Prescreening Slash Pine and Cronartium Pedigrees for Evaluation of Complementary Gene Action in Fusiform Rust Disease

Stelzer, H. E., Doudrick, R. L., Kubisiak, T. L., and Nelson, C. D. 1999. Prescreening slash pine and Cronartium pedigrees for evaluation of complementary gene action in fusiform rust disease. Plant Dis. 83:385-389. Single-urediniospore cultures of the fusiform rust fungus were used to inoculate seedlings from 10 full-sib families of a five-parent slash pine diallel at two different times in 1994. The presence or absence of fusiform rust galls was recorded for each inoculated seedling at 9 months postinoculation, and percent infection levels for each family-inoculum-time combination were used for detecting differences among host families and fungal cultures and for identifying differential interactions. The existence of differential interactions between two or more fungal cultures and two or more host families verifies that complementary gene action does exist in this pathosystem. Some host families may be excluded from more detailed interaction studies on the basis of their redundancy and lack of participation in differential interactions. Additional keyword: Pinus elliottii Despite 40 years of phenotypic selection and breeding in both loblolly pine (Pinus taedu L.) and slash pine (Pinus elliottii Engelm. var. elliottii) against the fusiform rust fungus (Cronartium quercuum (Berk.) Miyabe ex Shirai f. sp. fusiforme (Hedge. & N. Hunt) Burdsall & G. Snow), the genetic basis of this host-pathogen interaction is not clearly understood (4,24). Recent efforts to characterize the genetic interactions involved in the development of this disease on loblolly and slash pines have resulted in the hypothesis that the interaction may conform to a complementary genetic system (16,17,21,30,32). A principal argument against the complementary genetic system is that family and race specificity are artifacts of plant breeding and thus are very unlikely to occur in a natural pathosystem such as that of P. elliottii var. elliottii and C. quercuum f. sp. fusiforme (2). Such specificity, however, has been demonstrated in wild populations of Glycine canescens and G. argyrea and the rust pathogen Phakopsoru pachyrhizi (3). Furthermore, it has been argued that much of what is called “general” or “quantitative” resistance is the result Corresponding author: T. L. Kubisiak E-mail: kubisiak @datasync.com Accepted for publication 10 December 1998. Publication no. D-1999-0128-02R This article is in the public domain and not copyrightable. It may be freely reprinted with customary crediting of the source. The American Phytopathological Society, 1999. of specificity and can be explained by the interactions among complementary gene pairs (19). The complementary genetics model for fusiform rust disease has three basic tenets (1530). First, low levels of reaction (i.e., resistance) in the host and of pathogenicity (i.e., avirulence) in the fungus are considered dominant traits. Second, if the genotype of the complementary gene from either the host or pathogen is homozygous recessive, then a high infection type (e.g., a gall) will occur. Finally, when two or more pairs of complementary genes are present, the gene pair that imparts the lowest infection type (e.g., no gall) is epistatic to all other gene pairs. The fusiforrn rust pathosystem, however, presents some challenges in verifying the model and identifying complementary gene pairs. First, the nonrepeating spermatialaecial stage of the fusifoxm rust fungus is the one that causes the economic damage we wish to control. An in vitro inoculation technique in which single-genotype haploid cultures of the fungus were used has been reported for infecting loblolly pine hypocotyls (13,31). However, in vitro efforts to infect epicotyls or micropropagated shoots were not successful (31). Recent attempts to inoculate slash pine epicotyls by using agar plugs containing mixedor single-genotype hyphal cultures of the fungus have been highly variable, thus restricting at present the potential usefulness of the technique for research (R. Schmidt, personal communication). Singleaeciospore and single-urediniospore cultures exhibiting varying degrees of pathogenicity have been developed for use in complementary genetics research of this pathosystem (6,17,21,30,32). The basidiospores that infect pine, however, have been shown to be haploid products of meiosis and hence represent a segregating population (57). Significant problems with loblolly and slash pine include the long time it takes to generate F, and F2 populations (compared with agronomic species) and the inability to clonally propagate large numbers of selected host genotypes. Rooted cuttings are currently the most efficient means of vegetative propagation. Rooting success, however, is still dependent upon the genotype of the individual. In addition, there are reports of rooted cuttings exhibiting an apparent increased resistance to the rust fungus (9,lO). Genomic mapping and bulk segregant analysis have been used recently to identify regions of the loblolly host genome that confer a low infection type when singleaeciospore cultures of the rust fungus that possess the corresponding complementary gene are used for inoculation (17,32). These techniques, in conjunction with the basic hypotheses of the complementary genetics model, are being used to identify reaction genes in the host (H. Amerson, personal communication). This work has utilized a megagametophyte DNA approach that can discern only genes of a heterozygous nature in the maternal host parent. It has been demonstrated in