Newsletter Summary 98-99
This is the beginning
of a newsletter service managed by Dag Lindgren. I will share documents,
information and thoughts connected to forest tree breeding (such things there
information is not easily available through other sources and which I have easy
access to). I feel that there is too little discussion and communication about
what is essential in forest tree breeding in this world, and want to do a
contribution in a way which is rather easy for me and others. If someone wants
to spread a comment or raise a matter for discussion or something else through
this mailing list it is warmly welcome (currently I believe it just to press
the answer all button). Relevant, interesting and informative discussions have
appeared in this way in many other fields, unfortunately more seldom in forest
genetics. Hopefully this service will stimulate breeders in their thinking and
operations and scientists (like me) to design relevant studies. These letters
will be sent at irregular intervals. It is sent to some of those I have
identified as potentially interested. Those who ask for being added to the
mailing list will get the future letters. Probably I will expend the circle who gets it when sending next letter (the first letter is
sent just to some breeders and some people around me at Umeå). For those who
feel that they have been unwillingly put up on the mailing-list, just return
this mail to me with a "Quit sending me E-mails of this kind", and
you will be removed from the list.
Rotation time in
the breeding population
People like me (Studia Forstalia
Suecica 166 p16) and many others have earlier argued
that the key problem is to maximise gain per year. However,
when we did not consider diversity.
We loose diversity in generations shifts by genetic
drift. There is no way around that (even if there are counteracting forces like
mutations acting in a long term perspective). Were we not interested in
diversity we would keep breeding programs and production populations much more
narrow than we do, so we care for diversity. How should we consider diversity
when discussing rotation time in the breeding population?
Some possible options are listed:
*Increase the size of the breeding population so the diversity loss at
generation shifts becomes insignificant;
*Start with a larger breeding population;
*Maximise a goal, which weights gain and diversity, a fixed time ahead;
*Get as much gain combined with as much diversity as fast as possible;
*Get as much gain as possible for a given diversity loss.
*Get as much gain as possible per unit diversity lost and unit time.
*Run a short generation program including hefty inbreeding to adapt the
population to high inbreeding and low diversity in the same time as we get
gain.
*Make the production population rotation short also so we become aware of
possible side effects faster and thus probably forget about diversity faster.
Sites of interest
The main website of interest for
forest geneticists and forest tree breeders is
"Forest Genetics and Tree
Breeding"
http://www.metla.fi/info/vlib/forestgen/breeding.htm
From this site you can reach
almost all other interesting places. It is very well managed and updated (by
Matti Haapanen). Strongly recommended!!!!! The site is very
useful as a starting point for forest genetic net-surfing and for following up
the latest news.
The news and discussion club of
forest geneticists and tree breeders is FORESTGEN, information about that is
found at the site mentioned above. From a discussion point of view it is mediocer, but a subscription results in announcements on
meetings, positions, books etc; so I suggest forest geneticists/tree breeders
to subscribe.
I have frequently commented on
the lack of creative visible discussions among forest geneticists in
Contributions to
the discussion
Effective size
I discourage using the concept "inbreeding effective population
size" in tree breeding.
If need for the concept is felt,
one of the following concepts may be more informative:
·
Variance effective population size
·
Status number
·
Inbreeding (F)
·
Selfing rate
·
Expected increment in inbreeding.
I am thinking at the following expressions at the moment
·
effective clone number (only to
be used when the offspring (the seed crop) is not of much interest)
·
effective parent number to a
population
Are those concepts suitable, is their another existing terminology which
is preferable?
Population semantics revisited
Some time ago we discussed
population semantics on FORESTGEN; what should we call the probability that two
genes taken at random from the gene pool of a population are equal by descent?
The conclusion following the discussion was that "group coancestry"
was a preferable term, because of the link to Cockerham.
This link was pointed out by two different discussion contributors. Cockerham's intention was not to coin a concept, and his
concept can actually be used "between populations" as well.
The background of that
discussion was that I disliked the term "average coancestry" because
it is ambiguous and interpreted in different ways (is self-coancestry included
or not? does self-coancestry get equal or half weight compared to
cross-coancestry? is it average for individuals or groups? etc.). The
distinction is important and confusion connected to the term has led to the
suggestion of algorithms, which do not do what they are assumed to do. Still we
first used the term "average coancestry" and the selection method
(GMS) was initially called "coancestry-adjusted selection", which is
reflected in Zheng et al (1997). "Adjusted"
I also saw disadvantages with in the associations it gives.
