Degree Name

Doctor of Philosophy


Department of Biological Sciences - Faculty of Science


Increasing rates of interspecific hybridisation can be detrimental to the viability of small plant populations and increase the risk of local extinction. This may occur by means of demographic swamping if the production of unfit hybrids reduces seed set and recruitment of the pure-bred genotypes of the rare species. Alternatively, if hybrids are fertile, this may also occur by genetic swamping whereby introgression erodes the genetic integrity of the rare species resulting in its assimilation by more abundant congeners. The rates of hybrid production can vary widely between individuals and populations due to local differences in ecological and population parameters (e.g. phenology, isolation distances, plant-pollinator interactions, relative frequency of parentals). For small populations, alterations of these parameters may result in more frequent hybridisation with cross-compatible congeners, which may affect the conservation of rare species because the frequency of hybridisation is expected to directly influence time to local extinction.

Habitat fragmentation can play a critical role in altering a number of ecological and population parameters by reducing population size and plant densities, altering the frequencies of parentals, and modification of soil characteristics which can shift the flowering times of co-occurring species. All of these factors can potentially increase rates of interspecific pollen flow thereby increasing the rate of hybrid production within remnant populations. However, few studies have empirically explored the extrinsic ecological and population conditions influencing hybrid production in fragmented landscapes. Examining the processes promoting hybridisation in fragmented systems is critical for understanding the relationship between ecological/population parameters and the scale and rapidity of interspecific hybridisation. Such knowledge on the conditions promoting hybridisation and the subsequent effects on population fitness may be useful for understanding how anthropogenic activities can affect important evolutionary and ecological interactions and important for managing the demographic and genetic viability of remnant populations. I examined the influence of habitat fragmentation and associated ecological and population parameters on interspecific pollen dispersal, hybrid production and reproductive success of remnant populations of the uncommon tree Black Gum (Eucalyptus aggregata; Myrtaceae) in South-Eastern Australia. Black gum putatively hybridises with two more common species, Candlebark (Eucalyptus rubida) and Ribbon gum (Eucalyptus viminalis) that co-occur at most E. aggregata sites. The three species often exist in mosaic distributions with E. aggregata inhabiting poorly drained flats and frost hollows, E. rubida on surrounding rocky to well drained skeletal soils, and E. viminalis on moist soil along watercourses. The extent and size of most E. aggregata populations has been reduced due to land clearing for agriculture and grazing. They range from scattered trees in farmland to large undisturbed open woodland populations and exhibit a variety of population densities, absolute population sizes and relative frequencies of parental species at both local and landscape scales. This range provided a good opportunity to identify the ecological and population parameters influencing hybrid production at the individual, within population, and inter-population scales.

An array of evidence including, morphological intermediacy, genetic admixture identified by both allozyme and microsatellite markers and direct interspecific paternity assignments, unequivocally indicates E. aggregata is hybridising with E. rubida and E. viminalis. With allozymes I identified hybrid seedlings in 83 % of the 18 E. aggregata populations, indicating it readily hybridises with E. rubida and E. viminalis when they co-occur. Morphology correctly identified 71 % of hybrid seedlings due to expression of intermediate phenotypes between parentals, but became less accurate when dealing with backcrosses (50 % success). Genetic markers are clearly the preferred method of hybrid identification in this system as morphology was likely hampered by variable modes of genetic control of traits and by past introgression.

The frequency of hybridisation in this system was high and varied considerably between populations for both seed cohorts and established adult populations. Estimations of hybrid numbers in seed cohorts using both allozyme and microsatellite markers revealed average hybrid seed production for populations of between zero to 36.0 %. The maximum levels of hybridisation are substantially higher than those reported by most other Eucalyptus studies. Within adult populations, the proportion of adult hybrids (identified with microsatellite markers) was substantial and ranged from 10.9 % in a large population (Bendoura), to 28.8 % and 24.0 % of the trees in the two remnant populations of Duck Flat and Norongo respectively. These hybrid trees were fecund and produced viable offspring. The flowering time of E. rubida, E. viminalis and hybrid adults significantly overlapped with E. aggregata suggesting that ample opportunities exist for the formation of F1 and backcross hybrids. Indeed allozyme, morphological and direct paternity analysis indicated that F1 (first generation) and backcrossed hybrids are produced by pure-bred E. aggregata.

At the population scale, the frequency of E. aggregata in relation to its congeners (relative population size) was negatively correlated with hybrid production (R2 = 0.59, P < 0.01) across 17 populations. When congeners were more abundant, seed from E. aggregata adults consisted of up to 31 % hybrids. This relationship was even stronger for E. aggregata populations only sympatric with E. viminalis (R2 = 0.88, P < 0.01), but there was no significant relationship when considering only populations co-occurring with E. rubida. Hybridisation had a substantial impact on the genetic diversity of seed cohorts. Both genetic diversity and allelic richness increased with hybridisation rate in seed cohorts (R2 = 0.42, P < 0.05; R2 = 0.29, P < 0.05 respectively) and with decreasing relative population size (R2 = 0.52, P < 0.01; R2 = 0.42, P < 0.01). In populations out-numbered by congeners, this suggests genetic diversity may be maintained by frequent opportunities for hybridisation and introgression.

