This chapter presents the results of the authors’ separate and conjoint studies of the palaeobotany, stratigraphy and palaeoecology of the Latrobe Valley Coal Measures, a thick sequence of coal seams and interseam elastics in the Gippsland region of southeastern Australia (Figure 14.1).

The Latrobe Valley Coal Measures are of Middle Eocene-Middle Miocene age and were deposited after the separation of Australia from Antarctica, a time of significant climatic change for the Australian continent and a period during which major global changes in sea levels and climates occurred. The Eocene-Miocene was an important time for the development of the modern Australian flora and the Latrobe Valley Coal Measures provide a window into this critical period. The study of the fossil assemblages of the coals has led to a detailed understanding of the vegetation that formed the coal measures and contributes significantly to our knowledge of the development of the modern Australian flora.

Previous workers have directed their efforts to particular aspects of the nature and relationships of the coal swamp vegetation. Early workers (Chapman, I925a,b; Deane, 1925; Cookson, 1946, 1947, 1950, 1953, 1957, 1959; Cookson & Duigan, 1950, 1951; Cookson & Pike, 1953a,b; 1954a,b; Pike, 1953; Patton, 1958) concentrated upon the taxonomic affinities of macroscopic and microscopic plant remains. Duigan (1966) studied selected micro- and macrofossil plant taxa from a palaeogeographical, ecological and evolutionary viewpoint. Baragwanath & Kiss (1964), Partridge (1971) and Stover & Partridge (1973) were more concerned with establishing stratigraphic relationships with the microfossils. The last of these studies, of pollen and marine microfossils, developed a long-accepted biostratigraphy for the basin.

Recently the biostratigraphy has been revised using more extensive sampling and better correlation with the geology of the basin (Haq et al., 1987; Holdgate, 1985, 1992; Holdgate & Sluiter, 1991). These studies also demonstrated the significance of eustatic sea level changes in influencing sedimentation within the basin and also that significant marine transgressions can be traced in both the lithofacies and the microfossil record.

The Quaternary is the period of modern life in which all the kinds of plants and animals still living have evolved, or have continued from the Tertiary unaffected by new environments. Technically, the beginning of the Quaternary has been defined as the last phase of the Matuyama reversed magnetism epoch after the Olduvai event, which finished at 1.62 million years (Ma). Others relate it to the first appearance of arctic marine faunas about 2.2 Ma in the mid-latitude North Atlantic (Bowen, 1978). The period therefore breaks with the definitions of older periods based on widespread evolutionary changes and instead uses climatic changes, such as the growth of icesheets and spread of cold surface water, to establish the chronology. Cores from the sea-floor show that parts of the tropical Pacific were little affected, but the lock up of snow in icecaps left the oceans enriched in the heavy isotope of oxygen (18O), so that a record of ice on the earth is well preserved in marine sediments. This record shows a series of cyclic changes on time scales of about 100 ka, with phases of maximum and minimum ice extent (stadial and interglacial periods, respectively), each occurring for 10% of each cycle, the remainder being cool interstadials (Chappell & Shackle - ton, 1986). Over 20 of these alternations have been identified, showing that for the last 2.3 Ma the earth has been very sensitive to minor changes in thermal equilibria, forced, in part at least, by variations in the season and amount of solar radiation reaching the earth (Chappell & Grindrod, 1983).

There are only two divisions in the Quaternary: the Pleistocene from 2.2 (or 1.62) Ma to only 10 ka ago, and the Holocene, our present interglacial, which is the last 10 000 years. The Pleistocene represents the period of establishment of our present landscapes, climatic patterns and types of variability, and the adaptation of the Tertiary biota to these new environments. The term Holocene means life as at present and this is generally true in relation to evolution - the time has been far too short for major new species to appear.

This chapter reconstructs Tertiary vegetation and phytogeography from the fossil spore and pollen record. Plants are constrained by their environment: they cannot grow outside of acceptable environmental limits, and climate is the most important of all environmental factors. For this reason, past distributions are considered in conjunction with the appropriate past climate and other environmental factors that may have been very different from those of today. Some wellestablished phytogeographical hypotheses are discussed in the light of the fossil record.

