PPRI, Private Bag X5017, Stellenbosch, 7599, South Africa

Background and objectives
The daisy bush Chrysanthemoides monilifera, native to South Africa, is a serious weed of natural areas in eastern Australia. A search for biological control agents in South Africa is currently being undertaken, with various insects having been released in Australia or currently under evaluation for release [1]. The microcyclic rust fungus Endophyllum osteospermi [2] is also considered to be a potentially useful biological control agent. It systemically infects its host, producing witches' brooms which survive for up to 4 years. Heavily infected bushes appear to have their growth and seed production greatly reduced in comparison to uninfected bushes. The host specificity is presently being tested. The work reported here was undertaken to better understand the biology and host penetration mechanism of the rust in preparation for these specificity trials.

Results and conclusions
Excised young leaves (ca 1.5 cm in length, mature leaf length 4-6 cm) of Chrysanthemoides monilifera were carefully rubbed to remove their covering of hairs, and placed an moist tissue paper in 9-cm-diameter petri dishes. Aecial-teliospores of Endophyllum osteospermi were dusted over the surface of the leaves, and sprayed with water using an air brush. The leaves were incubated in a growth room at ca 16C for periods of 24, 48, 72 and 96 h. After these periods the epidermis of the leaves were sliced off using a scalpel, cleared in Carnoy's solution for 15-30 min, stained for 2 min in lactophenol cotton blue, and observed using a light microscope. After 24 h the aecialteliospores had germinated and developed to produce 2 (sometimes 3) basidiospores which had themselves germinated with a thin germ tube whilst still attached to the promycelium. At 48 h penetration pegs had developed directly penetrating host epidermal cells. These developed from a slight swelling at the end of the germ tube, but as no septum separated this swelling it can not be considered to be an appresorium. At 72 h, a small vesicle had developed in the host epidermal cell which had expanded by 96 h. It is assumed that further development consists of the rust growing through the leaf to penetrate the stem (shown by galling) and the axillary bud at the base of the infected leaf. The hormonal balance of the bud must then be changed for its growth and development into a witches' broom. Examination of cross sections of the leaf (assumed original infection site) adjoining young witches' brooms produced in the field confirms that these are systemically infected. Field observations show that occasionally the rust may infect one or two of the closest axillary buds to produce a multiple witches' broom.

From field observations it is known that the development of the witches' broom is delayed for a long period after infection. This is shown by new witches' brooms developing at the beginning of the growth season (July , August) from host branches produced during the previous growth season, or even occasionally from 2 or 3 years previously. Those that develop towards the end of the growth season (November) may arise from host branches produced at the beginning of that same season. This long development time was confirmed by inoculating whole plants in a dew chamber in a manner similar to that described above. Witches' brooms developed on 10% of the bushes inoculated, the earliest developing 7 months after inoculation and the latest after 2 years.

1. Adair RJ, Edwards PB, 1996. Proceedings of the IX International Symposium on Biological Control of Weeds, pp. 429-434.
2. Wood AR, 1998. South African Journal of Botany 64 (in press).