Evidence of plant-soil feedback in South Texas grasslands associated with invasive Guinea grass

17 Plant-soil feedback (PSF) processes play an integral role in structuring plant communities. In 18 native grasslands, PSF has a largely negative or stabilizing effect on plant growth contributing to 19 species coexistence and succession, but perturbations to a system can alter PSF leading to long20 term changes. Through additions of novel root exudates and litter which alter soil microbial 21 communities and nutrient cycling, invasion by non-native plants has a strong impact on 22 belowground processes with broad shifts in historical PSFs. Guinea grass, Megathyrsus 23 maximus, an emerging invasive in South Texas, can efficiently exclude native plants possibly 24 due to its fast growth rate and high biomass accumulation, but its impacts on belowground 25 processes are unknown. Here, we provide a first look at PSF processes in South Texas savannas 26 currently undergoing invasion by Guinea grass. We addressed the question of how the presence 27 of the invasive M. maximus may alter PSF compared to non-invaded grasslands. Under 28 greenhouse conditions, we assessed germination and growth of Guinea grass and the seed bank 29 in soil collected from native grasslands and grasslands invaded by Guinea grass. We found that 30 Guinea grass grown in soil from invaded grasslands grew taller and accumulated higher biomass 31 than in soil from non-invaded grasslands. Plants grown from the seed bank were more species 32 rich and abundant in soil from non-invaded grasslands but had higher biomass in soil from 33 invaded grasslands. In South Texas savannas, we found evidence to support shifts in the 34 direction of PSF processes in the presence of Guinea grass with positive feedback processes 35 appearing to reinforce invasion and negative feedback processes possibly contributing to species 36 coexistence in non-invaded, native grasslands. Future work is needed to determine the 37 mechanisms behind the observed shifts in PSF and further explore the role PSF has in Guinea 38 grass invasion. 39 Author-formatted, not peer-reviewed document posted on 01/06/2022. DOI:  https://doi.org/10.3897/arphapreprints.e86933

belowground processes with broad shifts in historical PSFs. Guinea grass, Megathyrsus 23 maximus, an emerging invasive in South Texas, can efficiently exclude native plants possibly 24 due to its fast growth rate and high biomass accumulation, but its impacts on belowground 25 processes are unknown. Here, we provide a first look at PSF processes in South Texas savannas 26 currently undergoing invasion by Guinea grass. We addressed the question of how the presence 27 of the invasive M. maximus may alter PSF compared to non-invaded grasslands. Under 28 greenhouse conditions, we assessed germination and growth of Guinea grass and the seed bank 29 in soil collected from native grasslands and grasslands invaded by Guinea grass. We found that 30 Guinea grass grown in soil from invaded grasslands grew taller and accumulated higher biomass 31 than in soil from non-invaded grasslands. Plants grown from the seed bank were more species 32 rich and abundant in soil from non-invaded grasslands but had higher biomass in soil from 33 invaded grasslands. In South Texas savannas, we found evidence to support shifts in the 34 direction of PSF processes in the presence of Guinea grass with positive feedback processes 35 appearing to reinforce invasion and negative feedback processes possibly contributing to species 36 coexistence in non-invaded, native grasslands. Future work is needed to determine the 37 mechanisms behind the observed shifts in PSF and further explore the role PSF has in Guinea 38 grass invasion. 39

