Tag Archives: Architecture

Rauhs Model

Rauh’s model represents the Cupressaceae family (cypress, juniper & redwood), some Araucariaceae, the Pinaceae family including most Pinus species, the Podocarpaceae family as well as angiosperms such as oak, maple and ash. It is a very common model for trees we encounter in bonsai.

The architecture according to Rauh’s model includes a monopodial trunk (one which continues to extend, and does not terminate) which grows rhythmically (on a seasonal cycle) and so develops tiers of branches, the branches themselves morphogenetically identical with the trunk (ie. they develop in the same way). Because the branches are identical, the trunk can be less dominant in this form and another stem can take over if the trunk is removed or damaged. Flowers and reproductive organs are always lateral and without effect on the growth of the shoot system.ref Often these are on short shoots.


An essential feature of Rauh’s model is that branches develop mainly by ‘prolepsis’, from dormant lateral buds close to the resting terminal bud.ref Prolepsis in this context means ‘the discontinuous development of a lateral from a terminal meristem to establish a branch, with some intervening period of rest of the lateral meristem’. So basically there is a gap or period of dormancy before the bud extends to form a branch. Whilst this might seem obvious to European readers, actually this mode of development is not what happens in other parts of the world, particularly the tropics, where continuous growth occurs, and this difference creates differences in the tree architectures visible in those different places.

It was noted in one study that Apple trees follow Rauh’s model during their juvenile phase but a different one during their reproductive phase (ie. their flowers terminate shoots and affect the branching after this point).ref

Massarts Model

This architectural model is associated with many conifer families including Abies, Picea, Sequoia, Metasequoia, Cedrus, Taxodium, Taxus, Cephalotaxus, Ginkgo & Ilex aquifolium. The pattern for this architecture is a vertical, dominant trunk with rhythmic growth and which consequently produces regular tiers of branches at levels established by the growth of the trunk meristem. Branches are plagiotropic (horizontal) either by leaf arrangement or symmetry. The position of flowers/cones is not significant in the definition of the model (which means they don’t terminate the branches and have any effect on the structure).


One study in Indonesia looked at rainfall ‘stemflow’ and ‘throughfall’ (basically how much water runs off the tree into the ground causing erosion) and found that the Massart’s model tree (Pterocarpus indicus) had the highest leaf surface area and caused the least erosion from water runoff.ref However the study used angiosperms with broadleaves and not gymnosperms with needles.

Tree Architectural Models

Halle & Olderman in the 1970s created a model of 23 types of architectural models to which all tree species are believed to conform. They started with the idea that the shoot apical meristem/s (“SAM” – the primary growing tip) is/are the ‘treemakers’.ref The behaviour of the SAM over time determines the form of the tree. They identified four types of meristems (active growing shoots) which led to different forms – those with a single meristem (like a palm), those with modular construction which follow a precisely repeating pattern, trees with trunk/branch differentiation and those in which the meristem changes direction to produce both trunk and branch.ref These depended on factors like whether the SAM is reproductive or vegetative, whether it grows vertically or horizontally, whether growth of the SAM is continuous or rhythmic (ie. has a period of dormancy or a growth pause) and the chronology of growth of each meristem.ref

Below is an updated version which includes some models added later (and one which was removed as nobody found a real tree which conformed to the theoretical model).


Each architectural model has a unique combination of growth, branching, axis differentiation and position of sexual structures.ref Not all of these are relevant to bonsai – as you can see Holttum & Corner don’t branch and Tomlinson & Bell branch underground. Many of these models are relevant for palms, cycads and tropical trees which aren’t commonly used for bonsai.

Two of the architectural models represent many of the trees used in bonsai at least in the Northern Hemisphereref – Massart’s and Rauh’s model – their main difference is in the branching angle with Massart’s at an angle to the stem and Rauh’s growing upwards. In both models the trunk is monopodial – it keeps extending upwards and is dominant.

