Early Pleistocene vegetation change in upland south-eastern Australia more

Journal of Biogeography (J. Biogeogr.) (2011) 38, 1456–1470 ORIGINAL ARTICLE Early Pleistocene vegetation change in upland south-eastern Australia J. M. Kale Sniderman* School of Geography and Environmental Science, Monash University, Clayton, Vic. 3800, Australia ABSTRACT Aim This study aims to improve our understanding of the late Cenozoic history of Australian rain forest and sclerophyll biomes by presenting a detailed pollen record demonstrating the floristic composition and orbital-scale patterns of change in forest communities of upland south-eastern Australia, during the Early Pleistocene. The record is examined in order to shed light on the nature of the transition from rain forest-dominated ‘Tertiary’ Australian vegetation to opencanopied ‘Quaternary’ vegetation. Location Stony Creek Basin (144.13° E, 37.35° S, 550 m a.s.l), a small, infilled palaeolake in the western uplands of Victoria, Australia. Methods A c. 40-m-long sediment core was recovered from the infilled palaeolake. Palynology was used to produce a record of changing vegetation through time. Multivariate analyses provided a basis for interpreting the composition of rain forest and sclerophyll forest communities and for identifying changes in these communities over successive insolation cycles. Results Early Pleistocene upland south-eastern Australian vegetation was characterized by orbital-scale, cyclic alternation between rain forest and sclerophyll forests. Individual intervals of forest development underwent patterns of sequential taxon expansion that recurred in successive vegetation cycles. Diverse rain forests included a number of angiosperm and gymnosperm taxa now extinct regionally to globally. Sclerophyll forests were also diverse, and occurred under warm and wet climate conditions. Main conclusions The Stony Creek Basin record demonstrates that as recently as c. 1.5 Ma diverse rain forests persisted in southern Australia beyond the modern continental range of rain forest. The importance of conifers in these rain forests emphasizes that they have no modern Australian analogue. Alternation in dominance between these forests and diverse, sclerophyllous open canopied forests was apparently driven by changes in seasonality, and may have been promoted by fire. Keywords Australia, insolation, palaeoecology, Pleistocene, pollen, rain forest, vegetation history. *Correspondence: J. M. Kale Sniderman, School of Geography and Environmental Science, Monash University, Clayton, Vic. 3800, Australia. E-mail: kale.sniderman@monash.edu INTRODUCTION In Australia, the timing and nature of the contraction of mesic, ‘Tertiary’ vegetation, and its replacement by more cold-, aridity- and fire-adapted ‘Quaternary’ vegetation, is poorly understood. This transition involved the loss of formerly widespread rain forests, and the diversification and expansion 1456 http://wileyonlinelibrary.com/journal/jbi doi:10.1111/j.1365-2699.2011.02518.x of a range of open-canopied vegetation types. However, because few fossil records provide a detailed picture of this critical interval (Kershaw et al., 1994; Macphail, 1997), it has been difficult to understand how and when Australian rain forest, even in the most favourable climates, became largely restricted to relict stands (Webb & Tracey, 1994) surrounded by fire-dependent and fire-promoting vegetation, mostly ª 2011 Blackwell Publishing Ltd Early Pleistocene vegetation in south-eastern Australia dominated by the sclerophyllous genus Eucalyptus (Myrtaceae) (Ashton & Attiwill, 1994). The processes through which these juxtaposed fire-dependent and fire-sensitive communities are maintained over ecological time-scales (Jackson, 1968), and the antiquity and environmental drivers of these relationships over evolutionary time-scales (Kershaw et al., 1994; Bowman, 2000) have long been the subjects of intense interest. However, our understanding of the interactions between fire, vegetation and climate remains limited by the lack of long records of vegetation change. One promising source of information about the late Cenozoic history of rain forest and of rain forest–sclerophyll forest dynamics is Stony Creek Basin (SCB), an Early Pleistocene palaeolake in upland south-eastern Australia. Sniderman et al. (2007) used pollen analysis to show that vegetation dominance at SCB alternated between rain forest and sclerophyllous forest communities over orbital time-scales for c. 280 kyr of the earliest Pleistocene. However, those authors presented only a summary of major pollen taxa. Here, the floristic diversity of the vegetation is analysed through provision of the full pollen record for SCB. This provides a basis for understanding historical relationships of components of the present-day flora and vegetation, and their conservation status. The early Pleistocene Stony Creek Basin record Stony Creek Basin (144.13° E, 37.35° S, 550 m a.s.l.) is a small (c. 10 ha), infilled palaeolake, probably of volcanic origin (Graham et al., 2003), in upland south-eastern Australia (see Fig. S1 in Appendix S1 in the Supporting Information). SCB lies in Victoria’s western uplands (Joyce, 1992), a band of dissected ranges and plateaux forming part of the continental divide, ranging from 300 to 1000 m a.s.l. The region experiences cool, wet winters, and warm, dry summers, governed by the seasonal migration of the Southern Hemisphere subtropical anticyclone (Sniderman et al., 2009). Natural vegetation, where not cleared for agriculture, consists primarily of open forests dominated by Eucalyptus, with grassy, shrubby or heathy understories. Limited areas with annual rainfall > 1000 mm support more productive, tall wet Eucalyptus forests, but rain forest is absent from the western uplands, which lie beyond the continental range of rain forest in Australia. Previous work on the SCB record includes leaf fossils recovered by McCoy (1876) and Patton (1928). Cookson (1953) and Cookson & Pike (1954) illustrated several fossil pollen types from the site, for which Cookson (1953) inferred a Pliocene age. In 2000, a c. 40-m-long sediment core was recovered from the richly organic silty clay sediment (Sniderman, 2007). Sniderman et al. (2007) presented a summary record of major pollen types, for which they developed a floating chronology based on analysis of sedimentary laminations, then preliminarily anchored this chronology within the geological time-scale using zircon fission track ages and a palaeomagnetic record. They showed that diverse rain forest Journal of Biogeography 38, 1456–1470 ª 2011 Blackwell Publishing Ltd and sclerophyll plant communities alternated at SCB with a c. 23 kyr rhythm over c. 280 kyr during the Early Pleistocene, and argued that this implies SCB vegetation change was driven primarily by precession-dominated summer insolation. They then inserted the record into the astronomically calibrated geological time-scale, by pattern-matching the amplitudemodulated rain forest angiosperm pollen record to amplitudemodulated summer insolation, placing the record between c. 1.835 and 1.555 Ma (Fig. 1). From samples of the 40-m sediment core and from sediments recovered during excavation of the site in 2002, Core Depth polarity (m) ai n Po fore do st ca an rp gi ac os ea pe rm e R Insolation 21 Dec W/m2 500 550 s 450 Age Ma 151 • • • • 153 155 157 159 5 1.60 10 Matuyama • • • • 161 163 1.65 15 165 167 1.70 20 • 169 171 173 25 1.75 Olduvai 30 35 • • • • • • • • • 175 177 0 0.05 1.80 0 % 20 40 Eccentricity Figure 1 Summary of Stony Creek Basin rain forest taxa, plotted on the time-scale of Sniderman et al. (2007), showing pollen percentage values of the rain forest angiosperm sum and the rain forest gymnosperm sum. Correlation of rain forest-rich intervals with Southern Hemisphere summer insolation maxima at 38° S (insolation cycles of Lourens et al., 1996) is shown with grey bars. 1457 J. M. K. Sniderman Jordan et al. (2007) described a diverse fossil leaf assemblage of Styphelioideae (Ericaceae), suggesting that sclerophyll floras at SCB were much more diverse than indicated by pollen. Sniderman et al. (2009) used beetle fossils recovered from sediments of rain forest- and sclerophyll-rich pollen zones to generate quantitative palaeoclimate reconstructions for the site. Their reconstructions indicated that annual and summer precipitation were much higher than today, during both rain forest- and sclerophyll-rich intervals, and that temperatures were consistently c. 2 °C warmer than present. This analysis failed to detect a climatic difference between the strikingly different rain forest- and sclerophyll-rich vegetation intervals. Sniderman et al. (2009) offered several possible explanations for this failure, noting that actual temperature