Introduction 43
Invasive species are an increasingly widespread concern due to their negative impacts on 44 ecosystems and difficulty in controlling their spread (Assessment 2005, Pyšek   In South Texas, Guinea grass, Megathyrsus maximus (Jacq.) B.L. Simon and Jacobs, is emerging 78 as a problematic invasive (CABI 2021). A perennial bunchgrass native to Africa, Guinea grass 79 has been introduced in tropical areas globally as a pasture grass due to its fast growth, high 80 biomass accumulation, and stress tolerance, but these same traits also make it a successful 81 invader (Rhodes et al. 2021a). For instance, the fast growth rates and high biomass accumulation 82 of Guinea grass results in displacement of many native species through direct competition for 83 space and resources (Ho et al. 2016 invasion have focused on removal of Guinea grass with herbicides, burn treatments, and grazing 93 followed by reintroduction of natives from seeds or out-plantings, but results have been mixed 94 Our goals for this study were to assess the study system in South Texas for evidence of PSF in 107 native grasslands and grasslands invaded by Guinea grass specifically addressing the question: 108 how does the presence of invasive Guinea grass alter PSF compared to non-invaded grasslands? 109 We hypothesized that germination and growth of Guinea grass would be higher in soil from 110 invaded grasslands than soil from non-invaded grasslands due to an overall shift toward a 111 positive PSF in the presence of Guinea grass. In contrast, plants from the seedbank will not 112 experience a similar increase in germination and growth in soil from invaded grasslands possibly 113 due to inhibition by Guinea grass (Chou and Young 1975). In non-invaded, native grassland 114 soils, growth and germination of both native plants and Guinea grass will be lower than in soils 115 from invaded areas, but species richness of plants from the seedbank may be higher than in soils 116 from invaded sites due to the presence of negative PSF processes in native grasslands 117 between 4 to 8 km apart spanning an area of approximately 5.5 km 2 . Soil from the three sites 130 sampled in this study was composed predominantly of sand (mean 92% ± 1.8%) with minor 131 amounts of silt and clay (mean 5.7% ± 0.8% and 2.3% ± 1.5%, respectively). Two of the sample 132 sites were in grasslands that had remained intact at least since the 1980's, while the third site had Within each of three sites, we sampled soil from plots invaded by Guinea grass and non-invaded 138 plots (i.e., predominantly native with no Guinea grass present) that were located within 10 m of 139 each other to minimize the confounding effects of distance on soil microbial communities or soil 140 traits ( Supplementary Fig. S1). We collected two sets of soil from invaded and non-invaded sites: 141 a) bulk soil for use as the growth medium and b) soil for use as additional inoculum. For both 142 sets of soil, we removed the litter layer and excavated the soil using a hand trowel to a depth of 143 15 cm. Bulk soil was collected from two locations in each plot. For the additional inoculum, we 144 collected five soil cores from each plot with individual cores located approximately 1 m apart. 145 Additional inoculum soil was collected individually in plastic bags and stored in a 4°C fridge. 146 Bulk soil (hereafter referred to as whole-soil inoculum) was stored at room temperature in a 147 climate-controlled building (~20-22°C). Within one week of collection, we sieved all the soil 148 (i.e., whole soil inoculum and additional inoculum) using a 2 mm soil sieve to remove leaf litter 149 and plant roots. Between each use, the sieve was sterilized with 0.5% NaOCl for five minutes, 150 washed with tap water, and allowed to air dry. 151 152 For our experiment, we chose to use whole-soil inoculum due to concerns that autoclaving 153 impacts soil nutrient availability and composition/abundance of microbial communities. To 154 confirm the effect autoclaving has on soil nutrient availability, we conducted a small assessment 155 on soil nutrients in the whole-soil inoculum pre-autoclaving and after two autoclave times (30 156 minutes and 60 minutes). We found that autoclaving increased levels of phosphorus, sulfur, 157 sodium, and electrical conductivity with autoclave time (ANOVA results in Supplementary 158 Table S1, also see Skipper  Since we were unable to refrigerate the whole-soil inoculum due to its large quantity, we added 165 inoculum that was kept at 4°C to counter any changes in the microbial community in the whole-166 soil inoculum. For this, we created two sets of additional inoculums: a pooled inoculum referred 167 to hereafter as a mixed soil sample (MSS) and an unpooled inoculum referred to as individual 168 soil sample (ISS). To create the MSS inoculum, we pooled inoculum based on soil origin 169 (invaded or uninvaded grasslands) for each of the three sites to create a common inoculum that 170 was applied to replicates (n = 6 inoculum pools used for MSS treatments). For ISS inoculum, we 171 used distinct (i.e., unpooled) soil cores for each replicate.  Pots were randomized in the greenhouse to account for variation in temperature and lighting. We 179 matched the whole-soil inoculum and the additional inoculum by soil origin (site and invasion 180 status), i.e. MSS and ISS inoculum treatments from invaded sites were added to bulk soil also 181 from the same invaded site. Soil samples from each of the treatments were submitted for nutrient 182 analysis at the Texas A&M AgriLife Extension Service Soil, Water, and Forage Testing 183 Laboratory. Soils were analyzed for pH, nitrate, phosphorus, potassium, electrical conductivity, 184 calcium, magnesium, sodium, and sulfur (Schofield and Taylor 1955, Mehlich 1984, Rhoades 185 1984. 186

