Content uploaded by Jon Luly
Author content
All content in this area was uploaded by Jon Luly on Aug 10, 2014
Content may be subject to copyright.
http://hol.sagepub.com/
The Holocene
http://hol.sagepub.com/content/23/2/305
The online version of this article can be found at:
DOI: 10.1177/0959683612460792
2013 23: 305 originally published online 9 November 2012The Holocene
Silvie Moravek, Jon Luly, John Grindrod and Russell Fairfax
The origin of grassy balds in the Bunya Mountains, southeastern Queensland, Australia
Published by:
http://www.sagepublications.com
can be found at:The HoloceneAdditional services and information for
http://hol.sagepub.com/cgi/alertsEmail Alerts:
http://hol.sagepub.com/subscriptionsSubscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
http://hol.sagepub.com/content/23/2/305.refs.htmlCitations:
What is This?
- Nov 9, 2012OnlineFirst Version of Record
- Jan 24, 2013Version of Record >>
at James Cook University on August 10, 2014hol.sagepub.comDownloaded from at James Cook University on August 10, 2014hol.sagepub.comDownloaded from
The Holocene
23(2) 305 –315
© The Author(s) 2012
Reprints and permission:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0959683612460792
hol.sagepub.com
Introduction
Grassy balds are patches of grassland in forested montane land-
scapes. Most are subject to gradual attrition by the encroachment
of trees into the grassland and there is an urgent need to under-
stand their history and dynamics before attempting to conserve
them and the economic and biodiversity values they support.
Some grassy balds can be easily understood as products of the
effect of cold air drainage (Paton, 1988), waterlogging of soils
(Smith, 1975) or as transient echoes of some environmental distur-
bance (Rogers, 1994; Scott, 1977). The origin and history of many
other grassy balds remains obscure. These ‘mysterious’ grassy
areas are found worldwide (Bruce, 1988; Coop and Givnish, 2007;
Wells, 1963). The most thoroughly studied examples are in the
Appalachian Mountains in the eastern USA. There, balds are vari-
ously considered to be the products of climatic, edaphic and topo-
graphic phenomena, with secondary roles played by anthropogenic
fire, tree clearing and grazing (Billings and Mark, 1957; Knoepp
et al., 1988; Lindsay and Bratton, 1979; Mark, 1958; Weigl and
Knowles, 1999; Whittaker, 1956).
In Australia grassy patches within closed forest can be found in
the rainforests of the Wet Tropics of northern Queensland (Tracey,
1982), in the dry rainforests of inland Queensland (there generally
known as ‘pockets’; Fensham and Skull, 1999), and in the lowland
subtropical rainforests of northern New South Wales (where they
are known as ‘grasses’; Boyd et al., 1999, Stubbs, 2001). Particu-
larly striking examples of grassy balds are found in the Bunya
Mountains of southern Queensland, about 160 km northwest of
Brisbane (Figure 1) where natural grasslands are embedded in a
rainforest, vine thicket and eucalypt woodland matrix that cloaks
cool basaltic uplands.
Attempts to explain the grassy balds of the Bunya Mountains
began with Herbert (1938) and Webb (1964) who considered the
balds to be relict vegetation preserving elements of herbfields
adapted to cold, dry, low CO2 environments of last glacial maxi-
mum times. Fensham and Fairfax (1996a) ruled out soil depth,
temperature, aspect and elevation as significant controls on the
occurrence and distribution of grassy balds. Instead, they proposed
that the grassy balds are either remnant vegetation maintained as
grassland by fire since the Pleistocene or are the result of cataclys-
mic events which breached the forest canopy, changing microcli-
mate and predisposing to the occurrence of intense fires and
occupation of forest gaps by grassland. They argue that grassland
patches could then be maintained by recurrent fire of natural or
anthropogenic origin. Whilst the role of fire is emphasised in most
accounts of the history and persistence of grassy balds, experi-
mental burns in the Bunya Mountains suggest that fire has limited
success in controlling tree incursion into grassland (Fensham and
Fairfax, 2006) and under modern fire regimes, the area of grass-
land in the Bunya Mountains has declined significantly (Fensham
and Fairfax, 1996a).
In this paper we report a palaeoecological reconstruction of
vegetational history of a swamp adjacent to the Dandabah grassy
bald in the Bunya Mountains National Park. We seek to exam-
ine vegetational change at the site and to determine whether the
Dandabah grassy bald is an ancient relict community or the
beneficiary of some local environmental perturbation. We also
reconstruct aspects of the fire regime that allegedly underpins the
survival of a grassy bald in the face of invasion by trees, indepen-
dent of its mode of origin.
460792HOL23210.1177/0959683
612460792The HoloceneMoravek et al.
2012
1James Cook University, Australia
2Monash University, Australia
3Queensland Herbarium, Australia
Corresponding author:
Jon Luly, School of Earth and Environmental Sciences, James Cook
University, Townsville Queensland 4811, Australia.
Email: jonathan.luly@jcu.edu.au
The origin of grassy balds in the Bunya
Mountains, southeastern Queensland,
Australia
Silvie Moravek,1 Jon Luly,1 John Grindrod2 and Russell Fairfax3
Abstract
Montane grasslands, or grassy balds, are enigmatic features of mountains worldwide. Their origins are often obscure. Pollen, phytolith and charcoal
analysis of Dandabah Swamp in the Bunya Mountains in southeastern Queensland, Australia suggest that there, grassy balds comprise a relict vegetation
maintained in the face of postglacial tree invasion by fire. The balds are not the product of edaphic phenomena or natural or anthropogenic cataclysms
and will require intensive management efforts to be conserved in a world of increased woodiness, rising atmospheric CO2 and changing climate.
