The Economics of Free Plants
Let us begin with a number: $380. That is what a mature Ficus lyrata — the fiddle-leaf fig, that totem of interior-design aspiration — costs at the flower market on Prince Edward Road West. It is also roughly what you would spend on a 10 cm pot of Monstera deliciosa ‘Thai Constellation’, the variegated cultivar that has colonised every plant-shop window from Sheung Wan to Sai Kung. Now consider what it costs to produce either plant yourself: a cutting taken from a friend’s plant. A plastic nursery pot you already own. Potting mix, perhaps $15 worth. Time, which you were going to spend anyway. The arithmetic is not subtle. Propagation is the closest thing gardening has to a free lunch, and yet remarkably few gardeners do it systematically.
This is not an article about the romance of watching roots appear in a jar of water — although there is that, and we will get to it. This is an article about propagation as a gardening discipline: the techniques that work reliably, the ones that waste your time, and the quiet satisfaction of populating an entire garden without ever swiping a credit card at a nursery. I have propagated thousands of plants over two decades — Salvia cuttings taken between my fingers while walking through the Botanical Gardens, Hemerocallis divisions that now fill three borders, Pelargonium that have outlived the original plants by a decade. I have also killed more cuttings than I care to count. The line between the two outcomes is thinner than most gardening books admit.
Before we go further, a word on what propagation is not. It is not the timid placing of supermarket basil stems in water on a kitchen windowsill, although that is technically propagation and I will not stop you. It is not the wholesale cloning of garden-centre stock for resale, which is both illegal for patented cultivars and, more to the point, tedious. Propagation as I mean it is the deliberate, seasonal practice of increasing the plants you love — the ones that thrive in your particular light, your particular soil, your particular neglect — so that your garden becomes a self-renewing system rather than a recurring expense. A garden that propagates itself is a garden that has achieved some measure of independence.
The Five Methods: A Practical Taxonomy
All propagation falls into one of five categories. Understanding which applies to which plant — and when — is the single most useful piece of knowledge a gardener can acquire. The categories are: seed, cutting, division, layering, and grafting. We can dispense with grafting immediately. Unless you have a rootstock collection, a grafting knife sharpened to surgical standards, and the patience to wrap union points in Parafilm while holding your breath, grafting is best left to nursery professionals. The other four methods are available to anyone with a sharp knife, some pots, and a willingness to fail.
Seed is the democratic method: cheap, abundant, and capable of producing variation that no cutting can match. It is also slow. A Davidia involucrata (handkerchief tree) from seed will take a decade to flower. An Agapanthus from seed will take three years to reach flowering size and will almost certainly not match its parent. For annuals — Zinnia elegans, Cosmos bipinnatus, Nicotiana alata — seed is the only sensible method, and the rewards come within months. For perennials, seed is a gamble worth taking when you want quantity and are relaxed about genetic outcome. For trees, seed is a gift to your future self, who may or may not still live in the same house.
Cuttings are the workhorse of propagation: fast, clonal, and satisfying in the way that only visible root growth can be. A cutting is a piece of stem, leaf, or root that, placed in the right medium under the right conditions, produces an independent plant genetically identical to its parent. The key phrase is “right conditions.” Most cutting failures come down to three things: the cutting dried out before it could form roots, the cutting rotted because the medium was too wet, or the cutting was taken at the wrong time of year from the wrong part of the plant. We will return to this.
Division is the simplest method and the hardest to get wrong: you dig up a clump, you pull it apart, you replant the pieces. It works for any plant that forms multiple crowns or rhizomes — Hemerocallis, Hosta, Iris, Agapanthus, ornamental grasses, Zingiber (ginger), Alpinia. Division has the highest success rate of any propagation method because each division already has roots, shoots, and stored energy. The only way to fail at division is to make the pieces too small or to do it in midsummer heat.
Layering is the method nobody uses and everyone should. It is the technique of rooting a stem while it is still attached to the parent plant, so the cutting never experiences the trauma of separation until it already has roots. For plants that are difficult from stem cuttings — Magnolia, Rhododendron, Camellia, Wisteria — layering succeeds where cuttings fail. The trade-off is time: layering can take six months to two years. But the success rate, for plants that resist other methods, approaches 100 percent.