187
In each pot, we sowed approximately 0.015 g of Guinea grass seed (approximately 15 seeds) 188 collected from the same area and time in South Texas. Although we were unable to quantify the 189 seed bank, we standardized the amount of soil that went into each pot to normalize the seed 190 bank. During the sieving process, we homogenized the whole-soil inoculum based on site and 191 soil origin as described above, then placed the same amount of whole-soil inoculum and 192 additional inoculum as stated above into each pot. We visually assessed the sieved litter for seeds 193 to assess whether larger seeds were removed during soil sieving (i.e. size sorting of seeds), but 194 noted only plant leaves and roots in the material removed during sieving. 195 196

Germination and growth of Guinea grass 197
After three weeks, we counted the total number of Guinea grass seedlings and thinned them to a 198 single seedling per pot. We did not normalize Guinea grass seedling number as the number of 199 seeds put into each pot was normalized by weight (see Sampling and experimental design). We 200 monitored growth of these seedlings over the course of the experiment (14 weeks), after which 201 plants were carefully removed from pots to keep as much of the root intact as possible. We 202 measured the plant height at the end of the experiment, then separated the aboveground tissue 203 from roots at the root collar and placed both in a drying oven at 65°C for 3-5 days in labeled 204 paper bags. We measured the dry weight of both above-and below-ground tissue. 205 206 Germination and growth of seed bank 207 Plants germinating from the seed bank were monitored in the same pots as Guinea grass. We 208 monitored the total number of plant seedlings sprouting from the seed bank weekly. At the end 209 of the experiment, we counted the number of plants present within each pot noting how many 210 were monocots and dicots. We were unable to identify seedlings to species as the plants were 211 juveniles and did not have flowering structures. Therefore, to quantify species richness, we used 212 phenotypic differences to distinguish morphospecies within each pot (hereafter, referred to as 213 species richness). To measure dry weight (total biomass) of the seedbank community, we placed 214 above-and below-ground tissue in drying ovens at 65°C for five days before weighing. 215 216

Statistical analyses 217
All statistical analyses were conducted in R and code is available for reproducibility (see Code 218 availability). To assess the effect of soil origin (invaded or uninvaded grasslands) and soil 219 handling method on Guinea grass growth and germination, we used a mixed effect model to 220 analyze germination, height, root length, and dry biomass. We treated soil origin and soil 221 handling method as fixed variables and site as a random variable. We considered Guinea grass 222 germination as the total number of seedlings and did not normalize this number as we used the 223 same mass of seeds (0.015 g) per pot. We evaluated all data for normality and homogeneity of 224 variance prior to analysis. Germination, height, and biomass data were log-transformed prior to 225 analysis. Three pots had no Guinea grass growth and were removed from analyses. 226

227
The effect of soil origin and soil handling method on germination and growth of the seedbank 228 plant community was also assessed using mixed-effects models as above. Here we also treated 229 germination as the total number of seedlings that germinated as the amount of whole-soil 230 inoculum and additional inoculum used was the same across all treatments and replicates. As 231 above, all data were assessed to see if they met the assumptions for parametric analysis. 232 Germination counts and plant abundance were log-transformed prior to analysis, whereas species 233 richness and biomass were transformed using the formula log (x + 1). 234

235
To assess for differences in soil characteristics as a function of invasion, we used a t-test and 236 included only data from unautoclaved soil (n = 6 samples; 3 from invaded sites and 3 non-237 invaded sites). Electrical conductivity, phosphorus, and sulfur were log transformed prior to 238 analysis. 239 240

Soil nutrients 282
We found that no significant difference between soil nutrients in invaded and non-invaded sites, 283 although some nutrients trended higher in invaded sites ( Fig. 4; Supplementary Table S2). 284 285

