Hybridizing accomplishes that mixing of genes much more efficiently. Isolating a particular characteristic can sometimes be done by back-crossing F2's, but almost always results in sterile offspring. They can be propagated by cuttings, tissue culture, etc., but are also usually less vigorous than non-sterile plants. But not always. Of course Acer p. takes ~15 to 20 years to flower...
I'm hoping that you're talking about maples only, because sterile F2's are pretty uncommon in the plant world.
As for grafting, the 'mixed' genes usually aren't mixed genes, but they are mixed cell cultures: individual cells from both plants, mixed in a single physical structure. Meaning that pluripotent cells (like stem cells, and plants have a lot of those and move them around a lot) are moved between the two varieties. When testing a sample of cells, it's pretty hard to isolate a single one. More often, it's clumps or clusters of cells that are used. Those can show up in genetic analysis as 'originating from both plants' because by default, that's how genetic testing works: we use a known sequence that differs between the type 1 and type 2.
Let's do a small exercise, but in language instead of genetic code.
We need a start codon, we type this out as START
We need a genetic code that varies between the two, let's say GENCODE1 and GENCODE10001.
We need a terminate or stop sequence, let's say STOP
We need a mix to polymerise (synthetise) the sequences. So we can check them on elektrophoresis gels or sequences.
What do we want to know?
1. We want to know if every single type is present.
2. We want to know if both types are present.
3. We want to check if our mix works, so we use a piece with random code as a control.
Mix 1: Polymerizes type 1 only.
Mix 2: Polymerizes type 2 only.
Mix 3: Polymerizes type 1 and 2.
Mix 4: Contains our control sample for mix 1
Mix 5: Contains our control sample for mix 2.
Mix 6: Contains our control sample for mix 3.
Mix 7: Contains no DNA but all the other components to check for contaminations/flaws.
Results:
Mix 1: Positive. The code we sequenced is START-GENCODE1-STOP
Mix 2: Positive. The code we sequenced is START-GENCODE10001-STOP
Mix 3: Positive. The code we sequenced is START-GENCODE1-STOP and START-GENCODE10001-STOP
Mix 4: Positive, our mix works. The code we sequence is START-RANDOMCODE-STOP
Mix 5: Same as 4. The code we sequence is START-RANDOMCODE-STOP
Mix 6: Same as 3. The code we sequenced is START-GENCODE1-STOP and START-GENCODE10001-STOP
Mix 7: Negative, no DNA has been polymerized.
Now due to limitations in sequencing techniques, Mix 3 will most likely be processed as something like START-GENCODE100-STOP. Background of sequencing: some types of sequencing make use of the DNA being polar. In a way, DNA is dragged across a laser beam, letter by letter, to generate a sequence of the code. Like some old telegraphs work actually. Sequencing sometimes works by calculating the average of all parts that have crossed that molecular-sized laser beam. If it reads different letters on the same location at the same time, it will put out a weird, non-DNA letter like S or R. A technician later has to decide, based on how the signal 'looks' on what letter was actually read. I have done this for hours, weeks on end. And it's hard to see. But if there's a "00" added to a code, nothing screws over that signal. It will read out as a hybrid, but it's actually just a sample containing DNA from mixed sources. I too have had that happen.
A practical way to solve this, would be to isolate single cells, let those grow until you have enough DNA to harvest and to actually get a solid output, and then sequence them. This takes 6 weeks at least, and will cost a couple of thousand dollars more.
This is when only sequencing is used. When using electrophoresis gels, we can distinguish lengths between DNA strands. This has an accuracy of around 10 basepairs - It's been a long time, it might be more accurate nowadays. Meaning that if one code is 10 letters longer than the other, it will show. But if it's less than a 10 letter difference, we just end up in the same blob.
Our example codes differ just by 4 letters, meaning that if we use the polarity to electrically drag it through a gel (like sifting gold particles; same sized particles end up on the same places in the seive), they basically have the same 'weight' and drag, so they'll end up in the same location in the end.
Then there's RNA, the actual message originating from DNA that gets sent around to do stuff. Like, the DNA is the main chief and the RNA is the guy on the shop floor doing all the work instructed by the chief. Well, as you might know, the people on the shop floor move around a lot, from department to department and back. The same happens with RNA in plants, and even humans - loads of RNA in our body isn't from human origin. This means that when testing for RNA, you might even find a bacteria code in a plant. Does that mean the two have hybridized? I believe not.
But, the fun thing is protoplast fusion. This is actually melting two pollen together. And it's friggin awesome. Companies are using this to incorporate sterility into their seed stock, so that no farmer in the world can copy their genetics.
I hope this makes any sense, further reading can be done by googling:
- Genetics
- Sequencing
- PCR
- Elektrophoresis
- Biochemistry
- Or any of the terms used.