Outdoor Nursery Propagation of Kava (Piper methysticum) by Air Layering

How to read this paper — This is a proof of concept, not a comparison. There was no untreated control, no auxin-only or inoculum-only arm, and no parallel cutting-propagation arm — the protocol was tested as an undifferentiated package at a single site. Every comparison to another method is against historical published figures with different endpoints, materials and assessors. The paper says so repeatedly, and so does this page.

A method everyone assumed worked, that nobody had measured

Kava propagation has historically relied on nodal stem cuttings taken from mature plants — a practice that constrains planting material and ties propagation to harvest timing. The alternatives that exist, soft-stem cuttings in peat cubes or aeroponics and tissue-culture micropropagation, generally require controlled environments, specialized infrastructure or technical inputs that put them out of reach in low-infrastructure nursery and smallholder settings.

Air layering (marcotting) is widely used in other woody and tropical crops, including the economically important Piper species P. nigrum, P. betle and P. longum. Its application to kava is described qualitatively in the monograph literature — Lebot and Cabalion note that marcotting is straightforward in kava and that runner nodes contacting the ground root spontaneously — but the authors could locate no peer-reviewed study quantifying air-layering outcomes in kava using standardized rooting criteria, or that benchmarks them against other kava propagation systems.

The technique here draws on two propagation traditions. The first is ground layering, in which lateral stems are bent to contact moistened soil and secured with a stone, inducing adventitious roots at the node–soil interface. In black pepper, runners trailed over a raised soil mound root at the nodes and the rooted pieces are harvested and potted on, giving 90.7% establishment against 62.9% for conventional propagation. The second is the Hawaiian practice of propagating kava in moss beds, which the authors have observed in ʻawa cultivation between 2010 and 2025. This method combines the two: fresh sphagnum moss inoculated with mycorrhizal and bacterial amendments, packed around hormone-treated nodes and enclosed in a two-piece pod that creates a stable rooting microenvironment.

Method

Plant material

Layers were established on Kali-ISA and Kali-Hiwa, obtained from Kali Kava, Sacramento in October 2024 as tissue-culture-derived nursery stock, propagated under greenhouse conditions at the Kali Kava research nursery in Boca Raton and then grown on in nursery containers held outdoors on open benches under shade cloth. All plants were 12 to 18 months old at pod installation. Three replicates of 40 layers each (n = 20 per variety) were drawn from the same maintained stock-plant population.

Two identity caveats, stated up front — The number of individual stock plants supporting the 120 layers was not recorded, and stock plants were drawn from the same population in all three replicates, so some plants likely carried layers in more than one replicate. Neither the plant count nor the layer-to-plant mapping is available. Separately, Kali-ISA and Kali-Hiwa are trade designations used in the American kava nursery trade rather than formally registered or morphotype-verified cultivars, and no herbarium vouchers or genetic accessions were deposited. Variety identity is provenance-based.

Air-layer preparation

1. Layering performed on intact nodes without wounding or girdling — mid-canopy stems 0.5–1.0 cm in diameter, first through third node below the apical meristem, non-senescing with full turgor and uniform dark-green colour.
2. A thin layer of Clonex® Rooting Gel (0.31% w/w indole-3-butyric acid with micronutrients) applied directly to the stem surface at the node.
3. Approximately 10–15 g of inoculated sphagnum moss — pre-hydrated, saturated but not dripping — packed around the treated node.
4. Enclosed in an opaque black two-piece snap-together polypropylene pod, nominally 4.7 in (about 12 cm) outer diameter, closing at both ends around the stem.
5. Left undisturbed outdoors for the full 21-day evaluation period, and never re-watered. Pods sanitized between replicates in 10% sodium hypochlorite with a clean-water rinse.

No greenhouse, propagation chamber or other controlled-environment enclosure was used at any point. Shade cloth and natural canopy shading from adjacent kava foliage reduced desiccation and direct insolation — that was the whole of the environmental management.

Inputs

A Bacillus amyloliquefaciens inoculant (Hydroguard®, labeled minimum 1.0 × 10⁴ CFU mL⁻¹) at 2 mL per gallon, together with Plant Success® Great White Premium mycorrhizal inoculant at 3 g per gallon, combined in 5 gallons of distilled water at 25 °C. Moss was soaked for 4 hours and wrung out immediately before packing. Inoculum viability was not independently assayed. The paper also notes that the manufacturer's marketing copy does not reconcile with the itemized guaranteed analysis on the lot used, and reports the label statement rather than treating it as a stable product specification.