Keywords
fire, grassy balds, Holocene climate, phytolith analysis, pollen analysis, tree incursion, vegetation change
Received 25 May 2012; revised manuscript accepted 30 June 2012
Research paper
at James Cook University on August 10, 2014hol.sagepub.comDownloaded from
306 The Holocene 23(2)
The site
Dandabah Swamp (26°52′57″S, 151°35′41″E, elevation 975 m
a.s.l.) is located in the headwaters of Saddle-Tree Creek, close to
the township of Dandabah. The swamp is about 70 m wide and cov-
ers approximately 0.6 ha. It is bounded on one side by a formerly
natural grassland, now converted into the lawns of the Dandabah
picnic and camping area, and mature rainforest on the other. Rem-
nants of the natural grassland remain immediately upstream of
Dandabah Swamp. The swamp surface gently slopes from the
northwest and an ill-defined channel of Saddle-Tree Creek runs
along the southeastern edge.
Climate
The regional climate is strongly seasonal with tropical weather
systems bringing regular heavy rain and thunderstorms in the sum-
mer months. Winter (April–September) is cool and relatively dry.
Rainfall at Mt. Mowbullan (1 km east of Dandabah) averages 1075
mm/yr (Bruce, 1988) with an estimated additional 10% of the total
moisture budget contributed by fog drip (Fairfax, personal com-
munication, 2003). The mean daily maximum and minimum tem-
peratures recorded at Mt. Mowbullan are 23.2°C and 15.4°C in
January, and 13.1°C and 4.5°C in July. Frosts occur approximately
20 times per year (Bruce, 1988) and snow on rare occasions.
Vegetation
The Bunya Mountains support a rich and varied flora. Araucarian
rainforest predominates at higher elevations, Eucalyptus wood-
lands dominate the ridges, and drier rainforests and vine thickets
occupy the mid-slopes and gullies. On the western foothills,
Casuarina cristata and Acacia harpophylla communities occur
and eucalypts dominate the eastern foothills (EPA, 2001). Grass-
lands occur across the whole massif in a range of topographic
positions and regardless of surrounding vegetation type. Woody
species, especially Acacia and eucalypts, are actively colonising
the grasslands (Fairfax et al., 2009).
The grassy balds have relatively high species diversity, with 70
grass species recorded by Sparshott (2003). The common genera
Poa and Themeda dominate many of the balds, however rarer
grasses, including Bothriochloa bunyensis and Thesium australe,
are also present (Fensham and Fairfax, 1996b). Subtropical rainfor-
est with an open understorey and Araucaria bidwillii as an emer-
gent are interspersed with vine thickets of Acacia irrorata and
assorted rainforest trees on basaltic soils and plateau areas. The
rainforest is species-poor relative to other rainforests of the region,
and species present are comparatively drought-tolerant (Butler,
2003). Eucalypt open forests dominated by E. tereticornis and E.
eugenioides are found on drier slopes (Bruce, 1988; EPA, 2001).
The Dandabah bald is enclosed by rainforest and swamp veg-
etation. Grassland in the grassy bald is dominated by Poa labil-
lardieri with Poa sieberiana. A narrow fern-land fringe dominated
by Cyathea australis with Alocasia brisbanensis separates grass-
land from rainforest. Swamp vegetation on the sodden down-
slope end of the bald is overwhelmingly dominated by the swamp
reed, Phragmites australis, with a subsidiary component, includ-
ing Cyperaceae spp., Pteridium and other ground ferns occupying
openings in the dense reed bed.
Human history
The Bunya Mountains have a particular importance to both
Aboriginal and European occupants of southeastern Queensland.
Landscape features, such as Mt. Mowbullan, derive their names
from the indigenous languages – Mowbullan is apparently a con-
flation of mau, meaning head, and balan meaning bald, in refer-
ence to the grassy balds near the summit (French, 1989; Steele,
1983). This derivation suggests the grassy balds have been in
existence for some time. Large gatherings, estimated by French
(1989) to range between 2000 and 20,000 participants, were held
approximately triennially by Indigenous people to coincide with
peak production of bunya nuts (the seeds of Araucaria bidwillii).
Aboriginal populations declined dramatically in the 1830s due to
smallpox epidemics (French, 1989) and ethnographic accounts
probably underestimate the scale of Indigenous use of the
Bunya Mountains. European observers of the bunya festivals
remarked upon the diversity foods available, many of which
were obtained from grassy balds or the ecotones between balds
and forest (Steele, 1983).
Timber harvesting in the Bunya Mountains dates from 1863
(Jarrott, 1995). The settlement at Dandabah, then known as The
Lucerne Patch, was established in the late 1800s as a forestry
camp and to support pastoral enterprises which grazed cattle in the
mountains. Since 1908, parts of the Bunya Mountains have been
conserved in the Bunya Mountains National Park.
Methods
Cores collected with a Livingstone corer were taken along a tran-
sect running approximately east–west across the lower third of
the Swamp (Figure 2). An 80 mm diameter percussion core was
collected from the deepest part of the swamp. Material from this
core (DB-3) was used for radiocarbon dating and for pollen and
phytolith analyses. Sediments were described following Kershaw
(1997).
Tauber pollen traps (Tauber, 1967) were placed in representa-
tive stands of major vegetation communities of the Bunya Moun-
tains National Park to gauge the pollen signal yielded by each of
the different vegetation types. The trapping period was April to
October 2003. Each trap was placed at a height of 1.5 m and filled
with 100 ml of 1:1 formalin/glycerol mix. Traps were sealed for
transport after collection. Trap contents were filtered and the few
insects present were removed prior to processing. The contents of
each pollen trap were processed as per Faegri and Iversen (1964)
Figure 1. (a) Location of the Bunya Mountains. (b) The Bunya
Mountains massif showing approximate locations of pollen traps and
vegetation types. 1: Araucaria-dominated rainforest; 2: vine forest;
3: Eucalypt forest; 4: natural grasslands; and 5: exotic grassland. (c)
Approximate dimensions of Dandabah swamp. The maximum width
of the swamp is 80 m, and the northerly patch of exotic grassland is
introduced pasture on former grassland.
at James Cook University on August 10, 2014hol.sagepub.comDownloaded from
Moravek et al. 307
and mounted on slides in silicon oil. Counts were conducted at
400× magnification to a pollen sum of 100 arboreal pollen grains.
