Transcript The number of clones was presented as a slide show at a meeting

"Optimum number of tested
clones in seed orchards “
Iceland, Aug 28, 2003
Dag Lindgren
Department of Forest Genetics and Plant Physiology
Swedish University of Agricultural Sciences
S-901 83 Umeå, Sweden
Reservations
Clone number is an easy and managable
concept, but this concept leads to non-optimal
seed orchards. A more relevant formulation is
to what degree each clone should be
represented (unequal proportion deployment).
Do not expect the definite answer why
diversity is needed, it will not be a main focus
now.
How many clones…?
 One
of the first and most evident questions
 A “key” question
 Easy to adapt to the answer
 Used for control, certification and monitoring
Literature on number of clones in seed orchards

Two previous studies has the agenda in the title:
Lindgren-D 1974. Aspects of suitable number of
clones in a seed orchard (proceedings); Hattemer-HH
et al. 1982. Klonanzahl forstlicher Samenplantagen
und genetische Vielfalt (in German).

Few citations to these studies (mostly self-citations),
indicates no-one searched the literature on the issue.

I have not found a quantitative in depth discussions in
a journal paper. Surprising and a bit upsetting,
considering the practical significance!

So, I decided to make a new study Xmas 2002.
Clone number, statistics for 255 conifer
seed orchards (source: Kang et al 2001 New Forest)
Location
Finland
Sweden
Korea
Species
Number of
Seed
orchards
statistics
based on
P sylvestris
176
P abies
25
P sylvestris
22
P abies
11
P koreniensis 13
P densiflora 8
Average
Number of
Clones
139
75
80
71
70
94
Number of clones in pine seed
orchards in south US
Source: McKeand et al. 2003
 Rough
average
 24
clones in lob SO
 42 clones in slash SO
6
with only 5-10 clones
First seed orchards

For early seed orchards with untested clones, many
clones may be right.
 The selected plus trees differ little in expected
breeding value.
 Its easier to get support for selecting plus trees and
testing clones if they are in seed orchards. Thus
many clones in seed orchards was a way to get tree
breeding financed.
 Material for grafting may sometimes be limiting.
More plus trees means more material.
 Seed orchards functioned as clonal archives.
Thus, there has not been important reasons to argue
for low numbers until recently.
Arguments for diversity
Genetic diversity in is likely to favour biological
production:
• One genotype demands the same things at the same time, several
use the site resources better!
• In a mix another genotype may take over the ecological space left
by a failed one.
• A disease spreads faster in a uniform crop.
This expectation has generally been confirmed by a number of
experiments with different agricultural crops

Public relations, green certification etc.
has an evident market value, and staff is
more happy if operations are sustainable
and in reasonable harmony with
environment.

From this point of view it is a large
advantage if the situation appears to be
monitored and controlled.

Calculating Status number (=gene
diversity = group coancestry) contributes
to filling that function.
Some clone considerations…

Evolutionary potential for
an individual planted forest
stand should not be seen
as a consideration (it must
be based on national
considerations).
Allele
richness has a value only if it can save
the stand, alleles in a frequency below 15% cannot
save the stand, thus clonal number > 4.
Similarity
with natural conditions could be an
argument. Clonal numbers 10 – 25 may best
mimic the natural half sib groups around seed
trees.
Non-overlapping
few clones, > 4.
flowering can be a problem if
Too much diversity in plantations?!
Most crop- and many forest managers do not like diversity,
 Uniform trees = better economy, more valuable crop, simpler
management;
 Genetic superiority of the best clones is much larger than the
likely loss in biomass production by uniformity;
 Frequently occurring demand for diversity in intensively
managed forests is – in my opinion - unreasonable
expensive in lost future gain;
 Most of the possible quantitative benefit with diversity is
obtained by five genotypes.
Few clones = high gain
G  ir A
G = genetic gain
i = selection intensity
r = correlation between measured “height at 6 years at some test
environments” and goal “value for forestry on future environments
where the seed orchard crop will be used”
A = variation in desired goal character (CV additive)
if r and A are given, G depends only on i.
The fewer clones, the higher i and the higher G!
Higher gain is the major argument for few clones!
The fewer selected, the higher the gain!
May be 1 is
the best??
2
1
0
1.0
0.8
0.6
0.4
Selected proportion
0.2
0.0
Selection intensity
3
Our study on optimum clone
number
•Objective was to predict optimum clone
number for tested seed orchards by the aid
of computer simulation
•The tool: “SeedOrchardManager” at
www.genfys.slu.se/staff/dagl
Model considerations
 Genetic
gain of seed orchard crop
 Breeding value of available candidates
 Gene diversity
 Selfing
 Fertility differences
 Pollen contamination
 Close neighbor pollination
Scenarios
Swedish Scots pine
American loblolly pine
Most reasonable values.
In Sweden these values are
meant to be typical for Scots
pine seed orchards to be
established in the coming
decade.
Scale and chosen entries
G  ir A
G = BV related to average of candidates in “value for
forestry”
i = i(n,300) n clones selected from 300 tested candidates
r = 0.7 for Sweden
r = 0.8 for America (“older” and more uniform field
trials and the “value for forestry” refers to a closer
future)
A = 0.125 (proportion from the total in BP)
Diversity measures
Gene diversity can be measured as status number.
Effective population size of an orchard is best
described as foreseen status number of the seed
orchard crop. But group coancestry is more
convenient for the mathematics.
Group coancestry (Status number)
Gives the probability for interaction between related
tree, e.g.
 competition for resources
 complementing shortcomings
 mating (inbreeding in plantations)
 disease or pest “jump” to the same substrate
 rise in group coancestry = loss of gene diversity
compared to the wild forest.
Thus a meaningful quantification of genetic diversity
Quantification of low
diversity loss
A high group coancestry, , can be assumed
to cause reduced production, increased risk
for calamities and more severe epidemics,
but how much (Loss)?
The loss is proportional to group coancestry,
but what is the coefficient? I call it
DivCoeff. The loss (on the same scale as
gain) is
Loss = DivCoeff x 
How large is DivCoeff?
Thus how important gene diversity?

