1317 Int. J. Plant Sci. 162(6):1317–1326. 2001.  2001 by The University of Chicago. All rights reserved. 1058-5893/2001/16206-0015$03.00 INTERSPECIFIC HYBRIDIZATION WITHIN EUCALYPTUS (MYRTACEAE): SUBGENUS SYMPHYOMYRTUS, SECTIONS BISECTAE AND ADNATARIA K. L. Delaporte,* J. G. Conran,† and M. Sedgley 1, * *Department of Horticulture, Viticulture, and Oenology, Waite Agricultural Research Institute, Adelaide University, Glen Osmond, South Australia 5064, Australia; and †Department of Environmental Biology, Adelaide University, Adelaide, South Australia 5005, Australia The potential for interspecific hybridization within the genus Eucalyptus was investigated through controlled pollination and measurement of seedling leaf morphology. Eucalyptus gillii and E. socialis (subgen. Symphy- omyrtus sect. Bisectae ser. Subulatae) were used as the female parents, and pollen was sourced from 16 Eucalyptus species from a number of series within sections Bisectae and Adnataria (subgen. Symphyomyrtus). Thirty-four out of 36 crosses produced seeds; however, the percentage of seeds produced per flower pollinated varied considerably between crosses, as did germination percentage and seedling survival. At 3 mo of age, all surviving seedlings were measured for 15 leaf and stem characters. Multivariate analysis (ordination) of the data from each cross placed the seedlings in relation to their parents, with most crosses intermediate, albeit closer to the maternal parent. Successful hybridization occurred within sect. Bisectae ser. Subulatae and between ser. Subulatae and sect. Bisectae ser. Kruseanae, ser. Levispermae, ser. Curviptera, ser. Erectae and sect. Ad- nataria ser. Heterophloiae. In contrast, crosses between sect. Bisectae ser. Subulatae and sect. Adnataria ser. Aquilonares and ser. Melliodorae did not produce hybrids. Crosses between closely related species showed a greater degree of success than those between distant crosses, as did those between species with similar flower size. Keywords: Eucalyptus, controlled pollination, multivariate analysis, ordination. Introduction The genus Eucalyptus L’Herit. contains over 700 species (Brooker and Kleinig 1999), and Pryor and Johnson (1971) proposed a classification into a hierarchy of subgenera, sec- tions, series, subseries, superspecies, species, and subspecies, with groupings based on shared diagnostic morphological characteristics of the constituent taxa. Their classification was modified recently by Brooker (2000), and his system is the one followed here, although his inclusion of the genera Angophora and Corymbia within Eucalyptus sens. lat. is controversial (La- diges and Udovicic 2000). Subgenera are reproductively iso- lated, and the likelihood of a successful cross increases with closeness of taxonomic affinity: intraseries crosses have a greater chance of success than interseries crosses, which them- selves have a greater chance than intersectional crosses (Griffin et al. 1988). The potential for natural hybridization between species in the genus has been discussed by Ladiges (1997), with respect to genetic isolation between subgenera, sections, and some series and by Potts and Wiltshire (1997) regarding the evolutionary significance and taxonomic, structural, and phys- iological barriers to crossing. Some sections show greater af- finity than others to crossing. For example, crosses between sections Bisectae and Adnataria have a high chance of success 1 Author for correspondence; e-mail margaret.sedgley@ adelaide.edu.au. Manuscript received February 2001; revised manuscript received June 2001. (Ellis et al. 1991); however, wide hybridizations between dis- tantly related sections (such as Bisectae and Maidenaria) are more likely to fail prefertilization. Hybrids may exhibit in- creased fitness (hybrid vigor) because of heterosis or decreased fitness because of the cross being too distant (Eldridge et al. 1993). This has implications for species improvement through crossing programs: potential crosses must be considered not only for the combination of desirable characters but also for taxonomic affinities and possible structural and physiological barriers to hybridization. Controlled pollination, or manipulated/artificial hybridiza- tion, is used for the generation of interspecific hybrids for plant improvement and is widely used in the development of euca- lypts for forestry and ornamental horticulture (Eldridge et al. 1993; Wirthensohn et al. 1999). The technique is also useful to test the probability of natural hybridization between species (Pryor 1951, 1956). Offspring from both natural and manip- ulated crosses may show morphology intermediate between the two parent taxa, as well as impaired reproductive capa- bilities and high levels of variability in the second generation offspring (Hopper 1995). The standard technique of controlled pollination for euca- lypts varies with researchers, but generally, the flowers are emasculated before anthesis, bagged to exclude external pol- len, pollinated manually some days later, and left to develop. Research has found that emasculation of flower buds can dis- courage pollinators such as bees (Hodgson 1976), bagging can have a detrimental effect on the survival of the pollinated flow- ers (Pryor 1951), and successful pollination can be achieved with neither bagging nor emasculation (Beardsell et al. 1979).