differences between rain forest- and sclerophyll-rich intervals may have been, within error, up to 2 °C. For their precipitation estimates, they suggested that relatively subtle differences in precipitation seasonality may have driven changes in fire regime which determined the relative abundance of firesensitive rain forest and fire-tolerant sclerophyll forests. This is consistent with modern Australian rain forest occurrences, which even in the wettest climates are typically juxtaposed against fire-dependent sclerophyll forest communities (Bowman, 2000). Hence the inability of Sniderman et al. (2009) to discern a climatic difference between sclerophyll- and rain forest-rich vegetation intervals may indicate that the importance of fire in dictating vegetation boundaries has a long history in Australia. In summary, the SCB record indicates that climates with high summer rainfall, and associated diverse rain forests now absent in the region, persisted in south-eastern Australia until at least 1.55 Ma, approximately one million years into the Early Pleistocene (2.588–0.781 Ma). MATERIALS AND METHODS In February 2000, a c. 40-m-long sediment core was recovered from near the centre of SCB using truck-mounted, hollowauger equipment. Sixty-eight individual cores varying in length from c. 30 to 85 cm were recovered and extruded on site, with core recovery of c. 96% (Fig. S2). In the laboratory, 3986 cm of the mostly uniformly black, silty clay sediment core was sampled for moisture, organic content and pollen analysis at approximately 20-cm intervals. All samples responded negatively for carbonates with 10% HCl. Organic content of samples was determined by the loss-on-ignition (LOI) method, in which weighed, air-dried samples were combusted at 500 °C for two hours then reweighed, the loss expressed as a proportion of oven-dried weight. Samples for pollen analysis were spiked with exotic Lycopodium spores in tablet form, dispersed in warm 10% Na4P2O7, and then sieved through 210 lm and 7 lm mesh. Samples were then treated with 10% KOH followed by acetolysis (Moore et al., 1991) and heavy liquid flotation with Na6(H2W12O40) made up to specific gravity 2.0 (Munsterman & Kerstholt, 1996). Resulting organic residues were dehydrated in ethanol, suspended in glycerol, and mounted on glass slides. 1458 Pollen grains, pteridophyte and selected other spores, microcharcoal and Lycopodium marker spores were counted along transects at magnifications of 400· and 1000·, on a Zeiss Axiolab compound microscope fitted with E-PL 10 · /20 objectives. Dryland pollen counts were calculated as percentages of a pollen sum (average sum 446 grains, minimum of 350) excluding aquatic and unknown types, and all spores. Proportions of pteridophytes and other spores, and of indeterminate pollen types, were calculated as percentages of pollen sums that included these types, but aquatics were excluded from all sums. Pollen and spore, and microcharcoal concentrations were calculated by reference to associated exotic Lycopodium counts. Two forms of pollen deterioration were observed: degradation, in which sculptural and structural exine features uniformly thin or fuse (Cushing, 1967); and corrosion, characterized by localized etching of the ektexine (Wilmshurst & McGlone, 2005). Deteriorated grains were separately tabulated for Eucalyptus, Myrtaceae and Casuarinaceae. Pollen clumps – tetrads of normally monad types, as well as larger clumps (some with > 30 grains) – were frequently encountered. These presumably result from short distance dispersal of anthers, floral tissue or faecal pellets, and were counted as one dispersal unit for the purposes of the pollen sum. Pollen identifications were made by comparison with the modern reference collection held at the School of Geography and Environmental Science, Monash University, and with regional pollen floras from Indo-Australasia and other parts of the Southern Hemisphere (Heusser, 1971; Moar, 1993; Wang et al., 1995). Because many types in the SCB pollen flora could not be matched with pollen of extant Australian taxa, fossil specimens were also compared with regional Cenozoic fossil pollen studies. Description and illustration of SCB pollen and spores will be presented elsewhere. Microcharcoal particles, defined as angular, opaque objects > 10 lm along their longest axis, were counted on all pollen slides, and expressed as particles cm)3. Pollen percentage diagrams were produced using Psimpoll v. 4.25 (Bennett, 2005). In order to facilitate their interpretation, diagrams were divided into local pollen zones representing episodes of relative palynological homogeneity. Zonation was carried out using taxa that achieved ‡ 2% of the dryland sum in at least one sample, after square-root transformation. Zonation used optimal splitting based on information content (Birks & Gordon, 1985) to divide the diagram into the maximum number of statistically significant zones (Bennett, 1996). To investigate relationships between fire and vegetation change at SCB, charcoal and pollen values were smoothed, the uppermost eight samples omitted because of strong taphonomic biases in these samples, and then detrended to eliminate spurious correlations resulting from secular trends. Rate of change analysis (Bennett & Humphry, 1995) was carried out in Psimpoll v. 4.25, by calculating differences between successive pollen spectra divided by the time interval between consecutive samples, using an information statistic and the chronology of Sniderman et al. (2007). Journal of Biogeography 38, 1456–1470 ª 2011 Blackwell Publishing Ltd Early Pleistocene vegetation in south-eastern Australia The pollen data were analysed using the multivariate analysis techniques non-metric multi-dimensional scaling (NMDS) (Clarke, 1993) and cluster analysis, implemented in Primer v. 5 (Clarke & Gorley, 2001). These techniques provide an objective basis for interpreting the pollen diagram by revealing recurrent pollen spectra independent of the stratigraphically constrained zonation. NMDS was used to reduce the multivariate pollen data to two dimensions, and performed, along with cluster analysis using group average linking, on a sample dissimilarity matrix based on the Bray– Curtis (B–C) coefficient (Clarke & Warwick, 2001), using taxa that achieved ‡ 2% in at least one sample, after square-root transformation. RESULTS Two hundred and eight samples, extending from 9 to 3951 cm, were pollen-rich (sandy silts below 3951 cm contained very low pollen concentrations). Pollen/spore preservation was generally excellent, and concentration usually > 100,000 grains cm)3. Pollen types were assigned to rain forest or sclerophyll forest types by reference to the ecological affinities of the majority of modern relatives of the pollen types. The pollen diagram was divided into 22 statistically significant local pollen zones. Pollen percentage values of major pollen types are presented in Fig. 2 (see Fig. S3 for full details), arranged in the sequence: rain forest gymnosperms (Podocarpaceae and Araucariaceae); rain forest angiosperms; ecologically variable mesophytic Winteraceae and Pomaderris; sclerophyll taxa including Eucalyptus and select other Myrtaceae types, Callitris (Cupressaceae), Casuarinaceae (with values of corroded grains shown superimposed); open forest woody understorey taxa followed by predominantly herbaceous taxa; pteridophyte spores; and microscopic charcoal counts, detrended to remove a trend associated with upward-decreasing sedimentation rate (cf. the raw charcoal curve, Fig. S3f). For sediment description, see Fig. S2. Below, features of the pollen diagram are briefly described in terms of stratigraphic patterns including those emphasized by the zonation. However, because the record illustrates repeated intervals of forest development, and because the pollen spectra are difficult to relate satisfactorily to modern Australian vegetation analogues, multivariate analyses are used as a basis for quantitative comparison and grouping of pollen spectra, independent of their stratigraphic zone affinities. This approach provides an objective means of comparing successive intervals of forest development in long continental records. Similar approaches to the identification of recurrent pollen spectra, and of differences between phases of forest development, have been explored by Van’t Veer & Hooghiemstra (2000) and Tzedakis & Bennett (1995), respectively. Alternation between rain forest and sclerophyll forest To first order, the 22 local pollen zones define rhythmically alternating rain forest-rich and sclerophyll forest-rich zones, so Journal of Biogeography 38, 1456–1470 ª 2011 Blackwell Publishing Ltd that the overall impression from the diagram is of a bimodal vegetation ‘see-saw’ (Figs 1 & 2). This alternation