The palynological record is comprehensive and well suited to this purpose, but it has limitations. Not all pollen preserves as fossils. For example, the Lauraceae, an important Australian family and common in the macrofossil record, is absent from the fossil pollen record. When micro- (spores, pollen) and macrofossil (leaves, flowers, fruit) floras are compared, there is a core of common taxa, but some taxa are restricted to one or the other flora. For example, of the total flora identified in one deposit by Graham & Jarzen (1969), 27% were found as microfossils only and 17% as macrofossils only, with a core of 56% found as both. Furthermore, micro- and macrofossils from six Eocene lenses at Anglesea (Victoria) contain spore and pollen floras that vary somewhat, but are generally similar. However, macrofossil percentages are markedly different between lenses, and include taxa not identified as microfossils (Christophel et al.y 1987). The microfossil assemblage samples a broader area and gives a more general picture of the vegetation, whereas macrofossils present more localised and variable facets. Obviously, if both micro- and macrofossils are studied, a better picture of the vegetation emerges than from either one alone, but this has rarely occurred. It is not always possible to study both in the same deposit. Sediments deep underground are accessible by bores, which is a great advantage, especially in the Australian context where most of the landscape is so flat. Microfossils may be extracted from bore samples which are unsuited for macrofossil recovery.

During the last two decades, vertebrate palaeontological research in Australia has entered a new phase of development, with more investigators backed by a significant increase in financial support from government and private financial sources. The consequences of this accelerated phase of investigation has been rapid growth in information about vertebrate diversity, phylogenetic relationships, biocorrelation, palaeobiogeography and palaeoecology. In this review, we consider highlights of the developing late Mesozoic- late Cenozoic record of Australian terrestrial mammals, in part because the Cenozoic record of these is better known than that for any other group of vertebrates and in part because the ability to infer aspects of palaeohabitats from anatomical features is perhaps greatest for this group.

Most modern orders of mammals underwent adaptive radiations between the Late Cretaceous and late Paleogene subsequent to the Early to mid-Cretaceous diversification of angiosperms. For this reason many aspects of the history and structure of Australia's mammalian herbivores reflect the requirements of harvesting and consuming particular groups of flowering plants. In so far as this correlation holds, it is possible to infer from the structure of the dentition of extinct herbivores aspects of the vegetation upon which they fed. Although experimental studies (e.g. Sanson, 1989) of the function of the teeth of living Australian herbivores are few, deductive analysis of the diets of extinct forms based on diets of living species enables hypotheses about the timing of key mid-late Tertiary changes in the structure of Australia's terrestrial communities.

Higher-level systematic nomenclatures used here follow those of Aplin & Archer (1987; marsupials), Watts & Aslin (1981; rodents) and Walton & Richardson (1989; bats and other mammal groups). Biostratigraphic nomenclature, unless otherwise indicated, follows those of Woodburne et al. (1985) and Archer et al. (1989, 1991). The positions of the major fossil sites discussed in this chapter are shown in Figure 6.1 and the current understanding of the ages of the sites is shown in Figure 6.2.

AUSTRALIAN MAMMAL DIVERSITY

There are 12 groups of ordinally distinct endemic Australian mammals.

The Neogene was a time of transition both in the development of the present vegetation and the palynological study of it. The vegetation cover changed from one dominated by rainforest, which is traditionally regarded as ‘Tertiary’, to one in which rainforest became very reduced in extent. The nature of this change has been difficult to document due to an increasingly arid landscape with a concomitant reduction in suitable pollen preservation sites. The difficulty has been compounded by a relative lack of palynological study on the period. Stratigraphic palynologists have focussed on the earlier part of the Tertiary and there is no formal or well dated biostratigraphy, for much of the period under consideration, that is applicable to Australian terrestrial environments. Palynologists concerned with vegetation reconstruction have largely restricted their attention to the later part of the Quaternary period and have had variable success when venturing back into the Tertiary, as the vegetation then was frequently very different from that of today. Consequently, the database from which we piece together this critical period in Australia's vegetation history is very fragmentary and of varying quality.

In keeping with the problematic documentation of vegetation, there are difficulties in defining the period itself. There is general agreement on its beginning - the Miocene began about 25 million years (Ma) ago, although this does not necessarily hold any palynostratigraphic or biogeographical significance - but there are different views on the best location of the end of the period, i.e. the Pliocene/Pleistocene boundary. Conventionally this boundary is placed at the top of the Olduvai palaeomagnetic event dated to 1.6 Ma (Berggren et aL, 1985) but there is increasing pressure to reposition this close to the Gauss/Matuyama palaeomagnetic reversal boundary, around 2.4 Ma, as this reflects more closely the beginning of the substantial cooling and climatic fluctuations that characterise the Pleistocene period (Zagwin, 1985; Kukla, 1989).