Timing: The Variable That Determines Everything
I have a friend — let us call him Michael, because that is his name — who insists on taking cuttings in August. Every August, he trims his Hibiscus rosa-sinensis, sticks the prunings in pots of garden soil, places them in full sun, and every August they die. He has been doing this for seven years. He has concluded that hibiscus are “difficult from cuttings.” They are not. Hibiscus root readily from semi-hardwood cuttings taken in late spring, placed in a free-draining medium, and kept in bright indirect light with high humidity. Michael’s problem is not his technique. It is that he is trying to root cuttings in the hottest month of the year, in direct sun, using heavy soil that holds water like a sponge and rots the stems before a single callus cell can form.
Timing is not a suggestion. It is the variable that determines whether a cutting forms roots or forms mould. The calendar that follows is for Hong Kong’s subtropical climate, but the principles translate: take cuttings when the parent plant is in active growth but not flowering heavily, when temperatures are warm but not punishing, and when humidity is naturally high.
Spring (March to May): The primary propagation window. Softwood cuttings — the fresh, flexible growth at the tips of stems — root fastest now. This is the time for Salvia, Coleus, Plectranthus, Pentas, and virtually all herbaceous perennials. Semi-hardwood cuttings of shrubs like Gardenia jasminoides, Osmanthus fragrans, and Michelia also take well. Division of summer-flowering perennials — Hemerocallis, Agapanthus — should happen now, before they commit energy to flowering. Seed sowing of warm-season annuals and vegetables begins in March.
Early summer (June): Still viable for softwood cuttings if you can provide shade and consistent moisture. The humidity is your ally — ambient levels of 80 percent reduce transpiration from unrooted cuttings. This is the window for tropical plants that need warmth to root: Bougainvillea (semi-hardwood), Plumeria (hardwood, allowed to callus for a week before planting), Cordyline (stem sections).
Late summer to autumn (September to November): Semi-hardwood and hardwood cuttings. The summer’s growth has matured; the wood is firm but not yet lignified. This is the window for Camellia, Azalea, Osmanthus, Lagerstroemia (crepe myrtle), and many evergreen shrubs. The lower temperatures reduce stress, and the soil is still warm enough to encourage root formation. Division of spring-flowering perennials — Iris, Hosta, Chlorophytum — happens now.
Winter (December to February): Hardwood cuttings of deciduous plants taken during dormancy. Rosa (roses), Hibiscus syriacus (rose of Sharon), Ficus carica (fig), Punica granatum (pomegranate). These are the easiest cuttings of all — stick them in the ground, walk away, and by spring most will be growing. No misting, no humidity dome, no fuss. The plant is asleep and doesn’t know it’s been cut. Seed sowing of cool-season annuals — Calendula, Papaver, Lathyrus odoratus (sweet pea) — begins now for spring flowers.
The Cutting: Anatomy of a Technique
There are perhaps three people in Hong Kong who sharpen their secateurs before taking cuttings. I have met one of them. Her name is Mrs. Chan, she is seventy-four, and her rooftop garden in To Kwa Wan contains more healthy plants per square metre than any space I have ever seen. “If the cut is ragged,” she told me once, holding up a pair of Okatsune secateurs she has owned since 1982, “the wound heals ragged. If the wound heals ragged, the roots come ragged. The plant remembers.” She is right, in the literal sense. A clean cut reduces the surface area of damaged tissue, minimises the entry point for fungal pathogens, and produces a smooth ring of callus tissue from which roots emerge more uniformly. The tool matters.
The process, in detail:
1. Select the right material. For softwood cuttings, choose non-flowering shoots that snap cleanly when bent — they should be firm enough to hold their shape but not so woody that they resist bending. The ideal cutting is 8–12 cm long, with at least three nodes. The node is where the leaves attach; it is also where adventitious roots will form. Cut just below a node, because the highest concentration of undifferentiated cells — the cells that can become roots — is at the node. Remove the lower leaves, leaving two to four at the top. If the remaining leaves are large (Hydrangea macrophylla, I am looking at you), cut each leaf in half horizontally with scissors. This reduces transpiration — water loss through the leaf surface — without removing the leaf’s ability to photosynthesise and fuel root production.
