Mapping Soil Microbiome Shifts During Pythium Disease Development in Soybean Seeds and Seedlings Under Different Management and Soil Conditions

Experiment Design Overview

See subsequent sections for more details on treatments, rhizoboxes, soil preparation, inoculation method, disease severity rating, sample collection, sequencing & analysis. 

Experiments were conducted at the Penn State Greenhouse Complex on the University Park Campus and Rock Springs Research Farm. Table 1 below provides details on the differences across experiments. All rhizoboxes only had one (1) seed in them and the round pots had six. Rhizoboxes were numbered and color tagged by treatment to ensure they were set up correctly with the proper soil and treatment amendments. Pots were included in the first experiment to provide additional post-emergence disease data in case significant germination failure or non-emergence occurred. However, this ended up not being an issue and the pots added to the intensity of the experiment for the small research team so these were not included in the subsequent growth room experiment. These pots do provide an additional point of comparison for field experiment row plants, which were also grown in close proximity to other plants, that underwent disease rating.

In the field experiment, the (3) plot sites were measured to the same size in a flat area in the middle of a larger field selected based on the soil pH. There were two experimental rows per treatment, with a buffer row of soybeans planted between them and on the outer sides. Holes were dug in the plot site at equidistant intervals within the rows and soybeans were planted in the rows between the boxes. The soil from each hole filled the specific rhizobox that was placed there in the sheath that was implanted in the hole. Excess soil filled the gaps between the sheath and the hole edges for a snug fit.

The greenhouse, which had windows opening to the outdoors, and field experiments had less environmental control over temperature and humidity. The growth room was fully indoors and allowed for environmental conditions more favorable to Pythium disease. For the indoor experiments, plants were watered as needed based on visual assessment and soil moisture readings, but only soil moisture treatment plants received daily watering. All plants were watered at the same time. Soil moisture was measured but only soil moisture treatments received water (daily) in the field experiment, as soybean farmers would normally rely on nature and not irrigate fields. Two Bluetooth Govee Dual Temperature & Hygrometer Readers were set on opposite sides of the table in the indoor experiment and data was downloaded after harvest on Day 14. Button sensors were set in the middle of each (3) field plot, as the Govee Devices are for indoor use. These devices allowed for a continuous record of temperature and humidity over the experiment timeframe. Plants were randomly assigned to a new location in the greenhouse each day to avoid location bias, although all plants fit on a single table in the indoor experiments and were in close proximity to each other regardless. In the field, plants were returned to their specific numbered sheath implanted into the soil. 

Plants were grown for 14 days from planting and harvested on day 14. After final soil samples were collected and plants were harvested, leftover soil was discarded and the rhizoboxes cleaned and bleach sterilized in preparation for the subsequent experiment. 

Table 1. Experiment Set-Up Overview

Treatments

This experiment considers four different management or soil conditions present in PA soybean production systems: soil type, pH level, soil moisture, and fungicide seed treatment. These were selected based on environmental factors that support Pythium disease development and are described in Table 2 below. 

Table 2. Treatment Overview

Rhizoboxes

The novel use of 3D printed rhizoboxes has the advantage of enabling multi-day non-destructive rhizosphere soil sample collection. Boxes can be use over multiple experimental runs with bleach sterilization, as the filament would melt in an autoclave. The filament is biodegradable polylactic acid (PLA), which is also more environmentally friendly to produce than other plastics. Boxes were designed using Tinkercad and 3D printed at the Media Commons at the Penn State University Library. 3D printing the boxes allowed for customization and uniformity of the final boxes. The design file can also easily be shared with other researchers for reproducibility. 

These opaque 2 (l) x 2 (w) x 5.5 (h) inch square boxes were designed based on the specific project needs, such as number of sampling holes, and expected soybean growth over the experiment timeframe for dimensions. A grid of small holes at the bottom allows for water drainage. Pieces of bleach-sterilized plastic straws cut in half and 'ribbed' (slightly cut in the middle so water or microbes could pass though) were placed in holes in the side of the box, protruding slightly on each side so they could be easily pulled out. Straw size selected was based on the soil sample size to have enough for DNA extraction (500 mg). Care was taken when adding soil to the boxes to ensure the straws were filled with soil. 

Twist ties were inserted into small holes at the top to create handles that allowed for easy handling of the boxes without disturbing the sampling straws protruding from the sides of the boxes. All rhizoboxes were covered on the sides with black plastic (cut from dog poop bags; top and bottom open) held on by rubber bands to help prevent outside contamination and help with moisture retention in the small boxes, especially once straws were removed for sample collection and open space/exposed roots were present within the rhizobox. 