Results are presented in Figure 3.
Samples for pollen and charcoal analyses were taken from 0.5
cm thick slices cut at 5 cm intervals from the percussion core. The
outer rim of each slice was removed to minimise the risk of con-
tamination. For each pollen sample 4 cm3 of sediment were pro-
cessed. Pollen extraction and slide preparation were carried out at
Monash University using the standard methods of potassium
hydroxide digestion, hydrofluoric acid treatment and acetolysis
(Fægri and Iversen, 1964). A Lycopodium tablet was added to
each sample as an exotic spike to enable absolute abundances of
pollen and charcoal to be calculated (Stockmarr, 1971). Pollen
residues were mounted on slides in glycerol and the slides sealed
with wax.
Samples were counted at 10 cm intervals. Multiple slides of
single intervals were often required to reach the pollen sum of 100
pollen from arboreal taxa. Pollen were counted on a Leitz Dialux
20 EB light microscope under 400× magnification. Identification
of woody taxa was carried out to generic level. Asteraceae pollen
were subdivided into tribes Tubuliflorae and Liguliflorae. Tubuli-
florae were considered a shrub taxon because of the frequent pres-
ence of woody genera such as Olearia and Cassinia in the modern
vegetation.
Charcoal was counted by the point counting method (Clark,
1982). Completely opaque black fragments were considered to be
charcoal for the purposes of this analysis (Swain, 1973; Tinner
and Hu, 2003). Counts are expressed as mm2/cm3. A minimum
confidence level of 85–90% was selected for charcoal counts. Fos-
sil pollen of major taxa and charcoal data are depicted in Figure 4.
A summary diagram is provided as Figure 5.
Phytoliths from the stems, leaves and flowers of the swamp
plant Phragmites australis and the leaves of Poa growing in the
grassy bald were extracted by chemical digestion with Schultze
solution and mounted in Naphrax (refractive index 1.73).
Sediment samples analysed for phytolith were taken from 0, 5,
10, 20, 40, 80, 120, 160, 200, 240, and 270 cm in the main core.
Phytoliths were extracted from the sediment by acid digestion of
Figure 2. (a) Location of Dandabah Swamp and the grassy bald. (b) Stratigraphic cross-section across Dandabah Swamp with core sites marked.
at James Cook University on August 10, 2014hol.sagepub.comDownloaded from
Moravek et al. 309
Figure 4. Percentages of major fossil pollen taxa and charcoal from Dandabah Swamp (core DB-3). Pollen spectra are expressed as percent of a pollen sum of 100 pollen grains from arboreal taxa and charcoal as
mm2/cm3. A complete diagram is available on request.
at James Cook University on August 10, 2014hol.sagepub.comDownloaded from
Moravek et al. 311
the organic fraction, followed by density separation following
Rovener (1971). Phytolith-bearing residues were dried and
mounted in Naphrax. A total of 200 individual phytoliths were
identified on each of the slides, following Piperno (1988) and
Ollendorf et al. (1988). Identification of fossil phytoliths was
based on modern reference material from Phragmites australis
and Poa sp. and by comparison with published illustrations
(Bowdery, 1996; Carter and Lian, 2000; Twiss et al., 1969). Pol-
len and phytolith diagrams were constructed with TILIA (Grimm,
1995). Phytolith counts are shown in Figure 6.
Three samples were submitted for radiocarbon dating at the
University of Waikato Radiocarbon Laboratory. Sample D1S
could be dated by standard 14C assay. Samples D1P and D1B
required AMS radiocarbon dating to yield ages. All samples
received the standard acid wash and NaOH pre-treatment regime
and the NaOH insoluble fraction was used for dating. Radiocar-
bon ages were calibrated (cal. BP) with CALIB version 5 (Stuiver
and Reimer, 1993). The Southern Hemisphere correction
(shcal04.14c) was applied for radiocarbon ages less than 10,000
BP. The Northern Hemisphere default calibration (intcal104.14c)
was employed for older samples (Stuiver and Reimer, 1993).
Results
Depositional environments and chronology of
sedimentation
The stratigraphy of Dandabah Swamp suggests a three-phase depo-
sitional history. Deposition of basal red clay reflects downslope
movement of basalt-derived soil and indicates instability in a cur-
rently forested, relatively stable catchment. Radiocarbon dates
(Table 1) show that this phase of deposition is pre-Holocene and
finds counterparts in Australian landscapes responding to increas-
ing effective precipitation as a result of climatic amelioration after
the LGM (Bowler et al., 1975; Genever et al., 2003; Harrison, 1993;
Kershaw, 1970, 1976). Gleying of clays above 130 cm suggests ris-
ing moisture levels and impeded drainage in the valley from about
9000 BP (cal. 10,000 BP), again consistent with early-Holocene cli-
matic reconstructions from a variety of sites in eastern Australia.
Organic deposition in a swamp or fen begins at about 80 cm (c. 2000
BP) and is marked by rising organic content of sediments, reaching
about 50% of dry weight between 40 cm depth and the surface.
These events imply rising swamp productivity, reduced flux of
sediment from the catchment, or some combination of the two.
Figure 6. Phytolith frequency in Dandabah Swamp. 200 phytoliths were identified in samples to a depth of 160 cm.
at James Cook University on August 10, 2014hol.sagepub.comDownloaded from
312 The Holocene 23(2)
Modern pollen
Pollen belonging to 49 families of higher plants were recovered
from pollen traps. Three taxa, Callitris, Casuarinaceae and Poa-
ceae were found in all pollen traps. As plants producing these pol-
len types do not grow at every site, the pollen types are considered
to be part of the regional pollen rain.
Pollen trap catches in the rainforest sites were dominated by
Araucaria (42%) and Rutaceae (26%) with minor contributions
made by rainforest taxa in Sapindaceae, Meliaceae and Araliaceae.
