2007 Nut Report

This week I completed the analysis of the nut samples from the trees selected for on-planting in 2007: I had collected all the nuts from these trees, cleaned and dried them in the usual way, then selected three nuts at random per tree for later cracking. These three-nut samples were then forced-air dried in paper bags at outdoor ambient temperatures for about six weeks before being transferred indoors, and left to stabilize in a warm, low-humidity environment for another six weeks. All nuts were cracked with a Master nutcracker. Nuts were cracked as replicates, i.e. one nut per tree was cracked until all trees were assessed, then the second nut in the same way, and finally the third nut from each tree. The order of cracking within each replicate was random.

Before presenting the results, it is important to note the characteristics of the 2007 harvest. A very late spring frost (May 19^th) caused widespread damage to emerging flowers. Nutting was considerably reduced when compared to 2006. Selection of trees was made initially on low expressed leaf-disease scores, then adjusted in total number for the number of trees that actually gave a sample big enough to ensure sufficient nuts for on-planting. The area selected for planting of this F1 material was big enough for a randomized complete block study of 25 lines replicated 11 times. The planted sample (two nuts per location) thus consisted of nuts from the 25 maternal trees with characteristics (traits) considered most interesting (low disease scores, high nut number in a low nut number year, precocious trees – nutting at a small size/young age, range in nut sizes, etc). As no cracking analysis had been done to date on-farm, there was no prior knowledge of kernel weights or percentages.

The mean nut weight (NW) was 15g, and the mean kernel weight was 2.5g. By themselves, these values don’t tell much about the sample. The following histogram indicates the distribution of KW. NW ranged from about 7.6g to 18.8g. Kernel weight (KW) similarly ranged from 1.7g to 3.9g (line 1-1-17), though NW and KW are not directly correlated, i.e. lowest NW did not exhibit lowest KW, nor highest NW the highest KW.

But I’ve left the best to last. The sample showed a mean kernel percentage (K%) of 22%, calculated from KW as a percentage of NW. Also shown as a histogram, we can see that almost 25% of the trees showed a very respectable K% of 25% or more. The highest recorded K% was over 28% (line 1-9-23). This was far higher than I expected to find in this sample. I had expected a mean K% of about 20%, with perhaps the top line approaching 24%. These expectations are the results of earlier surveys I’ve done off-farm where I’ve had many more trees in the sample, and mainly where lowest K% was far lower ( ~12%). But it is actually quite exciting to find 28% kernel, because the probability is that we shall eventually find higher values. Again, K% is not correlated with NW or KW. In fact, the highest K% was found in a line with mean NW <10g,>

If we complete the analysis by calculating kernel yield per tree (KY, by multiplying KW by the number of nuts per tree; no of nut data is lacking for two lines), we once again find a re-ordering of lines. Line 1-1-14 yielded 1.34kg of kernel. The tree with highest KW (1-1-17) yielded 0.45kg, and the tree with highest K% (1-9-23) yielded 0.32kg. As economic yield will be more closely related to KY, the other parameters of KW and K% become less critical in our selection focus.

Grafted named selections in the US have yielded K% of up to 39%. Our experience in 2007 indicated the existence on the farm of lines of up to about 75% of this potential K% in common walnut obtained from our own seed zone. Far less has been published on nut yield per tree. We have planted an F1 trial of these 25 lines to examine heritability in K%, and the other traits measured on each tree. Almost all of the 25 lines had good disease expression scores (lower incidence); it will be interesting to observe heritability in this trait.

The premise underlying our biomass approach to nut production is that there will be enough trees in the overall population with traits of sufficient value to exploit multiple income streams. To date we can characterize these streams as kernel, shell, and sequestered carbon. We have no intention of focusing on a single trait in our selection program but have yet to construct the functional model which will guide us in optimizing multiple selection criteria. All in all, our economic yield is likely to be defined by how we market the different carbon streams partitioned by the tree. So this is an interim report.