the barleymildew (Hordeum vulgare-Erysiphe graminis f. sp. hordei) disease association that the genetics of the pathogen can be inferred from a detailed genetic study of the host (20). A similar approach could be used for studying the southern pine-fusiform rust disease association. By inoculating an array of host pedigrees from a full-diallel mating scheme with basidiospores from several single-urediniospore fungal cultures and using the tenets of the model, hypothetical genotypes in both organisms for each complementary gene pair could be derived. Once the initial host and pathogen genetic arrays have been rogued of uninformative pedigrees, the resources required for subsequent studies involved in mapping reaction genes in the host or pathogenicity genes in the rust fungus could be substantially reduced. The objective of this study was to determine whether seedlings Plant Disease/April 1999 385 from full-sib families of slash pine could be used to prescreen host plant material and cultures of the fusiform rust fungus for their potential to exhibit complementary genetic interactions. MATERIALS AND METHODS Host seedlings. Five first-generation selections of fl elliottii var. elliottii were control pollinated in a full-diallel crossing design. All five selections were chosen from the U.S. Forest Service’s Harrison Experimental Forest in Harrison County, MS. In previous research, in which bulk inocula of the rust fungus was used, these selections were classified as follows: parent 8-7, resistant; parents 9-2 and 18-27, moderately resistant; and parents 18-26 and 18-62, susceptible (14,27). Twenty of the 25 full-sib families were propagated from seed (8) (selfs were excluded because of insufficient seed). Seed germination was staggered so that the seedlings were 8 weeks old at the time of each inoculation series. Pathogen cultures. Four single-urediniospore cultures of C. quercuum f. sp. fusifonne were used in this study: CCA2.SSl.SS1, wLP-10-2.ss1.ss1, CZ.SSl, and LM-S.SSl.SSl (5; U.S. Forest Service SRS Study Plan 2.01, unpublished, available upon request). All four cultures originated from aeciospore collections made in 1984: CC-A-2 on planted Livingston Parish loblolly pine in Madison County, FL; WLP-IO-2 on planted Livingston Parish loblolly pine in Livingston Parish, LA; CZ on native slash pine in Washington Parish, LA; and LM-5 on native slash pine in Jones County, MS. Single-urediniospore cultures were then developed from the aeciospore collections (5,23). Artificial inoculations. All inoculations were performed at the Southern Institute of Forest Genetics in Gulfport, MS, on 22 to 25 May and 11 to 14 July 1994. For each inoculation time and rust fungus culture, an average number of 36 seedlings from each full-sib family were randomly divided into four replications. In each replication (day), families and cultures were randomly ordered prior to the start of that day’s inoculations. A forced-air apparatus (29) was used to deliver 12 to 18 basidiospores/mm* of each fungal culture to the juncture of the stem and base of the terminal tuft of juvenile needles on each seedling. Spore density was monitored after inoculation of every tenth seedling and adjusted to ensure close adherence to the target density. After inoculation, the seedlings were incubated in the dark at 20 to 22°C and 100% relative humidity for 24 h. After incubation, the seedlings were returned to the greenhouse and grown under an 18-h-light photoperiod provided by 1,000-W metal halide lamps. Two weeks after inoculation and every second week thereafter, the seedlings were fertilized (20-20-20 N-P-K, 200 ppm N). Data collection. The presence or absence of fusiform rust galls on the seedlings was recorded 9 months after inoculation. The percent infection was then determined for each replication within a given full-sib family x inoculation time x inoculum treatment combination. Given the simplest scenario of only one complementary gene pair operating in this biological array, infection levels can range from 0 to lOO%, depending upon the segregation ratios of the complementary genes in each organism. If the single-urediniospore culture is homozygous for high pathogenicity (i.e., aa), then all the haploid basidiospores would be virulent and the expected infection would be lOO%, regardless of segregation in the host. If the culture is homozygous for low pathogenicity (i.e., AA), then all the haploid basidiospores would be avirulent and the expected infection could be 0, 25, 50, or lOO%, depending on whether the R:r segregation ratio is 1:0, 3: 1, 1: 1, or 0: 1, respectively. Given a 1:l segregation of avirulent and virulent basidiospores resulting from a heterozygous single-urediniospore culture and R:r host segregation ratios of l:O, 3:1, 1: 1, or 0: 1, the expected infection levels could be 50, 62.5, 75, or lOO%, respecTable 1. Analysis of variance for the effects of replication (REP), family (FAM), cross direction (RECIP), inoculation time (TIME), and inoculum (INOC) on the arcsine of the square root of percent infection of slash pine seedlings after inoculation with basidiospores from four single-urediniospore cultures of the fusiform rust fungus” Unpooled datar Pooled data2 Effect d f M e a n s q u a r e P d f Meansquare P FAM 9 1.9251 0.0001 9 1.9069 0.0001 RECIP(FAM) 9 0.1056 0.1042 INOC 3 0.8579 0.0001 3 0.8727 O.&l1 FAM*INOC 27 0.2553 0.0001 27 0.2529 0.0001 TIME 1 0.0036 0.8123 1 0.0036 0.8129 REP(TIME) 6 0.1469 0.0363 7 0.0946 0.1311 FAM*TIME 9 0.4087 0.0001 9 0.4228 0.0001 INOC*TIME 3 1.1029 0.0001 3 1.0783 0.0001 FAM*INOC*TIME 27 0.1038 0.0297 27 0.1089 0.0216 x Mean squares and P values were determined by the SAS general linear model procedure. Y R2 for unpooled reciprocal cross data = 0.57. z R2 for pooled reciprocal cross data = 0.56. 