What better than a sunny Thanksgiving to celebrate a successful harvest! This past week we were busy harvesting and measuring a small evaluation plot of “pre-bred” sunflower lines. These are all lines that were developed by the USDA. For the most part they are backcross 2′s or 3′s between a wild helianthus and a common elite line. Soon I will do some genotyping of this material to determine what parts of these wild genomes made it into these lines. I also wanted to add to the phenotypic information associated with these lines as well as just take a good look at them myself.
What we did: We grew these lines, plus HA89 (the common elite parent), in a complete random block design with 15 blocks each containing one individual from each line. We used a totally custom hand seeder to plant rows 80cm apart and plants 8 inches apart. Three seeds were planted at each spot and a barcoded tag was stuck in beside them. We recorded how many plants germinated and thinned to one plant per tag. We collected tissue for genotyping, surveyed for flowering time over the summer and recorded a number of other agronomically informative traits over the summer and especially last week during harvest.
There was an impressive amount of diversity and variation apparent early on. This picture captures shows some of the variation there was in flowering time, height, head size etc. (don’t be mislead though, there are 100% wild annuus genotypes in the background).
A quick look at flowering time:
Breaking it down by lines you can see some are early, some are late and some have a great deal of variation within a line. In about the middle, in black, is HA89, the common parent. Colouring by wild species we find some appear to be consistently earlier, and others later.
Lessons learned: 1) Planning goes a long way, not really ground breaking, but having seed packets, each with 3 seeds laid out in the right order and tied together into blocks allowed us to hand plant about 1000 plants in more or less an afternoon, with just two people. 2) More interestingly, although we got set up for it initially and I was really excited about it, barcoded phenotyping was not as great as I thought it would be. After all the effort of making the barcodes and programming the scanners, it ended up being less work to go out with a spread sheet and a pencil (especially if you have 2 people). This may largely be a product of the barcode scanners we have, they are pretty slow (they take a second or two to scan) and some times fail to recognize barcodes if they are dirty or if its too sunny out. I understand the theoretical advantages of using a barcode system but they were not realized this summer. 3) Random block design is key, although I have yet to really crunch the numbers on this. Just looking at my plot you could tell there was variability in the environment across the plot, and my original layout design (not complete random block) would have made a few lines look really bad. 4) Squirrels are evil.
This is obviously a first look at this data, and once the genotyping gets done we will be able to learn a lot more.