The selection method which
considers both gain and group coancestry can be called Population-merit
selection (PMS) or Group-merit selection (GMS). Population-merit was used by
Lindgren and Mullin (1997), which can be considered a strong argument.
Actually, Population-merit was initially suggested by a referee of the Lindgren
and Mullin (1997) article. Apart from the history, I
now feel Group-merit selection may be slightly superior as it leads
associations to Group-coancestry and associates to something different and
looser than population, group selection is a concept in evolution and this use
is not too different. I may also use that term in the future if others do. A
selected group from a population of candidates seems possible to talk about.
Group-merit can be seen as the result of a semantic evolution during some years
and will be used in future paper by Andersson, Rosvall and others.
We see here how a disparate
terminology is born and evolve and some of us are actively involved in this
process and this arises in spite of an active, frequent, friendly and intimate
discussion. A way to see it is as an evolutionary process, when we successively
search and find better semantics. As far as I see we all still mean the same
thing.
Some additional comments:
1) The concept Population-merit selection is not
connected to the algorithm used for finding the optimum (e.g. the optimum can
alternatively be found by integer linear programming).
2) To search the optimum by Lagrangian
multipiers or linear deployment (as I have done in
many previous studies) has the disadvantage that it does not constrain the
optimum to integers, this may be severe for long-term
breeding applications.
3) Dr Mullin recommends ___-selection, thus to
connect the words with a binding sign.
4) Population-merit has a much wider application
as a goal of breeding than selection.
5) There are many others who develop similar
ideas, so semantic influence from elsewhere may be important, and efforts from
us to standardise are anyway predicted to have limited influence.
6) I guess the word "status
number" will stay, the word has spread beyond our group, and the similar
concept "founder genome equivalents" is ambiguous (Lindgren and Kang
1997). Maybe "effective" could be added, "status
effective number" (Lindgren et al 1996).
7) In Swedish I have used
status-nummer (following discussion with the editor
of SkogsFakta).
"Effective parents" or "status
number"?
Lindgren and Mullin (1998) coined status number in
seed orchard crops.
I think that was a good idea. It is also a good
idea to use the concept as is done
in Andersson, El-Kassaby and
Lindgren (in review).
For these studies it is important that parents may
be related or inbred.
I know that I have encouraged "status
number" in orchard or stand related papers by Kjaer,
Bila, Kang etc etc. But now
I still ask myself if it is useful to always use status number for describing
differences in flowering or ramet number between genotypes. The strength but
also problem with status number is that it refers to a reference population,
and sometimes this is not informative and even misleading.
It seems that "effective number of clones in
a seed orchard" could be more relevant when fertility variations are only
based on ramet number (it is actually the gene pool of the orchard, not its
crop, we are discussing). cf (Kang, Harju, Lindgren, Nikkanen, Almqvist.....1999?). But
in general it is misleading to talk about effective number of clones, as often
what is thought of is the crop of the seed orchard or the relationship of the
orchard clones to the crop.
In studies where predictions of status number of
the crop are based only on flowering (or seed production), thus the fertility
of the parents is the only important variation, would it not give better
associations to use "effective number of parents to predicted crop"?
It leads associations and focus towards the parents and their characteristics
as parents, thus fertility. It is rather clear it is a state, not a rate.
By using other terminology than status number we
need not make the assumption that parents are unrelated and not inbred. We
could instead point out that if parents (or clones) are non inbred and
unrelated, the concept is identical to status number. But the concept itself
would be unrelated to that assumption.
The real problem comes when we go
multigenerational. If fertility variations are the same, still the status
number of a constant population declines over time as trees grow more related.
For this situation it is not good at all to link status number to fertility
variations.
"Effective
parents" or "status number"?
Ola Rosvall has made a comment:
Det är inte bra
att förorena statusnr.begreppet genom att flytta referensen till plantageträden
med ett antagande att de är obesläktade och särskilt inte på sikt. Jag tycker
att det är viktigt att behålla founderanknytningen i statusnr. Vad man skall
kalla nedgången i diversitet pga obalansen i plantageträdens genbidrag vet jag
ej. Ditt förslag är förklarande, vilket är bra. Kanske kan det byggas upp på
nåt annat sätt: Efficient crop parent number fast ditt är nog bättre.
Is environmentalism short-term commercial, while
production is long-term idealism?