386 Plant Disease / Vol. 83 No. 4 tively. The array of expected percent infection levels expands as more complementary gene pairs are added to the model (30). Statistical analysis. The data were initially analyzed by the general linear model procedure (6th ed., SAS Institute, Cary, NC) to test for significant effects. Because percent infection data are based upon a binomial response and some data points lie outside the stable variance range of 30 to 70%, all data were transformed to the arcsine of the square root of the percent infection (1). Family (FAM), inoculum (INOC), and inoculation time (TIME) were all considered to be fixed effects. Reciprocal cross effects were treated as a nested effect within family (RECIP[FAM]) and considered fixed as well. The replication effect was nested within inoculation time (REP[TIME]) and regarded as random. Differential interactions. Within a particular inoculation time, a significant family x inoculum interaction effect will suggest differential interactions between two or more fungal cultures and two or more host families. These differential interactions are characteristic of the complementary genetics system (19) and have been reported in earlier research on half-sib families of loblolly pine inoculated by using bulk inocula collected from single galls (25). For each inoculation time, a series of twoway contingency tables for individual combinations of two fungal cultures and two families were constructed, and significant (P = 0.05) differential interactions were identified by &i-square analysis (22). RESULTS Analysis of variance revealed no significant reciprocal cross effects. Therefore, forward and reciprocal cross data were pooled (Table l), after which the replication effect was no longer significant (P = 0.1311). Hence, replications were pooled. Increasing the average sample size for a given family-inoculum-time combination to 72 greatly increased the power of estimating percent infection levels. Family, inoculum, and family x inoculum effects were all significant (P = 0.0001) and accounted for 39% of the observed variation. These data suggest that infection type reversals exist. While the time main effect was not significant (P = 0.8129) all higher order interactions involving time implied that time did play some role in the development of disease symptoms. Percent infection data for all familyinoculum-time combinations are given in Table 2. For inoculations conducted in July, any host family having 8-7 as a parent differentiated the four inocula. On these families, culture CCA-2SSI averaged more than 50% infection, while LM-5SSl averaged 18%. Cultures CZ.SSl and WLPlo-2SS 1 were indistinguishable, each scoring less than 5% infection. The May inoculations yielded no host families capable of differentiating the four inocula. None of the fungal cultures were able to identify any general difference patterns among the host families. Differences in percent infection data between the May and July inoculations were observed in several families for each of the four fungal cultures (Table 2). When challenged with CZ.SSl and WLP-lo-2.SSl inoculum, all host families having 8-7 as one of the parents exhibited a noticeable reduction in infection levels for the July inoculations compared with those in May. For CCA-2.SS1, the percent infection was greater for the May inoculation series, regardless of host pedigree. Differential interactions. Several significant (P = 0.05) differential interactions among particular pairs of host families and fungal cultures were identified in both the May and July inoculation series (Table 3). The standard error of the mean percent infection data was 6%, but this error did not interfere with the detection of a differential interaction. In May, the host family pair 8-7 x 18-62 and 18-26 x 18-62 exhibited an infection type reversal for fungal cultures CCA2.SSl and CZ.SSl. Another family pair, 87 x 18-62 and 9-2 x 18-26, identified differences between fungal cultures CCA2.SSl and WLP-lo-2.SSl. A third family pair, 8-7 x 18-62 and 8-7 x 18-26, differentiated fungal cultures LM-S.SSl and WLP-lo-2.SS 1. A fourth differential interaction involved families 8-7 x 18-62 and 18-26 x 18-62 and fungal cultures CCA2SSl and LM-5SSl. These same two families also interacted differentially with fungal cultures CCA-2.SSl and WLP-lO2.SSl in the May inoculation series. Four significant differential interactions were found among the July inoculations. All four involved the fungal culture pair CCA-2SSl and WLP-lo-2.SSl. In addition, all four differential interactions shared the common family 18-27 x 18-62. The other host families involved were 8-7 x 92, 8-7 x 18-26, 8-7 x 18-62, and 9-2 x 1826. Within family 9-2 x 18-26, a significant differential interaction was also observed involving the two fungal cultures CCA2SSl and WLP-lo-2.SSl and the t w o inoculation times. DISCUSSION Pathogenic variability was shown within C. quercuum f. sp. fusiforme when bulk aeciospore isolates from different geographic sources were tested on half-sib families of loblolly and slash pine (27,28). The observed variability was particularly striking among progeny from putative resistant parents, in which there was in some cases a complete breakdown of resistance Table 2. Percent infection levels (mean f SD) of 10 full-sib slash pine families inoculated with basidiospores from four single-urediniospore cultures of the fusiform rust fungus on 22 to 25 May and 11 to 14 July 19942