PRO 406 briefing · for Sharuk · 30 Aug 2026

Mustard-green fermentation: what is being done, why, and how

Gundruk-style mustard leaves. 30 jars. Salt × cutting. Count lactic acid bacteria vs Enterobacteriaceae. Time to pH below 4.6.

Why this project

Household gundruk in Nepal is usually made with little or no salt. Newer work shows that cutting salt slows lactic acid bacteria (LAB) and lets Enterobacteriaceae hang around longer. Cutting the leaves finer speeds acidification. Nobody has put those two controls on mustard greens, in Nepal, with enough replicate jars and plate counts.

The question is not “can vegetables ferment?” It is: which combination of salt and cutting lets LAB take over fast enough to cross the safety pH of 4.6, and suppress Enterobacteriaceae?

That is original local data, wet-lab microbiology, agricultural but not soil, and a numbered criterion a viva can defend.

What is being done

Working title

Effect of salt concentration and cutting size on lactic acid bacteria versus Enterobacteriaceae during gundruk-style fermentation of mustard greens, scored as time to pH below 4.6.

Hypothesis

Jars with lower salt and/or rougher cuts take longer to reach pH < 4.6, keep higher Enterobacteriaceae counts for longer, and show a lower LAB:Enterobacteriaceae ratio in the first week than jars with more salt and thin shredding.

Design (n = 30 experimental units)

3 salt levels × 2 cutting sizes × 5 independent jars = 30 jars.

FactorLevels
Salt (% of veg + brine)0% (traditional gundruk style), 1%, 2.5%
CutThin shred (~2–3 mm) vs rough pieces (~5–8 cm)
VegetableOne batch of mustard greens (rayo / Brassica juncea leaves)

What is measured (the microorganisms)

  • Presumptive LAB (CFU/g) on MRS or cheaper GYP-Ca agar
  • Bile-tolerant Enterobacteriaceae (CFU/g) on VRBA / VRBG (MacConkey only if VRBA is unavailable — note that in the report)
  • LAB : Enterobacteriaceae ratio
  • Brine pH every 24 h → hours/days to pH < 4.6
  • Optional: titratable acidity as % lactic acid

No PCR. No species names. Call them presumptive LAB and presumptive Enterobacteriaceae. Product is not eaten.

How it works (the biology)

Raw leaves carry many bacteria. Enterobacteriaceae usually bloom first. Then LAB (often Leuconostoc, later more acid-tolerant types) ferment plant sugars to lactic and acetic acids. pH falls. Acid plus low oxygen knocks Enterobacteriaceae down. That hand-over is microbial succession.

Functions of each piece

  • LAB — make acid; that is the preservation engine and the safety mechanism.
  • Enterobacteriaceae — early plant-associated group; some are spoilers; high counts for too long is a safety/quality problem. They are not “the cholera bacterium.” Do not claim pathogen IDs from these plates.
  • Salt — pulls juice out of the leaf (food for microbes), adds osmotic stress. Lower salt = slower LAB takeover (Eilers et al. 2026).
  • Cutting — more cut surface → sugars leak faster → LAB acidify faster; Enterobacteriaceae die sooner (Valence et al. 2025).
  • pH 4.6 — Codex acidification target: below this, growth of dangerous spore-formers such as C. botulinum is not supported in acidified plant foods. It is a process-safety number, not an E. coli 0/100 mL water standard.

What’s the use / importance

  • Nepal food culture — gundruk is a real product; the science of how it becomes safe is still thin compared with kimchi/sauerkraut.
  • Less salt — WHO pushes lower sodium. The project tests whether “less salt” still lets LAB win in time.
  • Household practice — people shred or chop differently. Cutting is free. If shredding is safer, that is advice people can use.
  • Skills for work — aseptic sampling, serial dilution, selective plates, pH, a factorial experiment, statistics. That is food-micro / QC language, not an app.
  • Social impact (Annex 3) — safer indigenous fermentation without extra machines.

Say this honestly in the proposal: Codex CXS 260-2007 asks for equilibrium pH < 4.6 in pickled fruits/vegetables. It lists green mustard, but it also excludes kimchi and sauerkraut. Gundruk is closer to those exclusions. Use pH 4.6 as the acidification safety target, not as “this jar is a Codex pickle.”

Materials involved

Budget band NPR 20–30k. Confirm the media with the department store before buying.

Must have

  • Mustard greens, one market batch, ~8–10 kg
  • Food-grade salt (weigh accurately)
  • 30 glass jars ~250–500 mL with lids (boiled or autoclaved)
  • Distilled or boiled-and-cooled water for brine
  • pH meter (calibrated 4.0 and 7.0 buffers)
  • Autoclave, incubator (~30 °C and 37 °C), balance, pipettes or sterile pipettes, spreaders
  • Peptone/saline diluent (e.g. 0.85% NaCl + 0.1% peptone)
  • MRS or GYP-Ca agar (LAB)
  • VRBA or VRBG agar (Enterobacteriaceae)
  • Petri dishes, loops, gloves, 70% ethanol, autoclave bags
  • Candle jar or CO₂ jar if MRS is incubated that way; if not, incubate MRS and write it as a limitation

Nice if present

  • 0.1 M NaOH + phenolphthalein or pH 8.3 endpoint for titratable acidity
  • Stomacher / sterile filter bags (or sterile blender jar, then discard)
  • Gram stain + catalase (spot-check a few LAB-looking colonies)

Salt recipe (per jar)

Target ~200 g wilted leaves + ~200 g liquid = 400 g total.