A switch in commercial versus idealistic motives
has occurred. "Environmental concern" is now one of the strongest
commercial market forces. An investment in "environment concern" has
the potential of a fast (often immediate) return on investment.
"Production concern" on the other hand is a concern for idealists,
who worry about the future of mankind (or some smaller part of it, like people
living in remote forested areas), but production is not so interesting for
forest managers any more. An investment in "forest production
concern" is doubtful from an economic-commercial point of view and often
not likely to give a "reasonable" return on the investment (at least
not on typical forest land in
Status-nummer etc på
svenska?
I vardagligt tal kanske vi helt enkelt kan säga "effektivt
antal"?
Men då skall det av sammanhanget vara klart att det är status-nummer vi
menar.
Kan vi kalla "group coancestry" för "genomsnittligt
släktskap"?
Vad skall vi kalla "group merit selection"?
may use the list for shorter
announcements also in between, e.g. I announced a seminar November 20.
Seminar
Prof. Steve McKeand,
Deployment philosophy, scale, gains,
experiences and risks connected to half-sib plantations with P taeda.
ure and time for discussion)
Background
There is a spectacularly striking difference between EU and much of the new world concerning the view on need of diversity in the forest regeneration material. The EU (Swedish) policy is (somewhat oversimplified): At least 30 clones in a seed orchard and selective harvesting is banned.
Virtually
all plantations with P. taeda in the
southern
How is it possible that systems arrive at so radically different conclusions? Are there scientific reasons for the difference? If so, why do scientists not have a more coherent view on what is favourable and what is dangerous? The reasons for these large differences in deployment strategies needs to be debated considering that the matter is of central importance to tree improvement, and its possible dangers or blessings.
We have
PAM, positive assortative mating, but what to call
the opposite? Random mating? Ola claims that it is
common that authors use that as a contrast to PAM.
Number of representation of initial plus trees in the forming of a
closed breeding population
I have
since long argued that it would be more beneficial for tree breeding to
transmit genes from more of the tested plus trees but in the same time
let the plus trees be differently represented in an optimal fashion
allowing more offspring from the best (in contrast to the base alternative in
Swedish breeding, which is to let a fixed number of tested plus trees (50) be
equally represented in the next generation of the breeding population). I have
felt it important that the reasons for this opinion are well understood. Now I
feel pleased to express that what is possible to do to meet this desire from me
has been done. Förädlingsrådet has discussed the
issue in December 1998. I made a presentation and my presentation is in the
enclosed document. After the meeting I wrote down my comments on a number of
questions and comments about unequal representation, which were raised at the
meeting, this is also enclosed. A grant was given to study the phenomenon from Föreningen Skogsträdsförädling,
and Ola Rosvall was subsequently asked to write about the results from this
project in the recent yearbook of Föreningen. I
enclose this article. I see it as important that it is well documented why
decisions and actions are made and relevant arguments commented in connection
with that, so now my opinion is well documented and properly discussed. This
matter concerns the transition from an open to a closed breeding
population, and is not dealt with in Ola Rosvall's
thesis. This matter is of more immediate interest for breeders' actions than
what happens in later generations.
The issue
of unequal representation (unbalance) in a long term closed breeding
population with particular attention to the Swedish program has been thoroughly
discussed in Ola Rosvall's thesis. It appears that
the possible advantage with unbalance, at least considering the specific plans
for Norway spruce, is smaller and more doubtful than I would have guessed.
Sigfrid Blomqvist avled 6 mars 1999. En av den moderna svenska
skogsträdsförädlingens pionjärer.
A fine Award
At IUFRO's next World-Congress in Malyasia next summer, IUFRO will at its opening cermony give its "Outstanding Doctoral Research
Award" to Run-Peng Wei. That is in competition with all PhDs in the World
for the past five years within Forest Genetics and Plant Physiology. This is a
large honour to Run-Peng. But it is also a clear indication that
Literature by Dag Lindgren et al with summaries of
highlights!
Lit list 1997
Below I list and comment some of
the papers I have been involved in which were published 1997,
it is only papers which I consider of interest for forest tree breeding. The
points which I find most relevant to breeding are emphasised in the summary
(thus the summaries are not the authors summaries).
Lindgren D, Gea LD, & Jefferson PA 1997. Status number for measuring genetic diversity.