implies dramatic swings in the dominance of forest within the pollen catchment of SCB. However, other features indicate that both biomes maintained their presence within the pollen source area throughout the c. 280 kyr record. For example, several rain forest pollen types persist at trace levels through many sclerophyll-rich zones (Fig. 2a); the sclerophyll dominants Eucalyptus, Casuarinaceae and Callitris maintain pollen values ‡ 1.5% in every sample (with the exception of samples in oxidized zone S-1 and S-2); and pollen of the majority of sclerophyll understorey taxa (Fig. 2b) is consistently present at low values, hardly varying with the dominant pollen types. In addition, local persistence through vegetation cycles is also evidenced by the distribution of pollen clumps (Fig. S4). These indicate that Eucalyptus was locally present even at times when it achieved quite low pollen percentage values, and that types which are rarely recorded above trace values, such as Agathis and Araucaria, were nevertheless locally present. A number of pollen types and environmental indices exhibit secular trends. Some of these, such as upward increases in the values of pollen of aquatic plants (Fig. S3e), deteriorated pollen, and charcoal and pollen concentrations (Figs S3c & S3f), can plausibly be attributed to changes in sedimentation regime with progressive infilling of the palaeolake. However, dryland pollen types exhibit a combination of upwardincreasing (Ilex, Cunoniaceae, Macaranga, Winteraceae, Corymbia, Callitris, Poaceae) and upward-decreasing (Microcachrys, Dacrycarpus and Araucariaceae) trends, more likely to reflect real changes in regional to local vegetation, and by implication, secular climate changes over the duration of the record. A striking example is provided by Callitris, which exhibits very high amplitude, short-lived peaks in the upper c. 11 m of the record, that seem to indicate very rapid changes in population size in the source vegetation. Identification of recurrent assemblages Combined ordination and cluster analyses provide a means of objectively evaluating the similarity of the pollen spectra, independently from the stratigraphically constrained zonation. An NMDS ordination of the 208 pollen samples is presented in Fig. 3a. Bubble plots (Fig. 3b–f) illustrate the contributions of different taxa to the construction of ordination space, and show the dominance of spectra variously by Callitris, Podocarpaceae, rain forest angiosperms, Casuarinaceae and Eucalyptus, but they also reveal differences in the behaviour of these taxa. For example, rain forest angiosperms have a small number of large bubbles concentrated within the left side of ordination space, but only very small bubbles elsewhere, indicating a strong separation between spectra with very high and low rain forest angiosperm values. By comparison, the bubbles for Eucalyptus, Casuarinaceae and Podocarpaceae have more consistent or more gradually changing diameters, indicating more consistent representation. The configuration of sclerophyll-rich spectra is tighter than that for rain forest-rich spectra, which primarily 1459 J. M. K. Sniderman (a) Rain forest gymnosperms Rain forest angiosperms Age Ma Depth cm D ac ry di um Po do m sp ic o ro rit sa es cc c at f. D us ac ry ca M rp ic ro us Ag ca at chr Ar his ys au cf car . W ia N olle ot h mi N ofa a ot gu h s Ile ofa (B x gu ra s s (L so op sp C un ho ora on zo ) ia ni ce a) ae C un (3 C on ) Sy ia m ce p R loc ae ap o ( 2 C s a ) M ne yr a s El ine ae o Q ca ui r p n u Ac tinia s m e Sy na zy R giu ho m d Ar am al ni i a C ace /R up a ho a e do M nia m ac e/ yr tu a c s M ran f. A or ga lec ac /M try cf ea . L e all on o o / cf ma Urti tus . O ti ca a ce Be rite ae au s Pi pre cr a o Vi de ta nd ce ra C ae cea ha e Ph yl lo Age Po d 1.60 500 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 10 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • 1.65 1000 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • C alabr ian • • • • • • • • • • • • • • • • • • • • • • • • 1500 1.70 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 0 0 • • • • • • • • • 0 0 −0.4 0 0.4 • • • • • • • 2000 • 1.75 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 2500 • • • • • • • • • • • • • • • • • • 3000 1.80 • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • Gelasian 3500 • • • • • • • • • • • • • 0 • • 0 % 10 20 30 0 10 20 0 • • • 10 0 0 • • 0 0 0 0 0 0 0 0 0 0 0 0 0 bi a/ An pt o go Ac spe ph ac rm o ia um ra Ba nk si Pr a ot <3 e G ac 0 µ re ea m v e H illea un ak e / dif St a ( Hak . irl gr in a ea H gia nul al or / C ate C ag on ) op od o r e s Er osm nd per r m ic a a on ea Po cea e ac e ea e Le C Age Depth cm Ma Age Eu or C C C on oc ar pu e t s he go ubu lif no lo Ap p ra od ia e ce /A C ya ae ma ra th nt ea .> 10 D ick P s O on sm ia Lo und ph ac G oso eae le ic ria he ni Pt a er id iu m ea ro de e) d) (b) ac or ss e (c re ea up ac us (C rin rin pt Open forest (sclerophyll dominants and understorey taxa) ea e su m e ac ea ris /G ra c er or ag ra c ad ris ly ua ua ca te Po m ym lit te as as al al As W Pl H • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • • • 1.60 500 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 0 10 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 C an in ta ea is • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 1.65 1000 • C al ab r i an • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 1500 1.70 • • 2000 • • • • • • • • • 2500 • • • 3000 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • 1.75 • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • • • • • • • • • • • • 0 • • • • • • • • 1.80 • • • • • • • • • 0 Gelasian • 3500 • • • • • • • • • 0 % 10 20 30 40 0 10 20 30 40 50 0 10 20 30 40 50 60 0 • • • • • 0 0 • • • • 0 0 0 0 Figure 2 Stony Creek Basin pollen diagram, plotted by depth. The time-scale of Sniderman et al. (2007) is also shown along the left vertical axis. Taxon values are pollen percentages relative to a dryland sum: (a) rain forest gymnosperms and rain forest angiosperms, and detrended charcoal counts, (b) sclerophyll and open-forest understorey taxa, and pteridophyte spores. 1460 Journal of Biogeography 38, 1456–1470 ª 2011 Blackwell Publishing Ltd rc o (d al c et on re c nd en ed tra ) tio n Pteridophytes • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 S−20 S−1 S−2 S−3 S−4 S−5 S−6 S−7 S−8 S−9 S−10 S−11 S−12 S−13 S−14 S−15 S−16 S−17 S−18 S−19 S−21 S−22 s pu oc ar cl a du s Pollen zone S−1 S−2 S−3 S−4 S−5 S−6 S−7 S−8 S−9 S−10 S−11 S−12 S−13 S−14 S−15 S−16 S−17 S−18 S−19 S−20 S−21 S−22 Early Pleistocene vegetation in south-eastern Australia NMDS, all samples (a) Podocarpaceae Stress: 0.16 (b) Rainforest angiosperms Figure 3 (a) Non-metric multidimensional scaling (NMDS) plot of Stony Creek Basin pollen samples. One hundred random restarts were used to gain confidence that no lower stress solution could be found. The stress value, 0.16, represents the degree of scatter about the line of a nonparametric regression of interpoint distances on their corresponding Bray–Curtis dissimilarities. Clarke & Warwick (2001) consider that ordinations with stress values between 0.1 and 0.2 are useful, as long as care is taken not to over-interpret details of the plot. (b–f) Bubble plots of major pollen groups, illustrating the distribution of major pollen types within ordination space by using circle diameter to represent each sample’s rank value for a pollen type. For example, in Fig. 3f the 367-cm sample, having the highest Callitris values in the dataset, has the largest bubble, and serves as the scalar for all the other Callitris bubbles. (b) Podocarpaceae, (c) rain forest angiosperms, (d) Casuarinaceae (yellow), with Casuarinaceae (corroded grains) superimposed (beige), (e) Eucalyptus, (f) Callitris, 367-cm sample with the highest Callitris values labelled. Note that the position of the zones S-1 and S-2 spectra reflects their very high Casuarinaceae values, but also their high values for corroded Casuarinaceae and very low values for Eucalyptus, in these oxidized, near-surface samples. (c) Casuarinaceae (d) Eucalyptus Casuarinaceae (corroded) (e) Callitris 367 (f) reflects the lower number of pollen types contributing to the analysis of the sclerophyll-rich zones. In order to compare the distributions of pollen spectra within ordination space to the stratigraphic patterns captured by the zonation, in Fig. 4a first- and second-order groupings produced by cluster analysis are superimposed on the sample ordination. The first hierarchical level (at B–C dissimilarity > 50), separates the majority of the pollen spectra (group 1a) from the oxidized, near-surface spectra of zones S-1 and S-2 (group 1b). At the second hierarchical level (at B–C dissimilarity > 40), clustering differentiates