The box was designed so that the top two straws were 1" beneath the soil surface, which is the depth soybeans are planted at, so that at planting the seed simply had to be placed between the top straws in the middle of the box to achieve uniform depth of planting. As the plants grew, the roots would grow straight down between the straws and out into the straws such that soil collected was relevant to where disease would be occurring. 

For the field experiment, sheaths were also designed that the rhizoboxes could be securely slipped into to stabilize them and provide space for the protruding straws not to be disturbed. The outer portion of the sheath prevented surrounding field soil from falling into the rhizobox hole, which would have made it challenging to replace the boxes back in the ground after sampling each day. Sheaths were also tagged with labels on twist ties to match rhizobox labels to ensure boxes were returned to their exact place in the field each day. Sheaths had holes in the bottom to ensure drainage. Some design features, like the triangles on the outer part of the sheath, were to minimize cost (calculated based on grams of filament used) while ensuring the sheath could still perform its essential functions and remain sturdy. 

Soil Preparation

For the indoor experiments, soil was collected from local fields based on the treatment requirements. The (2) pH treatment soils and the silty clay (proxy control) soil was collected from the Penn State Rock Springs Research Farm. This site did not have sandy soil, which is why for logistical purposes this treatment wasn't included in the field experiment which was exclusively held on the research farm. Sandy soil was collected from a production field in Patton Township, PA. 

Soil was collected into plastic tubs from a small area in the middle of the field and was taken from the top 6 inches which is relevant to where seedling growth and Pythium disease would be occurring. Soil cores were collected with a soil probe in a randomized zig-zag pattern within that smaller portion of the field. All four soils underwent testing at the Penn State Agricultural Analytical Services Lab for fertility, OM and N content, and soil particle size testing to provide context for final results and treatment values. Field soil was collected for treatments based on historical tests; however, actual pH levels varied from what was expected. Soil was stored in covered plastic storage bins in a cold room until it was ready to use for experiments. 

Soils were sifted to remove large rocks and plant material which could cause sample loss when removing the straw. Soil was then mixed with sand to a volume ratio of 20:80 to prevent compaction in the greenhouse which could influence sample collection and emergence issues unrelated to Pythium. Mixing the soil also made the initial microbiome as uniform in the treatment set boxes as possible. Most soil used in these experiments was unsterilized, but sterile soil for controls in the indoor experiments was autoclaved at 121 degrees Celsius for 30 mins in autoclavable biohazard bags after mixing in the sand. A small amount of bleach-sterilized perlite (<1 cm) was placed on top of the soil (not mixed in) in the indoor experiments after the seeds were planted to protect the topsoil. 

No sand was added in the field experiment, as soil from sheath holes was directly placed in the boxes. Care was taken to omit any large rocks or plant material without sifting the soil. No perlite was used, as this would not be used in a regular soybean field production setting. Field sites used were not the same fields from which soil was collected for the indoor experiments. 

Inoculation

Soils used in this experiment were collected from field sites with a history of soybean production and were expected to already have Pythium present. However, to ensure pathogen presence, soils were inoculated with Pythium sylvaticum (syn. Globisporangium sylvaticum) which is pathogenic on soybeans. The pathogen isolate originated from a field in Pennsylvania and was obtained from a collection at Franklin & Marshall College, with the species re-verified prior to being shipped for use in these experiments. P. sylvaticum was grown at 25ºC on half-PDA plates containing no antibiotics for four days prior to each experiment. This media was used to minimize its effects on the soil microbiome. Extra plugs were stored in the -80ºC freezer for later use. 

Rhizoboxes were inoculated by placing two 5 mm agar plugs 1 inch apart between the top two straws. Non-inoculated soils received media plugs without anything growing on them, in case the media impacted the microbiome. Boxes were topped off with soil and watered. Seeds were planted the next day in between where the plugs were placed between the top two straws by making a small hole with a probe. In the pots, a metal hook was placed into the soil to indicate where the plugs had been placed so that the seed could be set in the middle. In the field rows, plugs were placed about 1 inch apart down the entire row without any indication on their exact location. A hoe was used to make a small furrow for planting the day after inoculation and this inoculated soil was put back on top of the seeds. 

Disease Severity Rating

On day 14 plants were harvested and visually assessed for disease symptoms. Rating was done separately for roots and shoots of each plant. Since visual assessment is challenging both a 0-9 and 0-4 scale adapted from the literature and the same single evaluator for each experiment was used to limit bias. As shown in Table 3, root symptoms for visual ratings include germination failure, lesions, brown discoloration, and necrosis, while seedling symptoms focused on brown spots, chlorosis, and wilting or lost leaves. 

