Starting in the late 1880s in Germany, agronomists used steam to eliminate plant and animal pathogens found in soil (Mihajlović et al., 2017; Arancibia, 2020). The soil steaming method was introduced to the United States (US) in 1893 and for the first half of the 20^th century was predominately used for small scale production of high value crops such as tobacco, tomato, and greenhouse production of ornamentals (Johnson, 1946; Hansen et al., 2011; Arancibia, 2020). In the 1950s newly introduced chemical pesticides and fumigants became the main method for soilborne pathogen management (Wilhelm, 1966; Arancibia, 2020). Since the Montreal Protocol in 1987 there has been a slow yet steady phasing out of fumigants due to their broad toxicity and negative impacts on the environment and human health (Arancibia, 2020; Gambhir, 2020). The mandated disuse of methyl bromide for soilborne pathogen management in important crops resulted in an urgent need to find alternative, more sustainable phytosanitary approaches (Roux-Michollet et al., 2010; Samtani et al., 2012; Schweigkofler et al., 2014). Soil steaming disinfection kills pests and soilborne pathogens without the toxic side effects because water, without chemicals or other imputs, is turned into steam and applied to the soil at high temperatures (~70–90°C) (Katan, 2000; Fennimore and Goodhue, 2016; Arancibia, 2020). Recently, soil steaming for plant disease management has increased in demand and provides a promising management approach in agriculture (Atallah et al., 2011; Mihajlović et al., 2017). Steam is used to kill plant pathogens mainly in indoor settings, although the method is growing traction for use in high tunnel farming and is in the process of being tested for field crops (Schweigkofler et al., 2014; Arancibia, 2020). The American Plant Health Inspection Service (APHIS) reports steam disinfection as an official management option for treating soils in plant nurseries (USDA APHIS, 2020). The successful management of these pathogens lies at the intersection of agriculture, ecology, ecological restoration, and environmental remediation. There is an urgent need to understand the underlying science of how soil-steaming disinfection disturbs the soil microbiome and whether there are alternate trajectories for the microbial community diversity and persistence during soil recovery. Understanding these patterns and the underlying functions of these microbes that contribute to soil fertility (e.g., nitrogen and phosphorus cycling, organic matter) is critical for growers to improve soil and plant health.

A major challenge today is to understand how certain ecological disturbances, both biotic and abiotic, can shift the composition and abundance of microbes that constitute the soil microbiome. Soil microbiomes are susceptible to disturbance events (Shade et al., 2012) and their recovery is known to be impacted largely by abiotic factors such as temperature, salinity, pH, or moisture (Shade et al., 2012; Thakur and Geisen, 2019; Kaminsky et al., 2021). It is known that biotic factors affect the microbiome, but the impact of microbial interactions on soil microbiome dynamics is far less studied (Thakur and Geisen, 2019). As the soil microbiome recovers post-disturbance, it can reach pre-disturbance dynamics or a new alternative stable state where microbial communities persist that are different from the pre-disturbance communities (Shade et al., 2012). Steaming for pathogen management shifts the composition and abundance of microbes; the question that follows is: will the recovering microbiome reach the pre-steam stable state or a new stable state, and can the community trajectory (abundance and diversity) be impacted by the early colonization of pathogenic and/or beneficial microbes?

Previous soil steaming research provides insight into post-steaming microbial recruitment patterns. Diversity in the bacterial microbiome spikes immediately post-steaming, then diversity and abundance levels off within 1–2 months (Crandall et al., 2022). However, the soil community composition doesn’t necessarily resemble pre-disturbance conditions. Certain communities of bacteria, such as Firmicutes, are most abundant 1–2 weeks post-steaming; this is not surprising as Firmicute endospores have been shown to survive other high temperature disturbances such as forest fires or autoclaving (Kaminsky et al., 2021; Fox et al., 2022). Soil fungal microbiomes, in contrast, exhibit a shift in composition and diversity post-disturbance with certain taxa blinking in and out of existence in the soil microbiome (Crandall et al., 2022). For this study we focus on fungal microbiome dynamics. Our rationale is: (1) a marked pattern of succession is found within the soil fungal microbiome as compared to bacterial microbiomes and (2) there are fewer studies that track fungal microbiome recovery.

