This is very cool. I think it will need to be extensively field tested in nature to see how well the results hold up. If it is a robust technique, then it is fascinating to think how it could be used to help reconstruct the paleoecology of areas. It could also be very useful for conducting wildlife censuses in combination with phototraps, aerial surveys, etc...may be especially useful for reclusive forest critters.
I would think that this technique might be useful for plants as well, although there are many challenges that plant genetics have that animals don't like multi-ploidy genomes and alkoloids that inhibit sequencing.