A Comparison of the Microbiome Population Induced by Different Microbial Foods for the Production of a Safe and Effective Aerated Compost Tea

Compost Tea Microbiology

• Feed type, not brew time, was the main driver of the compost tea microbiome. Molasses alone and molasses with humic acid were the only feeds rated non-pathogen-promoting and effective at both 12 and 24 hours, while every kelp-containing tea was pathogen-promoting by 24 hours.

  We sent 33 compost tea samples to the lab. Each tea came from the same base compost, but we fed the brews different things (alfalfa, molasses, kelp meal, fish hydrolysate, humic acid, and combinations of these). Samples were taken at 12 hours and 24 hours of brewing, plus one sample of the starting compost before brewing. AggregoData returned 2,280,904 DNA reads in total (38,239 to 77,760 per sample), classified down to genus.

  1. Big picture of "who's there"

  The lab looked at the bacteria by reading their DNA. In every tea, most of the bacteria belonged to a few big groups (phyla): Proteobacteria, Firmicutes, Bacteroidota, and Actinobacteriota.

  These groups showed up in all the teas, no matter what we fed them. The unbrewed compost (time 0) already had a mix of these groups, with Proteobacteria making up only about a fifth of it. Once we started bubbling and feeding the teas, the balance shifted toward Proteobacteria, but the same major groups stayed present.

  

  Figure 1. Relative abundance by phylum, grouped by brew time (0, 12 and 24 hours). Proteobacteria (light blue) rose from about 22% of the unbrewed vermicompost to as much as 85% in some fed teas. The unfed and humic acid teas stayed closest to the starting compost.

  
  

  Figure 2. Relative abundance by genus. In the unbrewed vermicompost, about 64% of reads fall in "Other", meaning many small genera. Fed teas are dominated by a handful of named genera, chiefly Acinetobacter, Pseudomonas, Aeromonas and Fictibacillus.

  
  
  
  

  Figure 3. Heatmap of phylum abundance (log10 read counts; grey = not detected). Rare phyla present in the vermicompost disappear from many of the brews, most often at 24 hours. 

2. Safety and effectiveness rating of every tea

Each tea was rated with the DNA-based definitions under Defining Terminology.
Values are the percent of DNA reads in each sample, shown as 12 hr / 24 hr.
NPP = non-pathogen-promoting or "safe" as per our definition above
PP = pathogen-promoting.

Source: AggregoData Bacterial Microbiome Report (October 2024) and genus-level Krona data. Salmonella and Enterococcus were not detected in any sample.

• 9 of 30 fed tea samples were non-pathogen-promoting and effective.
  Only molasses alone and humic acid + molasses qualified at both brew times.
  Their PGPB share (57 to 61%) came mostly from Fictibacillus (30 to 42% of reads),
  with Enterobacteriaceae at or below the unfed tea.
• 13 of 30 fed tea samples were pathogen-promoting.
  Kelp was in 8 of them and fish hydrolysate in 6.
  Every kelp-containing tea was pathogen-promoting by 24 hours, driven mostly by Klebsiella and Enterobacter.
• Molasses plus fish hydrolysate was the worst tea in the study, at 1.38% and 2.01% Enterobacteriaceae, 22 to 26 times the unfed tea, and was dominated by Aeromonas (45% of reads at 12 hours).
• Humic acid alone kept Enterobacteriaceae among the lowest and the community closest to the starting compost,
  but it lifted PGPBs only to 5.8%, well short of the 30% effectiveness line.
  Alfalfa alone came closest to the line without crossing it (26 to 29%).

3. What changed with different feeds

When we zoomed in to the genus level:

• Common names that showed up across samples included the following:
  • Acinetobacter
  • Pseudomonas
  • Aeromonas
  • Fictibacillus, and others. Their amounts changed from treatment to treatment, but they kept showing up across the board.
    
    
• Humic acid teas looked most like the unfed tea and the starting compost.
  Their top genera were the same as the control (Gimesiaceae, Amaricoccus), with a modest lift in Pseudomonas. On the PCoA(Principal Coordinates Analysis) plot Fig. 11 below,  their dot markings are shown close together, meaning they have very similar mixes of bacteria.  
• Molasses teas were taken over by Fictibacillus, a spore-forming genus identified as a potential plant growth promoter, which made up 30 to 42% of the bacterial reads in molasses alone and humic acid + molasses.
  These 5-gallon buckets were strongly enriched in beneficial bacteria while staying low in Enterobacteriaceae.
• Fish hydrolysate and kelp combinations were taken over by Aeromonas (up to 45% of reads) and Acinetobacter (up to 43%). Acinetobacter is counted as beneficial; Aeromonas is not, since some of its species are opportunistic pathogens.
  These tea mixes also carried the most Enterobacteriaceae.

