You are here: Home» Membership» Special Interest Groups» Cypress Development Group» Cypress Development Group reports» Towards Commercialising Cypress as a High-Value Plantation Forest Species» Methods

Towards Commercialising Cypress as a High-Value Plantation Forest Species

Report Date: March 2017

Author: Dean Satchell, Sustainable Forest Solutions, R.D. 1 Kerikeri, Northland 0294
+64 21 2357554

In memory of Allan Levett, cypress enthusiast and advocate for undertaking this research.

Special thanks and acknowledgement go to:

  • MPI Sustainable Farming Fund
  • Cypress Development Group (NZFFA)
  • Marika Fritzsche
  • Brian & Barbara Gibson
  • Ben McNeil & family
  • Andy & Tinks Pottinger
  • Vaughan & Jude Kearns
  • Angus Gordon
  • Jeremy Thomson & Sashil Dayal
  • Glenn Crickett & Catherine van Paassen
  • Allan & Gail Laurie
  • Neil & Pam Cullen
  • John and Robyn Fairweather
  • John Moore
  • Charlie Low
  • David Henley


Performance Criteria

Trees were individually identified and evaluated from 11 trial sites through New Zealand.

Four criteria were evaluated in living trees. These were:

  • Tree size
  • Tree form
  • Stem quality
  • Foliage density

These criteria were each rated for the individual trees assessed on a scale of 0-10, with 10 being the maximum performance score.

All factors influencing the rating for each criteria were evaluated at once, resulting in a single score for each criteria for each tree.


Tree size

Diameters over bark were measured at breast height, converted to overall basal area and then back to diameter (because some trees had double and sometimes multiple leaders from the base). For consistency between sites, diameters were then adjusted for age (given the oldest trees were 22), using the equation:

(((22-(2016 - year established)/22)*DBH/2)+DBH

Diameter was also then converted to a growth rating score out of ten by categorising diameters into ten categories (1-10) for each trail site. The difference between maximum diameter and minimum diameter for each site (and all sites) was divided by ten and this number was subtracted from each categories minimum diameter for the category below's minimum diameter.

Tree form

A subjective evaluation of form was made by observing the branching pattern of each tree, looking for ramicorns and heavy branches. These were evaluated on the basis of how log value would be influenced by the form of the tree. For example heavy branching was not assessed to be as bad as ramicorns because these tend to impact less on sawlog recovery and value. Stem straightness was evaluated from straight, to "wobbly" (slight waviness to the stem not having a major impact on stem value) to "wonky" and "kink" (which would require cross cutting resulting in losses to salvage value out of the stem). In evaluating stem straightness the question was always asked "How much does the waviness of the stem affect sawn recoveries and value?". Where the stem was pruned, the straightness of the pruned stem was evaluated, but equally the crown above the pruned stem was evaluated.

For example the form was rated:

  • between 8 and 10 if branching was light and stem straight with a single leader;
  • between 6 and 8 if wobbliness of stems did not have much impact on value, or branchiness had only a low level of heavy branching that did not significantly impacting on value.


Stem quality

Level of fluting was assessed as potential sawn recovery. The affect that fluting had on sawn recovery influenced the score allocated to stem quality. Although fluting is not necessarily associated with stem cankers, both impact on sawn recoveries and therefore log value. The question was asked "How would a sawmiller evaluate the value of the log?". Evaluation was made primarily of the buttlog, carefully inspecting this for stem defect. Fluting and cankers were also observed further up the stem to evaluate the whole log recovery from the tree.

For example:

  • Slight fluting or 1-2 smaller cankers only (that would not significantly affect sawn recovery) = 8/10
  • Fluting/cankers that definately affect sawn recovery, but still okay as a sawlog = 7/10
  • Cankers/fluting starting to impact more heavily on sawn recovery, not very nice sawlog but useable = 6/10
  • Fairly heavy fluting/cankers, maybe 50% recovery loss in the sawlog = 5/10
  • Heavy fluting/canker, depending on evaluation of expected recovery = 0 - 5


Foliage density

Foliage density was evaluated from the ground and was a subjective representation of the level of canker in the crown. Because canker is not necessarily present at any one time, but loss of crown and associated growth is a key concern, a subjective evaluation was made on how thin of dense the crown was, influenced also by the presence of foliar canker and dead branches.

Thus the rating for each tree subjectively describes the level of canker in the crown of the tree, but not including the influence suppression by other trees had on crown health.

For example:

  • Thin crown but no obvious canker disease present = 7 - 8
  • Some canker evident in the crown, but mostly healthy crown = 6 - 7
  • Reasonable levels of canker and fairly low crown density, not particularly healthy = 5 - 6
  • Heavily diseased or very thin crown (depending on level) = 0 - 5


Growth and Form

Ratings for growth and form were assembled by averaging growth rating and form rating for the individual tree.

Tree Health

Ratings for overall tree health were assembled by averaging stem quality and foliage density for the individual tree.

Overall Performance Score

Ratings for each of Tree size, Tree form, Stem quality and Foliage density were averaged for one score representing overall performance.

Statistical Analysis

A total of 2542 trees were measured and assessed and 1812 trees were assessed as unmeasurable (either gone, toppled or dead). Variables were subsetted by filtering out unmeasured trees and trees that were in less than three sites, for statistical analysis.

Because the trials were a mix of row plots (usually 5 trees per row) and single tree plots, given the age of the trees and the variability between clones observed in the field, because very little interaction between adjacent rows and trees within rows was observed, it was decided to class all trees as single tree plots.

Trials and clones

Analysis of variance was undertaken using the "R" statistical software and the "lm" package to fit linear models.

Response variables were:

  • Diameter (continuous variable)
  • Diameter Score (ordinal variable)
  • Form (ordinal variable)
  • Stem (ordinal variable)
  • Foliage (ordinal variable)
  • OverallScore (ordinal variable)
  • HealthScore (ordinal variable)
  • GFScore (ordinal variable)

Explanatory variables were:

  • Trial Location
  • Clone ID
  • Mean Annual Rainfall
  • Mean Winter Temperature
  • Mean Annual Temperature
  • Mean Summer Temperature
  • Mean Summer Rainfall
  • Mean Winter Rainfall

Ordinal response variables (ratings out of ten) were treated as continuous variables. Statistical significance level was set to 5%.

Toppling and nutrient deficiency experiments

For binomial data (toppling and nutrient deficiency, where there were two possible outcomes) the glm package (binomial logistic regression) in the "R" statistical software was used:

  1. The data was subsetted to existing trees in the trial location being studied;
  2. A group data object was created by block or row and then summarised by block or row;
  3. the number of trees by row or block was counted;
  4. a dataframe was created with number of toppled trees, then merged back with summarised data;
  5. a two column matrix was created of successes and failures for use in the model;
  6. the model was fitted to the data.

Response variables were toppling (Waimate site, block planted) and yellowing (Ribbonwood site, row planted). Explanatory variables were Clone ID.


Farm Forestry - Headlines

Article archive »