Genetic thinning of clonal seed orchards using linear deployment

1
Genetic thinning of clonal seed orchards using
linear deployment

• Forest Genetics and Tree Breeding in the Age of
  Genomics Progress and Future November 1-5, 2004
  Charleston
• Wednesday 130-540 PM
• Concurrent Session II Advances in Reproductive
  Biology and Seed Orchards
• Moderator Clem Lambeth
• 425-450 Seed orchard Thinning Using Linear
  Deployment of Clones
• Dag Lindgren, SLU, Sweden

2
The authors
Mohan Varghese Dag Lindgren Finnvid Prescher
3
Presents a genetic thinning algorithm

• Known
• Ramet number and breeding value for each clone
• Result
• Number of ramets to be rouged for each clone

4

• The algorithm combines the two desires
• High effective number of clones
• and
• High genetic gain

5
Linear deployment is optimal for establishment!

• No other deployment combines higher gain with
  higher effective number

6
(No Transcript)
7
At thinning ramets cannot be added, just
withdrawn. The algorithm has to be modified.
Linear deployment for genetic thinning
8
(No Transcript)
9
The optimal line is the same for all clones

• Bondesson, L. and Lindgren, D. 1993. Optimal
  utilization of clones and genetic thinning of
  seed orchards. Silvae Genet. 42 157-163 .

10
Math
11
More math
The linear deployment thinning algorithm
maximizing G at Ne is as follows
The algorithm results in an optimal combination
of G, Ne and ramets remaining, but there are many
optimal combinations. The specific solution is
given by the choice of g0 and b. g0 and b are
chosen to result in desired combination of values
for G, Ne and ramets remaining. (Bondesson and
Lindgren 1993). Note that linear deployment can
be seen as a solution searching for a problem,
and not as usual a problem asking for a
solution. This presentation shows three practical
applications
12

• Genetic thinning characteristics
• Remaining ramets
• Genetic gain (breeding value)
• Effective clone number
• Linear deployment thinning is optimal
• No other deployment can increase one of these
  three factors without decreasing another

13
This is first presentation of applications of the
algorithm published 1993!
14
Put into a worksheet
Input breeding values, ramet numbers
Output Gain, effective number, remaining ramets
per clone
Linear Deployment at www.genfys.slu.se/staff/dagl
15
Three objects
Place Species Type BV from
Lagan, Sweden Norway spruce Seed orchard, Cuttings Clonal test
Maglehem, Sweden Norway spruce Seed orchard, Grafts Progeny test
Coimbatore, India Eucalyptus camaldulensis Clonal test converted to seed orchard The site itself
16
At a suitable thinning intensity
The graph is generated by trying different lines
17
Result Lagan, linear deployment thinning
Before Thinned
Clones 32 32
Ramets 5351 3644
Gain ( ) 106.0 108.5
Effective number 20.0 22.0
Truncation
24
3644
109.0
16.8
Marginally higher gain, but many clones lost,
effective clone number substantially reduced!
Substantial improvement for both Gain and
Effective number!
Practical thinning resulted in almost full
optimality!
18
Genetic thinning Maglehem
19
Thinning at Maglehem
Truncation Truncation
28
1565
0.36
27.0
Before Thinned
Clones 36 32
Ramets 2006 1260
Gain -0.03 0.48
Effective number 34.9 26.8
Truncation selection that preserves the effective
number results in much lower gain!
20
Thinning at Maglehem
Truncation Truncation
23
1261
0.56
22.1
Before Thinned
Clones 36 32
Ramets 2006 1260
Gain -0.03 0.48
Effective number 34.9 26.8
Truncation with the same number of ramets
results in a little higher gain, but much fewer
clones and effective number
21
Thinning at Maglehem
Linear
32
1260
0.49
26.8
Before Thinned
Clones 36 32
Ramets 2006 1260
Gain -0.03 0.48
Effective number 34.9 26.8
The optimality remains!
22
Eucalyptus clone trial at Coimbatore
A clonal test of Eucalyptus camaldulensis
established at Coimbatore in south India
comprising 87 clones (selected from 7 seedling
seed orchards and commercially available clones).
There were 15 ramets of each clone arranged in 3
tree plots with 5 replications. The test was to
be converted to a clonal seed orchard based on
height assessment in the trial at three years.
23
The Eucalyptus clone trial at measurement and the
ramets at planting
24
Thinning Coimbatore
25
Linear Deployment Same ramet Truncation selection
Clones 72 43
Eff number 57.3 42.4
Ramets 573 573
Height 7.49 7.56
At the same thinning intensity there are much
higher retained number and effective number, but
marginal loss in gain,
26
Linear Deployment Same Gain Truncation selection
Clones 70 43
Eff number 50.5 42.4
Ramets 429 573
Height 7.56 7.56
At the same genetic gain there are much higher
retained number and effective number, but a more
intensive thinning is requiered.
27
Linear Deployment Same Ne Truncation selection
Clones 62 43
Eff number 42.4 42.4
Ramets 396 573
Height 7.65 7.56
At the same effective number of clones there are
a higher retained number and more gain, but a
more intensive thinning is required.
28
Conclusions

• Linear deployment at thinning is theoretically
  optimal!
• The loss from optimality because of practical
  difficulties is marginal and the added
  flexibility may offer advantages!
• The added practical difficulty is marginal.
• The increase in gain and clones retained at the
  same effective clone number are substantial!
• It is sometimes possible to make significant
  increases for both gain and effective clone
  number with a moderate genetic thinning. These
  entities have earlier been seen as incompatible!