Non-rainforest taxa contributing pollen to the rainforest traps
included Chenopodiaceae and Poaceae, though in very small num-
bers. The vine thicket community, which is in essence a simpli-
fied rainforest, produced pollen spectra with a relatively large
number of rare pollen types. Nine rainforest taxa are included in
pollen trap catches from vine thicket sites whilst Acacia polyads
are abundant, reflecting proximity of the traps to stands of Acacia
irorata. Pollen trap catches from eucalypt communities reflect the
relatively low diversity and open character of the vegetation.
Eucalyptus and Callitris pollen dominate catches. Poaceae (10%)
and Cyperaceae (5%) are the most important non-arboreal pollen
present as befits their presence in the understorey.
The arboreal pollen spectra of pollen traps in grassy bald sites
are dominated by Callitris (90% of arboreal pollen in the Westcott
trap and 20% in the Dandabah trap) though this taxon is not pres-
ent in or near the balds. Callitris produce prodigious quantities of
well dispersed pollen (Luly, unpublished data, 1992) and the
abundance of Callitris pollen in grassland traps can be seen in the
light of high regional production of Callitris, and a lack of flower-
ing of local grasses – a surprising feature of pollen spectra from
the grassy bald pollen traps was the scarcity of Poaceae pollen.
Though Araucaria pollen occur in the Westcott pollen trap, rain-
forest pollen are gratifyingly scarce in the grassy bald pollen
traps, in keeping with published accounts (Kershaw and Hyland,
1975) which stress low pollen production and limited wind dis-
persal of the pollen produced in Australian rainforest settings.
Pollen traps set on Dandabah Swamp sampled a wide range of
pollen types from a variety of settings. Poaceae pollen were abun-
dant, reflecting the abundance of Phragmites at the trap site whilst
wetland vegetation was represented by Typha, Polygonaceae,
Ranunculaceae pollen, and Cyathea spores. The Dandabah
Swamp pollen trap sample contained rainforest elements. Arau-
caria and Rutaceae pollen were found both in the Dandabah
Swamp trap and in the adjoining rainforest pollen trap but six
other rainforest pollen types were found only in the swamp trap.
It is likely that these taxa grow in rainforest to the northwest of the
swamp and were not part of the rainforest stand in which the rain-
forest pollen trap was placed.
Modern phytoliths
All Poa and Phragmites reference samples yielded phytoliths.
The Phragmites stem produced two main phytolith types – elon-
gate and variously shaped bulliforms, both of which are consid-
ered structurally specific rather than taxon specific (Krishnan et
al., 2000; Ollendorf et al., 1988). In Phragmites leaf samples,
several different phytolith types were identified. These comprised
large numbers of chloridoid and fan bulliform shapes with smaller
numbers of bulliform, elongate, and festucoid shapes. The fan
bulliform shapes were considered diagnostic of Phragmites aus-
tralis by Bowdery (1996). Festucoid shapes have not been found
in Phragmites australis (P. communis) previously (Brown, 1984;
Ollendorf et al., 1988), and it is possible that the festucoid phyto-
liths identified in the Phragmites from Dandabah Swamp were
actually rotated chloridoid forms.
The Poa sample contained three phytolith types – numerous
festucoid phytoliths, a lesser number of dumbbell shapes, and a
few elongate shapes. This is consistent with the views of Twiss
et al. (1969), who found festucoid type phytoliths abundant in Poa
pratenis, and classed dumbbell shapes as panicoid. More recently,
Lentifer et al. (1997) reported trichome and trapezoid phytoliths
to also be present in Poa spp.
Owing to its abundance in Phragmites australis and its absence
in the Poa sp., the fan bulliform shape was considered a key phy-
tolith morphotype indicative of the presence of Phragmites at
Dandabah Swamp. The indicator phytolith for Poa sp. is the
dumbbell shape which, although not abundant, is distinctive and is
not produced by Phragmites. Poa phytoliths are likely to be less
abundant in the sediment profile than Phragmites derived forms as
not only are they less abundant in the reference plant material, but
the grassland is located beside the swamp while Phragmites grows
on it.
Fossil pollen
Four local pollen zones have been delineated in pollen diagrams
from Dandabah Swamp. Phytoliths counts are used to distinguish
between influence of the aquatic grass Phragmites and the dryland
grasses that occupy the grassy balds. Charcoal is taken to be a
proxy for fire activity.
Zone D1 280–230 cm (11,140–9280 BP; 13,200–12,900 cal.
BP). A total of 26 pollen taxa were identified in this zone, four of
them trees (Callitris, Acacia, Eucalyptus and Casuarinaceae). At
the uppermost boundary eucalypt pollen exceeds 40% of the pol-
len sum and Casuarinaceae about 60%. Herbaceous pollen in the
zone are dominated by Asteraceae (Tubuliflorae). Other herba-
ceous taxa include Haloragaceae and Liliaceae. Triglochin, the
only true aquatic pollen type, and Azolla, an aquatic fern, have
higher percentages in this zone than in any other. Charcoal is
comparatively sparse. A modest charcoal peak (reaching about 35
mm²/cm³) half way through the zone precedes a rise in Eucalyptus
percentages, which persists to the top of the zone.
Zone D2 230–200 cm (9282– ~ 7000 BP; 10,550–7500 cal. BP). A
total of 37 taxa were identified in zone D2, with five rainforest
pollen types (Syzygium, Araucariaceae, Oleaceae, Sapindaceae
and Rutaceae) appearing for the first time. The rainforest tree pol-
len percentages are very low. Eucalyptus and Casuarinaceae pollen
share dominance of the zone. Shrubs also become evident, with
Solanaceae, Chenopodiaceae, Apiaceae, and Hibbertia present,
and Acacia and Pimelia more abundant in this zone than any
other. Tree ferns become well established, with spores present in
Table 1. Results of the radiocarbon dating from Dandabah Swamp, showing the conventional age (Stuvier and Polach, 1977) based on the
Libby half-life of 5568 yr with correction for isotopic fractionation. Calibrated ages calculated with CALIB 5.0 using SHCal04 calibration set
(Stuiver and Reimer, 1993).