What the main beneficial bacteria do. Three genera made up most of the PGPB share in our teas. In each, helpful and harmful species share the same genus, and our sequencing identified bacteria only to genus, so these are what the genus can do, not proof of what our teas contained.

Because the molasses result rests mostly on Fictibacillus, it is the least certain of our effectiveness findings. AggregoData's new species-level pipeline could settle which species were present.

The Krona plots below show each sample's community from phylum (inner ring) to genus (outer ring). A balanced tea looks like a wheel of many thin slices; a tea that one group has taken over shows one or two wide wedges.

Figure 4. Unbrewed vermicompost (time 0). The most even wheel of any sample, no genus above about 6%, and no Enterobacteriaceae detected. Shannon 6.06.

Figure 5. Unfed tea (No Feed Control), 12 hours.
The reference community: many small slices, no genus above about 4% (Gimesiaceae, Blrii41, Amaricoccus). Enterobacteriaceae 0.06%, PGPBs 3.9%, Shannon 5.22.

Figure 6. Humic Acids, 24 hours (Non-pathogen-promoting).
Nearly the same community as the unfed tea, with Pseudomonas lifted to 3%.
Enterobacteriaceae 0.03%,
PGPBs 5.8%, Shannon 5.96, the most rich tea in the study.

Figure 7. Alfalfa Meal, 12 hours (Non-pathogen-promoting).
Acinetobacter (12%) and Pseudomonas (10%) grow, but the rest of the wheel stays broad.
PGPBs 29.0%, just under the 30% effectiveness line. Shannon 5.54.

Figure 8. Molasses, 12 hours (Non-pathogen-promoting & Effective).
Fictibacillus, a potential plant growth promoter, takes 42% of the tank.
PGPBs 61.4%, Enterobacteriaceae 0.04%, Shannon 4.28.

Figure 9. Kelp Meal, 24 hours (Pathogen-promoting).
Fictibacillus (36%) and Cellvibrio (23%) dominate;
Shannon 3.99 is the lowest in the study, and Enterobacteriaceae reach 0.44%, mostly Klebsiella.

Figure 10. Molasses + Fish Hydrolysate, 12 hours (Pathogen-promoting). Aeromonas (45%) and Acinetobacter (19%) take over, and Klebsiella is visible as its own 1% slice. Enterobacteriaceae 1.38%, rising to 2.01% at 24 hours.

Interactive Krona plots for all 33 samples are available when you download the HTML files on this drive.

4. 12 hours vs 24 hours

• Some recipes looked fairly similar at 12 and 24 hours.
• In other recipes, a few bacteria became much more dominant by 24 hours, which likely means they liked that particular food and had more time to multiply.
• Enterobacteriaceae rose between 12 and 24 hours in 10 of the 15 fed teas. They held flat or fell in 4 of the 5 humic acid teas (humic alone, with alfalfa, with molasses, and with fish hydrolysate) and in molasses alone. Dr. Jones notes that putting compost into a liquid likely favors enteric organisms to some degree, since many of them can swim.
• Overall, time mattered, but feed type mattered more. On the PCoA plot, the teas mostly grouped by what they were fed, with 12- and 24-hour samples from the same feed sitting closer together than samples from different feeds.

So, within the 12 to 24 hour window that most farmers use, the choice of feed ingredient has more impact on community "shape" than the extra 12 hours of brewing, and a shorter brew is the safer choice for most feeds.

Figure 11. Principal coordinates analysis (PCoA); samples closer together have more similar communities.
Axis 1 explains 50.2% of the variation and Axis 2 17.3%.
The dots for the unbrewed vermicompost, unfed teas and humic acid teas are shown close together at the far right, meaning they have very similar mixes of bacteria.
Molasses teas (and kelp at 24 hours) sit at the top, driven by Fictibacillus.
Fish hydrolysate and kelp combinations sit at the lower left, driven by Aeromonas and Acinetobacter.
Light blue = 24 hours, dark blue = 12 hours.

5. Biologically Rich (how many different kinds of bacteria)

Shannon index counts two things at once: how many kinds of bacteria are in the sample, and how evenly they're spread out.

HA = humic acid,
AM = alfalfa meal,
FH = fish hydrolysate,
ML = molasses,
KM = kelp meal.

The unfed tea measured 5.22 at 12 hours and 5.54 at 24 hours.