This work examines the top-down effect that Trichoderma harzianum and Fusarium solani have on the soil microbiome when introduced post steaming. The majority of work on microbiome recovery after a disturbance focuses on the bottom-up regulation that plants have on soil microbe dynamics (Thakur and Geisen, 2019). Recent literature identifies microbial interactions as regulators of the soil microbiome structure, particularly the competition for resources (Bahram et al., 2018). Predation and competition can be key determinants of diversity in a community, especially when the competitor is a fast growing virulent (hereafter “aggressive”) plant pathogen (Roy, 1997; Roncero et al., 2003; Chesson and Kuang, 2008; Schmitz et al., 2010; Fukami, 2015). The objective for this research was to investigate the impact of an aggressive fungal pathogen and beneficial fungus on the soil fungal microbiome as it recovers post-steaming.

Our model organisms were the plant pathogen Fusarium solani, the cause of root rot in soybean (Glycine max) and Trichoderma harzianum, a known beneficial to soybean (Roy, 1997; Kabir et al., 2022). It had been shown that Fusarium spp. have an impact on the soil microbiome. Studies using fumigation to facilitate the removal of Fusarium spp. suggest that soil microbiomes were able to rebuild a resistance to Fusarium spp. post-fumigation. This indicates that the initial Fusarium spp. infections (prior to steaming) were suppressing the emergence of resistance to a pathogen (Shen et al., 2018; Xue et al., 2019; Deng et al., 2021; Zhang et al., 2021). There are few to no studies that quantify how plant pathogens like F. solani colonize the soil and influence microbiome recovery after a disinfestation event, such as steaming, in-vivo. There is also a growing body of literature that investigates the interaction between microorganisms used as biocontrols for soilborne pathogens (Sood et al., 2020). Specific fungi in the genus Trichoderma show antagonistic interactions toward common and virulent soil pathogens such as F. solani (Daniel and Filho, 2007; Pimentel et al., 2020; Sood et al., 2020; Bellini et al., 2023). Trichoderma spp. can elicit a bottom-up effect on the soil microbiome due to their positive interaction with plants, but the focus of this study was on the soil microbiome pre-planting. How microbes interact with one another is context dependent and within this study the microbiomes were assessed in the context of recovery post-steaming (Zhou et al., 2021).

We asked: how does the soil microbiome recover after a steaming event when the soil is inoculated with a fungal plant pathogen and a beneficial biocontrol fungus? Does the presence of these two fungi change the composition, diversity, and function of the soil fungal microbiome? We questioned if T. harzianum, in its capacity as a beneficial, would mitigate the effects of F. solani. We hypothesized that the early introduction of both fungi would significantly alter the development of the soil fungal microbiome after steaming. We expected that T. harzianum would suppress the growth of F. solani when added together. Our last question was: Is there a significant effect on the microbial community composition when a beneficial microbe is added to the soil in order to suppress a pathogen? Due to its capacity to parasitize F. solani, we hypothesized that T. harzianum would decrease the abundance of F. solani regardless of when it was applied. The usefulness of steam in managing soil pathogens has been validated through other research studies, (Atallah et al., 2011; Schweigkofler et al., 2014; Mihajlović et al., 2017), but the impacts on the microbial community remain unclear. Our work provides insight into how growers and land managers can best use beneficial fungi such as T. harzianum post-steaming. The purpose of this research is to provide baseline information in order to build predictive models to understand how the biotic composition of the soil can either facilitate or hinder the potential for plant germination, establishment, and survival.

An experiment was conducted in the cold frame of a greenhouse (outdoor area) to understand how the fungal microbiome rebuilds in potted soil after steaming and to investigate the effects of the early introduction of a beneficial (T. harzianum) and the soilborne pathogen (F. solani). Field soil was collected from the Pennsylvania State University, Plant Pathology Farm. The top 15 to 30 cm of soil was harvested using standard shovels from the edge of a fallow field in early July 2021. The soil was collected from a single 3 m square patch of soil. Field soil was thoroughly mixed with sand and perlite at the Pennsylvania State University (PSU) College of Agriculture greenhouses for a final ratio of 3:1:1 respectively. The prepared soil mixture was dispensed into 120 four-inch (10.16 cm) pots. These pots were placed three pots by five pots in to eight shallow trays (51.44 x 31.12 cm). The trays were distributed evenly along the central steam line in a metal steaming table. The steaming table was covered with a tarp and the soil was steamed at a maximum temperature of 82°C (180°F) and was held at this temperature for one hour ( Figure 1 ) (Fennimore and Goodhue, 2016; Arancibia, 2020).

