Is there a publication comparing DNA yield and / or PCR-amplifiability after extraction from fruit flesh (like apples, oranges, cherries etc) in comparison to seeds of the same fruits ? I would prefer a reference, but your unpublished experience is also interesting for me.
It depends on the seed and fruit you're talking about.
The amount of DNA in your crushed up sample of plant matter depends on only one thing--how many cells are in your sample. Look at this diagram of a kernel of corn (U of Indiana):
There are different cell densities in each area--in the endosperm, the cells can be huge (and therefore have a lower density of DNA when compared with the embryo).
Let's look at a stained mustard seed (U of London):
As you can see, the cells here are very dense. More dense means more DNA per unit sample.
As far as fruit is concerned, the same rule of thumb applies. Verner, 1938 shows some cool apple tissue histology that might give you an idea of cell density in fruit tissue (again, it will be different in different fruits--search for
However, now that we've talked about density, we need to actually think about getting the DNA out. The extraction process is actually much more important in terms of your final yield.
Many plant seeds contain polyphenolics, which can cause DNA degradation. This protocol (Cornell U) details many of the possible shortcomings:
Overall, there are definitely species-specific differences in fruit/seed plant DNA content. Practically, however, the most important factor in DNA yield and viability for future experimentation is sample purity and contamination. Specific fruits and seeds may contain one or more of the above contaminants, so the answer to "Seeds or fruits?" definitely depends on the species of plant you're working with.