Table 3. Disease Rating Chart

Additionally, notes were made on when plants had emerged during the experiment to identify delays. Photos were taken daily from each set, starting when at least one plant in the treatment/control set emerged, to capture disease development and plant growth. Plant shoot and root lengths in mm were measured, which verifies that the plant roots went deep enough into the box that all straw samples would be relevant. Plants were then cut in half, with the roots and shoots put in separate paper bags labelled by plant number. These bags were put in the greenhouse drying oven for about 2 days before measuring root and shoot dry biomass in grams which might be impacted by disease severity. Plant lengths and biomass were assessed for all indoor experiment plants, both in pots and rhizoboxes. In the field, a representative plant was selected from the middle portion of the numbered row area for assessment. 

Disease ratings are linked to the microbiome in analysis and also were used to select soil samples for sequencing. A variety of analyses including significance tests, linear discriminant analysis, correlation and co-occurrence network analysis were conducted using a combination of the open-source software R and Netshift. These analyses are used to identify microbiome communities related to final Pythium disease severity. They also identify highly interactive “hub” microbes and determine their potential for being key microbiome members driving Pythium disease severity during this soybean developmental stage by quantifying directional changes in individual taxa interactions. ROAR, the Penn State Supercomputer, was used to allow the processing of all the microbiome data (over 300 samples) in a single batch within a reasonable amount of time. 

Sample Collection

Samples were collected daily over the 14-day experiment timeframe to capture continuous changes in the microbiome during this dynamic soybean growth period and Pythium disease development. Table 4 below describes the soil samples collected for this experiment to support the research objectives. The initial and final timepoints, Days 1 and 14, were hypothesized based on the literature to be the most important timepoints so more samples were collected those days. For the field rows, bulk soil was taken from down the sampling row area (about 1.5 ft between boxes) to create a representative bulk soil sample. 

For the indoor experiments, soil samples were collected in the greenhouse facilities. For the field experiments, boxes were brought indoors for straw sample collection in the lab on-site at the farm to limit contamination and for ergonomic reasons. In the first experiment, straws samples were put directly in smaller sterile long-term storage tubes. However, this was very time consuming and the subsequent experiments used longer sterile capped test tubes that the straws could immediately be placed into and later, when convenient, the soil in the straws could be transferred to the smaller tubes. Samples were stored in the -20ºC freezer at the end of each day until use for DNA extraction. 

Environmental data, including temperature & relative humidity, were continuously collected throughout the experiment using the Govee and button sensors previously mentioned. Soil moisture was taken prior to watering each day starting with Day 1. The probe was sanitized with ethanol between samples; however, measurements were not taken for any sterile soil controls to ensure they wouldn't get contaminated and it was assumed that each soil type would have similar water retention to their measured non-sterile counterparts. For the pots, probes were set in the center of the pot to collect the soil moisture reading. For the field rows, the probe was set in the center of the row sample site in a similar spot each time. 

Table 4. Soil Sampling Overview

Sequencing & Analysis

Due to resource constraints, a subset of samples expected to be the most valuable in answering the research questions were sequenced. Sequencing focused on the straw rhizobox samples for comparability and only non-sterile samples were sequenced since we want to track the natural microbiome under Pythium disease pressure. One plant of each treatment set was not sequenced at all and stored for back-up purposes. Plants included for sequencing were selected using standardized selection parameters across treatments based on experiment plant growth and disease severity.

All plants were sequenced for days 1, 7, & 14, which were hypothesized to be the most meaningful timepoints. Seven (7) days is about when emergence occurs for soybeans. All remaining samples are kept in -20ºC freezer storage in case they are needed for future research or analysis building off this project. Table 5 below shows the specific days each plant in a treatment had sequenced. 

Table 5. Sampling Scheme for Each Experiment

DNA extraction of (4) plates of soil samples was done using the vacuum pressure protocol of the Macherey-Nagel NucleoSpin 96-well Soil DNA Extraction Kit. Extract plates were sent to the University of Michigan Microbiome Core Facility, which offers more full-service sample preparation at a reasonable price per sample, for MiSeq sequencing of the 16S (bacteria) and ITS (fungi) regions. Samples with poor MiSeq results were removed prior to data processing. Remaining MiSeq results were processed using the dada2 pipeline in an interactive R Studio interface on the Penn State ROAR Supercomputer, which allowed the processing of all 300+ samples together in a reasonable amount of time. Various analysis such as Shannon diversity indices and relative abundance was used to visualize soil microbiome communities. Principal components analysis visualizes differences in communities based on environmental and soil characteristics, while principal coordinates analysis and significance tests highlight differences between treatments. Altogether, this genomics data was used to visualize the abundances and diversity of fungal and bacterial microbiome community members throughout this early soybean development period under Pythium disease pressure.