The four-category scale

Outcome definitions, applied at day 21 by trained nursery staff from a common written rubric.

• Rotted — Complete tissue decay or necrosis at the treated node; no viable tissue
• Alive, no roots — Green, turgid stem, with or without new leaf expansion, and no visible root emergence
• Poorly rooted — One or more visible roots, but fewer than three reaching 2 cm — a residual class, so a layer with two roots of 3 cm scores here
• Well rooted — Three or more roots, each at least 2 cm, forming a branching root mass

Poorly rooted and well rooted are combined as “total rooted” throughout. Root length measured with a steel rule to the nearest 0.5 cm. Scoring was neither blinded to variety nor independently duplicated, so inter-rater reliability was not established.

The three weather windows

Environmental conditions, recorded as observed daily ranges from existing nursery instrumentation.

• June 2025 — summer wet — Daily highs: 33–35 °C · Overnight lows: 24–26 °C · Relative humidity: 80–88% · Character: Sustained heat, frequent afternoon thunderstorms
• October 2025 — fall transitional — Daily highs: 27–28 °C · Overnight lows: 20–23 °C · Relative humidity: 75–80% · Character: Modest diurnal swings, several rain events
• February 2026 — winter dry — Daily highs: 25–26 °C · Overnight lows: 18–19 °C · Relative humidity: 65–75% · Character: Coolest and driest; intermittent frontal passages

A continuous calibrated data logger was not deployed, which limits the precision of any climate–outcome inference.

Results

Root formation was observed in both varieties in all three replicates. Aggregate rooting was 75.0% in June, 80.0% in October and 72.5% in February — and the three-way test across seasons was clearly null.

Total rooting by variety and replicate

Figure A.

• June 2025 — Kali-ISA: 80 % · Kali-Hiwa: 70 %
• October 2025 — Kali-ISA: 85 % · Kali-Hiwa: 75 %
• February 2026 — Kali-ISA: 80 % · Kali-Hiwa: 65 %
• Pooled — Kali-ISA: 81.7 % · Kali-Hiwa: 70 %

Percentage of layers scored poorly rooted or well rooted at 21 days. n = 20 per variety per replicate; n = 60 per variety pooled.

Pooled outcome distribution by variety

Figure B.

• Rotted — Kali-ISA: 8.3 % · Kali-Hiwa: 13.3 %
• Alive, no roots — Kali-ISA: 10 % · Kali-Hiwa: 16.7 %
• Poorly rooted — Kali-ISA: 13.3 % · Kali-Hiwa: 18.3 %
• Well rooted — Kali-ISA: 68.3 % · Kali-Hiwa: 51.7 %
• Total rooted — Kali-ISA: 81.7 % · Kali-Hiwa: 70 %

All three replicates pooled, n = 60 per variety. Total rooted is poorly rooted plus well rooted.

Table 4. Pooled rooting outcomes across all three replicates (n = 60 per variety; n = 120 total).

• Rotted — Kali-ISA (n = 60): 5 (8.3%) · Kali-Hiwa (n = 60): 8 (13.3%) · Combined (n = 120): 13 (10.8%)
• Alive, no roots — Kali-ISA (n = 60): 6 (10.0%) · Kali-Hiwa (n = 60): 10 (16.7%) · Combined (n = 120): 16 (13.3%)
• Poorly rooted — Kali-ISA (n = 60): 8 (13.3%) · Kali-Hiwa (n = 60): 11 (18.3%) · Combined (n = 120): 19 (15.8%)
• Well rooted — Kali-ISA (n = 60): 41 (68.3%) · Kali-Hiwa (n = 60): 31 (51.7%) · Combined (n = 120): 72 (60.0%)
• Total rooted — Kali-ISA (n = 60): 49 (81.7%) · Kali-Hiwa (n = 60): 42 (70.0%) · Combined (n = 120): 91 (75.8%)

Yates-corrected χ² = 1.64, df = 1, p = 0.20; Mantel–Haenszel common OR = 1.92 (95% CI 0.81–4.52), p = 0.14. Wilson 95% CIs: Kali-ISA 70.1–89.4%; Kali-Hiwa 57.5–80.1%; combined 67.4–82.6%.

Tables 1–3. Rooting outcomes by replicate (n = 20 per variety per replicate).

