Stratified Sublining
1
D. Lindgren1, S. Ruotsalainen2,3 and M. Haapanen2,4
Swedish University of Agricultural Sciences, 901 83 UMEÅ, Sweden, 2The Finnish Forest
Research Institute, 3 58450 Punkaharju, Finland, 4 01301 Vantaa, Finland
Stratified sublining arranges the breeding material in unrelated sublines ranked for breeding
value. As a direct result of the stratification, the distribution of breeding values of the best
unrelated clones, which will be used in seed orchards, is amplified. The strategy can be seen as
“elite-main” or “PAM” driven to its extreme.
Stratified sublines can be constructed as follows (Ruotsalainen and Lindgren 2000):
1. Rank tested founders (F0 breeding population) for breeding value;
2. Mate adjacent founders (Positive Assortative Single-pair Mating which is equivalent of
creating F1 stratified sublines derived from 2 founders);
3. Test individuals in offspring for breeding value;
4. Select the two best offspring in each F1 family;
5. Rank the offspring pairs for their mean breeding value (F1 breeding population);
6. Mate the adjacent offspring pairs (the best with the best and the second best with the
second best);
7. Now F2 stratified lines have been formed, each subline comprising 4 individuals. Each
individual in the breeding stock will thus have 4 founders with a similar rank as
grandparents. F2-individuals in different sublines are not related whereas individuals
within sublines are either full sibs or double first cousins.
Stratified sublining is powerful in supporting clonal seed orchards, as well as many other ways to
transfer genetic gain from the breeding population to the production population, at least two
breeding generations ahead. In the first generation, stratified sublining is identical to positive
assortative mating (PAM), which has turned out to be effective in combining diversity with
genetic gain. Two generations ahead, compared with assortment to sublines at random, which has
been considered for Pinus taeda (McKeand and Bridgwater 1998), the superiority of stratified
sublining, mostly due to the effect of PAM, is above 15%. Stratified sublining is also highly
capable of supporting family forestry, as crosses of the best parents will be tested, and as
unrelated sublines with superior breeding value will be created. This will constitute a superb
starting material for developing clones for clonal forestry.
Starting with the maximum number of sublines (half the number of founders) makes it possible to
completely avoid inbreeding for at least three generations of breeding. As the development of
coancestry remains strictly confined to sublines, the number of unrelated selections available to
form a seed orchard is equal to the number of sublines in the breeding population. This number is
halved if all sublines are pairwise merged to obtain unrelated trees to regeneration matings. Since
inbreeding within sublines must be kept rather low to avoid harmful effects, the potential of
drawing unrelated selections to seed orchards may be significantly reduced due to the merging
after three or more (depending on the size of the founder population) generations of breeding. At
this point, different management strategies can be used. Feasible options involve, for example,
enriching sublines with fresh material, or allowing the accumulation of inbreeding within
sublines, or abandoning the sublining completely. The long term prospects are likely to be better
or at least as good as if stratified sublining had never been applied. In other words, it is not likely
to appear a penalty in the far future for applying stratified sublining in the first generations.
Stratified sublining will be implemented in the recently developed Finnish breeding strategy.
Therefore, the concept as a component in the Finnish breeding strategy and options to strengthen
the implementation were considered. In the following a summary is given of the planned way of
implementation of stratified sublines in Finnish breeding programs for Scots pine and Norway
spruce.
To further boost the effect of stratification and to obtain additional genetic gains from future seed
orchards, the Finnish breeding strategy involves the idea of distributing breeding, testing and
selection efforts unequally, making the effort positively dependent on genetic value of the
material being improved. This principle is implemented throughout the breeding cycle. In the
first generation turnover, the founders forming the first-generation breeding population (160
individuals) are single-pair mated with regard to breeding value (in the way described above).
Those of the founders that are ranked to the highest quarter are, however, double-pair mated to to
allow more options for recombination of their gene mass and reduce the risk that their genemass
is degraded by an unfortunate choice of partner, as well as a way to increase the number of
offspring. Furthermore, the target sizes of F1 families are larger (160 seedlings) for the best
quarter of the parents than for the average parents (120 seedlings) or for the lowest quarter of
parents (80 seedlings). The F1 families (the recruitment population) are grown in forward
selection trials that last from 5 to 10 years depending on species. At this age, the best individuals
within each full-sib family are phenotypically selected for further testing. Roughly three times as
many selections (candidates) are drawn from within the larger families (representing offspring of
the best parents) than from the smaller ones. In the two stage selection system applied, the
candidates are then clone or progeny tested for about 12 to 15 years to more accurately determine
their true breeding values. The genetic testing is essential for the new round of stratification to be
successfully carried out at the end of the second breeding cycle.
The number of candidates selected from each full-sib family to the new breeding population
varies in relation to the mean breeding value; three individuals are selected from the best full-sib
families (determined as the mean breeding value of the top 3 candidates), two individuals from
the average families, and one (possibly none) individual from the lowest ranking families
(Ruotsalainen and Lindgren 2001). As the result, the second generation breeding population,
comprising 100 individuals, will have an unbalanced structure where the size of the stratified
subline is six, four or two trees for the highest ranking, the average and the lowest ranking
candidates, respectively. This method results in a overrepresentation of the gene mass of the best
founders whereas a high number of low ranking founders will still be represented, but with
relatively little genetic contributions.
One of the many options for the management of stratified sublines could be to use gene mass of
the best founders to strengthen the lowest ranking lines. The relatedness between sublines would
probably not be too big a problem as the low ranking lines are anyway unlikely to support seed
orchards in the near future. Breeding material available has often characteristics which justify or
require modifications in the construction of sublines.
Literature:
Ruotsalainen, S. and Lindgren, D. 2000. Stratified sublining: a new option for structuring
breeding populations. Canadian Journal of Forest Research. 30: (4) 596-604.
Ruotsalainen S & Lindgren D 2001 Number of founders for a breeding population using variable
parental contribution. Forest Genetics 8:59-68.
McKeand, S. E. and Bridgwater, F. 1998. A strategy for the third breeding cycle of loblolly pine
in the southeastern U.S. Silvae Genet. 47: 223-234.
This document was published as
Lindgren D, Ruotsalainen S & Haapanen M 2004. Stratified sublining. In Li B & McKeand S
Eds Forest Genetics and Tree Breeding in the Age of Genomics: Progress and Future. Conference
Proceedings, pp 405-407.