Sample (lab code) Depth (cm) 14C age (BP) Calibrated age (1) Calibrated age (2)
D1S (Wk12926 – Standard 14C) 90–95 2404 ± 46 2330–2460 BP 2163–2695 BP
D1P (Wk12925 – AMS) 212–220 9282 ± 57 10,289–10,490 BP 10,247–10,556 BP
D1B (Wk129F24 – AMS) 275–280 11,141 ± 87 12,951–13,115 BP 12,902–13,201 BP
at James Cook University on August 10, 2014hol.sagepub.comDownloaded from
Moravek et al. 313
every sample. Other fern spore percentages increase towards the
top of the zone. Asteraceae (Tubuliflorae) remain prominent
along with Poaceae. Pollen percentages for Typha, an emergent
aquatic, begin to rise, while Triglochin percentages decrease as
compared to zone D1, and Azolla spores disappear. Charcoal lev-
els are consistently high, exceeding 50 mm²/cm³ in each sample
from this zone, and peak close to 60 mm²/cm³.
Zone D3 200–108 cm (~ 7000–2400 BP; 7500–2340 cal. BP). The
total pollen concentration markedly increases in zone D3. Casu-
arinaceae and Eucalyptus pollen continue to co-dominate, how-
ever rainforest taxa become more diverse and abundant. Araucaria
pollen contribute up to 15% of the pollen sum in the upper half of
the zone. Other rainforest taxa present include Syzygium, Sapinda-
ceae, Rutaceae, and Oleaceae. Pollen of herbaceous taxa and ferns,
especially the tree ferns and Davalliaceae, are common. Typ ha and
Poaceae reach their maximum percentages in this zone, and Bras-
sicaceae, Asteraceae, and Liliaceae are also abundant. Tri gloch in,
ranges between 5% and 20% for most of the zone, until about 4000
BP (4300 cal. BP) after which time it drops to below 5% and
decreases to almost nothing by the top of the zone.
After a marked reduction in charcoal levels at the beginning
of the zone, levels rise to approximately 50 mm²/cm³, before
declining gradually to their minimum level for this zone of ~
20 mm²/cm³.
Zone D4 108 cm to surface (2400 BP to present; 2340 cal. BP to
present). In zone D4, inorganic sediments characteristic of lower
parts of the core give way to peat in various stages of humifica-
tion. With 73 taxa present, zone D4 has the highest pollen com-
plexity in the diagram. All of the rainforest taxa identified in this
study are present in the zone. Araucaria is the most abundant of
them, peaking at about 50% of the pollen sum. Eucalyptus pollen
levels remain static at around 30% however Casuarinaceae pollen
levels decline steadily through the zone, from 20% to 5%, their
lowest level in the diagram. Shrub and herbaceous pollen percent-
ages remain steady throughout the zone. Fern spore percentages
decline rapidly in comparison with zone D3, and remain low
through to the present day. There is a marked input of pollen from
swamp taxa such as Apiaceae and Cyperaceae, as well as of Poa-
ceae. Typha declines in zone D4, and disappears altogether around
1000 BP. Pollen of the introduced tree Pinus sp. is present in the
uppermost sample.
Charcoal levels vary between 15 and 35 mm²/cm³ in the zone.
From about 500 BP, the charcoal concentration stabilises at
approximately 20 mm²/cm³ before declining to 10 mm²/cm³ at the
surface.
Fossil phytoliths
The fossil phytolith diagram shows the phytolith numbers in
each shape class for samples from the top 160 cm of the main
core (Figure 5). Below 160 cm, phytoliths became unrecognisa-
ble and were no longer counted. This depth coincides with the
change from inorganic to relatively organic clay sediment at c.
6000 BP (~6500 cal. BP). At the base of the phytolith diagram (c.
6000 BP), there are few key phytoliths indicative of either Poa or
Phragmites. The index Poa phytolith, the dumbbell, is scarce in
all samples and is missing from many samples entirely. Festucoid
phytoliths associated with Poa are abundant throughout the core,
but fall to less than half their usual levels in samples from 0 and
5 cm depth. The key Phragmites phytolith, the fan bulliform, is
scarce in lower levels of the core, but increases slightly towards
the surface. At 10 cm, fan bulliform phytoliths increase prodi-
giously and are the most abundant phytolith type in the surface
sample.
Interpretation and discussion
Four phases can be distinguished in the pollen and phytolith
record from Dandabah Swamp. In the earliest phase (11,141–9000
BP; 13,200–12,900 cal. BP to ~ 10,000 cal. BP), open vegetation
marked by a prominent Asteraceae understorey surmounted by a
few scattered Eucalyptus trees pervade the landscape at higher
elevations, while Casuarinaceae-dominated woodlands occupied
the foothills. Callitris, often viewed as an indicator of dryness
(Field et al., 2002) was also present on the lowlands and although
the Callitris representation is low, it is considered significant
owing to the difficulty of detecting Callitris in pollen prepara-
tions. The vegetation surrounding Dandabah Swamp was likely to
be sparse at this time, as indicated by the low organic content of
sediment, erosion of sediment from catchment slopes and the fre-
quency of open-ground taxa in pollen samples. As the slopes stabi-
lised, taxa such as Acacia, Eucalyptus, Haloragaceae, Liliaceae,
and ferns were able to gain a foothold. The aquatic fern Azolla first
appeared about 10,500 BP and Triglochin became established
some 300 years later. These aquatic markers indicate the pres-
ence of an open permanent water body which persisted until
about 9000 BP (~ 10,000 cal. BP).