2. Wound the base (for difficult species). For plants that are slow to root — Magnolia, Daphne, some Rhododendron — make a shallow vertical cut, about 1 cm long, through the bark at the base of the cutting. This exposes the cambium layer, the thin band of actively dividing cells between bark and wood, and increases the surface area available for root formation. Dip the wounded base in rooting hormone. More on that below.
3. Use rooting hormone, or don’t. The debate over rooting hormone — usually a talc or gel containing indole-3-butyric acid (IBA) or naphthaleneacetic acid (NAA) — is one of those gardening arguments that generates more heat than light. The purists say they never use it. The commercial growers use it on everything. The truth: hormone is insurance, not necessity. For easy rooters — Salvia, Coleus, Plectranthus, Tradescantia, Pelargonium, most succulents — you can skip it and lose perhaps five percent in strike rate. For moderate rooters — Hibiscus, Gardenia, Bougainvillea, Rosmarinus — hormone increases strike rate from roughly sixty percent to roughly eighty-five percent. For difficult rooters — Camellia, Daphne, Magnolia, Lavandula taken as semi-hardwood — hormone is the difference between occasional success and consistent failure. I use it on everything except succulents and Pelargonium, not because I am a purist or a commercial grower, but because a $45 jar of Clonex gel lasts two years and eliminates the single most common failure mode: the cutting calluses but never roots.
4. The medium. Garden soil is not a propagation medium, and using it as one is how Michael kills his hibiscus every August. A propagation medium needs three qualities: it must hold moisture without becoming waterlogged, it must be sterile enough that fungal pathogens don’t kill the cutting before it roots, and it must be fine-textured enough that emerging roots don’t hit an air pocket. The standard mix: one part perlite, one part vermiculite, one part coco coir or fine-grade peat. For succulents and other rot-prone plants: two parts perlite to one part coir. For hardwood cuttings in winter, which root slowly and risk rot: pure coarse sand or pure perlite. Do not add fertiliser. Unrooted cuttings cannot absorb nutrients, and fertiliser salts in the medium can burn emerging root tips.
5. Humidity. A cutting has no roots, which means it cannot replace the water it loses through its leaves. Until roots form — typically ten days to six weeks, depending on species and season — the cutting must be kept in an environment where the air is nearly saturated with moisture. The simplest method: a clear plastic bag placed over the pot and secured with a rubber band. Open it every few days for air exchange. The more elegant method: a propagation dome, essentially a clear plastic lid that fits over a standard seed tray. The method I use: a large clear plastic storage box, upturned, with a few ventilation holes drilled in the sides. It costs $35 at Japan Home Centre and holds twelve 10 cm pots.
One critical detail that is almost never mentioned: remove the humidity cover gradually. A cutting that has rooted under 100 percent humidity has soft, thin leaves adapted to a saturated atmosphere. If you remove the cover suddenly, the cutting will wilt within hours — not because it lacks roots, but because its leaves have no cuticle development and cannot regulate water loss. The process is called “hardening off”: remove the cover for an hour on day one, two hours on day two, four hours on day four, and only then permanently. It adds a week to the process. Skipping it adds a high probability of losing the cutting you just spent six weeks nursing.
Division: The Confidence-Building Method
If you have never propagated anything, start with division. It requires no equipment beyond a spade or a pair of hands, it works on a vast range of plants, and the divisions are already plants — they just need their own space. The failure rate is close to zero, provided you observe three rules.
First: water the parent plant thoroughly the day before you divide it. A hydrated plant is a resilient plant. Dry roots tear; hydrated roots flex. Second: each division must have at least three healthy shoots and a corresponding section of root system. A single shoot with a scrap of root is not a division; it is a cutting that happens to have roots attached, and it will struggle. Third: replant immediately. A division left on the soil surface for even an hour on a warm day can lose enough moisture through its exposed roots to set the plant back by weeks.