zones S-1 (group 2d) from S-2 (group 2c), which are rain forest- and sclerophylldominated, respectively, and separates the single sample at 367 cm (group 2a), strongly dominated by Callitris (62.4% of the dryland sum), from all remaining samples. At lower dissimilarity (Fig. 4b), clustering identifies two groups dominated by sclerophyll (group 3a) and rain forest (group 3b) pollen, which are closely juxtaposed in ordination space. The palynological composition of these groups is illustrated in Fig. 5, in which percentage values of selected pollen types have been plotted on a y-axis defined by the terminals of a cluster dendrogram, rather than by depth. This shows that the first order separation of the samples of zones S-1 and S-2 (group 1b, Fig. 4a) is clearly reflected by their very high Casuarinaceae values (including corroded grains). The separation of groups 3a and 3b (Fig. 4b) for the most part distinguishes samples that have high values of Eucalyptus and low values of rain forest, particularly rain forest angiosperms (group 3a), from those with low values of Eucalyptus and high values of rain forest (group 3b). Figure 4d, which illustrates a separate NMDS analysis of the mean pollen percentage composition of the 22 pollen zones, demonstrates that this corresponds well with the stratigraphically-constrained zonation. Journal of Biogeography 38, 1456–1470 ª 2011 Blackwell Publishing Ltd 1461 J. M. K. Sniderman (a) 2b 2a 367 (b) 1a 1b S-2 50 40 2d S-1 Stress: 0.16 2c Clusters 1b and 2a 3a 3b Dissimilarity >30 Dissimilarity (c) (d) 3b 10 4 6 15 13 8 17 21 3a 5 3 18 14 19 16 20 12 11 22 9 7 Clusters 4a 4b 4d 4c Dissimilarity 45 35 25 1 Stress: 0.06 2 Figure 4 (a–c) Non-metric multi-dimensional scaling (NMDS) plot of Fig. 3a, with groups defined by group-average cluster analysis superimposed. (a) Groups defined at the first (1a and 1b, ellipses drawn in solid lines) and second (2a–2d, ellipses drawn in dashed lines) hierarchical levels, corresponding to Bray–Curtis dissimilarities of > 50 and > 40, respectively. Group 1b corresponds exactly to zones S-1 and S-2; groups 2c and 2d correspond exactly to zones S-2 and S-1, respectively; group 2a comprises the single sample at 367 cm; group 2b embraces all remaining samples. (b) Groups defined at the third hierarchical level (B–C dissimilarity > 30), excluding groups 1b and 2a from further analysis. Groups 3a and 3b are dominated by sclerophyll and rain forest pollen types, respectively. (c) Groups defined at the fourth hierarchical level. Groups 4a and 4b primarily distinguish Casuarinaceae-dominated from Eucalyptus- and Callitris-dominated samples, respectively; groups 4c and 4d primarily distinguish rain forest-rich samples with high and low values for rain forest angiosperms, respectively. (d) NMDS plot of the mean composition of individual pollen zones, with groups defined by cluster analysis superimposed. Stress = 0.06. Sample labels correspond to pollen zones S-1 to S-22, with ‘S’ omitted. Pollen zones S-1 and S-2 separate from all other zones at the > 45 and > 35 dissimilarity level, respectively. Remaining zones separate at the > 25 dissimilarity level into two groups, primarily distinguishable as rain forest-rich (group 3b) and sclerophyll-rich (group 3a) zones. The associated cluster analysis (Fig. S6) first identifies as groups the distinctive zones S-1 and S-2, and then places the remaining pollen zones into either a rain forest-rich group (group 3b, composed of S-4, S-6, S-8, S-10, S-13, S-15, S-17 and S-21) or a sclerophyll-rich group (group 3a, composed of all other zones). Zone S-19, and to a lesser extent zone S-22, have high values for Podocarpaceae, but also have high values for Eucalyptus or Callitris. Their mixed compositions are highlighted by the central placement of these zones in ordination space, and their exclusion from the rain forest-rich group of pollen zones (Fig. 4d). Correspondingly, Podocarpaceae-rich samples from zones S-19 and S-22 are mainly placed together in group 7a (Fig. 5). At the next hierarchical level, rain forest-rich samples of group 3b are composed of two groups (Figs 4c & 5). Group 4c has very high values for rain forest angiosperms, and relatively low values for Podocarpaceae, while group 4d is a larger group with high, but quite variable representation of different Podocarpaceae genera, and variable contributions from rain forest angiosperms and tree ferns. The sclerophyll-rich samples of group 3a are composed of group 4a, with high values for Casuarinaceae; and a larger, variable group 4b, with generally 1462 high values for Eucalyptus. Group 4b includes Podocarpaceaerich samples which, at a still lower hierarchical level, define group 7a (Fig. 5). These analyses reveal that, after allowing for the effects of oxidation on samples from zones S-1 and S-2 (groups 2d and 2c), the major compositional distinction in the pollen spectra is between the rain forest- and sclerophyll-rich clusters 3b and 3a (Figs 4b, 4d & 5). This distinction corresponds well with the binary alternation of rain forest- and sclerophyll-rich intervals identified by stratigraphically constrained zonation, which charts the repeated expansion and contraction of broadly consistent rain forest and sclerophyll floras. Zones S-19 and S-22 represent exceptions to this pattern, by having relatively high values for Podocarpaceae, but also for Eucalyptus (S-19) or Callitris (S-22). Distinct intervals of forest development Although the cluster analysis reveals a pattern of recurrent pollen spectra in successive rain forest- or sclerophyll-rich zones, it also provides evidence for repeating patterns of floristic change within individual zones, and for differences Journal of Biogeography 38, 1456–1470 ª 2011 Blackwell Publishing Ltd ic e ae ce na Charcoal ri co r a os d rro cc ed at us Clusters −0.4 0 0.4 • • • • • 2d 2c • 1b 2d 2c • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 6d 5h 3b 4d 6d • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 4b 3a • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 7a • • • • • • • • • • • • • • • • • • • • • • 6b 14 18 11 20 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 19 (12/22/ 16) 1a 5d 7a • • • • • • • • • • • • • • • • • • • • • • • • 22 (12/21/ 19) 4b • • • • 6a 3 • • • • • • • • • • • • • • • • • • • • • • 0 • 0 0 • • • • 10 0 0 • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 0 0 0 • • • • • • • • • • • • • • • 10 20 0 • • • (10/12) 3a 4a 5c • • • • • 10 20 0 • • • • • 5 7 4a 2a 5a 6a • • • 5c • • • 5b • • • • 5a • • • • 2a • • • • • • 9 3 40 20 0 Bray-Curtis dissimilarity 0 0 0 0 10 20 30 0 10 20 30 0 10 20 30 40 0 10 20 Early Pleistocene vegetation in south-eastern Australia (a) (b) (c) % (d) Figure 5 Selected pollen types and detrended charcoal values plotted on the terminals of the cluster dendrogram. (a) Cluster dendrogram with select labelled nodes. (b) Clusters identified at a range of hierarchical levels, corresponding to groups of homogenous palynological composition. Illustrated clusters are significant at the 0.1% level in analysis of similarity ANOSIM, executed in Primer v. 5.2.7. (c) Selected pollen types and detrended microcharcoal. (d) Zone affinities (where a single zone is given, samples within a cluster occur only within that zone; where more than one zone is given, one or more samples in a cluster occur in other zones – these zones are shown in brackets or as vertical text). 3/5/7/9/2/8 5b 12/14/16/19/11/18/9/20/21 12 19 20 Journal of Biogeography 38, 1456–1470 ª 2011 Blackwell Publishing Ltd • 6c 5h 4d 6c 4c 5g 5f 5e • • • • • • • • • • • • • • • • • • • • • • • • 3b 5g 5f 4c 5e • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • 2b 5d 6b 7b 2b 7b • • • • • • • • • • • • • • • • • • • • • • • • • • • • • zone affinities 1 2 17 16(10) 15(8) 21(1) 13 21(19) 19(17/8) 22(19) 6(4) 13(10) 8 6 4 11 19 ) ra ia) s po on m ra os oz per ho ss h s .m ra Lop gio op cf s (B ( an ng es us ry s t s us us is a/A us rit t ia rr i ia gu gu s um arp ach s ad p a po cl car di c c hi car ofa ofa fore ea son ade mb lyp si r is ry cry ro at th os a llo o y h n h u nk llit ick om or uc ac a ic g ya ra Not Not Rai hy Pod od Ba Ca D P E D D M A A P C P C a ae ace n ce ra ari m te su su in a W C as C ua (detrended) 1463 J. M. K. Sniderman 5c 5f 5f 6a 5f 6d 6d * * * 5f S-4 S-6 S-8 S-10 5e 5e 5g 5g * 5g 5e 6d 6c 5g Figure 6 Non-metric multi-dimensional scaling (NMDS) plot of Fig. 3a, with individual trajectories of rain forest-rich zones S-4, S-6, S-8 and S-10 shown, and their samples labelled according to cluster affinity. The base of each rain forest-rich zone shown by an asterisk. Age yr BP Charcoal concentration (detrended) Rate of change (detrended) 3 5 S−1 S−2 S−3 −1 0 1 1600 S−4 S−5 S−6 S−7 S−8 1650 S−9 S−10 S−11 S−12 1700 S−13 S−14 S−15 S−16 S−17 S−18 