃, and weighed. Below ground biomass was sampled within the above ground biomass ring, the same day as the second above ground sampling, with a 5-cm diameter core at 3 depths . Roots were washed from the soil, dried, and weighed. We visually measured percent cover of each species with the Daubenmire method at three monthly time points during each growing season, ending once peak flowering of all species was captured.To test our first hypothesis and determine whether the exotic and natives differ in their effects on water holding capacity, total %C and %N, C:N, soil organic matter, and net rates of mineralization and nitrification at different depths throughout the soil profile, we fit linear mixed effect models for each soil property . No response variable required transformation to meet assumptions of normality. The fixed effects for each model included soil conditioning , depth , and their interaction, while the random effects were plot and block . Analyses were performed with the “lme4” and “emmeans” R packages. To further test our first hypothesis and determine whether potential feedbacks may be driven by the microbial community rather than solely soil chemical or physical properties, we looked at the net impact of soil conditioning on fungal and bacterial community composition. Analyses were performed on the entire soil profile as well as each depth separately, as the variation attributed to depth may overshadow differences from soil conditioning, for not only do rooting depths differ but physical soil conditions vary across depths and influence composition independently of plant conditioning . Community composition was examined with permutational multivariate analysis of variance using the Bray-Curtis dissimilarity matrix on normalized reads, with soil texture and soil conditioning as fixed effects. Soil texture was included additively , but not in interaction with soil conditioning due to the unbalanced sample sizes. Community dissimilarity was tested for homogeneity of dispersion with the betadisper function from the “vegan” package and visualized with non-metric multidimensional scaling using Bray-Curtis dissimilarity matrix on 3 dimensions. Because previous studies in California grasslands found that native and exotic-dominated grasslands differ in fungal guilds and nitrifying bacteria , we focused on these key taxa in addition to the overall fungal and bacterial communities. We compared total relative abundance of each fungal guild and nitrifying family with Kruskal-Wallis tests on the normalized reads but assessed composition only for the nitrifying community, as fungal guilds had too few ASVs for adequate analysis.To evaluate whether exotic vs native conditioning of the soil results in a plant-soil feedback , and whether the feedback is influenced by the competitive environment , we fit models for each performance variable according to the type and distribution of the data. Exotic and native grass groups were analyzed individually, since we were interested in how each group’s performance varied with soil conditioning. While other model terms such as year and depth may be significant on their own, we only addressed them further if they significantly interacted with the soil conditioning and community treatments. The traditional feedback design involves pairing up individual plants in ‘home’ and ‘away’ soil and then calculating a feedback effect variable for each pair . However, due to the complexity of our experimental design, which involves multiple time points and depths, and an interaction with the competitive environment, we tested for potential feedback due to soil conditioning by model fitting with the original data .