THEORETICAL BASIS OF RUST RESISTANCE TESTING 305 



also be affected by selection in the population of parasites. Much more 

 work is needed on the ecological, physiological, and genetical back- 

 ground of this trend in resistance; but there can hardly be any doubt that 

 it exists. H'e should look upon results of laboratory tests or nursery 

 experiments from a different angle. The character selected for in early 

 tests, i.e., resistance of seedlings or small plants after vegetative 

 propagation, should be correlated with the probability for infection 

 during the complete life cycle of the tree, the family, or the strain, at 

 least up to. an age where it is of economic importance. Genetic gain of 

 indirect selection, which means selection for a character that is easy 

 to assess and correlated with one or more characters of economic impor- 

 tance but difficult to assess, means genetic gain in the probability to 

 escape infection up to an age where infection is no longer important from 

 an economic point of view. This probability, referring to a larger pro- 

 portion of individuals of the family or strain, must be put into the 

 equation for genetic gain from indirect selection as given by Falconer 

 (1964). 



Calculations of genetic gain from indirect selection might be 

 further complicated in our case by genotype -environment interactions and 

 by incomplete control of environmental factors in laboratory tests or 

 nursery experiments. Resistance in any stage of development can be 

 subjected to interactions of this type. Bingham (1968 and these 

 proceedings) has shown pronounced effects of seed bed environment on 

 proportion of plants infected by white pine blister rust. 



DISCUSSION 



The common base of all methods for estimating quantitative-genetic 

 parameters of a population are covariances between relatives. There are 

 no principal differences between procedures to be applied to different 

 characters, the mathematical side having been discussed thoroughly in 

 textbooks of quantitative genetics and breeding. Therefore we have con- 

 centrated here on some of the major problems of model-building resulting 

 from the genetics of host-parasite relations. It will certainly never be 

 possible to put all the parameters needed for complete accounting into 

 one biometrical model. But we might be able to refine our models to make 

 predictions more reliable if we are aware of the biological peculiarities 

 of host-parasite systems and of possible fallacies which might result 

 from applying oversimplified models. 



There is certainly a good chance for employing the concept of genetic 

 gain also in breeding for resistance if the breeder can be sure his model 

 fits the particular case he is working on. The main additional assump- 

 tions to be made if predictions of genetic gain shall be calculated for 

 resistance have been given above. 



The breeder should try to verify whether these assumptions really 

 hold by using evidence from both observations in the forest and basic 

 experiments on physiological and genetical features which might be 

 relevant to understanding the host-parasite system and hence to under- 

 standing the background of resistance which he wants to establish or to 

 increase quantitatively in the host population. 



This combination of field and laboratory work seems to be the best 

 approach to all problems of ecological genetics (Ford, 1964). Results 

 from planned experiments following the rules in quantitative genetics can 