1750 S−19 1800 between individual zones. For example, the seven samples of group 5e and 5f (Fig. 5) represent rain forest angiospermdominated spectra that occur within three rain forest-rich zones (S-4, S-6 and S-8). Similarly, the six samples in group 5g, characterized by very high values for the extinct podocarp Podosporites cf. microsaccatus, are distributed across four different zones (S-4, S-6, S-10 and S-13). These distinctive groups represent transient states within recurrent trajectories of orbital-scale vegetation succession. Hence, early during zones S-4, S-6 and S-8 forest vegetation was particularly rich in rain forest angiosperms (groups 5e and 5f). Subsequently, these intervals, along with broadly similar zone S-10, evolved (represented in Fig. 6 by curves linking the samples of these zones) towards more podocarp-dominated compositions represented by groups 5g and 6d. The relative similarity of these zones is thus indicated by the similarity of their curves within ordination space (cf. Tzedakis & Bennett, 1995). Microcharcoal (Fig. 2a) was very strongly positively correlated (P < 0.001) with Callitris (r = 0.446) (Table S1) and with rate of change (r = 0.298) (Fig. 7), and was strongly negatively correlated (P < 0.01) with Casuarinaceae (r = )0.246) and Eucalyptus (r = )0.213). Correlations with Podocarpaceae and with rain forest angiosperms were not significant. Although fire–vegetation relationships probably involved lags between climate, vegetation and fire, the sampling resolution (typically 20 cm, representing c. 0.5–2.0 kyr), and the c. 20–100 years integrated within each 1-cm sediment sample, mean such ecological scale lags are unlikely to be detected here. Lagged correlations (not shown) between charcoal, rate of change and major pollen types either decreased correlations or failed to increase them significantly. DISCUSSION Spatial resolution of the Stony Creek Basin environmental record In order to gauge the fidelity of the SCB pollen record for understanding late Cenozoic vegetation and environmental history, it is necessary to evaluate its spatial and temporal Journal of Biogeography 38, 1456–1470 ª 2011 Blackwell Publishing Ltd S−20 S−21 S−22 −2 0 2 4 Figure 7 Charcoal and rate of change (both detrended) plotted on the time axis of Sniderman et al. (2007). Zone boundaries are as in Fig. 2, although note time- rather than depth-axis. 1464 Early Pleistocene vegetation in south-eastern Australia resolution. Generally fine-textured sediments (Fig. S2) indicate that the SCB palaeolake was well insulated against high energy fluvial or surface inwash, and the core site was continuously remote from the lake edge. Fine sediments are typical of maar lakes (Lorenz & Kurszlaukis, 2007) which, over a few hundred thousand years, typically accumulate dust, crater slope regolith, pollen and spores, autochthonous organic matter and opaline silica. Insofar as this model of sedimentation applies to SCB, it implies that both the modes of pollen transport, and the size of the pollen source area, were relatively constant throughout the life of the basin. Hence changing pollen signals faithfully represent changing pollen production and varying floristic composition of vegetation within a relatively constant catchment surrounding the palaeolake. However, several lines of evidence suggest that progressive basin infilling altered patterns of deposition in the SCB record, particularly in upper c. 12 m of the record. First, secular increases in pollen and charcoal concentrations (Fig. S3f) imply that clastic sedimentation rate slowed with basin infilling. This is confirmed by measurements of sediment laminations (Sniderman et al., 2007). Second, increased values for Leptospermum, small Myrtaceae types, deteriorated pollen, ¨ aquatic Isoetes and Botryococcus, and increased amplitude of organic matter cycles (Fig. S3e–f) all suggest that progressive lake infilling was associated with increased variability of water depth. Corroded and degraded grains, as observed in Casuarinaceae (Fig. S3c), are thought to indicate soil borne pollen transport (Wilmshurst & McGlone, 2005), and imperfect anoxia (Cushing, 1967), respectively. Both may reflect the migration of the lake edge closer to the core site with lake infilling, particularly during sclerophyll-rich zones that are associated with high aquatic values indicating low water level. Pollen source area is positively related to lake size (Jackson & Lyford, 1999) and varies between pollen types because of different deposition velocities. However, relationships derived from Northern Hemisphere vegetation to predict pollen source radii from lake area (Bradshaw & Webb, 1985) are of uncertain value in Australia, where a great proportion of forest tree taxa are animal pollinated (Walker, 2000). In Australia, pollen production and dispersal is apparently much greater in open forests than in tropical rain forests (Kershaw & Hyland, 1975; Kershaw & Bulman, 1994), where dispersal of the angiosperm component may be limited to a few tens of metres (Walker & Sun, 2000). Walker (2000) concluded that rain forest tree pollen percentages of 25–50% in maar lake sediments from north-east Queensland indicate that the nearby vegetation was dominated by rain forest. Hence, zones at SCB with 10–25% identifiable rain forest angiosperm pollen almost certainly indicate that rain forest was a major component of local vegetation in the landscape immediately surrounding SCB. However, the presence of a diversity of conifers at SCB is without close analogue in modern Australian rain forest. In New Zealand, conifers are important vegetation components, and only a few tall, wind-pollinated angiosperms are present; hence the majority of angiosperms are rarely encountered as pollen unless source plants are present locally (Macphail & Journal of Biogeography 38, 1456–1470 ª 2011 Blackwell Publishing Ltd McQueen, 1983; Randall, 1991). Pollen samples taken near conifer–broadleaf forests can include 70–90% conifer pollen (McGlone, 1988). At SCB, peak values of Podocarpaceae plus Araucariaceae of 40–50% therefore suggest that conifer density in the western uplands was substantial during rain forest-rich zones, if not necessarily as high as in modern New Zealand conifer–broadleaf forests. Pollen clumps (Fig. S4) represent a ‘gravity’ pollen source component (Jacobson & Bradshaw, 1981), presumably derived from within tens of metres of the lake’s edge, that, like plant macrofossils (Birks & Birks, 2000), demonstrate local presence of source plants. Pollen clumps were observed in both rain forest and open forest taxa mainly, but not exclusively, within zones in which each type achieves highest dispersed pollen values. This implies that the alternating dominance of rain forest and open forest pollen taxa in the SCB record represents vegetation changes primarily at local (perhaps < 1 km) scale. Locally dispersed taxa can be expected to show patchy pollen representation; where such types are consistently present in trace quantities (e.g. Wollemia, Beauprea, Fig. S3a), it is likely they were under-represented, rather than necessarily rare in source vegetation (Macphail et al., 1994). On the other hand, some apparently animal-pollinated taxa (e.g. Ilex, Symplocos, Myrsine) form distinctive curves in successive rain forest-rich zones (Fig. 2a), indicating that these pollen types adequately represent the changing size of source populations. It seems unlikely that the SCB pollen source area extended significantly into lowland Victoria, because the presence of several taxa which had disappeared from regional lowland basins by the Early Pliocene (e.g. Ilex, Podosporites cf. microsaccatus, Wollemia) (Macphail, 1997) is most plausibly explained if diverse rain forests had become confined to moist uplands by the Early Pleistocene. One exception is Nothofagus subgenus Brassospora, which, because it is typically over-represented (Hope, 1976) was probably sourced from surrounding lowlands. Early Pleistocene vegetation change in south-eastern Australia SCB pollen spectra include taxa, and combinations of taxa – such as diverse podocarps with Eucalyptus – that no longer occur together. Hence it is difficult to assign the fossil pollen spectra unambiguously to extant forest formations. Many SCB taxa (e.g. Phyllocladus, Cunoniaceae, Acmena) occur exclusively, or primarily, in fire-sensitive, closed canopy forests, but some of these may also occur under extreme climatic or edaphic conditions in very open, or non-forest communities. For example, the monotypic genus Microcachrys is confined to subalpine Tasmanian shrublands (Gibson et al., 1995), and Dacrydium, Araucariaceae and Cunoniaceae occur in very open vegetation on ultramafic soils in New Caledonia (McCoy et al., 1999). However, elsewhere in