Massart’s model represents Abies, Picea, Sequoia, Metasequoia, Cedrus, Taxodium, Taxus, Cephalotaxus, Ginkgo & Ilex aquifolium. More forms and discussion of Massart’s model are represented in this post.

Rauh’s model covers the Cupressaceae family (cypress, juniper & redwood), some Araucariaceae, the Pinaceae family including most Pinus species, the Podocarpaceae family, as well as angiosperms such as oak, maple and ash. This is shown in more detail in this post.

If you’re looking to understand the architectural model for your particular tree, you might consult this book – it mainly focuses on tropical trees but gives some pointers on working it out.

Some other models include Attim’s model for Eucalyptus, this is similar to Rauh’s model but follows a continuous growth pattern – in these trees as one leaf expands outside the bud it is replaced by a new bud initiated at the shoot apex.

Troll’s model is applicable for hemlock, acacia, beech, where “axes are all plagiotropic (ie. horizontal), the architecture being build by their continual superposition; main-line axes contribute part trunk, part branch, the proximal part becoming erected, most often secondarily after leaf fall” – it is believed that reaction wood is involved in determining this architecture (the type of wood created to stabilise a branch against gravity – compression wood developed under the branch in the case of gymnosperms, and tension wood developed above the branch in the case of angiosperms). Hemlock is a gymnosperm with this model.

Troll’s Model

Finally, trees can move from one model to another when they move from their juvenile vegetative phase to their reproductive phase. For example Apple has been found to conform to Rauh’s model when juvenile but Scarrone’s when reproductive:ref

Root structure and architecture

So we know what roots achieve for a tree, but how are they structured? To start with tree roots are either woody or non-woody. Woody roots have undergone secondary thickening and are long-lived, like the trunk and branches, and provide the structural framework for the tree.ref

The ‘root collar’ is the area on the tree’s trunk where the roots join the main stem, and where there is typically a root flare (the root collar is still part of the trunk though, which is why it shouldn’t be buried in soil).ref At the base of the root collar, there are usually five or more primary structural roots that “descend obliquely into the soil before becoming horizontal within a short distance of the trunk” and these taper rapidly within 1-2m of the trunk.ref These are known as lateral roots since they grow in a lateral (horizontal) direction.

In his book ‘Trees, Their natural history’, Thomas says that trees develop a root plate, which is wide and shallow (vs the commonly held view of a root ball, which is only applicable to certain trees). Having a wide root plate helps trees achieve two of their main goals – to support and strengthen the tree against wind & weather, and to access waster and nutrients which are concentrated in the top layer of soil.

According to Thomas, root systems are more variable than shoot systems because the underground environment is more variable than aboveground. When roots encounter an obstacle underground, they fork, and as they fork and expand underground the main lateral roots can fuse into each other. This creates a criss-crossing of roots, which provides greater structural strength than if the roots were not connected. Roots can also connect to other trees’ roots (and even detect if they are ‘kin’ or not).

Structural lateral roots can develop into buttress roots, which have been found to provide tension strength in high-wind situationsref – as a little girl growing up in Australia the best fun could be had climbing over the huge roots of the Moreton Bay Figs (Ficus macrophylla).

Ficus macrophylla in Kings Park, Perth Western Australia

In addition to lateral roots, most bonsai enthusiasts will have encountered the dreaded tap root. A tap root is the root generated by a new seedling (Thomas), which grows downwards and becomes a thick structural root. The tap root can become dominant in the root system and be a total pain for bonsai – it often generates its own lateral roots, creating a second root plate and makes it hard to get the tree into a bonsai pot. But luckily according to Thomas and others (and personal experience) the tap root isn’t necessary and can be removed. This is always best done sooner rather than later so that energy is not diverted to its growth vs the roots you do want to keep.

As well as tap roots, other structural roots trees create include sinker roots which go deeper into the soil (often to find water), can set up a secondary root plate, and also grow back upwards to create a ‘root cage’ (Thomas).

