332 RONALD J. DINUS 



This paper describes infection rates under artificial conditions and 

 compares them to those observed in field plantings at a number of locations 

 and after several years of exposure. 



METHODS AND MATERIALS 



Six slash pines were selected in Harrison County, Mississippi, for 

 use in artificial and field tests (Jewell and Mallett, 1967). Three were 

 rust-free (8-7, 11-6, and 18-27) and three were infected (18-40, 18-41, 

 and 18-62). The natural stands involved were about 75 percent infected. 

 Open-pollinated seeds were collected for several years from each of the 

 six selections. 



Dr. F. F. Jewell 1 artificially inoculated open- pollinated progenies 

 of the six trees in identical randomized block designs in 1963 and 1964. 

 Each family was represented by two row plots of 18 trees in each of five 

 blocks. Proportions of plants infected per row were determined 9 months 

 after inoculation. Results of one of the individual experiments were 

 reported previously (Jewell and Mallett, 1964). For presentation here, 

 the original data were reanalyzed as separate tests and as a single 

 experiment combining both years. The prop ortion o f plants infected was 

 transformed to degrees of angle = arcsin /percent and subjected to 

 analysis of variance for the randomized block design with intra-block 

 replication (Steel and Torrie, 1960). Family means were compared by 

 multiple range tests at the 0.01 level. 



Field tests of open-pollinated progenies were established by Jewell 

 in 1963 at Gulfport , Mississippi, and in 1964 on Crown Zellerbach 

 Corporation land near Bogalusa, Louisiana. The Gulfport planting contains 

 progenies from all six selections in a randomized block design. Each 

 family is represented by two row plots of 15 trees in each of seven blocks. 

 The Bogalusa planting consists of five blocks each containing two row 

 plots of 15 trees; five families (all but family 18-41) are represented. 



Rust infection was measured annually in terms of three indices: 

 number of stem infections per plant, total number of infections per plant, 

 and proportion of plants infected per row plot. The proportion infected 

 in the latest year was also analyzed after plants previously infected 

 but currently rust-free had been deducted. In each cas e, proportions 

 were transformed to degrees of angle = arcsin /percent . Plot means for 

 the four variables were subjected to analysis of variance for the random- 

 ized block design with intra-block replication. A combined analysis of 

 both field tests was not attempted as the effect would have been con- 

 founded with different periods of exposure. Family means were compared 

 by multiple range tests at the 0.01 level. In addition, simple phenotypic 

 correlations were calculated to determine which index best measured 

 infection. 



^Currently Associate Professor of Forest Pathology 3 Louisiana Polytehnic 

 Institute, Ruston 3 Louisiana. 