Australasia on ‘normal’ soils and in moderate climates, the latter are primarily confined to rain forest. Moreover, independent climate reconstructions for SCB indicating a high rainfall, warm temperate climate (Sniderman et al., 2009) demonstrate that at least one Early Pleistocene 1465 J. M. K. Sniderman Microcachrys species occupied much warmer temperatures than those to which the genus is confined today. Thus the collective presence of a diversity of these ‘rain forest’ taxa is accepted here as indicating the presence of a plant community that would be interpreted today as ‘rain forest’. For non-rain forest taxa, modern southern Australian Eucalyptus species have such consistent open-canopied physiognomy (King, 1997), and dependence on fire for regeneration (House, 1997) and health (Close et al., 2009), that it may be reasonable to assume Early Pleistocene Eucalyptus species were similarly fire-dependent dominants of open vegetation. If so, pollen types that co-vary with Eucalyptus may have been similarly fire-dependent, with high light requirements for regeneration. On the other hand, continuous regeneration in the absence of fire is reported for some species of Callitris and Allocasuarina (Bowman, 2001). However, if, as argued by Bowman (2001), what is called in Australia by the appellation ‘rainforest’ is merely the moist, tall, closed-canopy endmember of a continuum of fire-sensitive plant communities, it is likely that Early Pleistocene Callitris- and Allocasuarinadominated communities nevertheless occupied a different position from ‘rainforest’ along this continuum. Hence, for the purposes here, it is considered convenient to accept Eucalyptus, Casuarinaceae, and Callitris as ‘sclerophyll’ taxa. The SCB record reveals several aspects of the nature of southern Australian vegetation during the Early Pleistocene. First, the primary pattern, at orbital scale, is the alternation between rain forest and sclerophyll forests, indicating rhythmic changes in the dominance of vegetation. Moreover, the trace (or greater) presence of at least some rain forest pollen types within sclerophyll-rich intervals, and of pollen of both forest canopy and understorey sclerophyll taxa during rain forestrich intervals, implies that the local vegetation was never entirely given over to either of these forest formations. One interpretation of this is that the pollen record represents fluctuating dominance within a single, integrated forest community with both rain forest and sclerophyll components. For example, modern, ecotonal ‘mixed forests’ (Gilbert, 1959), in which rain forest occupies the understorey of tall Eucalyptus forests, may represent partial analogues for SCB pollen spectra. Certainly, SCB forests must have remained moist even when little rain forest was registered, as mesic Winteraceae and the tree fern Dicksonia are primarily found in zones dominated by Eucalyptus; and SCB forests must have remained relatively dense, as typically over-represented Poaceae and Asteraceae never achieve high values. However, as the majority of sclerophyll understorey taxa are intolerant of fully developed rain forest canopies, it seems more likely that topographic and edaphic variation hosted a vegetation mosaic in which Eucalyptus-, Casuarinaceae- and Callitris-dominated forests were closely juxtaposed with rain forest communities. The latter is supported by macrofossil evidence of an extremely diverse assemblage of Ericaceae subfamily Styphelioideae (Jordan et al., 2007) and of other micro-sclerophyllous taxa (G.J. Jordan, University of Tasmania, unpublished data), indicating that species-rich sclerophyllous understorey com1466 munities were locally present, at least during times of sclerophyll forest dominance. The nature and magnitude of the climate changes that drove these rhythmic vegetation responses is uncertain. Intuitively, the dramatic changes in pollen values suggest an equally dramatic climatic driver, and Sniderman et al. (2007) assumed that vegetation changes were primarily a response to changing summer rainfall, in turn driven by summer insolation changes. Establishing the dynamic mechanisms responsible for translating insolation changes into precipitation patterns at SCB requires climate modelling. However, one can speculate that insolation changes drove changes in hemispheric temperature gradients, regional sea surface temperatures, or land–sea heating differences, which in turn led to differences in annual or seasonal rainfall, which in turn led to two distinct climate states: one favourable to rain forest at the expense of sclerophyll forests, and the other favourable to sclerophyll forest at the expense of rain forest. The inability of beetlederived climate estimates (Sniderman et al., 2009) to differentiate the climates of rain forest- and sclerophyll-rich intervals, and the quite high rainfall reconstructed for both vegetation types, implies that moisture limitation per se may not have played a large role. Given the fire-tolerance of Eucalyptus and the fire-sensitivity of rain forest, it is likely that alternation between these vegetation modes was driven by differences in fire frequency and/or intensity, particularly during the cool half year, by analogy with the timing of burning season today in parts of eastern and northern Australia with high summer rainfall (Russell-Smith & Stanton, 2002). This might explain the beetles’ insensitivity to these differences, because the relative winter inactivity and/or subterranean juvenile stages of many beetle species may make them relatively insensitive to climate variability during the cool half-year (Tauber et al., 1998). Finer scale vegetation differences within and between individual zones suggest that some Early Pleistocene insolation cycles varied significantly from one another, but also that some successive cycles were apparently so similar that rain forest successions were replicated from one cycle to the next. The repeated evolution of zones S-10, S-8, S-6 and S-4 from rain forest-angiosperm-dominated to Podocarpaceae-dominated is an example of the latter, but an example of the former is uniquely provided by zone S-21. Dominated initially by diverse podocarps, Agathis and angiosperms, this interval later saw the expansion of Araucaria and Microcachrys, a change presumably responding to a biologically significant climate transition. The composition of open forest-rich intervals was also sometimes as predictable as for rain forest-rich intervals. For example, sequential peaks of Callitris, Casuarinaceae and Eucalyptus must represent the changing importance of these genera within the local vegetation, and the recurrence of this sequence, in zones S-9, S-7, S-5 and S-3, suggests that the regional climate, at sub-orbital time-scales, evolved in a repetitive manner during successive insolation cycles. Recurrent patterns of forest development are also known from the Plio-Pleistocene in southern Europe (Combourieu-Nebout, 1993; Tzedakis & Journal of Biogeography 38, 1456–1470 ª 2011 Blackwell Publishing Ltd Early Pleistocene vegetation in south-eastern Australia Bennett, 1995; Joannin et al., 2007), where ‘warm’ forest phases are typically followed by ‘cool’ forest phases, interpreted as out of phase moisture and temperature responses to changing orbital configurations (Combourieu-Nebout, 1993; Fusco, 2007). Such variability has not previously been documented for the Early–Middle Pleistocene in the Southern Hemisphere. Charcoal concentrations may shed light on the importance of fire during particular vegetation states and during transitions between different states. However, in small lakes, microcharcoal source catchments may be larger than those for pollen (Clark & Royall, 1995). Hence, the negative correlation of charcoal with fire-dependent Eucalyptus, and its positive correlation with more typically fire-sensitive Callitris, suggests that SCB microcharcoal registers regional, rather than local fire activity. The western uplands today includes a range of forest communities distributed along elevationally controlled climatic gradients; a similar pattern during the Early Pleistocene may have prevented the development of a linear relationship between regional fire and local vegetation at SCB. However, the positive correlation between charcoal and rate of change implies that fire activity at regional scale was greatest at times of most rapid and profound vegetation transitions at local scale. Broadly, high rates of change correspond to zone boundaries, which are times of first-order vegetation change driven by orbital forcing. Hence, the simplest interpretation of the charcoal data, despite probable differences in local-scale vegetation and fire regimes across the uplands, is that high regional fire activity coincided with changes in regional