PALAEOENVIRONMENTAL SETTING

Major factors influencing the whole of Australia during the Neogene include global climatic changes and the northward movement of the continent. The build up of ice on Antarctica, partly a result of the northward movement of Australia, which allowed the development of a circum-Antarctic ocean current, caused a steepening of the temperature gradient from equator to pole and development of the present atmospheric circulation pattern (Kemp, 1978).

The Tasmanian Cenozoic macrofossil record is relatively rich, and changes that have occurred in the vegetation of the region are becoming increasingly well understood. The record is essentially one of rainforest elements, especially in the Paleogene, but taxa that are now common in sclerophyllous heathlands and woodlands are increasingly prevalent in Quaternary sediments.

Extant Tasmanian rainforest is renowned for its beauty, and botanists have long recognised its marked taxonomic and structural similarity to other southern hemisphere ‘cool temperate’ forests of New Zealand and Chile. These are generally dominated by Nothofagus trees, their boughs laden with lichens and verdant shrouds of bryophytes. Other links are often made by phytogeographers to similar forests in high altitude regions of northern New South Wales and the much more species-rich vegetation of the generally montane regions of New Guinea and New Caledonia where Nothofagus also grows. A striking aspect of these forests is the presence of a variety of conifers, principally Podocarpaceae, but also Cupressaceae and Araucariaceae. In Tasmania the Araucariaceae are extinct, but the region is unique in the southern hemisphere in having a genus of Taxodiaceae, Athrotaxis. Athrotaxis spp. are often associated with Australia's only winter deciduous plant, Nothofagus gunnii, in montane regions of the island. The macrofossil record shows conclusively that the current diversity of Tasmania's woody rainforest flora is very much lower than at any other time during the Cenozoic. It confirms that there are strong floristic links to regions as widespread as eastern and southwestern mainland Australia, southern South America, New Zealand and New Guinea. In fact, Tasmanian Paleogene floras contain a wealth of taxa that are closely related to plants now confined to these regions.

Apart from the relatively large tracts of rainforest in Tasmania, closed forest lacking eucalypts is now confined to small patches along the east coast of Australia. In contrast to mainland Australia, Tasmania is relatively mountainous and has a well-developed woody alpine vegetation, dominated by shrubs of the Asteraceae, Epacridaceae, Myrtaceae and Proteaceae.

It is oft said that ‘Australia is an old continent’. Certainly Australia has many ancient landscapes which have been exposed to the processes of weathering and soil formation for very long periods under very stable tectonic conditions. The results of this are that many of our landscapes have well-leached and infertile soils, a major factor in the evolution of the Australian flora. Having said that, it is not true that all Australian landscapes are ancient. Many are very young, and of course all our landscapes are still undergoing modification, albeit some very slowly. Many landscapes that were thought to be comparatively young, such as the Southeastern Highlands (Andrews, 1911; Browne, 1969; Hill, 1975), have recently been shown to be comparatively ancient (Wellman, 1987; Bishop, 1988; Taylor et al., 1990a).

The Australian continent attained its present outline between 150 and 50 million years (Ma) ago (Wilford & Brown, Chapter 2, this volume), but many of our landscapes are even older. Comparison of the major landform regions (Figure 5.1) and the major geological structure of the crust or tectonic provinces (Figure 5.2) demonstrates this well.

The tectonic provinces of Australia can be divided into two along the Tasman Line (Figure 5.2; Veevers, 1984). West of this line the continent is dominated by Precambrian blocks and fold belts overlain by thin Phanerozoic basins, while to the east of it there are mainly Phanerozoic fold belts overlain by younger basins. Broadly these fundamental geological divisions correspond to landscape regions. The Precambrian blocks correspond to plateaux at elevations of up to about 500 m, the fold belts to upland areas up to 2000 m and the basins to lowland plains with elevations of generally less than 200-300 m (Figure 5.3). The distinction between these major landform regions is reflected in the present drainage networks. Most integrated drainage occurs along the coastal margins, particularly along the eastern Phanerozoic fold belt. The drainage systems in the western parts of the continent are, however, generally uncoordinated.

During the past decade, the three-element invasion theory that was initially advocated by Hooker (1860) to explain the present-day Australian flora has been questioned or dismissed (Barlow, 1981; Webb et al, 1986). Current concepts developed from ecological evidence indicate autochthonous differentiation from an ancient Gondwanan flora during the Late Cretaceous and early Tertiary (Webb et al, 1986). However, Truswell et al (1987) believed that the pollen record known to them favoured a Late Cretaceous- early Tertiary phase of floristic exchange between Australia and regions to the north, with dispersal occurring in both directions. Evidence that countered invasion from the north during the Late Cretaceous has since accrued (Dettmann & Thomson, 1987; Dettmann & Jarzen, 1988, 1990; Dettmann, 1989; Dettmann et al, 1990; Jarzen & Dettmann, 1990). From patterns of pollen introductions in separate regions of the southern Gondwana assembly, it was concluded that many elements of the Australian Cretaceous flora either evolved within the Austro-Antarctic region or entered Australia using an Antarctic route (Dettmann, 1989; Dettmann & Jarzen, 1990), as had been suggested previously (Dettmann, 1981).