The pots containing steamed soil were left to cool for 24-hours under the steaming tarp ( Figure 1 ). The pots were then moved to an outdoor gravel cold frame structure behind the greenhouses ( Figure 1 ). The experimental trial was initially kept outside as aerial spore dispersal is one way that new microbes can colonize the soil after the steaming disturbance and we assumed that the airborne fungal spore density is naturally greater outside than inside (Warcup, 1955; Cho et al., 2018; Martinez-Bracero et al., 2022). The pots were partially protected from rain by a 1.2 x 1.2 meter cube ramada, with a 0.61-meter roof, constructed of PVC piping with a polycarbonate roof to prevent cross contamination from rain splash ( Figure 1 ). Treatments were first applied 24-hours after being placed in the cold frame (48 hours post steaming) ( Figure 2 ). The experiment ran in the cold frame over the summer of 2021 (July–September). Over the course of the experiment each pot was watered twice a week with 20 ml of tap water. In September 2021 the pots were moved into a greenhouse where they were maintained for an additional year (Sept. 2022).

There were six treatments with 20 replicates each. The control (C) treatment was steamed soil with no further amendments. The remaining treatments were: steamed soil with Fusarium solani (F), steamed soil with Trichoderma harzianum (T), steamed soil with F. solani and T. harzianum at time one (FTT1), steamed soil with F. solani and T. harzianum at time two (FTT2), and steamed soil with F. solani and T. harzianum at time three (FTT3) ( Figure 2 ).

All treatments, excluding the control, were inoculated with F. solani 48-hours post steaming (48hps) ( Figure 2 ). The F. solani was isolated from a Pennsylvania soybean field and pathogenicity was confirmed with an in-vitro seedling assay (personal communication, Dilooshi Weerasooriya, Jun 9, 2021). Fifteen Potato Dextrose Agar (PDA) culture plates of F. solani were flooded with 5 ml of sterile 0.01% Tween20 deionized (DI) water and scrapped with a glass rod to harvest microconidia. The concentration of microconidia was calculated using a hemocytometer and diluted with sterile DI water to a final concentration of 2e6. The microconidia suspension was applied at a rate of 1 ml per 100 g of soil. The pots averaged 329 g of soil, and the F. solani inoculum was applied at 3.5 ml per 4-inch pot (Leslie and Summerell, 2008).

The commercially available product RootShield® was used for the Trichoderma spp. treatments. The active ingredient in RootShield® is T. harzianum. RootShield® is commonly used in the nursery industry and its efficacy against Fusarium spp. in greenhouse production has been studied (Lamboy and Call, 2001; Rose et al., 2003; Roberti et al., 2012). RootShield® was applied at the recommended concentration for nursery pots (3–4 oz of powdered RootShield® per 100 gallons of water applied at 4–8 oz per 3–6inch pot). This resulted in 0.264 grams of powdered RootShield® per liter of DI water applied at 120 ml per 4-inch pot. Treatment FTT1 had T. harzianum applied in conjunction with F. solani. Treatment FTT2 had T. harzianum applied two weeks after F. solani, and treatment FTT3 had T. harzianum applied three weeks after F. solani ( Figure 2 ). This was a completely randomized designed experiment conducted in plant pots. To decrease the influence of weather and wind patterns, the pots were randomly moved within the cold frame 3 separate times over the course of the experiment.

The day that the soil was mixed and steamed, a pre-steam soil sample was collected prior to steaming the pots. After steaming, samples were collected 24 hours after each treatment was applied, two weeks after the last treatment was applied, and around a full year after the initial steaming. The number of samples collected for DNA sequencing to identify microbial communities was limited by funding, resulting in the selection of 3 of the sampling time points; 3 days post steaming, 37 days post steaming, and 426 days post steaming. Soil samples were collected using sterile plastic transfer tubes and stored in small plastic bags. The tubes were washed and autoclaved before each sampling. The 4-inch pots were gridded three by three, and at each sampling two quadrants were randomly selected for sampling. The transfer tubes were used to collect samples that transected the entire soil column of each pot. Samples were stored on ice and then transferred to long term storage in a −20°C freezer. These samples were stored for downstream DNA extraction, sequencing and analysis to identify changes in the soil microbiome over time ( Figure 2 ).