Susan Day et alref say “although structural roots comprise most of the root biomass, they account for a small percentage of total root length and root surface area.” The remainder of the root surface area is comprised of fine roots, which are the main mechanism for the tree to extract water and nutrients from the soil. Connecting the main structural roots to the fine roots are a network of tapering roots which branch off the structural roots.

A study of nine North American tree species found that in eight species roots <0.5 mm in diameter accounted for >75% of the total number and length of roots assessed.ref Thomas quotes a study on Douglas fir estimating that 95% of the total root length comes from roots <1mm and about half less than 0.5mm.

As noted above the fine roots are non-woody and don’t undergo secondary thickening – this means they die and are replaced by new roots. It’s quite hard to measure this and there is differing information about fine root lifespan, but the above study found the average fine root lifespan to range from an average of 153 to 359 days. This is also expressed as a ‘fine root turnover rate’ and based on this data table fine roots of gymnosperms turn over more slowly than angiosperms (some Pinus species 20% per year vs beech 100% per year).

The fine roots are concentrated in the top part of the root plate, where most of the nutrients and water are located (20-30cm of soil, and the leaf litter & humus if present). Like the stems aboveground, the roots are constantly developing and growing, with new root tips being created by the root apical meristem (RAM) (this is described below). How the root goes about absorbing water and nutrients from the soil is covered in this post: How roots absorb water & nutrients.

These fine roots are what we are trying to encourage in bonsai as they enable the tree to extract the most water and nutrients from their environment, while still fitting into a small pot. What we want in the fine roots is lots of branching and ramification – just like aboveground – read more about encouraging this in ramification of Roots (lateral root development).

The below diagram shows the ratios of leaf, stem and root biomass to total tree mass for a data set including 3700 ‘woody’ plants (ie. trees!)


As you’ll notice, the larger the tree gets, the more the stem (trunk) represents of the total biomass. However the ratio of roots to total biomass stays within a range from 16% to 40%. By comparison the ratio of leaf mass has a much wider range all the way from 60% down to 2%. So there is a certain baseline amount of root biomass needed to maintain a tree.

This mass is mainly made up of the structural roots, as although the fine roots comprise the vast majority of the root surface area, they are very light in comparison to the woody roots.

So bonsai nerds, what to make of all this? Key info is the fact that fine roots die and regrow on a regular basis – and – kill that tap root! Help your tree be more stable by encouraging a root plate of connected structural roots, and you won’t need a deep root ball or a tap root. Nebari and root mass should be around 20% of the mass of the tree for an old tree look.

Should I remove flower buds or fruit?

That depends what tree you have and what you are trying to achieve. Obviously if you have satsuki azalea, you probably want to leave the flowers on the tree! If you have a crabapple, personally I don’t think there is much point if you don’t let a few fruit form. And I am really partial to rose-coloured larch cones. All trees form some kind of reproductive organs, whether they be conifers with their strobili (cones, either pollen or seed forming), ginkgo with their ovules, or angiosperms with their flowers and fruit. Some are almost unnoticeable and others are right in your face. Bonsai wisdom sometimes says these should be culled or removed entirely in order to avoid draining the tree of its energy.

When considering this question we need to understand the idea of resource ‘sources’ and ‘sinks’ in plants. A source is a material producer and exporter, and a sink is a material importer and consumer.ref See the below table for sources and sinks in trees. As you’d imagine, leaves are a major source of carbon and a sink of inorganic nitrogen (nitrogen as a macronutrient). Roots are a source of inorganic nitrogen and leaves are a sink. So what about fruit, seeds, and flowers, which supposedly drain the tree? As you can see they are major sink organs – but not only sink organs…they are also source organs!