The plants that respond best to division in Hong Kong’s climate are, happily, also the ones most widely grown. Agapanthus africanus, which sulks when crowded, will flower far more freely after division — lift the clump, pull or cut it into sections each with a fan of leaves and a fist-sized root mass, replant with the crown at soil level. Hemerocallis (daylilies) are almost unkillable; you can divide them with a spade, with your hands, or, in one memorable case I witnessed, with a bread knife, and they will bloom the following season regardless. Chlorophytum comosum (spider plant) forms plantlets on stolons that root on contact with soil — division here means simply potting up the babies. Zingiber officinale (culinary ginger) and its ornamental relatives produce rhizomes that can be broken into sections, each with at least one “eye” or growing point, and replanted horizontally 5 cm deep. Alpinia zerumbet (shell ginger) and Hedychium (ginger lily) follow the same principle but on a larger scale: a mature clump can yield six to eight divisions, each capable of becoming a metre-wide plant within two years.
One plant that does not divide well, despite what the internet tells you: Lavandula. Lavender is a woody subshrub, not a herbaceous perennial, and its stems do not form separate crowns. Trying to divide a lavender plant will almost certainly kill it. Take semi-hardwood cuttings in autumn instead. This is a good rule of thumb: if the plant has woody stems at the base, it is a candidate for cuttings, not division.
Seed: The Long Game
There are gardeners who find seed-sowing beneath them — too slow, too uncertain, too much like farming. These gardeners are wrong, but they are wrong in an interesting way. The objection to seed is not really about time. It is about control. A cutting produces a clone: you know exactly what you are getting, because it is the same plant. A seed produces a genetic individual: it carries the genes of both parents (even if self-pollinated, recombination produces variation), and the result may be taller, shorter, paler, darker, more vigorous, or more susceptible than either parent. For some gardeners, this is intolerable. For others — and I count myself among them — it is the entire point.
Seed propagation in Hong Kong requires a recalibration of expectations. The seed packets sold in garden centres are overwhelmingly designed for temperate climates. “Sow in early spring after last frost” is not a useful instruction in a city where the last frost occurred during the Pleistocene. The adjustments are straightforward: treat “spring” as October to November (the beginning of the cool dry season) and “summer” as March to May (warm, wet, ideal for tropical germination). Many temperate annuals — Lathyrus odoratus (sweet pea), Papaver somniferum (opium poppy), Calendula officinalis — will only grow here as winter annuals, sown in October and finished by March, before the heat destroys them.
The seed-starting setup does not need to be elaborate. A seed tray with a clear lid, a bag of seed-raising mix (finer than potting mix, lower in nutrients), and a position that gets bright indirect light. The most common seed-sowing error — and I have made it more times than I can count — is burying seeds too deeply. The rule: sow at a depth equal to twice the seed’s diameter. For dust-fine seeds like Begonia and Petunia, that means surface-sow and press gently into the mix without covering. For larger seeds like Ipomoea (morning glory) and Ricinus communis (castor bean), 1–2 cm deep. The second most common error: letting the seed tray dry out even once. A germinating seed that dries out is dead. Bottom-watering — placing the seed tray in a shallow dish of water and letting the mix absorb moisture from below — is safer than overhead watering, which can dislodge seeds and create wet-dry cycles on the surface.
The plants most worth growing from seed in Hong Kong: Zinnia elegans (sow in March, flowers in June, keeps going until November if deadheaded), Cosmos bipinnatus and C. sulphureus (the orange sulphur cosmos is astonishingly heat-tolerant), Celosia argentea (the cristata and plumosa forms both thrive in our humidity), Gomphrena globosa (globe amaranth — drought-tolerant, long-flowering, excellent for cutting and drying), Tithonia rotundifolia (Mexican sunflower — will reach two metres in a single season and cover itself in blazing orange daisies), and Basella alba ‘Rubra’ (Malabar spinach — an edible ornamental climber with purple stems and glossy dark leaves that laughs at August heat).