Figure 12. Observed count and Shannon index for each sample.
Observed (left) counts the number of genera detected, from about 550 (Kelp + Fish Hydrolysate at 24 hours) to about 1,700 (Kelp + Humic Acids at 12 hours).
Shannon (right) also weighs how evenly reads are spread, from 3.99 (Kelp, 24 hours) to 6.06 (unbrewed vermicompost).

Humic acid buffered biological richness. The humic acid teas had Shannon values very close to the starting vermicompost, which means many genera were present and fairly evenly represented. In practice, humic acid, especially on its own or paired with alfalfa, seemed to buffer the brew and keep a broad, balanced community rather than letting one microbe dominate.

Molasses and kelp recipes were the least biologically rich. Six of the seven least rich teas contained molasses, and kelp alone at 24 hours was the least rich of all (Fictibacillus 36% and Cellvibrio 23%). Lower richness is not by itself a problem. In the molasses teas it reflects a bloom of Fictibacillus, a potential plant growth promoter, rather than a takeover by potential pathogens.
Biologically rich and dominated teas are different communities that may suit different uses.

Biological richness and beneficial bacteria are a trade-off. A biologically rich tea contains many kinds of bacteria, much like the compost it came from, but not necessarily more of the ones that help plants. In our study the two moved in opposite directions.  Meaning, the more my tea contained many different types of bacteria, the fewer beneficial bacteria (PGPBs) it seemed to have and vice versa.  See below:

A biologically rich tea (humic acid or no feed) carries a broad sample of the whole compost community into the soil.
An enriched tea (molasses) grows a large share of a few plant-helping bacteria at the cost of variety.
This study did not test which approach helps plants more.

6. What the microbiology tells us

• Molasses, alone or with humic acid, was the most effective feed. It was the only feed rated non-pathogen-promoting and effective at both 12 and 24 hours. This reverses our hypothesis that molasses would be the least safe feed.
• Humic acid alone was the most conservative feed. It was non-pathogen-promoting at both brew times, kept Enterobacteriaceae below the unfed tea, and stayed closest to the starting compost, but it added few beneficial bacteria. It was selected for the tomato trials using the earlier standards before these definitions were finalized with our technical advisor.
• Alfalfa alone was a middle ground, non-pathogen-promoting at both times with 26 to 29% PGPBs.
• Kelp and fish hydrolysate combinations carried the highest pathogen risk. This differs from the EdaPro data, where the alfalfa and kelp blend performed well; in our compost, kelp consistently raised Klebsiella and Enterobacter.
• Molasses paired with fish hydrolysate produced the least safe tea in the study, even though molasses on its own did not raise Enterobacteriaceae.
• Limits of the method: relative abundance shows each group's share of the DNA, not how many live cells are in the tank, and 16S sequencing usually cannot separate harmful from harmless strains within a genus.
  Each tea is also a single sample.
  
  

Nutrient Density in Tomatoes Applied with Compost Tea

Across two seasons, humic-fed ACT did not produce a statistically significant yield difference, and its nutrient effects were mixed: weak leaf sugar and nitrate status in 2025, but the highest mineral content in fruit in 2026.

2025: Leaf Extract Analysis (September 2, 2025)

Source: Apical Ag reports #35361, #35362, #35363.

• Across all treatments, the Leaf Extract Analysis indicated patterns consistent with constrained plant physiology rather than simple nutrient shortage. This was determined from data interpretation given by Apical Ag lab.
• The dominant signals were low photosynthetic energy status (very low sugars in new growth), low nitrate relative to ammonium, and multiple cation and micronutrient limitations in new leaves.
• These patterns point most strongly toward uptake and transport constraints driven by field conditions (water status, root function, temperature, soil aeration/Oxidation-Reduction Potential or ORP, and nutrient antagonisms).
  There was hardly 1 inch of rainfall in August and September when these samples were taken.
  The July 30 soil test also showed a low ORP (12 against an ideal of 80).

ACT vs control (what we can responsibly claim).

• At the September 2025 sampling, the ACT program did not produce a clear improvement in the primary "plant health" indicators compared to the control.
• ACT showed very low sugars in both new and old leaves and the lowest nitrate of any treatment, suggesting the program did not improve photosynthesis or nitrogen-form balance at that time point. ACT did carry more calcium than the control in both new and old leaves, and the most potassium in old leaves.
• The control showed the strongest sugar status at this sampling point, particularly in the older leaf tissue (2.46%). This raises the possibility that, under these field conditions and application method, the inconsistent treatment of compost tea did not improve plant energy status and may have contributed to additional physiological stress rather than alleviating it.
• Nitrogen results: The nitrogen treatment cannot be used as a valid fertility comparison because the applied rate was far above label guidance (a mixing error in the ammoniacal-N application). 
  
