Several possibilities:
- It's just oxidized. Tin dioxide is clear/colorless, tin oxide is variously (yellow to red to gray) colored, and any copper present can oxidize brown as well. Moreover, these are likely very thin layers (100s nm) so that, rather than being opaque, they are transparent, and interference effects -- as well as some dyeing effect from their natural absorption colors -- come into play. You often see a faint rainbow sort of coloration on clean, heated copper. There could still be tin there, it's just hard to see under the oxide(s).
- The tin is completely gone. Having oxidized as above, the flux has dissolved that oxide, leaving the bulk metal beneath. However, this has the problem that, the trace is still very dark, and usually flux leaves a pink (clean copper) surface. Perhaps insufficient flux was present, unable to clear away all the oxides and only the initial oxides were dissolved?
- The tin simply diffused into the base metal. Metal atoms are able to move around; that's pretty obvious for low-melting alloys like tin, but it's less obvious for copper which remains solid during the process. Normally during soldering, at the interface between solid copper and molten tin, an intermetallic layer forms. Intermetallics are compounds of metals, crystalline, and generally hard and brittle. The layer is thin, a few µm. But electroless tin can also be quite thin (comparable thickness). On heating, the metallic tin might simply disappear into the copper surface. (These intermetallics are still silvery colored, but if the tin diffuses deep enough, a yellow to pink "bronze" layer could be left: see also the classic golden penny reaction which uses zinc in a similar way.)
- Unlikely in my opinion, but perhaps still a possibility -- @Math Keeps Me Busy's suggestion of the tin beading up. The electroless layer should be very thin indeed, so the beads would be microscopic, and probably oxidize very quickly anyway.
I'd say some oxidation is still needed to explain the final color, so the exact answer will be a mix of these.
On a practical note, since electroless tin is so thin -- going over it with solder is still a good idea (as can be seen!). You can wipe on fresh solder quite quickly, and wipe off excess with solder wick if necessary.
And in that case, the initial tinning isn't necessary at all: you need to prepare a very clean copper surface for plating, but clean copper tins beautifully with a little flux and fresh solder!
I find hand-made boards, tinned, are still flat enough not to be annoying to place components and reflow solder, at least with a hot-air machine and some tweezers to adjust parts as I go. In fact I usually leave a little excess solder on the board, and let that bond the components; only flux is required (not paste), and maybe a little touch-up work with the soldering iron.
A note on flux chemistry: these are either acids, complexing agents, or both. A "complex" is when a metal atom is relatively strongly bound to some molecules, through a more unusual kind of chemical bond (technically, metal d-orbitals to lone pairs of O, N, etc., among other mechanisms). Acids dissolve oxides by direct reaction (forming a salt, which may dissolve in a solvent, or melt or evaporate), and complexes can dissolve oxides by attaching to metal atoms (to much the same end).
Heavier organic acids, such as those found in natural pine resin, happen to be very well suited to the tin-lead-copper system; usually with a few modifiers to make them form less gunk, stay better in place, maintain reactivity, etc. Hence the usual "activated" rosin types. Some lead-free fluxes may be partly/entirely synthetic; they use the same mechanisms, but different chemicals that perform better at the higher temperature.
Oxidation, is very much a normal and natural part of soldering, and is why flux is mandatory for almost all work (in air). Soldering can in fact be done in inert gas or vacuum, given perfectly clean metals -- metals "like" each other very much, and are more than happy to spread out over and bond together, given the chance. The problem is contamination, usually surface oxides.
Edit: some additional information some may find of interest: Wettable Flank Plating (particularly pp.5-6), gives micrographs of the plating and interface layer. The baking step at 150°C for 8hr will be equivalent to ballpark 10s of s at 250°C (assuming an Arrhenius extrapolation is valid over such a range of temperatures), so similar changes can be expected during soldering.
