Austral Ecology (2000) 25, 333–343 Palaeo-ecology of the Gap and Coturaundee Ranges, western New South Wales, using stick-nest rat (Leporillus spp.) (Muridae) middens V. ALLEN,1 L. HEAD,1* G. MEDLIN2 AND D. WITTER3 School of Geosciences, University of Wollongong, Wollongong, New South Wales 2522, Australia (Email: lesley_head@uow.edu.au), 2Mammal Section, South Australian Museum, Adelaide, South Australia and 3National Parks and Wildlife Service, Broken Hill, New South Wales, Australia 1 Abstract Pollen, plant and animal macrofossils recovered from nine Leporillus spp. (Muridae) middens found in the Gap and Coturaundee Ranges, western New South Wales were examined. By comparing current vegetation, pollen from modern surface samples and pollen from midden samples, general vegetation characteristics over the last 6500 years BP were reconstructed. Evidence shows that a greater shrub cover was apparent between 6500 and 5200 BP, while other aspects of the vegetation cover were similar to present. An increase in tree pollen, possibly indicating greater tree cover, occurred around 3400–2600 BP, while vegetation in the last 1300 years was similar to present. These interpretations, particularly from the older samples, are tentative due to spatial and temporal limitations. Animal macrofossils from the middens indicate that several mammal species now extinct or uncommon in western New South Wales have occurred in the area in the past 3000 years. This study also confirms that reconstruction of vegetation from Leporillus spp. midden evidence should be seen as separate ‘snapshots’, rather than continuous records over a stratigraphically defined timescale. Key words: Australia, environmental change, Holocene, palynology, semi-arid zone. INTRODUCTION Reconstructing the Quaternary palaeo-ecology of arid and semi-arid areas of Australia has been limited by the lack of sites with suitable preservation, particularly of pollen. Typically, arid and semiarid pollen records have been derived from playa lakes (Singh 1981; Singh & Luly 1991; Luly 1993), spring and swamp deposits (Boyd 1990) and cave floor sediments (Martin 1973). In recent years animal middens have provided crucial evidence in many arid regions of the world (Betancourt et al. 1990). The potential for Leporillus spp. (stick-nest rat) middens to fulfil this role in Australia has been recognised for some time (Green et al. 1983; Berry 1991; Pearson & Dodson 1993; McCarthy et al. 1996). Indurated urine, known as amberat, provides a good preservation medium for pollen and plant and animal macrofossils. Detailed studies in Central Australia and the Flinders Ranges, including extensive radiocarbon dating programmes, indicate that the taphonomy of the Australian deposits is complex, and careful attention must be paid to the sources of the various types of evidence (McCarthy 1996; Head et al. 1998; Pearson 1999). *Corresponding author. Accepted for publication November 1999. In this paper we present the results of a study of nine Leporillus middens from the Gap and Coturaundee Ranges, western New South Wales, Australia. Our main aims were to understand ecological change in the area over Holocene and post-European timescales. This requires, first, an understanding of the nature and timing of midden accumulation, and the implications of this for the preserved evidence, and, second, an understanding of the modern pollen representation. THE GAP AND COTURAUNDEE RANGES The Gap Range consists of the Ravendale sandstone formation, producing cuesta landforms, gorges, bouldery scree slopes and outwash plains (Rose 1974). A block of resistant sandstone forms the Coturaundee Ranges to the north-east. The ranges rise to a maximum of 150 m above the surrounding plains (Rose 1974). Average annual rainfall in the area is 200 mm, but is highly variable, falling mostly in the winter months (Gerritsen 1977). Chenopod shrublands dominated by Maireana pyramidata and Rhagodia spinescens form the dominant community on both ranges and plains, with scattered Casuarina cristata in more protected gullies and on small rises. Ptilotus obovatus, Sida calyxhymenia, Abutilon leucopetalum, Rhagodia ulicina and Cymbopogon ambiguus occur commonly in the 334 V. ALLEN ET AL. shrublands. The hillslopes