• Rotted — Jun ISA: 2 (10%) · Jun Hiwa: 3 (15%) · Oct ISA: 1 (5%) · Oct Hiwa: 2 (10%) · Feb ISA: 2 (10%) · Feb Hiwa: 3 (15%)
• Alive, no roots — Jun ISA: 2 (10%) · Jun Hiwa: 3 (15%) · Oct ISA: 2 (10%) · Oct Hiwa: 3 (15%) · Feb ISA: 2 (10%) · Feb Hiwa: 4 (20%)
• Poorly rooted — Jun ISA: 3 (15%) · Jun Hiwa: 4 (20%) · Oct ISA: 2 (10%) · Oct Hiwa: 4 (20%) · Feb ISA: 3 (15%) · Feb Hiwa: 3 (15%)
• Well rooted — Jun ISA: 13 (65%) · Jun Hiwa: 10 (50%) · Oct ISA: 15 (75%) · Oct Hiwa: 11 (55%) · Feb ISA: 13 (65%) · Feb Hiwa: 10 (50%)
• Total rooted — Jun ISA: 16 (80%) · Jun Hiwa: 14 (70%) · Oct ISA: 17 (85%) · Oct Hiwa: 15 (75%) · Feb ISA: 16 (80%) · Feb Hiwa: 13 (65%)

Fisher's exact test, total rooted vs. not rooted: June OR = 1.71, p = 0.72; October OR = 1.89, p = 0.69; February OR = 2.15, p = 0.48. All three replicate confidence intervals overlap substantially.

What the rooting classes look like

The photographs below were selected to illustrate the category definitions and were not drawn by random sampling from the scored units. They are documentary rather than quantitative evidence.

Post-transplant establishment

All rooted layers were severed at 21 days and potted into 1-gallon nursery containers with standard medium, then held under 30% shade cloth at the same site without supplemental humidity control. Across all three replicates and both varieties, 91 of 91 transplanted layers survived to the 30-day assessment (95% CI 95.9–100), with no observed transplant shock, wilting or mortality.

What 100% survival does and does not mean — Survival was recorded as a binary presence/vigor observation. No growth, biomass or root-development measurements were taken, and 30 days is too short an interval to speak to establishment in any agronomic sense. Surviving plants were later moved to open-field plantings, but that transfer happened after the observation window closed and no field outcomes are reported.

Against the published benchmarks

Aggregate rooting exceeded the values reported for the alternative systems. One-sample binomial tests against each reference proportion return p < 0.001 in every case. Those tests treat the published figures as fixed, known constants, which they are not — they show only that the difference is larger than sampling noise in this sample, not that a controlled between-method effect exists.

Table 5. Reported rooting or establishment outcomes across propagation methods. Rows are not concurrent controls; endpoints and materials differ.

• Air layering, outdoor nursery, South Florida — pooled (this paper) — Interval: 21 d · Reported outcome: 75.8% (91/120), CI 67.4–82.6 · Endpoint and caveats: Rooting at fixed interval; ambient outdoor nursery; vascular continuity maintained
• Air layering — October 2025 replicate (this paper) — Interval: 21 d · Reported outcome: 80.0% (32/40), CI 65.2–89.5 · Endpoint and caveats: Best single replicate; difference from other replicates not significant (p = 0.73)
• Soft-stem cuttings in peat cubes, greenhouse — Interval: 21 d · Reported outcome: 48% · Endpoint and caveats: Excised tissue; source not yet peer reviewed
• Soft-stem cuttings in aeroponics, greenhouse — Interval: 14 d · Reported outcome: 53% · Endpoint and caveats: Shorter interval; high technical and energy input; source not yet peer reviewed
• Nodal cuttings, field, South Pacific — Interval: To establishment · Reported outcome: 20–60% · Endpoint and caveats: Cuttings that formed plants, not a rooting score; strong node-position and planting-month effects

The nodal-cutting figures are Davis & Brown's Table 9: 20%, 49% and 60% for basal, middle and apical nodes planted in March, and 39%, 60% and 56% for the same positions in August. The full range is 20–60%; citing “49–60%” alone would select the more favorable subset, and the paper declines to do so.

Why it might work — offered as a hypothesis

One candidate explanation is the preservation of vascular continuity with the mother plant, and it is worth being explicit that this runs against the conventional account of adventitious rooting rather than following from it. Auxin is the central regulator, acting through canalization to responding target cells and a downstream cascade involving AUX/IAA proteins, TOPLESS, ARFs and SAUR-like proteins. But in cuttings, the early auxin accumulation that initiates rooting is generally attributed to the interruption of the basipetal auxin drain caused by excision, together with wound signaling. On that account an attached layer, which retains its drain, should have less auxin stimulus, not more — and the applied IBA presumably substitutes for it.