The second phase of vegetational development (9000–7000
BP; ~ 10,000 cal. BP to ~ 7500 cal. BP) coincides with a sudden
rise in the occurrence of charcoal particles in samples. Charcoal
levels remain high throughout this relatively short phase, reach-
ing maximum levels for the Holocene at about 8000 BP and
declining only slightly near the end of the phase, at about 7000
BP. Eucalyptus pollen during this time make up about half the
pollen sum, possibly replacing Casuarinaceae in frequently
burned lowland areas, but more likely as a consequence of expan-
sion of eucalypt forest or woodland in the swamp catchment. Vine
thicket elements (including Rutaceae, Sapindaceae, Pittospora-
ceae and Syzygium) also grew close to the swamp, suggesting that
closed canopy communities were becoming established amidst
the sclerophyll vegetation. It is not until the end of this phase that
Araucaria pollen appears, marking the development of mature
rainforest. Its scarcity prior to this time suggests that Araucaria
was confined to minor refugia, none of which was adjacent to
Dandabah Swamp.
The high charcoal levels seen in sediments at this time are
surprising. Rainforest is relatively fire sensitive, rainforest micro-
climates do not favour the spread of fire and expansion of rainfor-
est in the face of an active fire regime is unusual. The expansion
of rainforest towards Dandabah Swamp suggests that the fires
were predominantly local, propagating through flammable com-
munities but not severely disadvantaging the fire sensitive rain-
forest. In north Queensland, Kershaw (1975) argued that fire
impeded the colonisation of the Atherton Tablelands by rainforest
early in the Holocene although climatic conditions were suited to
rainforest expansion. According to Kershaw, maintenance of open
grasslands and woodlands in the face of rainforest encroachment
was a motivation for Aboriginal people to burn landscape to main-
tain the matrix of open forest and rainforest that provided a greater
diversity of resources than does rainforest alone. The ubiquity of
fire at a time of rainforest expansion on the Bunya Mountains may
indicate a similar response by Aboriginal people to environmental
change and the Holocene fires near Dandabah Swamp may also be
of anthropogenic origin.
Vegetation on the swamp began to change in this phase of
swamp history. Azolla spores disappear, and Triglochin percent-
ages decrease, suggesting that although rainfall was increasing,
the pond phase of swamp development came to an end as dense
stands of Typha became established. Ferns and dryland herbs and
shrubs including Asteraceae (Tubuliflorae), Pimelia, Hibbertia,
Solanaceae, Brassicaceae, and Plantago colonised open areas sur-
rounding the swamp basin.
at James Cook University on August 10, 2014hol.sagepub.comDownloaded from
314 The Holocene 23(2)
Between about 7000 BP and 2400 BP (~7500 cal. BP to 2695–
2163 cal. BP) fern spores and Poaceae pollen dominate the pollen
diagram. Local fires appear to have prevented tree taxa from
expanding in the immediate vicinity of the swamp and the Danda-
bah bald would have been a clearly defined feature of the land-
scape. The presence of rainforest adjacent to the swamp is evident
through the prominence of Acmena, Oleaceae, Rutaceae, Sapin-
daceae and Araucaria in the pollen diagram. With the exception of
Araucaria, rainforest tree pollen values do not increase throughout
this interval, suggesting the rainforest did not expand towards the
swamp. The increase in the Araucaria pollen may signal matura-
tion of vine thicket into rainforest or reflect gradually increasing
importance of these slow-growing, long-lived trees. Regionally,
fire may not be as prominent as it was near the swamp as rela-
tively fire-sensitive Casuarinaceae remain abundant in the pollen
record, especially in the first half of this interval. Poaceae pollen
percentages attain their maximum levels for the Holocene just
after the beginning of this phase and it is likely that this reflects
gradually increasing influence of the reed Phragmites australis in
the swamp proper. Festucoid phytoliths, belonging to grasses
such as Poa, dominate phytolith assemblages from about 4000 BP
(4300 cal. BP) onwards. The presence of the festucoid and dumb-
bell phytoliths suggest existence of a grassland near the swamp
throughout this time, whilst the swamp itself was occupied by
Phragmites australis, as indicated by presence of fan bulliform
and chloridoid phytoliths.
From about 2500 BP (~ 2390 cal. BP), at the beginning of pol-
len zone, D4, complex notophyll rainforest was fully established
on slopes around Dandabah Swamp. The representation of rain-
forest in the pollen diagram is similar spectra measured in pollen
traps, indicating that rainforest was comparable in character and
distribution to the present day. The dominance of rainforest
around the swamp coincides with a marked decrease in charcoal
levels from those seen in the previous two phases. Grassland sur-
vived adjacent to the swamp, as indicated by a strong Poaceae
pollen signal and the presence of festucoid phytoliths shapes
attributed to Poa sp. (Twiss et al., 1969) in sediment samples. The
swamp sediments take on the organic dominance characteristic of
reed swamp environments and contain pollen from typical swamp
taxa such as Apiaceae and Cyperaceae. There is a significant
decline in regional occurrence of Casuarinaceae, while Eucalyp-
tus pollen remains steady following an increase at the beginning
of this period. The decline in Casuarinaceae is not directly linked
with the effects of fire as charcoal was more abundant during
times with higher Casuarinaceae pollen levels. The reappearance
of Callitris suggests that the regional climate had dried slightly.
The uppermost sample contains pollen from introduced plants
(Pinus sp.), marking the influence of European settlement on the
mountain some 150 years ago. European fear of fire and abandon-
ment of Aboriginal land management contributed to reduced land-
scape burning and reduced flux of charcoal to the swamp. The
spectacular increase in the fan bulliform phytolith attributed to
Phragmites strongly suggests that the total dominance of swamp
vegetation by Phragmites is a very recent phenomenon.
Conclusions
At the regional scale, this sequence of changes is consistent with
patterns described throughout much of eastern Australia. Land-
scapes dominated by open woodlands with grassy understorey
were replaced by eucalypt woodland and rainforest which sur-
rounded and invaded grasslands and grassy woodlands under the
impetus of a warmer and wetter climate. The transition from open
Asteraceae-dominated vegetation to forest in the Bunya Moun-
tains is comparable with the climatically influenced transitions
from grasslands and herbfields to forest occurring elsewhere at
comparable timescales (see examples in Clark et al., 2001; Hopkins
et al., 1996; MacPhail, 1980; Miller and Halpern, 1998; Rochefort
et al., 1994; van der Hammen, 1974). Details of the change are still
to be fully explored. Zone D1 vegetation, in which Asteraceae
dominate the understorey of very sparse woodland, has similarities
with both subalpine and semi-arid landscapes in which a grass
component is prominent. The scarcity of grass very early in the
record requires further investigation, as does the mechanism by
which grass comes to dominate the balds early in the Holocene.