For trees and shrubs, the seed game is different: it is an investment. Bauhinia × blakeana, the Hong Kong orchid tree, is sterile and cannot be grown from seed — every specimen in the territory is a cutting from a single tree discovered in Pok Fu Lam around 1880. But Bauhinia purpurea and Bauhinia variegata produce viable seed, and a tree grown from seed, planted in the right place, will outlive you. Delonix regia (flame tree) seed, scarified with sandpaper and soaked overnight, germinates within a week and grows a metre in its first year. Cassia fistula (golden shower) is similarly obliging. These are not plants you grow for next season’s flower display. They are plants you grow because someone should, and you happen to be the someone with the seed, the pot, and the patience.
The Plants That Want to Be Propagated
Some plants resist every attempt at propagation. Daphne odora, that exquisite winter-flowering shrub with a fragrance that stops people mid-stride, is one of them — cuttings rot, layers take two years and succeed perhaps one time in three, and seed is rarely available. Others seem to want to be propagated. They root in water on the kitchen counter. They root when a stem touches the ground. They root when you put them in the compost pile by mistake and come back a week later to find them thriving. These are the gateway plants — the ones that teach you the rhythm of propagation before you attempt anything difficult.
The top five plants that want to be propagated, for a Hong Kong garden:
1. Coleus scutellarioides. Take a cutting of any length, strip the lower leaves, place in water. Roots appear within five days. Pot up after two weeks. The plant will grow so fast you will need to take more cuttings. This is not a problem; it is an opportunity. The range of available cultivars in Hong Kong flower markets — ‘Kong’ series, ‘Wizard’ series, the astonishing ‘Fishnet Stockings’ with its purple netting over lime-green leaves — makes coleus collecting a legitimate hobby.
2. Sansevieria trifasciata (now reclassified as Dracaena trifasciata, but nobody calls it that). Leaf cuttings placed in water or directly into gritty mix root slowly but inevitably. A single leaf can be cut into 8 cm sections, each of which will produce a new plant from the basal end. It takes months, but it works with such reliability that I have stopped buying sansevierias entirely. Note: variegated cultivars propagated from leaf cuttings usually revert to plain green, because the variegation is a chimeral mutation present only in the meristem. To preserve variegation, divide the rhizome instead.
3. Pilea peperomioides (Chinese money plant). Produces offsets — small plantlets that emerge from the soil around the base of the parent — with such enthusiasm that anyone who owns one mature specimen will, within a year, be giving away babies to everyone they know. Detach offsets when they are at least 5 cm tall, pot individually, and water sparingly until established.
4. Epipremnum aureum (pothos, devil’s ivy). The plant that survives everything. Stem cuttings with at least one node root in water within ten days. Cuttings with aerial roots root even faster. Once established in soil, the plant will trail or climb depending on what you give it. The ‘Marble Queen’ and ‘Neon’ cultivars are worth seeking out.
5. Portulacaria afra (elephant bush, spekboom). A South African succulent that in its native Eastern Cape forms vast thickets capable of sequestering more carbon per hectare than some forests. In Hong Kong, it is a container plant, but one that roots from stem cuttings with a reliability that borders on the absurd. Let cuttings callus for three days — this is important; un-callused succulent cuttings rot — then insert into dry gritty mix and do not water for a week. Roots will form at the callus site, and the cutting will begin growing within a month.
The Failure File
I keep a notebook — a cheap exercise book from a stationery shop in Sham Shui Po, the cover now faded to an indeterminate grey — in which I record propagation failures. It is, by some margin, thicker than any notebook in which I record successes. This is not because I am a bad propagator. It is because propagation, like sourdough baking or learning a musical instrument, involves a predictable phase in which your failure rate exceeds your success rate, and the only way through it is to keep going while paying attention.
Some entries from the failure file:
April 2019. Michelia alba, semi-hardwood, 12 cuttings. Medium: perlite/coir. Hormone: Clonex. Result: all rotted within 14 days. Diagnosis: medium too wet. The perlite-coir mix, prepared with what I now recognise as excessive enthusiasm for watering, held water like a wrung sponge. The cuttings, which need moisture but also oxygen at the cut surface, essentially drowned.