The earliest angiosperms in Australia were almost certainly immigrants. The pollen record emphasises a 5-10 million years (Ma) time lag between initial inceptions of angiosperms in the southern Laurasian-northern Gondwanan region (Hauterivian or earlier) than introduction in Australia (late Barremian-Aptian). This evidence argues against Australia as a cradle region of the angiosperms (Takhtajan, 1969) and provides little support for inception and diversification of earliest angiosperms on fragments of the Australian plate that rafted northwards during the Late Jurassic (Takhtajan, 1987). Migration routes taken by the early angiosperms to Australia probably involved southern Gondwana (Dettmann 1981, 1989; Truswell et al, 1987); dispersal to Australia from Southeast Asia via microcontinents detached from northern Australia (Burger, 1981, 1990) has scant support from the pollen record.

The issues of introduction and Late Cretaceous differentiation of angiosperms in Australia are explored here using up-to-date pollen evidence. This evidence, combined with knowledge of palaeogeography and palaeotemperatures, has provided a basis for interpreting floristics and structure of Cretaceous plant communities represented in Australia.

In examining the fossil record of the Australian flora since the arrival of angiosperms, several taxonomic groups loom large, either because they have an extensive and informative fossil record, or because they are prominent in the living vegetation and are selectively sought in the fossil record. The aim here is to consider some taxa that cover each of these areas in order to complement the vegetation reconstructions discussed in earlier chapters. There are several candidates that fall into the category of having an extensive fossil record, but the outstanding one is Nothofagus, which dominates many palynofloras, and is also well represented in the macrofossil record. The Podocarpaceae, Araucariaceae, Proteaceae and Casuarinaceae are also considered here because they have a mixture of good pollen and macrofossil records, and important evolutionary arguments can be based on these records. The prime example of a taxon that is prominent in the living vegetation and is actively sought in the fossil record is Eucalyptus, and its pollen and macrofossil record will also be considered, although it is much smaller than those of the other taxa. In choosing these taxa, several other notable groups have been excluded. In most cases this is because the record is biassed to either pollen (e.g. Acacia, Chenopodiaceae) or macrofossils (e.g. Lauraceae, cycads) and a combined data set cannot be supplied. There is no doubt that the greatest weakness lies with the macrofossil record, especially for those taxa that produce entire-margined, medium-sized leaves. Taxonomic research on these groups should be a priority for the future.

NOTHOFAGUS

Nothofagus has been described as the key genus in the study of southern hemisphere plant evolution and biogeography (van Steenis, 1971,1972). There are a number of reasons for this.

1. 1. It has a completely southern hemispheric distribution, whereas its closest relatives are, and probably always have been, northern hemispheric.

2. 2. It occurs in all the major Gondwanic land masses except Antarctica, where it has an extensive fossil record, and Africa and India, where it has never been recorded as an autochthonous fossil.

Climate is the primary framework within which plant populations grow, reproduce and, in the longer time frame, evolve. However, reconstructions of palaeoclimate are often based on the marine record (e.g. Quilty, 1984 and Chapter 3, this volume) or on computer modelling (Sloan & Barron, 1990). The latter type, in particular, suffer from their dependence on simplified scenarios and must be tested against the terrestrial palaeontological record. Individual species and whole plant communities are morphologically and physiologically adapted to their physical environment, most strongly to climate, and so plant macrofossils are a proxy record of past climates (Wolfe, 1971, 1985; Upchurch & Wolfe, 1987). Several studies have indicated that plant macrofossils provide a potentially accurate record of Tertiary terrestrial palaeotemperatures (Wolfe, 1979, 1990; Read et al, 1990). Interpretation of the climatic signal preserved in plant fossil assemblages is dependent on understanding (1) how plants (and vegetation) interact with climate, (2) how plant fossil assemblages were formed, and (3) how these assemblages relate to the original vegetation.