Increasingly active fire regimes allowed patches of Poa-
dominated grassland to persist within the increasingly wooded
landscape. The grassy bald at Dandabah represents one such
enduring remnant of pre-forest grassland and lends support to the
notion that grassy balds of the Bunya Mountains are of historical
origin, comprising a relict grassland maintained through time by
fire, much as originally posited by Webb (1964). The grassy balds
are not the product of cataclysms of human or natural causation,
though it is likely that human agency, in the form of fire manage-
ment, contributed to their survival. In this sense, it is a splendid
irony that a place selected for its natural qualities to be one of the
earliest protected areas in Queensland could best be regarded as
much of a cultural landscape as anywhere in Europe and that in
view of the complex modern relationship between forest trees
grassland and fire demonstrated by Fensham and Fairfax (2006),
very stern measures indeed may be required to maintain their
integrity into the future.
Acknowledgements
We are pleased to acknowledge financial support for this proj-
ect provided by the Australian Geographic Society. We thank
staff of the Queensland Environmental Protection Agency for
their help in the field and in provision of logistical assistance.
We thank Ursula Pietrzak at Monash University for assistance
with sample preparation and hospitality while Silvie Moravek
was resident in her laboratory. Students in the School of Earth
and Environmental Science at James Cook University provided
a willing labour force in trying conditions and we appreciate
their assistance.
Funding
The Queensland Herbarium provided funds for AMS dating.
References
Billings WD and Mark AF (1957) Factors involved in the persistence of montane
treeless balds. Ecology 38: 140–142.
Bowdery DE (1996) Phytolith analysis applied to archaeological sites in
the Australian arid zone. Unpublished PhD thesis, Australian National
University.
Bowler JM, Hope GD, Singh G et al. (1975) Late Quaternary climates of
Australia and New Guinea. Quaternary Research 6: 359–394.
Boyd WE, Stubbs BJ and Averill C (1999) The grasses of the Big Scrub district
of north-east New South Wales: A sedimentary record of late Holocene
grasslands in a subtropical forest landscape. Australian Geographer 30:
331–336.
Brown DA (1984) Prospects and limits of a phytolith key for grasses
in the central United States. Journal of Archaeological Science 11:
345–368.
Bruce RE (1988) The grassy balds of the Bunya Mountains, Southeastern
Queensland: A study of bald distribution and the physiognomic and floristic
species patterning on the balds. Unpublished Ph.D. thesis, University of
Queensland.
Butler DW (2003) Seed dispersal syndromes and the distribution of woody
plants in south-east Queensland’s vine forests. Unpublished PhD Thesis,
Department of Botany, University of Queensland.
Carter JA and Lian OB (2000) Palaeoenvironmental reconstruction from the
last interglacial using phytolith analysis, southeastern North Island, New
Zealand. Journal of Quaternary Science 15(7): 733–743.
Clark JS, Grimm EC, Lynch J et al. (2001) Effects of Holocene climate change
on the C4 grassland/woodland boundary in the Northern Plains, USA.
Ecology 82: 620–636.
at James Cook University on August 10, 2014hol.sagepub.comDownloaded from
Moravek et al. 315
Clark RL (1982) Point count estimation of charcoal in pollen preparations and
thin sections of sediment. Pollen et Spores 24: 523–535.
Coop JD and Givnish TJ (2007) Spatial and temporal patterns of recent forest
encroachment in montane grasslands of the Valles Caldera, New Mexico,
USA. Journal of Biogeography 34(5): 914–927.
Environment Protection Agency (2001) Park Guide: Bunya Mountains
National Park. Brisbane: Queensland Parks and Wildlife Service.
Faegri H and Iversen J (1964) Textbook of Pollen Analysis. New York: Hafner
Publishing Company.
Fairfax RJ, Fensham RJ, Butler D et al. (2009) Effects of multiple fires on
tree invasion in montane grasslands. Landscape Ecology 24: 1363–1373.
Fensham RJ and Fairfax RJ (1996a) The disappearing grassy balds of the
Bunya Mountains, south-eastern Queensland. Australian Journal of Bot-
any 44: 543–558.
Fensham RJ and Fairfax RJ (1996b) The grassy balds on the Bunya Mountains,
south-eastern Queensland. Floristics and conservation issues. Cunning-
hamia 4: 511–523.
Fensham RJ and Fairfax RJ (2006) Can burning restrict eucalypt invasion on
grassy balds? Austral Ecology 31: 317–325.
Fensham RJ and Skull SD (1999) Before cattle: A comparative study of Eucalyp-
tus savanna grazed by macropods and cattle in north Queensland, Australia.
Biotropica 31(1): 37–47.
Field J, Dodson JH and Prosser IR (2002) A late Pleistocene vegetation his-
tory from the Australian semi-arid zone. Quaternary Science Reviews 21:
1023–1037.
French M (1989) A History of Darling Downs Frontier. 1. Conflict on the Con-
damine: Aborigines and the European Invasion. Toowoomba: Darling
Downs Institute of Advanced Education Press.
Genever M, Grindrod JF and Barker B (2003) Holocene palynology of White-
haven Swamp, Whitsunday Island, Queensland and implications for the
regional archaeological record. Palaeogeography, Palaeoclimatology,
Palaeoecology 201: 141–156.
Grimm E (1995) Tilia and Tilia Graph. Springfield: Illinois State Museum.
Harrison SP (1993) Late Quaternary lake level changes and climates of Australia.
Quaternary Science Reviews 12: 211–232.
Herbert DA (1938) The upland savannahs of the Bunya Mountains, south
Queensland. Proceedings of the Royal Society of Queensland 49:
145–149.