What an attached layer does retain is the resource network that excision removes. Adventitious rooting in cuttings proceeds under isolation from the whole-plant resource and signal network, with wound-induced ethylene and jasmonate signaling, altered auxin conjugation and catabolism, and reduced invertase activity that impairs sink establishment at the rooting zone. A layer keeps receiving water and photosynthate from the mother plant throughout induction — which is the more plausible basis for any advantage, and is consistent with the fact that no layer required misting or humidity control.

This is a hypothesis, not a finding — The design tests none of it. No hormone, carbohydrate or transport measurements were made, and no excised-cutting arm was run in parallel. The account above is a proposal for a future factorial experiment. The higher absolute rooting rate reported here is equally consistent with differences in plant material, node selection, substrate and assessor between this trial and the historical comparators.

What the numbers support, and what they don't

Seasonal effects were not detected. October gave the highest point estimate and February the lowest, but the three-way test was clearly null (p = 0.73) and the intervals overlap. Rather than reading a season optimum into the ordering, the useful result is a negative one with practical value: the method did not fail in peak summer heat or in the cool dry winter window, and rooting stayed above 70% in every replicate. Temperature and moisture during induction are established determinants of adventitious rooting, so a season effect may well exist — this study simply lacked the resolution to find one.

The variety difference is inconclusive, not negative. Kali-ISA led by 11.7 points pooled, in a consistent direction across all three replicates, with a homogeneous stratum-specific odds ratio (common OR 1.92, test of equal odds p = 0.98). But neither the pooled chi-square (p = 0.20) nor the stratified test (p = 0.14) nor any within-replicate test reached significance, and the risk-difference interval (−3.5 to +26.8 points) is compatible with no difference and with a substantial one. With achieved power near 0.32, the consistency of direction is suggestive enough to justify a properly powered trial — about 209 layers per variety.

Nothing here speaks to the inputs. The auxin gel and microbial inoculum were applied together to every layer. No factorial arms, no auxin-only or inoculum-only treatment, no untreated control. Any statement that microbial inoculation improved rooting here would be unsupported, and the same applies to the moss substrate and the enclosure.

• 75.8% — Pooled rooting at 21 days (91/120)
• 0.32 — Achieved power for the variety comparison
• 209 — Layers per variety needed for 80% power
• 91/91 — Rooted layers surviving 30 days post-transplant

What it costs

Air layering shifts investment away from fixed capital and environmental control and toward per-plant labor. At US$25.00 per hour, installing a layer took roughly 3–10 minutes of hands-on time, with a further 2–7 minutes to detach and pot a rooted one.

Table 6. Estimated labor and material cost per air layer installed and per rooted layer potted. Labor at US$25.00/hr.

• Labor — air-layer installation — Time (min): 3–10 · Cost low: US$1.25 · Cost high: US$4.17
• Labor — detachment and transplant — Time (min): 2–7 · Cost low: US$0.83 · Cost high: US$2.92
• Total labor per rooted layer potted — Time (min): 5–17 · Cost low: US$2.08 · Cost high: US$7.08
• Consumables — hormone, inoculants, fasteners and sanitation — Time (min): — · Cost low: US$0.18 · Cost high: US$0.90
• Reusables amortized — pod and moss charge — Time (min): — · Cost low: US$0.08 · Cost high: US$0.25
• Total materials per layer installed — Time (min): — · Cost low: US$0.25 · Cost high: US$1.10
• Total per air layer installed — Time (min): 3–10 · Cost low: US$1.50 · Cost high: US$5.27
• Total per rooted layer potted — Time (min): 5–17 · Cost low: US$2.33 · Cost high: US$8.18

Reusables assume 10–20 pod reuse cycles that were not verified over a full service life; sanitization labor is not monetized.

Why these ranges are so wide — The site kept two independent time-and-motion records that do not agree — one logged 3–6 minutes per installation and 2–4 per transplant, the other 6–10 and 4–7. Neither was collected under a stopwatch protocol, so the figures span both records rather than selecting the more favorable one. They exclude stock-plant maintenance, overhead, sanitation labor, capital recovery on the pod inventory, and the inputs absorbed by the roughly 24% of layers that did not root. Success-adjusting for the 75.8% rooting rate gives about US$2.81–9.87 per rooted plant potted. Do not use these for commercial planning without validation at production scale.