  In short, I didn't really do a good job of carrying out treatments in 2025.  Which is why I retested in 2026.

2026: Harvest

Per-bed figures exclude 45 oz from a combined B3/C3 pick on September 18 that could not be split by bed (it is included in the Fertilized total).

The Tomato Trial Tracker can be found here.

• No treatment difference was statistically significant. On bed totals, one-way ANOVA gave F = 0.563, p = 0.597, and Kruskal-Wallis gave
  H = 1.422, p = 0.491.
  On yield per plant, ANOVA gave p = 0.640.
  With three beds per treatment, only a very large, consistent effect could have been detected.
  (An explanation of the statistical testing is found above Materials & Methodology/Statistical Analysis section)
  
  
• Compost Tea was the most consistent treatment. Its three beds ranged only from 508 to 584.5 oz (SD 40.3 oz), against 355 to 787 oz for Fertilized and 338 to 572 oz for Control.
  
  
• Ranking depends on how yield is expressed. Fertilized had the highest total, but it also had 9 plants against 6 in each of the other treatments. Per plant, Compost Tea was highest (276.8 oz), then Control (242.3 oz), then Fertilized (211.1 oz). Because plant count per bed ranged from one to three, neither total nor per-plant yield is a clean comparison.
  
  
• Timing differed by treatment. Compost Tea led mid-August and Control led through early September. Fertilized had almost no harvest until August 22, then had its largest days on September 9, 12, 21 and 25, finishing ahead. On August 11, fruit set was 20.5 fruit per plant for Control, 20.2 for Compost Tea and 10.1 for Fertilized.

Figure 13. 2026 season at Bethany Farm, July 1 to September 25: daily high and low temperature and daily precipitation at Dulles Airport (NWS station KIAD), and harvest weight by treatment on each harvest date.
Dotted lines mark the six application dates. Daily highs cover the full season; daily lows and rainfall are available from August 10 (lows missing September 17 and 18).
The hottest days came at the first application, 97 to 99°F from July 1 to 4, followed by 95 to 96°F on July 15 and 16.
Note the dry stretch of August 21 to 29 and the cool, wet final week (1.15 inches on September 22).
Weather data: NWS weather reports and the Iowa Mesonet archive for KIAD.

Pests and disease (2026).

• Hornworm: recurring on A2 (Fertilized) and C1 (Control), with damage on Compost Tea beds (B1, A3, C2) later in the season and on B3 and A2 at the final harvest. Parasitic wasp larvae killed hornworms on several dates.
• Stink bugs: on Control beds (A1, B2) and one Compost Tea bed (C2).
• Blossom end rot: on Compost Tea beds (B1, A3, and one fruit on C2) and one Fertilized bed (A2).
• Leaf curl and yellowing on B2 (Control): Apical's CNA field notes list Tomato Yellow Leaf Curl for the Control sample. B2 was also the only bed with a Kellogg's Breakfast plant, so its lower yield (338 oz) cannot be attributed to treatment alone.

The ACT did not show a clear disease-resistance benefit. Blossom end rot is a calcium-transport disorder linked to uneven watering, which fits the dry stretch of August 21 to 29.

2026: Crop Nutrient Analysis of Fruit (August 24, 2026)

Units as reported by Apical Ag.

Figure 14. 2026 tomato fruit nutrients by treatment.
Top: The three scores for each treatment.
Bottom: Each of the 21 beneficial nutrients as a percent of Apical Ag's desired value for tomato
The dashed line marks the target.

• Compost Tea fruit was highest in calcium, potassium, magnesium, iron, zinc, sulfur and boron, and scored highest on NAS. The same direction showed up in the 2025 leaf data, where ACT leaves carried more calcium than the control.
• Control fruit was highest in sugars, protein and phenols, which matches the 2025 leaf data where the control had the strongest sugar status.
• Fertilized fruit had the highest nitrate (96.07, above the desired 65) even though Tomato-tone is a low-nitrogen organic fertilizer.
• All three samples were free of detectable heavy metals. Aluminum was above Apical's 3.5 guidance in all three; Apical recommends leaf analysis during the season to track it.
• Limits: budget allowed one pooled sample per treatment (n = 1), so these differences cannot be tested for significance and should be read as descriptive.
  The NAS tier weights are our own; Apical's David Knaus supported the heavy metal index but noted that nutrient priorities vary from person to person, and Apical generally interprets nutrient ratios rather than summed scores.