Frozen soil samples were thawed and mixed to fully homogenize each sample. A subset of each sample was weighed out to be extracted, 250 to 300 grams of soil per extraction. Extractions were done using the QIAGEN DNeasy® PowerSoil Pro Kit (QIAGEN, Carlsbad, CA, USA) to the manufacturers specification (Lear et al., 2018; Pearman et al., 2020). The DNA yield in our samples (ng/μL) DNA were calculated for all 161 extracted samples using a Qubit for broad range dsDNA. The DNA concentration data were analyzed using a one-way ANOVA and post-hoc Tukey analysis. We used metabarcoding (amplicon-based sequencing) to determine the composition and relative abundance of the fungal microbiome in our soil samples. The ITS PCR primers (Forward: 5’CTT GGT CAT TTA GAG GAA GTA A-3’, Reverse 5’-GCT GCG TTC TTC ATC GAT GC-3’) (Gardes and Bruns, 1993) were used for PCR amplification with HotStarTaq Plus Master Mix Kit (Qiagen, USA) under the following conditions: 95°C for 5 minutes, 30 cycles of 95°C for 30 seconds, 53°C for 40 seconds and 72°C for 1 minute, and a final elongation step of 72°C for 10 minutes. PCR products were checked using a 2% agarose gel. Samples were multiplexed using unique dual indices and pooled together in equal proportions based on their molecular weight and DNA concertation. Samples were purified with calibrated Ampure XP beads prior to preparing an Illumina DNA library. Sequencing was performed at MR DNA (Shallowater, TX, USA) on a MiSeq Illumina Platform following the manufacturer’s guidelines.

The MR DNA bioinformatics pipeline was used to generate Amplicon Sequence Variants (ASV) and taxonomy tables which were further processed using the phyloseq package in R (MR DNA, Shallowater, TX, USA). The pipeline sequences were first joined, then sequences with less than 150bp and/or ambiguous base calls were removed. The sequences were then quality filtered using a maximum expected error threshold of 1.0 and dereplicated and denoised to provide a final ASV. Taxonomy was assigned using BLASTn against a curated database derived from NCBI (www.ncbi.nih.gov). Once taxonomy was assigned using the pipeline, we employed the phyloseq package in R to filter, sort, normalize, agglomerate, and graphically and statically analyze the genera counts in order to determine the effects of our treatments on the composition of the soil fungal microbiome (McMurdie and Holmes, 2013). To avoid spurious taxonomic designations, genera with a prevalence of less than 0.0000005 were filtered out, removing 51 genera. For abundance counts the top 20 genera were used for graphical analysis and the low abundance genera were grouped into a single category. The alpha diversity was graphically analyzed using the estimate_richness function for the observed index, Shannon index and Inverse Simpson index. We ordinated the genera data using Bray-Curtis distances in a principal coordinates analysis (PCoA). Statistical differences were analyzed by pairwise PERMANOVAs using the pairwiseAdonis package in R (Martinez, 2020). Three different sets of PERMANOVAs were conducted to make specific comparisons. The first PERMANOVA compared the effect of treatment, combining the treatment data from each timepoint. The second set of PERMANOVAs compared each treatment across time by creating one PERMANOVA for each treatment. Lastly, a set of three PERMANOVAs was run, one for each time point, to compare the treatments at each time point independently. Finally, in order to understand what the broad ecological functions were of the fungal microbiomes found in our experiment, we employed the FUNGuild tool (Nguyen et al., 2016). FUNGuild is an open annotation community resource tool based in python that is used to parse fungal ASVs by ecological guild (e.g., plant pathogen, saprotroph, beneficial) and draws from guild assignments from the literature. This pipeline functions independently of sequencing platforms and analysis pipelines (Nguyen et al., 2016). This database is accessible at https://github.com/UMNFuN/FUNGuild.

In order to assess plant response to the altered fungal microbiomes, a pilot soybean germination experiment was conducted. A single soy seed was planted in each pot after the final microbiome sampling (1 year post-steaming). Susceptible seeds of the SC9277R variety were used. A subset of seeds were destructively sampled 18-hours post-planting for further analysis in a different study. The remaining 12 replicates were visually observed until successful germination. Data was not collected post germination as the pots were too small to support plant growth. The percent germination rate and days to germination were both compared via one way ANOVA.

This experiment was conducted over a 14-month period. Specifically, the samples were collected at 4 time points: pre-steam (after the soil was mixed, immediately before steaming), 3 days post steaming, 37 days post steaming, and 426 days post steaming (hereafter, dps) ( Figure 2 ). We compared the taxonomic similarity between fungal soil communities using PERMANOVA pairwise comparisons.