Summary: Argues for use of
status number to monitor breeding, compares with inbreeding effective
population size (which is shown to give rather senseless information), offers
an algorithm for calculation of status number from a coancestry (relationship)
matrix, demonstrates how status number develops over
time in a real breeding program.
Lindgren D &
Summary: Reveiw
of lit on status number till end of 1997. Points out the
similarity with variance effective population size. Demonstrates
its use for describing seed orchard crops. Makes a discussion of the
concept "founder genome equivalents" used in animal gene
conservation. This concept can have two meanings, the one recommended by the
inventor is equivalent to status number, but the variant most used by others is
not. Remark: I believe our article is the last article
in the closing volume of this series.
Lindgren D & Lindgren K. 1997. Long distance pollen transfer
may make gene conservation difficult. In: Kurm M and Tamm
Y (editors), Conservation of
Summary: Genes are
able to migrate over several hundred kilometres with fertile pollen, where
are indications on a gene flow over some hundred kilometres from south to
north.
Lindgren D & Mullin TJ 1997. Genetic variance within a full
sib family.
Summary: Derives the additive
variance within a full sib family. This is a function of the initial variance
and the inbreeding of the parents, but the relatedness of the parents does not
matter, neither does the size of the family. This is a known relationship,
it is just the way of proving the relationship which is fresh.
Lindgren D
& MullinTJ 1997. Balancing gain and relatedness in selection. Silvae Genetica. 46:124-129.
Summary: Population merit (thus
a suitable scaled genetic gain minus group coancestry) is the entity to
maximise in breeding and selection. Technically this is done by an iterative
such algorithm. Population-merit selection is shown to be considerable superior
to restricted selection and the superiority remains over generations. Remark: The use of population-merit
can be seen as the optimal way of balancing gain and diversity. Restricted
selection (thus maximising the influence of individual parents) is a less
optimal way, which still is used in many of our studies, either because the
population-merit concept was not yet invented, or too difficult to apply.
Lindgren D & Persson A. 1997. Vitalization of results from
provenance tests. In: Mátyás
C (ed). Perspectives of
Summary: A list of possible uses
of mature provenance tests, several are interesting for breeding.
Lindgren D, Wei R-P & Lee S.
1997. How to calculate optimum family
number when starting a breeding program. For. Sci. 43(2): 206-212.
Summary: Argues to start with
offspring with many more plus trees than needed later. (Thus if a breeding
population size 50 is headed for in later generations, many more plus trees
should be included in the initial crosses). Argues that cost components are
more important for optimisation than variance components, thus the optimum is
not much dependent on heritability but very depending on the cost of additional
plus trees and crosses. Argues to invest a large share of
available funds in creating offspring from many plustrees
in the early phase of the breeding programme.
Spanos K, Andersson EW & Lindgren D,
1997. Multigenerational comparison
between restricted phenotypic and restricted combined index selection for gain
and diversity. In: Kurm M and Tamm Y (editors), Conservation of
Summary: To consider the
performance of sibs (combined index selection) when selecting leads to very low
diversity. However, combined index selection is still rather efficient when
heritability is low. Setting a limit on the highest contribution a parent can
give to the offspring preserves diversity. Remark:
Other studies reported here shows that it is possible to get considerable
better combinations of diversity and gain than obtained by restricting.
Wei R-P, Lindgren D & Yeh
FC. 1997. Expected gain and status number following
restricted individual and combined- index selection. Genome, 40:1-8.
Summary: Proper restriction can
reduce diversity loss to acceptable levels, they are
often urgently needed if selection is based on predicted breeding values. To
restrict phenotypic selection to the good families can improve gain
considerable. A sufficient family number is essential for high gain combined
with sufficient diversity.
Zheng YQ, Lindgren D, Rosvall O & Westin J. 1997. Combining genetic gain and diversity by considering
average coancestry in clonal selection of Norway spruce. Theor Appl Genet 95:1312-1319.
Summary: Applies population
merit selection to a real breeding material. Extends the
concept to finding optimal number of clones in a clone mixture.
Lit list 1998
Below I list and comment some of the publications
I have been involved in which were published 1998. Only studies which I
consider of interest for forest tree breeding are listed. The points which I
find most relevant to breeding are emphasised in the summary (thus it is not
the authors summaries).
Andersson EW, Lindgren D, Spanos
KA, & MullinTJ 1998. Genetic diversity after one
round of selection.
Andersson EW, Spanos KA, MullinTJ & Lindgren D 1998, Phenotypic
selection can be better than selection for breeding value. Scand. J. For. Res.