Let’s have an interesting little diversion – did you know that it’s not only leaves which photosynthesise? This fascinating studyref looked at the photosynthetic activity of (a) ears of wheat (b) sycamore seed pods (c) a green tomato (d) unripe and ripe strawberries (e) a greengage (f) unripe cherries; and (g) a green apple. The images below were taken using fluorescence imaging and anything with a colour indicates that there is photosynthesis taking place – with the red and orange areas the strongest. Check out the sycamore seed pods!


How the heck can this happen – well there are various theories about the mechanism (including recycling CO2 from respiration, and the presence of stomata on fruit) but the point is that maybe seeds and fruit, particularly if they have periods when they are green, don’t act as such as sink as we might think, and for a period are acting as a source and not a sink.

This study states that “reproduction in Beech does not deplete stored carbohydrates, but it does change the amount of nitrogen stored” and this study found that “fruiting is independent from old carbon reserves in masting trees”ref which basically means that fruit uses current year photosynthates/energy and doesn’t actually deplete reserves.

On the other hand this study found that Douglas fir tree rings were narrower in years when they bore many seed-conesref and this one mentions that “experiments with apple trees have shown that roots can die from lack of carbohydrate supply when they are over cropped”ref

All living things have processes for managing and balancing resource allocationref and this is likely an evolutionary differentiator. In trees, resource availability limits the amount of fruit which is allowed to develop – even pollinated flowers may not develop into fruit if the tree does not have enough resources available – these could include energy, or nutrients.ref So to an extent the plant itself manages the resource allocation.

To complicate matters further many trees use a ‘masting’ strategy for reproduction, which means they have years where many more seeds are produced, often synchronised with other trees of the same species. One theory for how this happens is that the weather influences how pollen is distributed – in beech windy conditions lead to mast years whereas in oak short pollen seasons do.ref Temperature and precipitation also affect pollen production and distribution (high temperature increases pollen production but high precipitation washes it away).ref In this study on Japanese oak, “high seed production never occurred in two successive years, but successive years of low abundance were observed several times between 1980 and 2000.”ref

Overall there are a lot of factors interacting when it comes to reproduction. Photosynthetic seeds or fruit can contribute to carbon production, and may use only current year photosynthates, so the tax may not be as high as thought, but there is some evidence that reproduction can divert energy from roots and foliage.

If you are really focused on trunk growth, branch structure or foliage development on your bonsai tree, you might want to divert the energy from reproduction to these areas by removing some or all reproductive organs, until you are happy with the trunk/foliage. At this point then you could then let the tree reproduce (noting that removing cones one year will cause more cones to develop the following year)ref.

ginkgo bud


Buds are the “small lateral or terminal protuberance on the stem of a vascular plant that may develop into a flower, leaf, or shoot.”ref Buds are responsible for primary growth, and are created by meristem tissue (a meristem is an area of stem cells which differentiates into different types of cells).

If you look inside a developing bud, you can see the starting points of the different cells which will arise – they can be vegetative buds (shoots & leaves) or reproductive buds (flowers in angiosperms or strobilus/cones in gymnosperms). Below is an image of a Jack Pine terminal bud which has many lateral vegetative buds on the sides.


When shaping your bonsai, you want to know where buds may appear, so that you can encourage the direction of growth and shape you desire. Predicting bud location is relatively easy in angiosperms, which follow a relatively reliable pattern in their growth. Bud growth is more unpredictable in gymnosperms, but many of the following guiding principles remain.

Firstly, there are different bud positions:

  • The terminal bud is at the end of a stem or branch and this is the growing tip which makes the plant grow larger.
  • Axillary buds develop along the stem during the annual growing season according to the architecture of the tree (see below for more); within this, preventitious buds are axillary buds which are dormant and then develop in a later season.
  • Adventitious or epicormic buds are buds which do not develop according to the repeating architectural pattern – they arise spontaneously from previously non-meristematic (growing) tissue which can be anywhere on the tree. They are unpredictable as described in this post.