The first PERMANOVA compared the effect of treatment on fungal community composition, combining the treatment data from each timepoint. This PERMANOVA showed that almost all of the experimental treatments (including the mock fungal community and the negative controls) were significantly different from one another in taxonomy. However, treatments that were not significantly different from one another were those that contained the pathogen F. solani: Fusarium (F) vs. Fusarium*Trichoderma Time 3 (FTT3), Fusarium*Trichoderma Time 1 (FTT1) vs Fusarium*Trichoderma Time 2 (FTT2), FTT1 vs FTT3, FTT2 vs FTT3.

The second set of PERMANOVAs were conducted to compare individual treatments across time points, including the pre-steam time point. These PERMANOVAs revealed that regardless of treatment, all steamed soil was significantly different than the pre-steam samples at every time point (0.001 ≤p≥ 0.036).

The third set of PERMANOVAs were employed to run pairwise comparisons between each treatment to quantify if fungal community composition was different at each time point during soil recovery (3dps, 37dps, and 426dps). At all three time points post-steaming, all pairwise comparisons were significantly different from each other except the treatments where F. solani was added. At 3dps F was not significantly different than FTT2 or FTT3, and FTT2 and FTT3 were not significantly different. At 37dps the F was not significantly different than FTT1 and FTT1, FTT2, and FTT3; these in turn were not significantly different from each other. By 426dps none of the fungal communities containing F. solani were significantly different from each other. At 426dps, fungal communities in F were significantly different than the control (steamed-only) (p = 0.001) while fungal communities found in the Trichoderma-only (T) treatment were only moderately different from the control (p = 0.024) ( Table 1 ).

A principal coordinate analysis plot (PCoA) was used to visualize whether the fungal microbiomes found within each of the treatments were similar to one another. We added 95% confidence intervals to our ordination as ellipses. The first PERMANOVA revealed no differences between the FTT1, FTT2, and FTT3 treatments. In the PCoA plot the three ellipses for these treatments were visually stacked upon each other, indicating that these communities were more similar in composition to each other than other microbial communities ( Figure 3 ). The F treatment was also grouped within these three treatments in the PCoA ( Figure 3 ). The F treatment was not significantly different than FTT3, but was from FTT1 (p=0.016) and FTT2 (p=0.035). However, these p-values maybe be skewed because two F replicate samples were clearly outliers. Without these outliers the F treatment may be more similar to FTT1, FTT2, and FTT3. The pre-steam replicates are grouped the tightest together, with the narrowest 95% confidence interval ( Figure 3 ). Although statistically different, the control treatment (steam only) and T treatments grouped together and were clearly distinct from all treatments containing F. solani ( Figure 3 ).

We created relative abundance bar graphs (rainbow plots) of the top 20 most abundant genera to visualize which fungal genera were present in our samples. This analysis was to determine the changes in the fungal microbiome over time and across treatments. The public database FUNGuild was used to assign the genera to ecological guilds; we conducted this analysis in order to better determine changes in broad ecological roles and functions (e.g., pathogens, saprotrophs, beneficials) of the soil fungal microbiome (Nguyen et al., 2016) ( Supplemental 1 ).

Before the soil was steamed there were four dominant genera (Mortierella sp., Fusarium sp., Alternaria sp., and Phallus sp.) and four genera that were naturally present at low abundancies (<5%, Trichoderma spp., Zopfiella spp., Spizellomyces spp., and Mycrothecium spp.) ( Figure 4 ). At time 3dps the four dominant pre-steam plant pathogen genera were present in the untreated control: Alternaria spp. was no longer found to be a dominant genus and Cladosporium spp. appeared. The FUNGuild analysis determined that the fungi present in the pre-steam soil and the control at 3dps were categorized as saprotrophs, endophytes, plant pathogens, fungal parasites, and other guilds that comprised a combination of these ecological roles ( Supplemental 1 ). A pairwise PERMANOVA returned an adjusted p-value of (0.001) indicating a significant difference between pre-steamed soil and the control at 3dps. The Shannon diversity and inverted Simpson indices were both higher in the pre-steamed soil than the control at 3dps ( Supplemental 2 ). The 95% confidence interval did overlap in the PCoA ( Figure 3 ).