13:7-11.
Summary: Results show that for moderate to high heritabilities unrestricted phenotypic selection is at
least as efficient as is combined index selection when compared at the same
level of diversity consumed. Combined index selection however has an advantage
in terms of selection efficiency in cases where the heritability is low
especially for large family sizes with the reliability of family means that
follow.
Andersson EW, Spanos KA, MullinTJ & Lindgren D 1998. Phenotypic selection compared to restricted combined index selection for
many generations. Silva Fennica, 32:111-120.
Summary: The consistency of the
results were verified for five generations. The methods compared were
unrestricted phenotypic selection and restricted combined index selection. A
variety of crossing schemes were compared as were family numbers and also the
introduction of a strong dominance component. Results were surprisingly stabile
and showed that apart from rather unlikely events, that combined index
selection has no or insignificant advantages over selection on the phenotype.
Comment: The technically easy
mass-selection (phenotypic selection) seems to have a place in many breeding
operations. If selection is guided by a known family structure (selecting good
individuals from good families) the advantage of high gain is approximately
balanced by the disadvantage of high relatedness. This is an observation I
think Run-Peng Wei deserves a very big heap of credit for, it was first noted
by Wei and Lindgren (1991). A closer examination of why it is so could be a
scientifically interesting study. The value of knowing the exact pedigree has
probably been over-stated by forest geneticists. I remember from my early days
in forest genetics, when we discussed a simple breeding program where forests
were planted from seed orchards and selections to new seed orchards were made
in these forests. We said that this was a stupid and unsophisticated idea, we are much more clever now and keep track of
pedigrees. I wonder how unsophisticated it really was, this deserves to be
studied more in detail.
Bila AD & Lindgren D 1998 Fertility variation in Milletia sthulimannii, Brachystegia speciformis, Brachystegia bohemi and Leucaena leucocephala and
its effects on relatedness in seeds.
Summary: Accumulated fertility in a tree population can be
reasonable well described by a power function (y=x^a),
where a is a parameter describing the variation in
fertility. Fertility variation can also be described as relative status number,
which varied between 0.38 and 0.67 in the objects studied. Thus the
"effective" number of trees can be considered to be between 38 and
67% of the census number of trees (but genetic drift is when not considered).
This can also be viewed as a measure of the relatedness of the seed crop. The
effect of keeping the female contribution constant (thus collecting equal
amount of seeds from each tree) could be quantified. This action has a large
effect (in spite of that paternal contributions are still unknown) and is
recommended for gene conservation purposes. The required number of trees for
getting estimates of fertility variation was derived.
Lindgren D. 1998. Balansen mellan produktion och genetisk mångfald. (In Swedish) Skogsfakta,
7
Summary: Popularised summary in
Swedish about group coancestry and status number, and how these entities may be
balanced to combine gain and diversity in an optimal way. The optimum for seed
orchard establishment and thinning is "linear deployment". It is
recommended that Swedish breeders or breeding related scientists make certain
that they get a copy of this.
Comment:
till now I thought Skogsfakta was a useable channel
for reaching concerned Swedes, now I have serious doubts!
Lindgren D & MullinTJ 1998. Relatedness and status number in
seed orchard crops. Canadian Journal of
Summary: Formulae are derived for group coancestry
and status number in seed orchard crops considering inbreeding and coancestry
of seed orchard genotypes, male and female reproductive success and pollen
contamination (including its relatedness to the seed orchard and to itself).
This method of characterising seed orchards (or rather their crops) seems to be
very general and useful.
Comment: I now lead a project which will estimate
the diversity connected to all Swedish seed orchards using this method as a
basis. Of course I use standardised estimates of most influences, so for most
seed orchards the only really important seed orchard specific input will be
clonal number.
Ruotsalainen S & Lindgren D
1998. Predicting genetic gain of
backward and forward selection in forest tree breeding. Silvae Genetica. 47: 42-50.
Summary: Backward selection of the parent and forward
selection of the best of its kids were compared. The better the parent, the
less likely it is that the best kid is better. Some type of combination of
backward and forward selection can be beneficial. High selection intensity when
plus trees were selected, low heritability and small progenies are factors
which favour selection backwards following wind-pollination.