Limitations

• Single site. All three replicates ran at one nursery. Site, soil, canopy, crew and stock-plant genetics are fully confounded with the outcome. Nothing here establishes performance anywhere else.
• No concurrent control or comparator arm. No untreated node, no auxin-only or inoculum-only arm, no parallel cutting arm. The protocol was tested as an undifferentiated package.
• Pseudoreplication. Multiple layers were installed on individual stock plants, so layers are not fully independent — and neither the number of stock plants nor the layer-to-plant mapping was recorded, so clustering cannot be modeled retrospectively. Because the same population served all three replicates, the replicates are not fully independent of one another either. The reported intervals and p-values assume independence and are therefore likely anti-conservative: the true intervals are wider than stated and the effective sample size is smaller than n = 120.
• Underpowered for the variety comparison. Achieved power roughly 0.32. The non-significant results are inconclusive, not evidence of equivalence.
• No randomization and no blinding. Layers were not randomized to canopy position or plant, and scoring was performed by staff who knew the variety and had also installed the layers. Categories 3 and 4 involve judgment at the margin, and no layer was scored by a second assessor, so observer bias cannot be excluded.

Where this goes next

The most valuable next experiment is a properly powered factorial trial. At roughly 209 layers per variety, crossed arms of auxin (present/absent) and microbial inoculum (present/absent) would resolve both the variety question and the contribution of each input. Adding a concurrent nodal-cutting arm scored on the same scale by the same blinded assessors would convert the benchmark comparison from historical to controlled.

• Multi-site replication across subtropical and tropical environments, including Pacific Island and continental American growing regions.
• Record layer position on the stock plant so within-plant clustering can be modeled with mixed-effects or GEE methods rather than assumed away, and deploy calibrated continuous environmental logging so a season effect can actually be estimated.
• Expand germplasm beyond two trade varieties, with vouchered accessions and recorded morphological covariates — stem diameter, internode length, node position, tissue maturity — so recommendations rest on traits rather than names.
• Systematically optimize the physical and biological parameters: hormone concentration, inoculant composition, moss type and volume, pod size and opacity.
• Track cohorts from transplant through to harvest, assayed for root biomass and kavalactone content alongside plants raised from cuttings and tissue culture, and rebuild the economic model from audited production-scale data.

That last point carries the most commercial weight. Rooted propagules were straightforward to detach and pot, and all survived to 30 days — a useful signal of handling robustness. But whether early rooting vigor translates into comparable growth rate, root-mass development, kavalactone accumulation or harvestable yield relative to plants raised from cuttings or tissue culture is untested here, and is the single most consequential open question for anyone considering the method commercially.

What this study supports

Air layering can be carried out on containerized stock plants under ambient outdoor nursery conditions in a subtropical climate — without greenhouse, misting or other controlled-environment infrastructure, without destroying mother-plant biomass, and with predictable per-plant labor — and rooted layers can subsequently be potted on and moved to open-field plantings.

It does not, on this design, establish that air layering is superior to cuttings or aeroponics, that the microbial inoculant contributed to the outcome, that any season is preferable, or that the varieties differ. Each of those requires a controlled, adequately powered, multi-site trial. What this study supports is a narrower and still useful claim: under the conditions tested, the method worked reliably enough to justify that larger investigation.

Disclosures and conflicts of interest

No external funding was received. This paper is published by the American Kava Association. The corresponding author is affiliated with Root of Happiness, a commercial kava retailer, and holds an interest in Kali Kava, a commercial kava nursery; the remaining authors are affiliated with Nakamal at Home, also a commercial kava retailer. The plant material was supplied by Kali Kava and all three replicates were conducted at the Kali Kava research nursery — so the trial site is a facility in which the corresponding author holds a commercial interest. These relationships are disclosed as potential conflicts. No manufacturer of any product named in this paper provided funding, materials, or input into the design, conduct, analysis or reporting of the study.

T.B. conceived the study, designed the protocol, performed the statistical analysis and drafted the paper. B.M. performed the majority of the air-layer installations and contributed to layer scoring and transplant handling. J.B. maintained the stock-plant population, installed and scored layers, recorded environmental data and contributed to transplant assessment. The authors thank the Kali Kava nursery staff in Boca Raton, and acknowledge the Hawaiian ʻawa cultivators whose moss-bed propagation practice informed the design of this method.