The shape and size of shells, kernels, and cracks —in a nutshell

148 accessions X-ray CT scanned       Qualitative crackability   Shell                     Air              Packing Tissue        Kernel

Preprint
DOI: 10.1101/2023.09.26.559651

Python Code
github.com/amezqui3/walnut_tda

Phenotyping walnuts with X-ray CT scanning

Erik Amézquita1

Michelle Quigley2 Liz Munch3 Pat Brown4 Dan Chitwood2

1. Plant Sciences & Technology, University of Missouri
2. Horticulture, Michigan State University
3. Computational Math, Science & Engineering, Michigan State University
4. Plant Sciences, University of California, Davis


Materials and methods

  • 1264 individuals → 149 accessions
  • 49 morphological phenotypes:
    ↪ lengths, areas, absolute and relative volumes, …
  • 12 traits of commercial interest:
    ↪ kernel weight, ease of kernel removal, shell strength

Allometry reveals biophysical limits

  • If nut volume increases by 2X
    ↪Then air volume increases by 2.3X
    ↪Walnut diameter capped at 15cm (6")
    ↪Diameter larger than 1.6cm (⅝")
  • If nut volume increases by 2X
    ↪Kernel volume only increases by 1.8X

Acknowledgements

This work is supported by NSF Plant Genome Research Program awards IOS-2310355, IOS-2310356, and IOS-2310357.

References

[1] Bernard, A., Hamdy, S., Le Corre, L. et al. (2020) "3D characterization of walnut morphological traits using X-ray computed tomography," Plant Methods, 16, 115.


Phenotyping → better walnut breeding

  • Stepwise linear regression to model commercial traits using only morphological traits
  • Determine shape traits that contribute the most to the predictive model
  • Perform a 70/30 train/test split to avoid overfitting
  • Use only traits with significant Spearman correlation
  • Relative tissue volume and thickness is all you need!
  • Inexpensive phenotyping platforms focused solely on volumetric analyses

Future directions: domestication

  • Earliest Himalayan accession is notoriously hard to crack open yet it is morphologically average
  • There must be a subtle yet fundamental morphological change when walnut was domesticated
  • Inexpensive platforms → More phenotyping
    → More data → Better math models
    → Insights into breeding and domestication