Comments: In
the situation wind-pollinated progeny and grafts of mother is the basis of the
breeding program, crossing between the best mothers seems the best way to
advance breeding, but the not top ranking mothers could sometimes be replaced
by selection forwards (especially if a high fraction of the families are
expected to contribute to long term breeding), and for some percentiles of
mothers it may be a good idea to advance the breeding population with both
types of selections.
Zheng YQ, Andersson EW & Lindgren D 1998. A
model for infusion of unrelated material into a breeding population. Silvae Genetica: 47:94-101
Summary: A model for introducing fresh material in
breeding programs based on group merit was developed. It may not be uncommon
that it occurs situations when it seems beneficial to inject unrelated material (e.g.
fresh plus trees or the best of the progeny tested plus trees which were
truncated at an earlier stage) to maximise group merit (thus gain at a given
diversity).
Comment: I guess, based
on this study and many others, one can foresee a situation in the current
Swedish program that it would soon turn beneficial to incorporate earlier
unused unrelated breeding material if the number of founders was truncated to
50 immediately to sustain long term breeding populations of size 50. This
problem can probably be avoided by using more than 50 founders from the
beginning.
Gain and Diversity in Multi-Generation Breeding
Programs
Erik W. Andersson
Akademisk avhandling som för vinnande av skoglig doktorsexamen kommer att
offentligen försvaras i Björken, SLU, Umeå, Fredagen den 21 maj 1999, kl 10.00.
Abstract
Progress in tree improvement comes from utilizing
the genetic diversity found in unimproved forests. The balance between genetic
gain and diversity is one of the most important considerations for all
breeders. The sustainability in gain extraction over generations of the
breeding program should be considered from its start. This thesis examines
various strategies for selection in breeding. Using coancestry and its
derivatives as a diversity measure, different methods are contrasted and
compared for their efficiency in giving response to selection, considering the
subsequent change in genetic diversity. It is concluded that restricted and
unrestricted phenotypic selection and restricted (individual and family)
index-selection, using data taking the performance of relatives into account,
are fairly equal in terms of selection efficiency. However, a genuine and
substantial improvement in selection response can be achieved by incorporating
information on the population structure into the selection criterion. A
possible way to enhance the efficiency in realized gain per unit decrease in
diversity is to balance selection with relatedness. This can be seen as
maximizing the allele containing capacity of the breeding population with
regards to constraints on relatedness. Two ways to accomplish this, group merit
selection and a linear programming method optimizing gain with a restriction on
group coancestry are exemplified in this thesis. Benefits from
coancestry-controlled selection are far from negligible, and can have a role to
play in tree improvement. The breeding population should be seen as a dynamic
entity regarding size and family contributions over time. A decision-model for
infusion of fresh unrelated material is presented. The conclusion is that there
often could be a place for refreshing the breeding population with new
selections in the first cycles of breeding. The diversity of regeneration
material affects forests over the long term. It is concluded that diversity of
species undergoing domestication must be monitored, with comparable measures
throughout the whole breeding program, including seed producing stages.
Key words: breeding,
coancestry, genetic diversity, selection, status number, tree improvement.
I see it as a part of a series of four
dissertations, which develops the art of a joint consideration of gain and gene
diversity in breeding and which comes in a logic and uninterrupted line, where
one thesis builds on the previous.
The first part was
Wei, Run-Peng 1995, Predicting Genetic Diverstity and Optimizing Selection,
which mainly discusses gain and diversity changes at the inititation of breeding.
The second part was:
Gea, Luigi D 1997. Genetic diversity and gain.
The concept of status number. PhD
thesis, school of forestry.
This thesis develops the methods making it
possible to analyse gain and diversity jointly over
generations (this dissertation is not much known, most other countries do not
expose their thesis to the surrounding world as
Erik Andersson is the third part:
Gain and Diversity in
Multi-Generation Breeding Programs.
The last part in this "series" is
planned to be defended Sept 3 by Ola Rosvall.
Eriks thesis is built on four papers, the abstract and comments are mine,
not Erik's
Andersson EW, Spanos KA, MullinTJ & Lindgren D 1998. Phenotypic selection compared to restricted combined index selection for
many generations. Silva Fennica, 32:111-120.
Summary:
The methods compared were unrestricted phenotypic selection and restricted
combined index selection for five generations. A variety of crossing schemes
were compared as were family numbers and also the introduction of a strong
dominance component. Results were surprisingly stabile and showed that apart
from rather unlikely events, that combined index selection has no or minor
advantages over selection on the phenotype.