At the individual scale, the frequency of hybrid production varied substantially between E. aggregata trees (zero to 77 % hybrids). This variation in hybrid production was strongly linked to local ecological and population variables in a remnant population, but was not found to be driving hybrid production in a large undisturbed population. For the remnant population (Duck Flat), local variables had good explanatory power (over 50 % of variation explained). Individual E. aggregata that were more isolated, were relatively out-numbered by E. rubida (within 180 m), and greater flowering synchrony with E. rubida tended to produce more hybrid seed (up to 48 %). With the large population (Bendoura), hybrid production ranged from zero to 77 % but there were no significant relationships with the array of local parameters assessed. These results suggest that the factors influencing the pattern of interspecific gene flow may differ in dispersed remnant populations compared to landscapes where there is a continuous distribution of plants. This could be due to the receipt of high levels of immigrant interspecific pollen by individuals within Bendoura compared to Duck Flat that would override the influence of local E. rubida pollen sources. Alternatively, hybrid production may be under genetic control and individuals may express variation in the strength of pre-zygotic interspecific isolation mechanisms.

Long distance interspecific pollen flow between remnants was an important process promoting hybridisation, especially in the smaller remnant populations. Using direct paternity and indirect pollen dispersal analyses I examined intra- and interspecific pollen flow amongst trees, and immigration rates within a large undisturbed site (Bendoura), a remnant (Duck Flat) and a small roadside site (Norongo). All populations had substantial levels of immigrant pollen, and this was greatest in the smallest population (Bendoura 23 %, Duck Flat 25 %, Norongo 44 %). Most intraspecific pollen originated from known near neighbours within the sites (Bendoura 83.1 % fathers known; Duck Flat 84 %, Norongo 88 %), with most of the potential E. aggregata fathers contributing pollen (Bendoura 64 %, Duck Flat 89 %, Norongo 40 %). In contrast, interspecific pollen originated from only a select few E. rubida and hybrid trees from the total potentially available within populations (Bendoura 6 %, Duck Flat 20 %, Norongo 0 %). The remaining hybrids were sourced from E. rubida paternal parents in populations up to 3 km away, with a substantial increase in interspecific immigration rates as population size declined (Bendoura 9.6 %, Duck Flat 11.5 %, Norongo 37.3 %). High rates of pollen immigration indicate that populations are well connected via pollen mediated gene flow across fragmented landscapes. Increasing rates of interspecific immigrant pollen movement into smaller remnants, suggests pollen swamping by congeners is occurring at a landscape scale across several kilometres.

I studied seed production, germination, and the survival and performance of seedlings across 17 populations in relation to population parameters (absolute and relative frequency of species), hybridisation rates, genetic diversity and levels of inbreeding. Of these parameters, the relative population size of remnants had the most consistent influence on fitness parameters, with significant positive relationships with seed production (R2 = 0.29, P < 0.05), germination (R2 = 0.27, P < 0.05) and seedling survival (R2 = 0.33, P < 0.05). Populations numerically out-numbered by congeners produced fewer seed (lowest mean 0.8/capsule) compared with populations where E. aggregata was in higher frequencies than congeners (mean up to 2.5/capsule). Germination and seedling survival was lower in small (42 % and 51 % respectively) compared to large E. aggregata populations (77 % and 85 % respectively). These patterns are likely due to increased interspecific pollen flow in populations when E. aggregata is numerically out-numbered and the action of pre- and post-zygotic barriers removing inviable hybrid genotypes. This suggests that E. aggregata remnants of small relative population size have lower seedling recruitment, and may be under increased risk of local extinction through demographic swamping.

There was no evidence of a reduction in plant height, leaf pair numbers or survivorship of F1 hybrids and backcross hybrid seedlings compared to pure-bred seedlings at least until eleven months of age. While seedling performance was assessed under benign glasshouse conditions, many hybrid genotypes must be viable under natural conditions because F1 and backcrossed hybrid adults were present at most populations and consisted of up to 28.8 % of adults at one population (Duck Flat). For pure-bred E. aggregata, I also found evidence of significant declines in seedling height with declining outcrossing rates, suggesting inbreeding depression may reduce the competitive ability of pure-bred offspring relative to hybrids.

This study has demonstrated that E. aggregata trees within small remnants are exposed to increased interspecific gene flow from E. rubida and E. viminalis compared with trees from larger undisturbed populations. This results in increased hybrid production, introgression, and reduced seed production, germination and survivorship of seed cohorts. Reductions in the reproductive success of pure-bred E. aggregata and subsequent introgression by established hybrids pose a real threat to the re-establishment, genetic integrity and long term persistence of remnant populations of E. aggregata. Management of remnant E. aggregata populations in regards to hybridisation will rest on assessing population and ecological parameters most strongly related to hybrid production. Evidence at local, population and landscape scales suggest that the frequency of E. aggregata relative to congeners is the most important factor influencing hybrid production. Local ecological and population variables (i.e. tree distributions, flowering) within populations may only become important for small and isolated populations. The potential adaptive benefits of hybridisation should also not be ignored, however, as genetic diversity was greater in seed cohorts from remnant populations dominated by congeners. This influx may provide new multi-locus genotypes that may be required for remnant populations to re-establish and adapt to rapidly changing fragmented landscapes and climate change.

02Whole.pdf (5754 kB)