Wood, especially live wood, is not a non-conductor at those high voltages. I had occasion once to watch a 10KV (14,000 peak Voltage)) oil burner ignition transformer start to burn a carbonized trace from a connection to its HV terminal through the seasoned wood bench top toward ground. I shut it off right quick.
Sure, wood shows an "infinite" resistance to your ordinary multimeter, but that's only putting out, max, say, 9V from its battery. And even in the old days, when wood was used in "breadboarding" circuits* at normal voltages (say, up to about 2-400 Volts) that was OK, but when you're talking millions of volts, the current will be enormous, especially through the live (sap-filled) trunk of a tree.
I point out that you can buy what is called a "Hi-Pot" tester**, which is used to test for "microscopic" leakages in printed circuits, etc. But these put out thousands of volts so that the tiny leakage currents can be sensed by the instrument and give you a readout.
So while well-seasoned wood can be pretty "nonconductive," it really isn't, and in the case of a blasted tree, much of the very large current mega-Amps) does indeed go through the internal wood of the tree causing instantaneous heating, causing instantaneous generation of lots of steam, which results in an explosive splitting of the wood.
Terry, 230RN
* An old-time "breadboarded" circuit. (Looks like a super-regenerative receiver circuit, but it's hard to tell.) It is obvious where the term "breadboard" originated.
http://tangentsoft.net/elec/bitmaps/ab0cw-breadboard.jpg** "Hi-Pot" stands for "High Potential." Remember that "potential" is another word for "Voltage," much as "tension" is... as in "high tension power lines."
ETA Tallpine beat me to it. Glad he put that smiley in after "jiggawatts." Sort of like my "microscopic" leakages.