One workaround would be in the last step, use smaller quantity of elution buffer. Also use more tissues in the first step, for plants I prefer young, tender leaves that are actively dividing and therefore guarantees consistent good yield!

In my experience, the kits work well for most plant groups if you make sure that the plant material is ground to a really fine powder before lysis. Sometimes this means going back to mortar and pistil if the retsch mill doesn' t do a proper job.

For a few plant groups the kits just don't work at all (because of polysaccharides, etc., that block the silica membranes) so one has to do the standard CTAB method "by hand". Good luck.

Small sized fragments on gel mean that your DNA is degraded. Please, be sure that the source material was fresh and properly stored before DNA extraction. Some plant species contain nucleolytic enzyme activities that would degrade DNA rapidly. Ideally, the freshly cut leaf is directly homogenised in liquid nitrogen.

 Try using around 100mg of starting plant material, or maybe even slightly less. I have often got better yields with less starting plant material. Also, give each reagent in the kit a little vortex before use, to make sure they are homogenous, especially if the kit hasn't been used in a while. Make sure there is no precipitate in buffer AP1 (again, more likely if the kit been unused for a while), by heating it in a water bath for up to 30 seconds while swirling. You can increase the incubation time after elution with buffer AE up to ten minutes, which sometimes improves yield. Finally, instead of eluting in 100ul, try extracting twice from the same sample if you can, and eluting both into the same final tube in 50ul each.

This worked for me:

Costa, C.M., and R.P. Roberts. 2014. Techniques for improving the quality and quantity of DNA extracted from herbarium specimens. Phytoneuron 48: 1–8.