The contributing fixed effects in the model are presented in Table \ref{209209},
with the remaining fixed effects tested, but not used in the model
presented in Supplementary Material 1 to avoid confusion. The fixed
effects of replication show some differences across the experiment but
do not show any particular trend (with the exception of diameter which
is due to the sampling constraints described in the method). In spite of
the high correlation between diameter and height, sampling type
(Replicates 6, 7 and 8 in Table \ref{209209} and Sampling Type in Supplementary
Material 1) only shows influence over diameter, this is because height
was measured pre-harvest on all (living) trees, while diameter was one
of the main visual indicators for selection during harvest, and measured
under bark post-harvest. It was not surprising that sampling type has
little effect on other traits, as (with the exception of height)
correlations with diameter are weak (-0.25, -0.3, -0.22, and -0.23), as
seen in Table \ref{725813}. Staking had a small effect on growth-strain, but this
increase is more likely due to tension wood development while the trees
were not staked, but had fallen over in the storm. Staking shows little
bias toward particular families, however is concentrated toward the
southern end of the experiment where wind funneled between two hedges.
The outer-most rows/columns (Edge effects) of the experiment show a
small effect in growth-strain, diameter and height (see Table \ref{209209}). It
should be noted there were a buffer rows at the southern end of the
block, but not on any other side. Remaining fixed effects all show
negligible influence (Supplementary Material 1).