Whenever someone first learns about fruit tree grafting, the idea of grafting multiple fruit varieties on a single tree very often immediately crosses his mind. This idea is indeed fascinating. I am no exception. When I attended my first grafting training, my first question to my instructor was on this subject. He confirmed to me that the vast majority of his trainees ask similar questions.

Can it be done?

To make a long story short: yes, it works. You can watch this TED episode, or read this wikipedia page, about an artist that grafted 40 different varieties of stone fruits onto a single rootstock.

Which species are intercompatible?

You cannot just graft any tree species onto any other random tree species. In the example presented in the TED talk about the Tree of 40 Fruits, all the fruits mentioned are stone fruits from the prunus genus. Trees from the same botanical genus are usually compatible and can be grafted together.

You can also graft together trees that are not from the same genus, but are from the same botanical family. For instance, in France we traditionally use quince trees , from the Cydonia genus, as dwarfing rootstocks for most pear varieties, Pyrus genus. This is because pears tend to grow way to tall to be harvested safely, and in the past we didn't know of any dwarf pear varieties. But not all pear cultivars are compatible with quince rootstocks. Sometimes the pear dies after a few years, sometimes the incompatible pear tree survives multiple decades but with a very low fruit yield and a much higher susceptibility to fungal diseases. In such cases, the solution is usually to first graft a compatible pear cultivar onto the quince, and then graft the wanted variety onto the intermediary pear.

Apple trees (Malus genus) are neither compatible with quince trees, nor with pear trees, although all 3 genera belong to the Rosaceae family. But all 3 genera are somewhat compatible with hawthorn rootstocks (Rosaceae, Crataegus), and so are many other Rosaceae trees. You can find a complete list of fruits compatible with hawthorns on the permies.com forum.

There are also reports of successful grafts of chestnut onto oak rootstocks, but most people who tried replicating the experiment failed. And many have tried, as grafting a chestnut onto an oak would enable producing chestnuts on limestone derived soils, which is normally impossible. This example illustrates the fact that you can find a compatibility between 2 very specific individual trees of different genera, while most other combinations of trees randomly chosen in these same genera will be incompatible.

What I'm getting at is that a lot of combinations of fruits on a single tree are technically possible, but that they are not always easy to achieve. Sometimes, intermediary grafts and marginal rootstock species must be used.

So, why can't you find multi-cultivar or multi-species fruit trees in most local nurseries? The answer to that question is that multi-grafting fruit trees is actually very impractical for nursery production. There are two main issues with this practice: the extra years required to produce marketable multi-fruit trees, and the complexity of pruning to maintain them in the long run.

Why is it more complicated to prune a multi-fruit tree?

When you plant a young single fruit tree, you can basically let it grow with no or very little pruning. Each year, the apical bud will grow a straight vertical shoot, while a few lower shoots on the previous year's wood will grow lateral branches.
The apical bud produces a hormone called auxin, that inhibits the growth of all the lower branches. Therefore the lower branches will grow sub-horizontally and will produce fruits rather than wood.

The natural shape of most trees, if allowed to grow without pruning, resembles a Christmas tree. All primary branches grow outwards from the trunk, and don't interfere much with each other. Pruning is limited to removing low branches to make it easier to drive machinery underneath, keeping the number of branches to a reasonable number so as to produce a lower number of larger caliber fruits, and removing primary branches that grow sub-vertically and might compete with the apical bud.

On the other hand, if you want to grow multiple cultivars on a single tree, this natural architecture will not work well. You should not keep any given cultivar in the apical position, as it would end up inhibiting the growth of all other cultivars. You will want to decapitate the tree to encourage all current primary branches to start competing for the apical dominance, and graft the cultivars on those primary branches, or on secondary sub-branches if you want to graft many cultivars.

As a result, you will have to basically manage multiple trees growing from a unique trunk. Some of the lateral sub-branches of each cultivar will grow towards the center of the tree, and will interfere with each other and shade each-other.

