1 University of Florida, 700 Experiment Station Road, Lake Alfred, FL 33850, USA; 2 Department of Plant Protection, University of Gukurova, Adana, Turkey

Background and objectives
Alternaria alternata (Fr.:Fr.) Keissl. causes citrus brown spot, a destructive leaf and fruit-spotting disease of citrus in all areas of the world where citrus is grown. The disease is particularly important on grapefruit-tangerine hybrids (Citrus paradisi X C. reticulata) such as Minneola tangelo but also affects the rootstock cultivar Rough Lemon (Citrus jambhiri). Isolates from Minneola and 'Rough Lemon' have been shown to be host-specific and produce host-specific toxins that are required for pathogenicity. The specificity of Alternaria isolates from other citrus hosts is not known. The objective of this study was to determine the genetic structure and host specificity of populations of A. ;alternata on 'Rough Lemon' (RL) and Minneola (MN) in Florida.

Materials and methods
Brown spot-infected leaves were randomly sampled from two abandoned citrus groves with RL volunteer root sprouts and from scions in a commercial Mn grove. Each grove was about 20 ;km from the other. Ninety-eight isolates of A. alternata were obtained from the three groves. DNA was extracted from all sampled isolates and screened for informative RAPD markers. Allele frequencies were scored for 16 RAPD loci produced with four primers. Nei's GST was used to estimate genetic differentiation among groves (populations). Genetic differentiation among isolates was estimated using Nei's genetic distance and dendrograms were generated using UPGMA. The pathogenicity of a subset of isolates from each population was determined using a detached leaf assay and a spray inoculation assay with intact shoots. All isolates were inoculated on RL and MN regardless of the host of original isolation.

Results and conclusions
Comparisons of RAPD allele frequencies among the three populations revealed a high level of genetic differentiation (GST=0.29). Almost all this differentiation was due to differences in allele frequencies between the RL and MN populations supporting the hypothesis that isolates of the pathogen are specialized on each of these hosts. A low level of genetic differentiation was observed between the two RL populations (GST=0.03) and 26 of 61 isolates (43%) from both populations had the same multilocus genotype. When isolates from the two RL populations were pooled and genetic distances were compared among isolates, three genetically distinct clusters of isolates were apparent. One cluster was highly genetically uniform with almost all (25/26) isolates having the same multilocus genotype. Most of the isolates within this cluster were pathogenic on RL and not on MN. One isolate with slightly different genotype was pathogenic on MN and not on RL. Another isolate with the dominant genotype was nonpathogenic on both hosts. The other two clusters were genetically distant from the first and much more variable within each cluster. Most of the isolates within these clusters were non-pathogenic on either host. Within the MN population, two genetically distinct clusters were observed and all isolates were pathogenic on MN but not on RL. When isolates from all three populations were pooled, three distinct clusters of isolates were apparent and RL and MN isolates were found together in two of the three clusters. These results indicate that an extremely high level of within-population genetic diversity exists among A. ;alternata isolates on citrus and may indicate the need for taxonomic revision of this fungus. Despite the apparent host specialization of the pathogen on RL and MN at the population level, related isolates were found associated with both hosts. It also appears that pathogenicity on a particular host is not limited to a single fungal lineage because at least one cluster contained isolates able to attack both hosts. In addition, all three clusters had both saprophytic and pathogenic isolates. We speculate that pathogenicity may have evolved multiple times in this fungus and that saprophytic and pathogenic isolates may be more closely related than pathogenic isolates from different citrus hosts.