The use of terms such as tropical, subtropical and temperate in palaeoclimatic discussions is potentially confusing, as these terms are rarely defined climatically and have geographical (latitudinal) connotations that are inapplicable for much of the Tertiary. Following Wolfe (1979, 1985; Upchurch & Wolfe, 1987), the terms megathermal, mesothermal and microthermal are used here to describe the temperature characteristics of the vegetation of the main climatic zones. The definitions of Nix (1982; Kershaw & Nix, 1988) are used here, although Wolfe's (1979, 1985) vegetation classification uses slightly different definitions.

This discussion is restricted to the Tertiary, since pre-Tertiary floras are dominated by plant groups with no or few modern analogues and so palaeobotanical indices of climate are (as yet) unreliable. Numerous accounts of the Tertiary climate history of Australia based on palaeobotanical evidence have been published (Kemp, 1978, 1981; Christophel, 1981, 1988; Nix, 1982; Truswell & Harris, 1982; Christophel & Greenwood, 1989), in some cases as part of global accounts (Axelrod, 1984; Wolfe, 1985). Many of these interpretations have been based solely on nearestliving- relative analogy (NLR) and provide qualitative reconstructions of climate.

Plant macrofossil deposits have been recorded from Australia for well over 100 years. Early reports were primarily from localities in Victoria and New South Wales; these are population centres and hence excavations likely to produce fossil material were commonest in those two states. Early work included reports by von Ettingshausen (1883, 1886, 1888), Deane (1895, 1900, 1902a,b,c, 1904) and von Mueller (1874, 1883), and was generally concerned with relating the fossils to extant plants on the basis of gross morphological features. Our current knowledge of continental drift throws doubt on the suggested affinities, found in the earlier papers, of the Australian fossils with northern hemisphere plants. Even those taxa with suggested Australian affinities unfortunately cannot be confirmed, as much of the material is either no longer available or is preserved in such a manner as to render it useless (Hill, 1988b). Therefore, these older reports and their deposits are not considered in this chapter. The reader interested in following them up can consult Hill (1988b) for general references or Christophel (1992) for specific references on Victoria. The locality that is an exception is Vegetable Creek, New South Wales. Hill (1988b) has reported that a number of the fossils from the original von Ettingshausen work on this deposit are still available and are excellently preserved. This flora, considered to be Late Eocene, is particularly important, as it contains the earliest macrofossils attributable to Nothofagus from continental Australia (Hill, 1988b).

Theoretically, the scope of this chapter is the entire early Tertiary, which includes the Paleocene, Eocene and Oligocene epochs. The reality of the situation, however, is that the chapter is basically an essay on the Eocene floras of Australia - primarily the Middle Eocene of southeastern Australia. This is due solely to the fact that most of the published accounts of early Tertiary macrofossil floras from continental Australia deal with Eocene localities. Indeed, only two Paleocene and two Oligocene floras (published or under current study) are known to the author, and none of these is extensive. It is the opinion of one biostratigrapher (B. McGowran, personal communication) that continental Australia has no Early Oligocene plant fossil assemblages, and hence this period, potentially of great evolutionary import after the global terminal Eocene climatic deterioration (McGowran, 1992), may never be documented.

The living Australian flora is a complex mixture of species with widely varying distributions and interactions, covering the range from arid zone grassland to rainforest, alpine heath to mangrove swamp. The latitudinal range of Australia spans tropical to cool temperate climatic zones, and this is reflected in the extant vegetation, which is enormously complex (see Groves, 1981). Attempts to explain the distribution of Australian vegetation based solely on prevailing variables have been less than satisfactory. Australia, in all its aspects, is a product of its past. That is especially true of its flora and, unless the fossil record is properly considered, all attempts to explain vegetation patterns will be incomplete. Despite the complexity of the living vegetation, there has been a tendency to consider the past vegetation, especially that of the pre-Quaternary, as consisting in widespread, monotonously uniform communities. The recent explosion in information on Australian fossil plants shows that this perception was largely the result of a highly incomplete database, where the unknown areas were assumed to be the same as the small areas that were relatively well understood. It is now clear that past vegetation complexity, at least during the Cenozoic, was as high as, or possibly higher than, that seen at present. The past complexity is abundantly illustrated in this book, which may well represent the last occasion on which a thorough review of such a large period of time can be accomplished for the whole of Australia. Data are accumulating at a rapid rate, and almost every new site produces much that is novel, causing a reassessment of the prevailing hypotheses.

There is a long history of attempts to explain the origin and evolution of the Australian flora; some of these are well known, others more obscure. Hooker's (1860) discussion of the Australian flora provided a base of the highest quality, which slowly evolved into the invasion theory, perhaps best argued by Burbidge (1960).