Comment: The technically easy
mass-selection (phenotypic selection) seems to have a place in many breeding
operations. If selection is guided by a known family structure (selecting good
individuals from good families) the advantage of high gain is approximately
balanced by the disadvantage of high relatedness. The value of knowing the
exact pedigree has probably been over-stated by forest geneticists. I remember
from my early days in forest genetics, when we discussed a simple breeding
program where forests were planted from seed orchards and selections to new
seed orchards were made in these forests. We said that this was a stupid and
unsophisticated idea, we are much more clever now and
keep track of pedigrees. I wonder how unsophisticated it really was. It should
be remembered that "phenotypic" in the sense of this paper means that
information about parents or sibs is not used, it is
not discussing when breeding values are based on offspring. This situation
corresponds to a selection with a higher heritability.
Rosvall O & Andersson EW 1999. Group-merit selection compared to conventional
restricted selection for trade-offs between genetic gain and diversity.
Abstract: The Group-merit selection
was devised by Lindgren and Mullin (1997), although they called the method
population merit selection. The superiority over a range of circumstances has
been investigated for a simulated long term breeding program based on single
pair mating. The largest superiority was obtained for low heritability. This
means that the superiority is not very high if the breeding program is based on
breeding values obtained from progeny-testing. Selection intensity was not
important for the superiority. The superiority was sustainable over
generations.
Comment: I have later found that for
more complicated designs the Lindgren & Mullin algorithm can miss the
optimum. I think this has little practical importance and it is most important
in situations of almost balanced selection. For the mating design used in this
paper the algorithm does not miss the optimum.
Andersson EW, Sanchez-Rodrigues
L & Andersson B. 1999. Group
coancestry controlled selection in a Pinus
sylvestris L. breeding program. TAG in press.
Abstract: Integer Linear Programming
was used to maximize gain when group coancestry was seen as a restriction. Real
OP-progeny test data were used to choose parents while the second generation
offspring from crosses between the selected was simulated. Selection which
allows the best-performing families to contribute a greater number was superior.
Comment: The method is expected to
give identical results to that of Lindgren and Mullin (1997). It has the
advantage of relaying on ready programs instead of home made, and it can find
the optimum directly for a given group coancestry, while the Lindgren and
Mullin (1997) algorithm has to find such a solution by iteration. On the other
hand ILP is useless in more complicated cases than this. I remind that
"linear deployment" is a good approximation (see Wei's
thesis).
Zheng YQ, Andersson
EW & Lindgren D 1998. A model for infusion of unrelated material into a breeding
population. Silvae Genetica:
47:94-101
Summary: A
model for introducing fresh material in breeding programs based on group merit
was developed. It may not be uncommon that it occurs situations when it seems
beneficial to inject unrelated material (e.g. fresh plus trees or the best of
the progeny tested plus trees which were truncated at an earlier stage) to
maximise group merit (thus gain at a given diversity).
Comment: I
guess, based on this study and many others, one can foresee a situation in the
current Swedish program that it would in the future turn beneficial to
incorporate earlier unused unrelated breeding material if the number of
founders was truncated to 50 immediately to sustain long term breeding
populations of size 50. This problem can probably be avoided by using more than
50 founders from the beginning. A possible, but considerable less favourable option, is to make certain that the highest ranking not
selected plus trees are kept in archives some decades ahead.
Ola Rosvall's thesis
Ola Rosvall's
thesis is now in print. The public defence is planned for Sept 3rd at
Umeå with Bailian Li as opponent. If you have not got it by Aug 20, you may ask
me or Ola Rosvall.
Its title is:
Enhancing Gain from Long-Term
Rosvall,
O., Lindgren, D. and Mullin, T.J. 1999. Sustainability robustness and
efficiency of a multi-generation breeding strategy based on within-family
clonal selection. Silvae Genetica: 47(5- 6), 307- 321.
Rosvall,
O., and Andersson, E.W. 1999. Group-merit selection compared to conventional
restricted selection for trade-offs between genetic gain and diversity.
Rosvall,
O., and Mullin, T.J. 1999. Positive assortative mating with selection restrictions on group
coancestry enhances gain while conserving genetic diversity in long-term
forest tree breeding. Manuscript.
Rosvall, O., Mullin, T.J., and Lindgren, D. 1999. Controlling parent contributions during positive assortative mating and selection increases gain in
long-term forest tree breeding. Manuscript.