Hopkins MS, Head J, Ash JE et al. (1996) Evidence of a Holocene and con-
tinuing recent expansion of lowland tropical rainforest in humid tropical
North Queensland. Journal of Biogeography 23: 737–745.
Jarrott JK (1995) History of the Bunya Mountains National Park. Brisbane:
The National Parks Association of Queensland (Inc.).
Kershaw AP (1970) A pollen diagram from Lake Euramoo, north-east Australia.
New Phytologist 69: 785–805.
Kershaw AP (1975) Stratigraphy and palynology of Bromfield Swamp, north-
eastern Queensland, Australia. New Phytologist 75: 173–191.
Kershaw AP (1976) A late Pleistocene and Holocene pollen diagram from
Lyn ch’s C rate r,no rth -eas ter n Que ensl and , Aust ral ia. New Phytologist 94:
669–682.
Kershaw AP (1997) A modification of the Troel-Smith system of sediment
description and portrayal. Quaternary Australasia 15: 63–68.
Kershaw AP and Hyland BPM (1975) Pollen transfer and periodicity in
a rainforest situation. Reviews of Palaeobotany and Palynology 19:
129–138.
Knoepp JD, Tieszen LL and Fredlund GG (1998) Assessing the Vegetation
History of Three Southern Appalachian Balds through Soil Organic Mat-
ter Analysis. Research paper SRS-13.
Krishnan S, Samson NP, Ravichandran P et al. (2000) Phytoliths of Indian
grasses and their potential use in identification. Botanical Journal of the
Linnean Society 132: 241–252.
Lentifer CJ, Boyd WE and Gojak D (1997) Hope Farm windmill: Phytolith
analysis of cereals in early colonial Australia. Journal of Archaeological
Science 24(9): 841–856.
Lindsay MM and Bratton SP (1979) Grassy balds of the Great Smoky Moun-
tains: Their history and flora in relation to potential management. Envi-
ronmental Management 3: 417–430.
MacPhail MK (1980) Regeneration processes in Tasmanian forests. Search 11:
184–190.
Mark AF (1958) The ecology of the southern Appalachian grass balds.
Ecological Monographs 28(4): 293–336.
Mill er EA and Halpern CB (1998) Effects of environment and grazing dis-
turbance on tree establishment in meadows of the central Cascade Range,
Oregon, USA. Journal of Vegetation Science 9: 265–282.
Ollendorf AL, Mulholland SC and Rapp G Jr (1988) Phytolith analysis as a
means of plant identification: Arundo donax and Phragmites communis.
Annals of Botany 61: 209–214.
Paton DM (1988) Genesis of an inverted treeline associated with a frost hol-
low in south-eastern Australia. Australian Journal of Botany 36: 655–663.
Piperno DR (1988) Phytolith Analysis: An Archaeological and Geological
Perspective. San Diego: Academic Press.
Rochefort RM, Little RL, Woodward A et al. (1994) Changes in sub-alpine
tree distribution in western North America: A review of climatic and other
causal factors. The Holocene 4: 89–100.
Rogers GM (1994) North Island seral tussock grasslands 1. Origins and land-
use history. New Zealand Journal of Botany 32: 271–286.
Rovener I (1971) Potential of opal phytoliths for use in palaeoecological recon-
struction. Quaternary Research 1: 345–359.
Scott GAJ (1977) The role of fire in the creation and maintenance of savanna in
the Montana of Peru. Journal of Biogeography 4: 143–167.
Smith JMB (1975) Mountain grasslands of New Guinea. Journal of Biogeog-
raphy 2: 27–44.
Sparshott KM (2003) Flora checklist for Bunya Mountains National Park.
Brisbane: Queensland Parks and Wildlife Service.
Steele JG (1983) Aboriginal Pathways in Southeast Queensland and the Richmond
River. Hong Kong: University of Queensland Press.
Stockmarr J (1971) Tablets with spores used in absolute pollen analysis. Pollen
et Spores 8: 615–621.
Stubbs BJ (2001) The ‘grasses’ of the Big Scrub District, north-eastern New
South Wales: Their recent history, spatial distribution and origins. Australian
Geographer 32(3): 295–319.
Stuiver M and Polach H (1977) Discussion of reporting of 14C data. Radiocarbon
19: 355–363.
Stuiver M and Reimer PJ (1993) Extended 14C database and revised CALIB
radiocarbon calibration program. Radiocarbon 35: 215–230.
Swain AM (1973) A history of fire and vegetation in northeastern Minnesota as
recorded in lake sediment. Quaternary Research 3: 383–396.
Tauber H (1967) Investigations of the mode of pollen transfer in forested areas.
Reviews of Palaeobotany and Palynology 3: 277–286.
Tinner W and Hu FS (2003) Size parameters, size-class distribution and area–
number relationships of microscopic charcoal: relevance for fire recon-
struction. The Holocene 13(4): 499–505.
Tracey JG (1982) The Vegetation of the Humid Tropical Region of North
Queensland. CSIRO Melbourne, 124 pp.
Twiss PC, Suess E and Smith RM (1969) Morphological classification of grass
phytoliths. Soil Science Society of America Proceedings 33: 109–115.
Van der Hammen T (1974) The Pleistocene changes of vegetation and climate
in tropical South America. Journal of Biogeography 1: 3–26.
Webb LJ (1964) An historical interpretation of the grass balds of the Bunya
Mountains, south Queensland. Ecology 45: 159–162.
Weigl PD and Knowles TW (1999) Antiquity of Southern Appalachian grass
balds: The role of keystone megaherbivores. In: Eckerlin RP (ed.) Pro-
ceedings of the Appalachian Biogeography Symposium. Virginia Museum
of Natural History Special Publication No. 7, Virginia, USA.
Wells BW (1963) The Southern Appalachian grass bald problem. Castanea
26: 98–100.
Whittaker RH (1956) Vegetation of the Great Smoky Mountains. Ecological
Monographs 26: 1–80.
at James Cook University on August 10, 2014hol.sagepub.comDownloaded from