59% of all loblolly is deployed as half sib family blocks
and no problems are reported (McKeand et al. 2003)!
The group coancestry in these plantations may be
approximately  = 0.14 (Kang et al 2002). 5% loss is
likely to observed. Thus DivCoeff < 0.35!
 Many experiments and some forestry with full sib
families. 10% loss at =0.25 would be observed, thus
DivCoeff < 0.4!
 Many experiments with monoclone plots and some
clonal forestry. 10% loss at =0.5 would be observed,
thus DivCoeff < 0.2!
 Mixes of agricultural lines (=0.5) probably 2.5%
superior to pure lines (=1), thus DivCoeff = 0.05!
DivCoeff
Swedish
scenario DivCoeff = 0.4
American scenario DivCoeff = 0.2
The difference is because south US gets
more experience with low diversity forestry,
less worries, less restrictions, shorter rotation
times, less “Nature” (industry forestry
replaced cotton, not old growth)
Gain and gene diversity are generally in
conflict and must be compromised
1.2
1.0
0.8
0.6
0.4
Gain
Diversity
0.2
0.0
0
20
40
60
Number of clones
80
Clones have different
reproductive success

Clones give rise to different amounts of seed and
pollen.
 Is expressed by sibling coefficient, , (related to the
probability that seedlings share a parent, Kang 2001)
 Consequences of a high :



Less gene diversity (more relatedness)
More selfing
Value: =2 (Kang et al. 2003)
=2 means the probability that seeds share a parent is double as
high as if all clones were equally reproductive successful
No to related clones
 We
used no related clones in the model;
 A few related clones will cause negligible
inbreeding depression, but reduce gene
diversity rather considerable;
 I suggest: it may sometimes be worth
considering to have fewer clones rather
than related.
Pollen sources
 The
ramet itself (part of selfing)
 Close neighbours (not the same clone in
the close neighbourhood)
 Ramets which are not close neighbours (a
share of the same clone causes some
selfing)
 Pollen contamination (outside)
Pollen contamination

Decreases gain
 Increases gene diversity
 Reduces selfing
 Estimate: 40 % pollen (that means almost 50% of
seeds)
 Estimate of breeding value of aliens: 0.94 for
Swedish; 0.85 for America (America more
advanced, but even contamination “improved”)
(average candidate =1)
Selfing

Pollen from the ramet itself (10% pollen).
 Pollen from other ramets of the same clone (40%
pollen)*(sibling coeff)/(clone number).
 Effective percentage of selfing much less than selfpollination
 Not the same clone in the close neighborhood (10%
pollen). Neither contaminating pollen (40%). Thus no
self-pollen in 50% of pollen cloud.
 Compared to outcrossing, selfing pollen is less efficient
(dead zygotes because of genetic load).
 Reduction in seed yield (rather yield of plantable)
seedlings to 0.3 for Swedish, 0.1 for America (compared
to outcrossing).
 Selfed seedlings reduce “forest value” 50%.
“Goodness” of seed orchard
FertT  SelfEff  PollwTree
(n  1   * SelfEff )  PollSO
 Poll Neighb 
 PollC
n
n
NS 
 (1  FertC / 2) 2
Goodness, B (“benefit”)=
 (1  G)(1  Fert Self (2Selfprod  1)  Fert Neighb  Fert SOOutc)  FertC GC 
B
 * 1  DivCoeff * 0.5 / N S 
2


The formulas used look like this, more detailed explanations ín the workbook on my web
Just to put these thoughts and numbers
into a worksheet…
Results
Seedorchardmanager at
www.genfys.slu.se/staff/dagl
Results
Swedish
scenario: optimal number 16
American scenario: optimal number 8
Main cause for this difference: less problems
with selfing and gene diversity in American
scenario
If the entries are different?

Pollination pattern not critical. Breeding value of
contaminates unimportant. Large contamination
favors few clones.
 Selfing important. Self-fertility favors many
clones.
 Wide genetic variation favors few clones
 Many candidates moderately favors many clones
 The value assigned to gene diversity is a very
important factor, a high value favors many clones.
Optimum seems smooth!
Benefit, gain and diversity
1.2
1.0
0.8
Benefit
Gain
Diversity
0.6
0.4
0.2
0.0
0
20
40
60
Number of clones
80
Conclusions

Around 16 maybe optimal but I am a bit conservative
and assumptions may be revised. So at the moment I
suggest 20 as a good choice for a type case of Scots
pine in Sweden with known breeding values.
 Selfing and gene diversity may have been given too
high weight and the sibling coefficient may be
overestimed, leading to lower optimal numbers, users
confident in such estimates may benefit from
somewhat lower clonal numbers.
 The math will be slightly different for unequal
proportions, which I strongly do suggest. Perhaps we
should talk about something like 25 clones deployed
in different proportions.
The computer simulator is on my web
You can use own parameters to answer this
not easy question …
Or you may want to study the formulas …
Probably also this presentation is there…
http://www.genfys.slu.se/staff/dagl/Breed_Home_Page/
Acknowledgement
Darius Danusevicius was very helpful in
constructing this slide show and made
much of the graphics