TreatmentSalt in that 400 g
0%0 g (water only)
1%4.0 g
2.5%10.0 g

Dissolve salt in the water first, then pour over packed leaves until covered. Leave almost no headspace. Press leaves under the liquid.

Procedure (this is the method the papers support)

This is a teaching-lab version of Valence et al. 2025 (jars, MRS, VRBG, pH) plus Eilers et al. 2026 (salt levels). No sequencing. No HPLC.

1
Confirm with Sharuk in writing that n ≥ 30 means 30 jars, not 30 market samples. Confirm MRS/GYP-Ca and VRBA exist. If VRBA does not exist, stop and switch medium with his signature.
2
Pilot (3 extra jars, 5 days). Learn dilutions so you do not waste 30 jars of plates. Typical LAB later: 10⁵–10⁹ CFU/g. Enterobacteriaceae early: 10⁴–10⁸ CFU/g.
3
Leaves. One batch. Discard rotten. No soap. Optional light tap rinse, drain well — same for every jar. Shade-wilt 12–18 h like household gundruk, all together.
4
Cut. Half the mass thin-shredded (2–3 mm). Half in 5–8 cm pieces. Mix each pile so jars of the same cut are comparable.
5
Pack 30 jars at random from those piles (5 jars per salt × cut). Weigh leaves. Add the matching brine. Press under liquid. Label: salt, cut, jar number 1–5. Room 20–25 °C, dark, do not open except for sampling.
6
pH every 24 h on every jar: sanitize probe, dip brine, wipe, ethanol. Record time to first reading < 4.6. Recap immediately.
7
Counts on days 0, 3, 7, 14. Aseptically take 10 g (about 5 g leaf + 5 g brine) into 90 mL diluent = 10⁻¹. Mix. Dilute. Plate 0.1 mL on LAB medium and on VRBA. Incubate LAB ~30 °C, 48 h; VRBA 37 °C, 24 h. Count plates with 30–300 colonies. Convert to log₁₀ CFU/g.
8
Optional checks. A few LAB colonies: Gram-positive rods/cocci, catalase-negative. That supports “presumptive LAB.” Do not invent species.
9
Analyse. For each jar: days to pH < 4.6; log LAB; log Enterobacteriaceae; ratio. Two-way ANOVA (salt × cutting) with n = 5. Graphs of mean ± SD over time. If a jar never reaches 4.6 by day 14, record that as failure to meet the target.
10
Kill and write. Autoclave all jars and plates. Do not taste. English report + Nepali abstract. In discussion: compare to Valence (cutting), Eilers (salt), Karki (gundruk pH ~4). Limitations: opening jars for counts, no anaerobic jar, no species ID, Codex 260 is an analogue not a gundruk standard.

Plate load: 30 jars × 4 days × 2 media is a lot. Use two well-chosen dilutions (from the pilot), one plate each, not triplicate. If the incubator cannot take it, plate all 30 for pH-critical days (3 and 7) and a subset of 2 jars per treatment on day 0 and 14 — and write that plan before you start.

Papers that back this procedure

Sharuk asked for a few. These three are enough to start. They are real, and they match this method (counts + pH + salt/cut), not a random gundruk story.

Paper 1 — the procedure (cutting, MRS, VRBG, pH)

Valence F, et al. (2025). The cutting type of vegetables influences the spontaneous fermentation rate. Peer Community Journal 5:e49.

They packed salted vegetables in jars, counted Enterobacteriaceae on VRBG and LAB on MRS, and measured pH and titratable acidity. Thin cuts fermented faster; enterobacteria fell sooner. That is the culture method this FYP copies (without their DNA sequencing or HPLC).

doi:10.24072/pcjournal.553

Open paper Direct PDF

Paper 2 — why salt levels matter

Eilers T, et al. (2026). From diversity to dominance: how salt and CO₂ shape LAB-dominated ecosystems in vegetable fermentations. Microbiology Spectrum.

Lower salt (0% and 1.25% vs 2.5%) delayed LAB dominance and left Enterobacterales around longer. That is why 0% / 1% / 2.5% is a real experimental factor, not decoration. You will not inject CO₂; that part is future work.

doi:10.1128/spectrum.03578-25

Open paper

Paper 3 — Nepal gundruk (the local product)

Karki T, Okada S, Baba T, Itoh H, Kozaki M. (1983). Studies on the microflora of Nepalese pickles Gundruk. Nippon Shokuhin Kogyo Gakkaishi 30(6):357–367.

Classic Nepal work: mustard/rape/radish gundruk, LAB succession, pH falling to about 4.0, acidity ~0.8–1.0% as lactic acid, fermentation ~one week. This is why mustard greens are the right vegetable. Your project adds salt × cutting and Enterobacteriaceae counts, which they did not set up as a factorial experiment.

doi:10.3136/nskkk1962.30.357

Open paper
Extra reading (not required this week)

Codex CXS 260-2007 — pickled fruits and vegetables; equilibrium pH < 4.6. Use as the numbered target, with the kimchi/sauerkraut exclusion noted above.

Lee et al. (2023) Food Chemistry: X — smaller radish pieces in kimchi changed the LAB community faster. doi:10.1016/j.fochx.2023.100950

Wuyts et al. (2018) — carrot-juice fermentation: Enterobacteriaceae first, then LAB, pH crash. Same succession story in a liquid model.

What to tell Sharuk in one breath

We will ferment mustard greens in 30 jars (3 salts × 2 cuts × 5). We count LAB and Enterobacteriaceae the way Valence 2025 did, vary salt the way Eilers 2026 showed matters, on the vegetable Karki 1983 already described for Nepal. Success is faster LAB dominance and pH below 4.6. Nothing is eaten. No PCR.