POPSIM vs Breeding Cycler

POPSIM considerations

I was asked to deliver considerations on selection forwards in polymix progenies for applications in POPSIM I did so in August 2007 and the writing when is glued below. I write a workreport about forward selections in polymix but this text is excluded from that and inserted here instead as it is not judged to be of long time relevance.

Comparisons POPSIM vs BreedingCycler is of interest for at least two reasons:
* Breeding cycler deals with cycle time and costs which POPSIM does not, but the desire is to insert these features in POPSIM
* There can easy be bugs in simulators as well as logic errors. If compatible results are obtained by POPSIM and BREEDING CYCLER for compatible inputs they back-up each other and are probably mainly correct. If incompatible results are obtained it indicates a bug or mistake in one of the simulators.

Projektet fokuseras mycket omkring POPSIM projektet med Ola som drivande kraft. Dokument med olika alternativa system som skall POPSIMmas från December 2007. EXCEL ark med förslag till cykeltider och plantbehov. Mainly Jon performs the runs. A POPSIM version useable for most alternatives was developed in February 2008 and runs were made in May 2008. Unfortunately POPSIM can not yet run strategy 5 in December 2008. Funding for POPSIM (Tim main contributor, Dag consultant) is available. POPSIM was commented by Dag at the meeting Uppsala May 14 2008.

Skogforsk has got a EU project Novel Tree since late 2008, which includes considerable POPSIM activities 2009-2011 including work with Tim Mullin (he is 30% employed by Skogforsk) and Johan Kroon

Last edit 090112

POPSIM APPLICATION DagLindgren August 2007

Background Late 2006 I was asked to suggest Polymix to SkogForsk. This enquiry has resulted in documents on

http://www-genfys.slu.se/staff/dagl/Foradlrevision07

It is based on discussions with Ola and Curt at two meetings and email exchange and the main suggestions has been presented to all relevant Skogforsk. The documents have been available since March 2007 for comments and I have invited for comments. I have also invited Tim to discuss March 2007. Thus the issue seems rather thoroughly discussed. My suggestion must be the most interesting to get exposed to POPSIM runs for Swedish breeding, as it is the most well-considered and little real arguments against. However, the POPSIM runs are not trivial and must be discussed and may even be forbiddingly difficult.

It is not easy, Ola suggested 25 pollen in the mix and 500 seed parents – and Tim 250 and I 5*25. I object to Ola, because it gives no room whatsoever for preferring good fathers. For POPSIM the important is just that it can be run, when we can write papers what is best.

The suggestion is that for one subpopulation (there are somewhat more than 20 in Sweden) among the offspring to the breeding population five polymixes are done, each with 25 pollen. Thus there are 125 field-selected candidates for pollen parents. Pollen is harvested in the trials from these. 250 seed parent candidates are field-selected and grafted. Trees serve either as pollen parents or seed parents. Mating relatives is avoided. Balance of the founder genomes is aimed at. A field trial with 10000 seedlings is established. 15 years later it is an option to only select the seed parents with the highest breeding values backward. But the main line for the simulation is to go on with PMX/WFA. In the best families a number of candidates are selected and subject to molecular father analyses. Based on these rather few father analyses estimate of breeding value can be done. A final selection of 250 seed parents and 125 pollen parents is done so the selections are reasonable balanced considering the founder genomes and so that genetic gain is high.

It seems complicated and a lot of specifications and considerations needed to squeeze the alternative into POPSIM. So it is something which need discussion. And actually it is a bit awkward to do all this job because at heart we believe there is a better and easier way – more controlled crosses.

Is there short-cuts? Yes, but if we analyze PMX/WFA, is it not right to try to find something close to optimal, and is it not right to start with what is considered the best option for Swedish breeding? The most important comparison is with a large number of pollens in separate crosses and when it is good to have that number, where 125 may be right.

I will not specify exactly what PMX/WPA details I believe most in until simulations are close to starting. Of course the system is not optimized and we may find better values for the numerics.