At 3dps the Trichoderma, Fusarium, and combination treatments all had relative abundances that were dominated by a single fungus. The Fusarium, FTT2, and FTT3 treatments all had Fusarium at >75% relative abundance. The Trichoderma treatment had Trichoderma spp. at >75% relative abundance. The FTT1 treatment had Fusarium spp. at >50% relative abundance and Trichoderma spp. >25% relative abundance ( Figure 4 ). The guilds present in these four treatments at 3dps were the same guilds present in the control and pre-steam treatments ( Supplemental 1 ). The Trichoderma treatment was significantly different than the pre-steamed soil (p=0.002), as was the Fusarium treatment (p=0.004), FTT1 (p=0.005), FTT2 (p=0.007), and FTT3 (p=0.002) ( Table 1 ). At 3dps the Fusarium, FTT2, and FTT3 treatments were not significantly different from each other, but all other treatments were (0.007 < p > 0.014) ( Table 1 ). The inverse Simpson index was low for all five treatments. The Shannon index was lower for all five than for the pre-steam and control treatments but it was slightly higher for Trichoderma ( Supplemental 2 ).

At 37dps, in all treatments where F. solani was applied, we found members of the Fusarium genus to have the highest relative abundance of any genus in the fungal microbiome. All treatments where both T. harzianum and F. solani were applied showed that Trichoderma spp. exhibited the second highest relative abundance ( Figure 4 ). At 37dps all treatments with Fusarium were the most similar. The Fusarium treatment was not significantly different than FTT1 or FTT3 ( Table 1 ). All other pairwise comparison resulted in significant differences (0.002 ≤ p ≥ 0.029) ( Table 1 ). When examining the functional guild data for the Fusarium treatment, plant pathogens constituted nearly 75% of the relative abundance of all taxa within the fungal microbiome ( Supplemental 1 ). The remaining microbiome comprised almost solely saprotrophs and possibly other plant pathogens. Endophytes, ectomycorrhiza, and arbuscular mycorrhiza were present at low relative abundancies <5% ( Supplemental 1 ). The only notable difference between the Fusarium treatment and the FTT1, FTT2, and FTT3 was the relative abundance of endophytes, which was attributed to the Trichoderma spp. The control (steam-only) had saprotrophs, plant pathogens, and “possible” plant pathogens at >75% relative abundance. The Trichoderma treatment possessed endophytes at >25% relative abundance (which was the Trichoderma spp.) and the relative abundance of saprotrophs, plant pathogens, and possible plant pathogens, dropped <75%. There was an increase in the relative abundance of arbuscular mycorrhiza compared to the control ( Supplemental 1 ).

By 426dps all soil treatments where F. solani was applied had roughly 40% relative abundance of F. solani ( Figure 4 ). When applied on its own to the soil, F. solani held a higher relative abundance than T. harzianum applied on its own. At 426dps, the Trichoderma treatment had a relative abundance of Trichoderma spp. of 25% ( Figure 4 ). There was no significant difference between any of the treatments that contained F. solani, while all other pairwise comparison resulted in significant differences (0.001 < p > 0.002) ( Table 1 ). The control treatment and Trichoderma treatment were significantly different but at a much higher p-value than all other comparisons (p=0.024) ( Table 1 ). At 426dps the diversity indices were evenly distributed across soil treatments ( Supplemental 2 ). The mean Shannon index was roughly even for the Trichoderma treatment and the control ( Supplemental 2 ). The inverted Simpson index was low for all treatments at 426dps, but it was just slightly higher for both the Trichoderma and control treatments ( Supplemental 2 ). Across all treatments there was a rise in saprotrophs and plant pathogens as the soil recovered from steaming ( Supplemental 1 ).

Soybean seeds were planted in the soil toward the end of the experiment at 480dps to measure both germination rate and time to germination for each treatment. Percent germination was equal across all treatments. A one-way ANOVA was conducted to access if there was a difference in days to germination from planting. Although not statistically significant, we observed that within the Fusarium treatment, the soybeans took longer to germinate (median time was 2 days longer) compared to the rest of the treatments ( Supplemental 3 ).

There was a significant difference in DNA yield (ng/μl) between the sampling time points (p=2e-16). However, there were no significant differences found within treatments or treatment interactions between the sampling time points. The Tukey Test analysis showed significant difference between each sampling timepoint ( Supplemental 4 ). Specifically, the pre-steamed soil had an average DNA yield of 80.16ng/μl, soil sampled 426dps had an average DNA yield of 21.74ng/μl, soil sampled 37dps had an average DNA abundance of 10.79ng/μl, and soil sampled 3dps had an average DNA abundance of 3.01ng/μl ( Supplemental 4 ).