vegetation that were coordinated by orbital forcing. Vegetation analogues The importance and diversity of gymnosperms in the SCB pollen spectra implies that extant Australian rain forests are inadequate structural and ecological analogues for SCB rain forest. This is because Podocarpaceae and Araucariaceae are typically forest dominants or canopy emergents with life spans centuries longer than Australasian rain forest angiosperm genera (Ash, 1983; Ogden & Stewart, 1995), and may represent ‘foundation species’ (Ellison et al., 2005) that control the successional and ecological dynamics of the forests they dominate. Today, conifers dominate Australian rain forests primarily in extreme edaphic or climatic environments and nowhere are they very species rich (Enright, 1995; Gibson et al., 1995), implying that this Early Pleistocene rain forest may be best interpreted as a distinct biome now largely extinct in Australia. The biological significance of this can be seen by considering that of the 12 rain forest gymnosperm genera represented at SCB, only three (Podocarpus, Microstrobos and Wollemia) are still extant on the south-eastern Australian mainland, all with relictual, highly localized or specialized habitats. Widespread pollen evidence for Podocarpaceae and Araucariaceae in both south-eastern and south-western Australia Journal of Biogeography 38, 1456–1470 ª 2011 Blackwell Publishing Ltd during the Pliocene (Kershaw et al., 1994; Macphail et al., 1995; Macphail, 1996; Dodson & Macphail, 2004) indicates the former wide extent of this biome. However, the SCB rain forest record is more diverse, and shows orbital scale alternation with sclerophyll forests much more clearly than any of the Pliocene records. The closest floristic analogues to SCB rain forest are found elsewhere in Australasia, in New Zealand, New Caledonia and New Guinea, where forests co-dominated by Podocarpaceae and/or Araucariaceae and rain forest angiosperms are common. However, in contrast with these extant forests, Australian Early Pleistocene rain forests were closely juxtaposed with sclerophyllous communities, highlighting the long history of ecological mingling between these biomes in Australia. Yet several modern Australian sclerophyll taxa (Eucalyptus, Casuarinaceae, Acacia) were also present in New Zealand during the Neogene and became extinct there only in the late Pliocene or Pleistocene (Lee et al., 2001). The regions also shared some extinctions, as Beauprea (Mildenhall & Alloway, 2008) and the Microcachrys clade (Mildenhall & Byrami, 2003) were present at SCB and in New Zealand during the Early Pleistocene, but are now absent from both areas. These connections reinforce accumulating appreciation of the botanical similarity of southern Australia and New Zealand during the Neogene (McGlone, 2006), and provide new evidence for how recently their floras have diverged. In summary, the transition to the Late Quaternary was not simply one of contraction of rain forest to its current small remnants but involved the loss of a distinctive, conifer-rich rain forest biome. It is unlikely that this biome was confined during the Early Pleistocene solely to Victoria’s western uplands, as their subdued topography offers little in the way of a climatic or fire refuge in the way that, for example, relict Wollemia nobilis has become confined to a few narrow gorges in the Blue Mountains of New South Wales (Peakall et al., 2003). It is thus more likely that this conifer-rich rain forest biome was widespread throughout upland south-eastern Australia during the Early Pleistocene. ACKNOWLEDGEMENTS This paper is based on work done by the author while the recipient of an Australian postgraduate award at Monash University. The author thanks J.R. Brown, A.P. Kershaw, M.S. McGlone, D.M.J.S. Bowman, and two anonymous referees for valuable comments. REFERENCES Ash, J. (1983) Growth rings in Agathis robusta and Araucaria cunninghamii from tropical Australia. Australian Journal of Botany, 31, 269–275. Ashton, D.H. & Attiwill, P.M. (1994) Tall open-forests. Australian vegetation (ed. by R.H. Groves), pp. 157–196. Cambridge University Press, Cambridge. 1467 J. M. K. Sniderman Bennett, K.D. (1996) Determination of the number of zones in a biostratigraphical sequence. New Phytologist, 132, 155– 170. Bennett, K.D. (2005) Documentation for Psimpoll 4.25 and Pscomb 1.03: C programs for plotting pollen diagrams and analysing pollen data. Department of Earth Sciences, University of Uppsala, Uppsala. Available at: http://www.chrono. qub.ac.uk/psimpoll/psimpoll.html. Bennett, K.D. & Humphry, R.W. (1995) Analysis of late-glacial and Holocene rates of vegetational changes at two sites in the British Isles. Review of Palaeobotany and Palynology, 85, 263–287. Birks, H.H. & Birks, H.J.B. (2000) Future uses of pollen analysis must include plant macrofossils. Journal of Biogeography, 27, 31–35. Birks, H.J.B. & Gordon, A.D. (1985) Numerical methods in Quaternary pollen analysis. Academic Press, London. Bowman, D.M.J.S. (2000) Australian rainforests: islands of green in a land of fire. Cambridge University Press, Cambridge. Bowman, D.M.J.S. (2001) On the elusive definition of ‘Australian rainforest’: response to Lynch and Neldner (2000). Australian Journal of Botany, 49, 785–787. Bradshaw, R.H.W. & Webb, T.I. (1985) Relationships between contemporary pollen and vegetation data from Wisconsin and Michigan, USA. Ecology, 66, 721–737. Clark, J.S. & Royall, P.D. (1995) Particle-size evidence for source areas of charcoal accumulation in late Holocene sediments of eastern North American lakes. Quaternary Research, 43, 80–89. Clarke, K.R. (1993) Non-parametric multivariate analyses of changes in community structure. Australian Journal of Ecology, 18, 117–143. Clarke, K.R. & Gorley, R.N. (2001) Primer v5: user manual/ tutorial. Plymouth Marine Laboratory, Plymouth, UK. Clarke, K.R. & Warwick, R.M. (2001) Change in marine communities: an approach to statistical analysis and interpretation. Plymouth Marine Laboratory, Plymouth, UK. Close, D.C., Davidson, N.J., Johnson, D.W., Abrams, M.D., Hart, S.C., Lunt, I.D., Archibald, R.D., Horton, B. & Adams, M.A. (2009) Premature decline of eucalyptus and altered ecosystem processes in the absence of fire in some Australian forests. Botanical Review, 75, 191–202. Combourieu-Nebout, N. (1993) Vegetation response to upper Pliocene glacial/interglacial cyclicity in the central Mediterranean. Quaternary Research, 40, 228–236. Cookson, I.C. (1953) The identification of the sporomorph Phyllocladidites with Dacrydium and its distribution in southern Tertiary deposits. Australian Journal of Botany, 1, 64–71. Cookson, I.C. & Pike, K.M. (1954) Some dicotyledonous pollen types from Cainozoic deposits in the Australian region. Australian Journal of Botany, 2, 197–219. Cushing, E.J. (1967) Evidence for differential pollen preservation in Late Quaternary sediments in Minnesota. Review of Palaeobotany and Palynology, 4, 87–101. 1468 Dodson, J.R. & Macphail, M.K. (2004) Palynological evidence for aridity events and vegetation change during the Middle Pliocene, a warm period in Southwestern Australia. Global and Planetary Change, 41, 285–307. Ellison, A.M., Bank, M.S., Clinton, B.D., Colburn, E.A., Elliott, K., Ford, C.R., Foster, D.R., Kloeppel, B.D., Knoepp, J.D., Lovett, G.M., Mohan, J., Orwig, D.A., Rodenhouse, N.L., Sobczak, W.V., Stinson, K.A., Stone, J.K., Swan, C.M., Thompson, J., Von Holle, B. & Webster, J.R. (2005) Loss of foundation species: consequences for the structure and dynamics of forested ecosystems. Frontiers in Ecology and the Environment, 3, 479–486. Enright, N.J. (1995) Conifers of tropical Australasia. Ecology of the southern conifers (ed. by N.J. Enright and R.S. Hill), pp. 197–222. Melbourne University Press, Melbourne. Fusco, F. (2007) Vegetation response to early Pleistocene climatic cycles in the Lamone valley (Northern Apennines, Italy). Review of Palaeobotany and Palynology, 145, 1–23. Gibson, N., Barker, P.C.J., Cullen, P.J. & Shapcott, A. (1995) Conifers of southern Australia. Ecology of the southern conifers (ed. by N.J. Enright and R.S. Hill), pp. 223–251. Melbourne University Press, Melbourne. Gilbert, J.M. (1959) Forest succession in the Florentine Valley, Tasmania. Papers and Proceedings of the Royal Society of Tasmania, 93, 129–151. Graham, I.T., Hollis, J.D., Sutherland, F.L. & Joyce, E.B. (2003) Insights into the newer volcanics province of Victoria. Specialist Group in Geochemistry, Mineralogy and Petrology Field Guide , Geological Society of Australia, Sydney. Heusser, C.J. (1971) Pollen and spores of Chile. Arizona Press, Tuscon, AZ. Hope, G.S. (1976) The vegetational history of Mt. Wilhelm, Papua New Guinea. Journal of Ecology, 64, 627–663. House, S.M. (1997) Reproductive biology of