Hülya

Concretely, for solid stock:

Thickness: a minimum of around 12 mm, and 18 to 20 mm is much better. Below that you are relying on a thin section to resist splitting at the screw, and the bevel becomes a knife edge.

Species: a hardwood if you can, for anything substantial. Something dense and reasonably stable rather than the softwood most people have in the offcut pile. Avoid species prone to splitting when screwed near an end.

Grain: straight, running along the length of the cleat. This is more important than the species. A board with grain running out at an angle across the width is a split waiting to happen, and it will look perfectly fine until the day it does not.

Movement: solid wood moves across the grain with humidity, plywood barely does. On a long cleat that is a real consideration — a two-metre solid cleat will change width slightly through the year, which is fine, and will also want to cup, which is not.

There is a further refinement about how the board was cut from the log, which is the difference between the two orientations people usually mean by "vertical" and "flat" grain.

Quartersawn stock — where the growth rings run roughly perpendicular to the face — is more stable, moves less with humidity, and is less prone to splitting along a ring line. Flatsawn stock is more common, cheaper, and moves more.

For a small shelf support none of this decides anything. For a long structural member, or anything where a split would be expensive, it is worth knowing how to read the end grain and choose accordingly.

The practical version: look at the end of the board. Rings running roughly straight across from face to face is the more stable orientation, and rings running in wide arcs across the face is the one that will cup.

On performance, setting the other issues aside: it is mediocre rather than terrible, and it is not doing what people think.

Oil-based treatments work mainly by repelling water, which reduces the moisture swings that cause splitting and slows fungal decay. Used engine oil does that reasonably well and it does penetrate.

What it does not do:

• It offers no ultraviolet protection, so the surface degrades in sunlight regardless.
• It is not a preservative in the sense of being fungicidal or insecticidal. Any claim that it protects against rot is confusing water repellence with preservation.
• It never fully cures. It stays oily, attracts dirt, transfers onto anything that touches it, and remains flammable.

And the important point for your project: your timber is already pressure treated. The preservative is in the wood. It does not need a preservative coating, and adding one contributes very little to longevity.

One thing worth reconsidering before any of this: whether to trim after glue-up at all.

The alternative is to cut the plywood accurately to size first and glue it in position. That removes the trimming problem entirely and it is how a lot of panel work is done. The cost is that it needs accurate cutting and careful alignment during glue-up, with no margin for the panel shifting as you clamp.

Oversize-and-trim exists precisely because that alignment is hard. So the question is whether you would rather solve an alignment problem before the glue or a trimming problem after it.

For a door with a bevel on the long edges, I would probably still go oversize and trim, because getting a bevelled panel positioned exactly during clamping is worse than trimming it afterwards.

On span: solid hardwood at typical desk thickness is quite stiff, and the spans in your dimensions are still substantial. Two things help without adding legs:

An apron or rail under the front and back edges of each top. A rail on edge is enormously stiffer than the same wood laid flat, and it is what stops a desk top from deflecting under a leaning elbow. On a knock-down design the rails can be bolted rather than glued.

Keep the overhang modest past the outermost legs. A long unsupported overhang at the end of an arm is where deflection is most noticeable, because it is the point furthest from any support.

For solid wood specifically, also allow for seasonal movement across the width. Fix the tops to the frame with fittings that permit the panel to expand and contract — slotted holes, buttons, or figure-eight fasteners. Bolting a wide solid panel rigidly at both edges is how tops crack, and a desk that gets moved between climates is exactly the case where it happens.