A two-phase plant-soil feedback experiment to explore biotic influences on Phragmites australis invasion in North American wetlands

Plants can cultivate soil microbial communities that affect subsequent plant growth through a plant-soil feedback (PSF). Strong evidence indicates that PSFs can mediate the invasive success of exotic upland plants, but many of the most invasive plants occur in wetlands. In North America, the rapid spread of European Phragmites australis cannot be attributed to innate physiological advantages, thus PSFs may mediate invasion. Here we apply a two-phase fully-factorial plant-soil feedback design in which field-derived soil inocula were conditioned using saltmarsh plants and then were added to sterile soil mesocosms and planted with each plant type. This design allowed us to assess complete soil biota effects on intraspecific PSFs between native and introduced P. australis as well as heterospecific feedbacks between P. australis and the native wetland grass, Spartina patens. Our results demonstrate that native P. australis experienced negative conspecific feedbacks while introduced P. australis experienced neutral conspecific feedbacks. Interestingly, S. patens soil inocula inhibited growth in both lineages of P. australis while introduced and native P. australis inocula promoted the growth of S. patens suggestive of biotic resistance against P. australis invasion by S. patens. Our findings suggest that PSFs are not directly promoting the invasion of introduced P. australis in North America. Furthermore, native plants like S. patens seem to exhibit soil microbe mediated biotic resistance to invasion which highlights the importance of disturbance in mediating introduced P. australis invasion.

To determine the strength and direction of plant-soil feedbacks, total biomass was normalized between 228 within each of the plant species (introduced P. australis, native P. australis, and S. patens) between all soil 229 conditioning treatments (introduced P. australis, native P. australis, S. patens, and sterilized soil). Nine pairwise 230 plant-soil feedback indices were estimated, 3 for each plant species including two heterospecific plant-soil feedback 231 metrics (PSF away ) as well as one sterilized plant-soil feedback metric (PSF ster ) using the methods of Reinhart (2012).
232 For example, we calculated two PSF away values for introduced P. australis, one indicating PSF in S. patens 233 conditioned soil versus conspecific soil and one indicating PSF in native P. australis soil versus conspecific soil.
234 Thus, the target soil conditioning species PSF away varied depended on plant species identity. Soil feedback would be 235 calculated using the formula, PSF = ln(X C /X H ), where X C represents the mean dry biomass of plant species "X" 236 grown in conspecific soil and X H represents the mean dry biomass of plant species "X" grown in a heterospecific or 237 sterilized soil. By observing direction and magnitude of PSFs in each of the plant species we evaluated the soil 238 conditioning effects experienced by each plant species. Positive PSF away would indicate that a plant experiences 239 positive PSFs, growing better in conspecific conditioned soils than in heterospecific conditioned soils. Negative 240 PSF away would indicate that a plant experiences negative PSFs, growing better in heterospecific conditioned soils 241 than in conspecific soils. PSF ster indicates overall soil biota effects and can indicate whether pathogens or mutualists 242 drive PSFs. A plant that displays a negative PSF ster has a conspecific PSF dominated by pathogens while a plant 243 with positive PSF ster has a conspecific PSF dominated by mutualists.

246
We performed substrate-induced respiration assays as an estimate of active microbial biomass and overall 247 activity (Fierer et al. 2003). In July 2018 (~ 2 months before harvest) soil samples were taken from the feedback 248 growth pots, 8 samples per species x soil conditioning treatment, and 4 samples per species for sterilized soils 249 (n=84). Approximately 10 g of wet soil was taken from the top 4 cm of each pot, making sure to extract soil closely 250 associated to the rhizomes and fine roots. Soil was placed in 55 mL glass headspace vials with rubber septa. Two 251 gas samples were taken in 30-minute intervals for wet soil without the addition of yeast extract broth to establish a 252 baseline respiration rate without substrate. CO 2 concentration was determined using an infrared gas analyzer (Licor 10 253 Model LI-7000, LI-COR Biosciences, Lincoln, NE, USA) configured for small sample injection with a six-port 254 valve and a stream of N 2 as the carrier. Ten ml of yeast extract broth (Sigma-Aldrich, St. Louis, MO, USA) (30 g 255 mL -1 ) was added to each vial containing the wet soil (Barreto et al. 2018). Yeast extract broth was added to provide 256 an excess of labile carbon and nutrients which allowed for quantification of maximal metabolic activity. After the 257 broth was added, headspace was sampled four times roughly at 30-minute intervals to calculate SIR rates. After the 258 final headspace sample, the caps of the septum vials were removed, and the soil slurry was placed in a 50°C oven to 259 dry for 48 hr. Soil mass per vial was calculated by subtracting (vial + dried broth + soil) -(vial + dried broth). Rates  295 each treatment the direction and magnitude of the soil conditioning can be assessed to confirm or reject our 296 hypotheses. Aside from total biomass, Tukey's HSD tests were used to assess differences between plant species and 297 soil conditioning treatment in other analyses and traits (SIR, fungal colonization, shoot elongation/ emergence, and 298 BGB:AGB). Tukey's HSD tests were performed using the function "HSD.test" in R package "agricolae" 299 (Mendiburu 2010). All figures were created using the R package "ggplot2" (Wickham 2009). While the main 300 response variable in this study is total biomass, which we deem to be a proxy for overall plant performance, the 301 other response variables tested provide insight into possible mechanisms of action that ultimately affect variations in 302 biomass in plants subject to various soil communities. All data was analyzed in R studio (Version 1.1.414).

318
Introduced P. australis displayed positive plant-soil feedbacks when grown in soil conditioned by S.
325 australis soil communities may be dominated by pathogens. Additionally, the same microbes that illicit negative 326 growth effects in introduced P. australis seem to also affect the native lineage. S. patens experienced much greater 327 growth in both P. australis-conditioned soils (PSF paa = -0.26  0.06, PSF pa = -0.29  0.05) than in its "home soil" 328 ( Fig. 2). However, growth in sterilized soil was much lower than in its "home soil" (PSF ster = 0.40  0.12) (Fig. 2), 329 suggesting that it benefits greatly from soil microbial mutualisms, especially in both P. australis soils.

332
The rates of soil respiration differed significantly between plant species (Table 2). Soil-conditioning 333 species was not a strong determinant of soil respiration across species (Table 2). The interaction of species x soil-334 conditioning species was marginally significant in explaining soil respiration rates ( Interestingly, the soils associated with introduced P. australis that were inoculated 338 with conditioned soil (introduced/native P. australis & S. patens) respired much less than sterilized soil (Fig. 4).
339 This suggests lower microbial biomass or community suppression in introduced P. australis associated soils, 340 especially in the pots inoculated with conditioned soil. Lastly, the variation in soil respiration rates was relatively 341 high in S. patens associated soils compared to the other two species (Fig. 4)