This is an interesting project. Lots of trials of unusual staple crops.
He does a lot of variety trials and handcrossing
Here is part of one of the blog posts.
"The first step is to decide which species are worth focusing your limited time and energy on. To begin it is generally worth growing a single variety of all available species of interest in order to assess compatibility with your local conditions. Native soil types have their own particular mineral balance and texture, which will strongly favour some species over others. Any species that doesn’t show moderate vigour in the absence of irrigation, excessive fertilisation or pest control is unlikely to be worth investing years of work into.
After the initial few years of throwing everything at the wall, eventually the time comes to assess what stuck. Crop species that demonstrated they were reasonably well suited to your conditions can be further improved by widening their genetics. Some niches in your system may take longer to find suitable candidates. Keep in mind the end result of having a reasonable diversity of crops with complementary end products. For me that looks like a dozen vegetable species (split evenly between cool and warm seasons) and a half dozen staple crops.
Different crop species can be broadly separated into those which naturally hybridise, and those that require hand pollination. I will talk about the natural outcrossers first since they are easier to work with. Once I have identified species with some potential I gather 6 to 12 distinct strains and do a side by side variety trial. I do not assume I can buy more packets of any particular variety (often seed sold under the same name will be distinctly different strains due to mislabelling or substitution). I follow a general rule of sowing less than half a seed pack in any one year, and making sure the rest is stored properly to maintain viability (in the fridge, inside multiple snaplock bags).
Store bought seeds are often dead on arrival or rather weak, and only present in small quantities, so I sow them in pots then transplant once they are large enough (which usually demands hand irrigation until established if it isn’t raining). I label the individual pots (using a chinagraph pencil on plastic tags). I label the date, variety and abbreviated source on one side, then put multiples of a large letter (A, B, C…) on the other side. This code is recorded in a notebook, since the details on the label often end up smudged in the field. When I transplant the seedlings I also record their relative positions in the beds as labels often end up pulled out during the growing season.
I only do this careful labelling during the variety trial years, as I find it to be a pain to manage long term. I only labelling during a variety trial so I can later selectively back cross my first generation hybrids to pure plants of the best varieties in the second year. Ultimately I am aiming for a balance between quality/purity and diversity. As always I only sow half my current seed stock in any season, as an insurance against disasters. Generally by the third season I have saved so much seed that I can afford to direct sow larger beds more thickly and then thin the seedlings to select for early vigour. I find this trait to be really useful since it lets the crop to compete with weeds during establishment.
In following years I enter a maintenance phase for the species, which involves ongoing selection of better performing individuals. I only save seed from the nicest looking individuals (which means getting in the habit of not harvesting the best looking plants). From time to time I will trial a small number of plants of a new variety or three alongside my established population, and if they are above average I will save some seed from the new strains to gradually blend into my main population. Over time selecting the best plants to reproduce in a small population will inevitably lead to inbreeding depression, so adding small amounts of fresh genetics every 5-10 years can delay this process indefinitely. This is one reason why many “pure” heirloom strains are degraded today, though they represent a source of interest diversity for creating new grexes. Another viable strategy is sharing your early, high diversity mixed population with other local growers, then swapping seed periodically to increase the effective population size.
For crops that need hand pollination the process is similar but slower since you need to drive every hybridisation event, but because you are in control you can have more fun deciding which crosses to make. Every species has its own timing and techniques, so be prepared to fumble around a little before you get a feel for what you are doing. (I’m currently learning sword-bean pollination, to be described in an upcoming post). If you are doing wide crosses, where there is likely to be unknown levels of incompatibility, it can be useful to mix pollen from multiple different sources to apply to every available stigma. In early years it makes sense to cast a wider net in order to get at least some hybrid seed to get you started. In later years you can afford to be more narrowly focused on getting seed from more deliberate crosses.
Generally I will take the species/strain with the best agricultural potential and apply pollen from everything else that is available (similar to my strategy with outcrossing species). That way if any seed form I know at least half its genes come from a high quality parent. This is the approach I took to breeding Canna. As long as I know the seed parent I don’t worry about keeping track of the pollen parent. Some breeders label both parents of every single cross, but for species where the chance of getting seed set is low I think your time is better spent doing more crosses, then allowing the hybrid seedlings to stand on their own merit"