PE RSI STANCE OF RUST RESISTANCE 563 



implied maturation of tissues at the infection court. Satisfaction cannot 

 be derived from this condition since the offspring of these trees, if used 

 for further breeding work, must pass through the most critical stage of 

 their life cycle again. Artificial inoculation will have to be applied 

 and we have heard about sophisticated equipment to test many small samples 

 of trees for resistance. An important point here is that both quality 

 and intensity of inoculation can be calibrated approximately (Bingham 

 et al. , 1969). There are indications that variation of infection intensity 

 can lead to inconsistent estimates of genetic variances and heritability 

 and of tentative gain. Such estimates, particularly the latter, would 

 diverge, and nobody would base all future operations on a single estimate. 

 Without some idea of the order of magnitude of such estimates for several 

 conditions, it is impossible to evaluate the prospects of various testing 

 strategies . 



Genetic correlations between various manifestations of resistance 

 possess two aspects that may be considered in more detail. One is the 

 correlation between early and adult behavior since in artificial inocula- 

 tion one wants to have the host organisms small for easier handling. 

 This means the host plants are younger (and also, in the case of blister 

 rust, more liable to infection) and are exposed to an amount of infection 

 that equals that of a longer period in the field. The other relates more 

 to the infection as such. The test environments may be as different as 

 the quality of the inoculum. Pathogenic races are almost invariably 

 found in obligate parasites such as rusts. If the pathogen genotypes 

 effective in the more or less heterogenous inoculum materials differ, the 

 two measurements of resistance may be only loosely correlated because 

 they measure different things. If the pathogen genotypes do not differ, 

 one might expect that resistance under natural and artificial conditions 

 measures the same thing and that only error variance is responsible for 

 low correlations. Host vigor may affect the inference on genetic varia- 

 tion of host resistance quite drastically (Walker, 1966) , since gene 

 action depends on the environment in the broadest sense of the word. 



It is clearly a matter of experimentation to find out which methods 

 are efficient and to investigate whether several testing techniques 

 should be used. One may wish to handle the various resistance measure- 

 ments somewhat like a selection index by assigning equal weight to every 

 such measurement. Fortunately, we deal with traits that are measurable 

 relatively shortly after one another. On the other side, one might think 

 of the higher selection differentials under some semi -natural, mass- 

 screening procedure that the index cannot fully utilize. But it is not 

 clear whether resistance always behaves like a quantitative trait with 

 additive gene-effects prevailing. This method of selection may not be 

 reasonable at all. Bingham ez at. (1969) have described a way of planning 

 and establishing tests that enables the breeder to make certain inferences 

 about the population handled. 



Both the measurement of the resistance reaction after inoculation 

 and the mere observation of field resistance can be "biased" in some 

 direction. We do not know which is more reliable, so we do best to 

 accept both. This will lead to material with a broad genetic basis of 

 resistance. 



A prerequisite of persistence of an achieved degree of resistance is 

 that the testing results show high repeatability over several physico- 

 chemical environments and modes of exposure to infection. Otherwise 

 the results may be simply artifacts in that the test conditions were in 