Below are some examples of angiosperm buds. The terminal bud is on the end of the shoot, this comes from the shoot apical meristem (SAM). Then there are axillary/lateral buds which occur along the shoot – in angiosperms these develop in the leaf axils (a position adjacent to where the leaf is attached).


Bud behaviour depends on a tree’s architecture, which is genetically determined – that is, it will be very similar for trees of the same species, albeit also affected by the environment. There is a lot of research out there about tree architectures, much of it pioneered by Halle & Olderman in the 1970s, there is even a mathematical model which can be used to represent the architecture of a given speciesref. As explained in this excellent articleref, “regular development of each plant represents the growth of repeating units – ‘phytomers’…a typical phytomer consists of a node, a subtending internode, a leaf developing at the node sites and an axillary bud (also called lateral buds) located at the base of the leaf”.

Each type represents a pattern consisting of a shoot with one or more leaves in the same arrangement. In some trees growth is repeated in a sustained way throughout the growing season (a single flush of leaves), whilst conditions are right. In others there are alternating growing and resting stages (multiple flushes of leaves). During the resting stage, new leaves and shoots are being created inside the budref. I’ve copied some of the main architectural models into this post: Tree Architectural Models

An important part of the phytomer pattern is the leaf arrangement, known as the phyllotaxis. Leaves can grow singly at one position on a stem, or they can grow in whorls where two or more leaves appear at the same position arrange around the stem. When leaves grow singly they spiral around the shoot to optimise their light capture – apparently using the ‘golden angle’ of 137.5o ref.

The leaf arrangement on your tree is important because each leaf axil (the base of the leaf) should be the location of an axillary bud (although in gymnosperms these can be missing). These are key to bonsai because they become new shoots (with leaves or flowers). They develop in the position just above where a leaf used to be; when it falls off, a scar is left and a bud generates above the scar.ref In fact what is happening is a continuous bud genesis, so when you have a bud about to burst, it already has embryonic buds developing at its base – this is why buds look like they form at the leaf axil (in fact they formed on the previous bud). Your new branches and leaves will generate from these positions, and dormant buds may be located here. Read more about buds in angiosperms here, and in gymnosperms here.

The growth of an axillary bud (and its embryonic buds) can be suppressed by its neighbours – this is how ‘apical dominance’ works. It used to be thought that in apical dominance, the shoot closest to the sun emitted hormones which suppress the growth of buds lower down the plant, ensuring that it gets the most resources. This research group at Cambridge University study the development of axillary shoots and their research says “shoot apical meristems compete for common auxin transport paths to the root. High auxin in the main stem, exported from already active meristems, prevents the activation of further meristems”ref. This results in axillary buds going dormant and becoming ‘preventitious’ buds, but they are still available to grow later if conditions change. According to this article, apical dominance in trees only works on buds in the current year of growth due to the slow movement of the hormone auxin through the treeref – meaning that current year buds on a branch are suppressed by the terminal bud on that branch and not by the main leader. HOWEVER, it has recently been found that auxin does not move fast enough to have this effect, and instead it is driven by sugar flows to the apical meristem.ref The effect of apical dominance remains, however it is now thought that sugar flow drives this and not auxin directly.

Encouraging axillary bud growth is a way of increasing ramification on a bonsai, as it can create multiple shoots instead of just the terminal buds. If the terminal buds are removed, axillary buds get the chance to grow, often more than one.

Application of exogenous cytokinins (benzyladenine) has also been shown to increase bud initiationref (see my post on ramification of Branches and Foliage for some substances containing benzyladenine).

Equally, looking at how the leaves are arranged, you can work out where new shoots will arise from existing stems. By removing the buds or shoots not meeting your design, you can encourage shoots to grow in the direction and position that you want. But it’s not enough to know about bud position, you also need to know what kind of bud is present – a vegetative or a reproductive bud, and you need to know the difference between short and long shoots – more here.