Discussion 286
We conducted an observational study to compare the effect of soil from Guinea grass invaded 287 and non-invaded, native grasslands on the germination and growth of Guinea grass, as well as 288 plants emerging from the seed back. Our experiment presents novel data on PSF processes in the 289 mesquite savannas in South Texas, the impact of Guinea grass invasion on PSF in native 290 grasslands, and the response of seedbanks to shifts in PSF. We found that, consistent with our 291 hypothesis, soil from grasslands already invaded by Guinea grass had a positive effect on 292 conspecific growth with plants growing taller and accumulating more biomass than Guinea grass 293 grown in soil from non-invaded grasslands (Fig. 1, Table 1). In contrast, plants germinating from 294 the seed bank had higher species richness (delimited based on plant morphology) and abundance 295 in soil from non-invaded grasslands. The observed decrease in species richness and higher 296 biomass accumulation of plants from the seedbank in soil from invaded grasslands could indicate 297 a release from negative PSF processes in non-invaded grasslands (Fig. 2, Table 2). Interestingly, 298 we found evidence of a broad phylogenetic signal in the response of monocots and dicots to 299 invaded and non-invaded soil (Fig. 3) indicating that Guinea grass may have a stronger negative 300 impact on more closely related plant species. These results suggest the presence of distinct 301 patterns of PSF in invaded and non-invaded grasslands in South Texas with evidence of positive 302 PSF on Guinea grass in invaded grasslands and an overall negative PSF in non-invaded, native 303 grasslands. Although we did not condition soil under controlled conditions making it difficult to 304 assign the difference in the direction of PSF to the presence or absence of Guinea grass, the low 305 spatial distance between the invaded and non-invaded grasslands we sampled suggests a minor 306 role of environmental factors, such as precipitation and temperature, in driving these differences. 307 308 For a non-native to be a successful invader, it needs to be able to colonize, establish, and 309 accumulation. This result in combination with the observed higher biomass of plants from the 329 seedbank in soil from invaded grasslands suggests that either the microbial community or soil 330 nutrients play a role in re-enforcing invasion. Although soil nutrients were marginally higher in 331 invaded soil than in non-invaded soil (Fig. 4, Supplementary Table S2), these differences were 332 not statistically significant. As even small differences could still be biologically significant, the 333 effect of soil nutrients as a possible contributor warrants deeper exploration. Invasion is 334 generally found to be associated with shifts in nutrient availability and cycling (reviewed in 335 increasing phylogenetic distance (Zhang et al. 2021), such that species of monocots should be 369 more negatively impacted than dicots in invaded grasslands. When we assessed differences in the 370 effect of PSF on monocots and dicots, monocot species richness and plant abundance were 371 higher in soil from non-invaded sites than invaded sites, whereas dicots showed no difference 372 (Fig. 3). Although overall PSF in soil from invaded grasslands was positive, these results 373 indicate that in invaded grasslands monocot species that are more closely related to Guinea grass 374 phylogenetically experience negative feedback. These results raise questions that we will test in 375 the future, such as whether negative PSF is driving species coexistence in non-invaded 376 communities, whether the switch to an overall positive feedback mechanism in invaded 377 grasslands is due to nutrient availability, and whether allelopathy or pathogen accumulation is 378 suppressing other monocot species post-invasion. In ongoing research, we aim to parse out the 379 effect of Guinea grass invasion on soil nutrients, allelopathy, and soil microbial communities to 380 better understand how Guinea grass impacts PSF processes and how this varies across the 381 heterogeneous landscapes of South Texas. 382 383

Conclusions and future directions 384
We found evidence for strong differences in PSF as a function of invasion with negative PSF in 385 non-invaded, native grasslands and positive PSF in grasslands invaded by Guinea grass. 386 Negative PSFs in non-invaded grasslands were associated with higher species richness and 387 abundance of the native plant community possibly contributing to species coexistence in native 388 grasslands. We found evidence to suggest that positive PSFs observed in invaded grasslands are 389 due to a combination of increased nutrient availability and a release of allelopathic chemicals by 390 Guinea grass which could be reinforcing establishment of Guinea grass, although the 391 contribution of each needs to be explored further. Our results represent the first time PSF 392 processes have been studied in South Texas savannas and show how Guinea grass, an emerging 393 invasive within the southern United States, influences these processes reinforcing its own 394 invasion. from i) grassland invaded by conspecifics and ii) non-invaded grasslands dominated by native 576 species. All data shown are non-transformed. 577 578 Fig. 2 Seedling count (a), abundance (b), biomass (c), and species richness (d) of native plant 579 community when grown in soil from Guinea grass invaded and non-invaded grasslands. All data 580 shown are non-transformed. 581 582 Fig. 3 Plant abundance (a) and species richness (b) as a function of invasion and plant group. 583 Monocot species richness and plant abundance were significantly higher in soil from non-584 invaded sites than invaded sites (species richness: Kruskal-Wallis Χ 1 2 = 13.4, p = 0.0002; plant 585 abundance: Kruskal-Wallis Χ 1 2 = 18.1, p < 0.0001), whereas species richness and abundance of 586 dicots showed no difference. All data shown here are non-transformed. 587 588 Fig. 4 Soil characteristics as a function of invasion. None of the soil characteristics were 589 significantly different based on soil origin although in general soil nutrients and characteristics 590 were higher in soil from invaded sites. All data shown here are non-transformed. EC is electrical 591 conductivity. 592 593