October 2020. Rosa ‘Pierre de Ronsard’, hardwood, 6 cuttings. Medium: coarse sand. Hormone: 0.3% IBA talc. Result: 5 callused but never rooted, 1 rotted. Diagnosis: cuttings taken from flowering wood. Roses root best from stems that have not flowered — the carbohydrate reserves are higher, and the auxin balance favours root initiation. The one cutting I took from a blind shoot rooted the following spring.
March 2022. Camellia japonica ‘Drama Girl’, semi-hardwood, 8 cuttings. Result: 0. Complete failure across two different media and with and without hormone. Diagnosis unknown. Some plants simply resist, and Camellia japonica cultivars — as opposed to C. sasanqua, which roots readily — are among the most stubborn. I have since learned to layer camellias, which works, slowly, or to buy grafted plants and accept the expense as the price of sanity.
June 2023. Plumeria rubra, hardwood, 5 cuttings. Result: 5 rotted. Diagnosis: I did not let the cuttings callus. Plumeria cuttings exude a milky latex when cut; if planted immediately, this latex blocks water uptake and provides a substrate for rot organisms. The correct method: place cuttings in a dry, shaded location for one to two weeks until the cut end is dry and callused, then plant in barely-moist sand. I knew this. I was impatient. The plants punished my impatience, as plants reliably do.
The pattern across these failures is consistent: water management. Too much water, too little air, too little patience. The cuttings that succeed are the ones given conditions that balance moisture and oxygen — which is to say, conditions that more closely resemble the airy, well-drained environments in which roots evolved to grow. The single most useful piece of equipment I own, after the secateurs, is a spray bottle. It allows me to maintain humidity around the leaves without saturating the medium. A humidity dome plus a light hand with the watering can is the combination that turned my strike rate from roughly fifty percent to roughly eighty-five percent over the course of about three years.
A Garden That Makes Itself
There is a garden in Lam Tsuen, in the New Territories, that I visit once a year. It belongs to a retired botanist — let us call him Dr. Lau — who bought a plot of abandoned farmland in 1998 and has spent the subsequent decades filling it with plants, none of which he purchased. Every plant in the garden is a gift, a cutting, a division, or a seedling from another garden. The Bauhinia by the gate came as a seed from a tree in the Botanic Gardens. The hedge of Duranta erecta was struck from twenty cuttings taken from a single plant at Kadoorie Farm. The Bambusa vulgaris grove began as a single division from the garden of a friend who has since died. “The garden remembers him,” Dr. Lau said to me once, not sentimentally but factually. He meant that the bamboo is genetically identical to the plant his friend grew, and that in some biological sense his friend is still here, photosynthesising in the afternoon sun.
This is the deeper argument for propagation: a propagated garden is a garden with lineage. Every plant has a story, a provenance, a connection to someone or somewhere else. A garden bought entirely from a nursery is a collection of commodities. A garden propagated is a collection of relationships — between gardener and plant, between plant and its parent, between gardener and the friend who donated the original cutting. I am not being romantic. I mean this literally: the act of taking a cutting and growing it on creates a chain of custody that runs, unbroken, from the parent plant through your hands to the soil in your garden. That chain is worth more than the plant, because the plant will die eventually — all plants die — but the lineage can persist indefinitely.
Practical advice, to finish. Start with one plant that is easy to propagate. Master it. Take ten cuttings instead of one, because some will fail and that is normal. Label everything — a pencil on a wooden stick, not a marker pen that fades in the sun. The label should include the plant name, the date, and the source. Keep a notebook, because your memory is not as reliable as you think it is, and the lesson you learn in March will be relevant again in October, by which time you will have forgotten it. Be patient. The plants are not in a hurry, and neither should you be. And when a cutting finally roots — when you tip it gently out of its pot and see those clean white roots curling around the edge of the root ball — take a moment to register the satisfaction. You have made a plant. You have, at no cost beyond your attention, added something living to the world that did not exist before. The nurseries will still be there if you need them. But you may find, as I have, that you need them less and less.