eucalypts. Eucalypt ecology: individuals to ecosystems (ed. by J. Williams and J. Woinarski), pp. 30–55. Cambridge University Press, Cambridge. Jackson, S.T. & Lyford, M.E. (1999) Pollen dispersal models in Quaternary plant ecology: assumptions, parameters, and prescriptions. The Botanical Review, 65, 39–75. Jackson, W.D. (1968) Fire, air, water and earth – an elemental ecology of Tasmania. Proceedings of the Ecological Society of Australia, 3, 9–16. Jacobson, G.L.J. & Bradshaw, R.H.W. (1981) The selection of sites for paleovegetational studies. Quaternary Research, 16, 80–96. Joannin, S., Quillevere, F., Suc, J.P., Lecuyer, C. & Martineau, F. (2007) Early Pleistocene climate changes in the central Mediterranean region as inferred from integrated pollen and planktonic foraminiferal stable isotope analyses. Quaternary Research, 67, 264–274. Jordan, G.J., Bromfield, K.E., Sniderman, J.M.K. & Crayn, D. (2007) Diverse fossil epacrids (Styphelioideae; Ericaceae) from Early Pleistocene sediments at Stony Creek Basin, Journal of Biogeography 38, 1456–1470 ª 2011 Blackwell Publishing Ltd Early Pleistocene vegetation in south-eastern Australia Victoria, Australia. International Journal of Plant Sciences, 168, 1359–1376. Joyce, E.B. (1992) The west Victorian uplands of southeastern Australia: origin and history. Earth Surface Processes and Landforms, 17, 407–418. Kershaw, A.P. & Bulman, D. (1994) The relationship between modern pollen samples and environment in the humid tropics region of northeastern Australia. Review of Palaeobotany and Palynology, 83, 83–96. Kershaw, A.P. & Hyland, B.P.M. (1975) Pollen transfer and periodicity in a rain-forest situation. Review of Palaeobotany and Palynology, 19, 129–138. Kershaw, A.P., Martin, H.A. & McEwan Mason, J.R.C. (1994) The Neogene: a period of transition. History of the Australian vegetation: Cretaceous to Recent (ed. by R.S. Hill), pp. 299–327. Cambridge University Press, Cambridge. King, D.A. (1997) The functional significance of leaf angle in Eucalyptus. Australian Journal of Botany, 45, 619–639. Lee, D.E., Lee, W.G. & Mortimer, N. (2001) Where and why have all the flowers gone? Depletion and turnover in the New Zealand Cenozoic angiosperm flora in relation to palaeogeography and climate. Australian Journal of Botany, 49, 341–356. Lorenz, V. & Kurszlaukis, S. (2007) Root zone processes in the phreatomagmatic pipe emplacement model and consequences for the evoluton of maar-diatreme volcanoes. Journal of Volcanology and Geothermal Research, 159, 4–32. Lourens, L.J., Antonarakou, A., Hilgen, F.J., Van Hoof, A.A.M., Vergnaud Grazzini, C. & Zachariasse, W.J. (1996) Evaluation of the Plio-Pleistocene astronomical timescale. Paleoceanography, 11, 391–413. Macphail, M.K. (1996) Neogene environments in Australia, 1: re-evaluation of microfloras associated with important Early Pliocene marsupial remains at Grange Burn, southwest Victoria. Review of Palaeobotany and Palynology, 92, 307–328. Macphail, M.K. (1997) Late Neogene climates in Australia: fossil pollen- and spore-based estimates in retrospect and prospect. Australian Journal of Botany, 45, 425–464. Macphail, M.K. & McQueen, D.R. (1983) The value of New Zealand pollen and spores as indicators of Cenozoic vegetation and climates. Tuatara, 26, 37–59. Macphail, M.K., Alley, N.F., Truswell, E.M. & Sluiter, I.R.K. (1994) Early Tertiary vegetation: evidence from spores and pollen. History of the Australian vegetation: Cretaceous to Recent (ed. by R.S. Hill), pp. 189–261. Cambridge University Press, Cambridge. Macphail, M.K., Colhoun, E.A. & Fitzsimons, S.J. (1995) Key periods in the evolution of the Cenozoic vegetation and flora in western Tasmania: the Late Pliocene. Australian Journal of Botany, 43, 505–526. McCoy, F. (1876) Prodromus of the palaeontology of Victoria. Decade IV. Geological Survey of Victoria Special Publication, 32 pp. McCoy, S., Jaffre, T., Rigault, F. & Ash, J.E. (1999) Fire and succession in the ultramafic maquis of New Caledonia. Journal of Biogeography, 26, 579–594. Journal of Biogeography 38, 1456–1470 ª 2011 Blackwell Publishing Ltd McGlone, M.S. (1988) New Zealand. Vegetation history (ed. by B. Huntley and T.I. Webb), pp. 557–599. Kluwer, Dordrecht. McGlone, M.S. (2006) Becoming New Zealanders: immigration and the formation of the biota. Biological invasions in New Zealand (ed. by R.B. Allen and W.G. Lee), pp. 17–32. Springer Verlag, Berlin. Mildenhall, D.C. & Alloway, B.V. (2008) A widespread ca. 1.1 Ma TVZ silicic tephra preserved near Wellington, New Zealand: implications for regional reconstruction of midPleistocene vegetation. Quaternary International, 178, 167– 182. Mildenhall, D.C. & Byrami, M.L. (2003) A redescription of Podosporites parvus (Couper) Mildenhall emend. Mildenhall & Byrami from the Early Pleistocene, and late extinction of plant taxa in northern New Zealand. New Zealand Journal of Botany, 41, 147–160. Moar, N.T. (1993) Pollen grains of New Zealand dicotyledonous plants. Maanaki Whenua Press, Lincoln. Moore, P.D., Webb, J.A. & Collinson, M.E. (1991) Pollen analysis, 2nd edn. Blackwell Scientific Publications, Oxford. Munsterman, D. & Kerstholt, S. (1996) Sodium polytungstate, a new non-toxic alternative to bromoform in heavy liquid separation. Review of Palaeobotany and Palynology, 91, 417– 422. Ogden, J. & Stewart, G.H. (1995) Community dynamics of the New Zealand conifers. Ecology of the southern conifers (ed. by N.J. Enright and R.S. Hill), pp. 81–119. Melbourne University Press, Melbourne. Patton, R.T. (1928) Fossil plants of the Stony Creek Basin. Proceedings of the Royal Society of Victoria (new series), 40, 88–90. Peakall, R., Ebert, D., Scott, L.J., Meagher, P.F. & Offord, C.A. (2003) Comparative genetic study confirms exceptionally low genetic variation in the ancient and endangered relictual conifer, Wollemia nobilis (Araucariaceae). Molecular Ecology, 12, 2331–2343. Randall, P.M. (1991) A one-year pollen trapping study in Westland and Canterbury, South Island, New Zealand. Journal of the Royal Society of New Zealand, 21, 201–218. Russell-Smith, J. & Stanton, P. (2002) Fire regimes and fire management of rainforest communities across northern Australia. Flammable Australia: the fire regimes and biodiversity of a continent (ed. by R. Bradstock, J. Williams and M. Gill), pp. 329–350. Cambridge University Press, New York. Sniderman, J.M.K. (2007) Late Pliocene and Early Pleistocene vegetation and climate cyclicity in the western uplands of Victoria, Australia. PhD Thesis, Monash University, Melbourne. Sniderman, J.M.K., Pillans, B., O’Sullivan, P.B. & Kershaw, A.P. (2007) Climate and vegetation in southeastern Australia respond to Southern Hemisphere insolation forcing in the late Pliocene–early Pleistocene. Geology, 35, 41–44. Sniderman, J.M.K., Porch, N. & Kershaw, A.P. (2009) Quantitative reconstruction of Early Pleistocene climate in 1469 J. M. K. Sniderman southeastern Australia and implications for atmospheric circulation. Quaternary Science Reviews, 28, 3185–3196. Tauber, M.J., Tauber, C.A., Nyrop, J.P. & Villani, M.G. (1998) Moisture, a vital but neglected factor in the seasonal ecology of insects: hypotheses and tests of mechanisms. Environmental Entomology, 27, 523–530. Tzedakis, P.C. & Bennett, K.D. (1995) Interglacial vegetation succession: a view from southern Europe. Quaternary Science Reviews, 14, 967–982. Van’t Veer, R. & Hooghiemstra, H. (2000) Montane forest evolution during the last 650 000 yr in Columbia: a multivariate approach based on pollen record Funza-I. Journal of Quaternary Science, 15, 329–346. Walker, D. (2000) Pollen input to, and incorporation in, two crater lakes in tropical northeast Australia. Review of Palaeobotany and Palynology, 111, 253–283. Walker, D. & Sun, X. (2000) Pollen fall-out from a tropical vegetation mosaic. Review of Palaeobotany and Palynology, 110, 229–246. Wang, F.S., Chen, N.F., Zhang, Y.L. & Yang, H.Q. (1995) Pollen flora of China, 2nd edn. Science Press, Beijing, 461 pp. Webb, L.J. & Tracey, J.G. (1994) The rainforests of northern Australia. Australian vegetation (ed. by R.H. Groves), pp. 87–129. Cambridge University Press, Cambridge. Wilmshurst, J.M. & McGlone, M.S. (2005) Corroded pollen and spores as indicators of changing lake sediment sources and catchment disturbance. Journal of Paleolimnology, 34, 503–517. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Appendix S1 Additional site details and analyses (Figs S1–S5 and Table S1). As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be reorganized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. BIOSKETCH Kale Sniderman is a palynologist and palaeoclimatologist interested in Cenozoic environmental change in the Southern Hemisphere. Editor: David Bowman 1470 Journal of Biogeography 38, 1456–1470 ª 2011 Blackwell Publishing Ltd