This situation will require a much more work-intensive annual pruning. Excessive pruning may favor wood growth the following year at the expense of fruit yield. Properly pruning a multi-axis fruit tree is a delicate balance, that requires years of experience, or a thorough training on techniques that are more of an art than a science.

How long does it take?

In the first year after sowing a seed or a planting a tree freshly produced from a cutting, you should expect a single shoot to grow straight up, without lateral branches, with leaves growing directly on the trunk. Initially the wood will be green and soft, and towards the end of the summer it will start turning into hardwood. The following year, the first lateral branches will start growing on the previous year's growth, while a new shoot will grow straight up again from the apical bud. These lateral branches will turn into hardwood towards the end of the second year. If at this stage you consider that you have enough branches to accommodate the number of cultivars that you want to graft, you can proceed with the grafts in the spring of the third year, and hope to sell the tree at the earliest in the beginning of the winter of the third year.

The tree will use up nursery space for 3 years before you can sell it. As a comparison, a regular single-variety fruit produced in a streamlined manner can use up nursery space for only a single year. The nurseryman buys and plants the rootstocks in the winter, grafts the wanted cultivar in the spring and sells the tree in the following fall.

I can imagine of a method that would reduce the multi-grafting timeline down to two years, but I have not tested it yet, and it is limited to a low number of cultivars per tree. All cultivars would have to be grafted directly on the trunk/shoot in the spring of year two, rather than grafted on higher branches in year 3. The grafting method would need to be shield-budding or chip-budding.

As you would need to graft multiples buds at different heights on the tree, it would not be easy to force the grafted buds to start growing a branch. Causing a bud to start growing is usually done by decapitating the tree, so as to stop the terminal bud from inhibiting lower branch growth. The decapitation would probably only cause the highest buds to start, and not the lowest ones. To force the lowest bud to start their growth, a method could be to use notching, as described in the following video starting at 15:45.

Notching works by shielding the bud from the phloem sap that flows down right underneath the bark, and transports the auxin hormone that inhibits growth of lower branches. The effect is that the bud right underneath the notch thinks it has become the terminal bud, and is incentivized to grow in height to ensure the tree's access to sunlight.

Why I believe it may be worth it anyway

In my opinion, there is probably an underestimated market for multi-fruit trees. Impractical as it may be, multi-fruit are probably easy to market. I believe that a large proportion of buyers of fruit trees are not buying because they thought thoroughly about the best strategy to produce the largest amount of food. Many buyers probably buy trees on impulse, because of the originality of the cultivar.

The first marketing argument I can think of, is that these trees allow to grow multiple cultivars in a small yard. A variant of this argument is to say that these trees allow to produce fruits over a longer period, as different cultivars ripen at different times of the year. For instance, you can find apples starting in the early summer for the earliest-ripening cultivars, and the latest-ripening cultivars will need to be harvested at the end of fall, before the first frost, and finish ripening for months in a cellar. Having multiple cultivars ripening at different times of the year means that you can continually harvest fresh fruits, without needing to store or to process them for preservation.

A second argument is that most fruit trees cannot self-pollinate. This means that you usually need to plant at least two trees with compatible flowering periods to obtain a harvest. Grafting 2 or 3 cultivars on the same tree is an compelling solution to this problem.

The third argument is the novelty factor. Showing your guests trees that produce at the same time plums, apricots and peaches is a good conversation starter.

It is also worth mentioning that multi-fruit trees can be used to preserve multiple heirloom varieties on a small piece of land. But this practice requires a very thorough labeling and mapping of each cultivar. If a label breaks and falls off, there is no chance that you will find the cultivar again on a 10 varieties tree in a 10 trees orchard. Even the best quality labels will eventually fall off the tree, after being wiggled for years by the wind.

The last reason to practice multi-fruit grafting that I will mention in this article, is that you can speed-up the fruiting by grafting a fresh shoot from a young tree onto an older tree. This can prove to be very useful if you are doing varietal selection. It can significantly speed up the process, as compared to just waiting for each seedling tree to produce its own fruits, which can take more than 10 years.