are sparsely timbered with Geijera parviflora, Myoporum platycarpum, Capparis mitchellii, Alectryon oleifolius and Flindersia maculosa. Eucalyptus camaldulensis lines the major water courses. The Gap and Coturaundee Ranges are now protected in the Mutawintji (previously Mootwingee) National Park and Coturaundee Nature Reserve respectively. METHODS Samples from five middens (GH 1–5) were collected from the main east-facing slope in the Gap Range study area (Figs 1,2) during a preliminary visit by Head and Witter in 1992. The remainder of the middens, Fig. 1. Location of the study area. including the Coturaundee middens, were found by Allen and Head in 1995, when vegetation sampling and surface pollen collections were also undertaken. These middens fall into three of six different types within the classification developed by Head et al. (1998). GH 1–5 and C7 were masses of sticks and other nesting material very weakly cemented with amberat (type 5). GH 6, GH 8 and C8 were amorphous middens composed almost entirely of amberat with little remaining nesting material intact (type 2). GHE was a combination of nesting material and solid amberat (type 3). The latter two types are extremely indurated and had to be sampled with a chisel and hammer. None of the middens had any visible stratigraphy. Every effort was made to take subsamples from the inner portions of middens, rather than the outside where modern contamination could occur. All plant species within a hundred metre radius (the approximate foraging range of Leporillus spp. (Copley 1988)) of midden sites were recorded, and their percentage cover determined along transects either side of middens. A surface sample of about 40 g of sediment, or moss cushion where available, was collected from close to each midden location to examine modern pollen representation. Indurated middens were processed by soaking and sieving according to methods adapted by McCarthy (1996). In non-indurated middens, soaking and sieving damages plant material, so these were disaggregated manually. All samples for dating were taken from inner portions to minimize contamination. Different components including wood, grass, Casuarina cladodes and faecal pellets were dated separately where possible. Sieved material was sorted under magilamp and dissecting microscope, and sent for specialist identifications where necessary. Pollen preparations were carried out using the decanted supernatant from indurated middens (McCarthy 1996) and finely sieved sediment from non-indurated ones. Samples underwent standard Fig. 2. Hillslope showing locations of GH 1, 3, 4, 5 and 6. S T I CK-N ES T R AT M I D D EN S KOH, HF, acetolysis and dehydration treatments (Moore et al. 1991) and were mounted in silicon oil. At least 200 pollen grains (or 200 of non-dominant taxa if one taxon dominated) were counted in midden samples. Due to poor preservation, surface sample counts were limited to 100 non-dominant grains. RESULTS Dating The results of 22 accelerator mass spectrometry (AMS) and conventional radiocarbon dates from these samples are shown on Table 1. Samples from middens with greater induration were generally older, with the less durable stick nests mostly dating to the last few hundred years, although a wood sample from GH 3 dated at 1300 6 40 BP (sample OZA-312 U). Multiple samples from GHE (Fig. 3) were taken to unravel the accumulation history of this large and complex deposit. While subsamples of faeces, Casuarina cladode and grass/wood within a single sample (GHE 3) all overlapped within their error ranges, those from another (GHE 5) differed by about 2000 years (1380 6 60 BP, sample Beta-91811 and 3430 6 70 BP, sample Beta89612). Further, there was considerable variability in dates from different parts of the midden, and stratigraphic anomalies with GHE 1 (5710 6 90 BP/Beta89609) lying stratigraphically above GHE 5. Table 1. 335 Modern pollen Surface pollen samples (Fig. 4) were dominated by Asteraceae and Chenopodiaceae/Amaranthaceae, demonstrating that these pollen taxa are were incorporated in middens solely by rat activity. The Chenopodiaceae/Amaranthaceae are well represented as they (particularly Maireana pyramidata, Rhagodia spinescens, R. ulicina and Ptilotus obovatus) also dominated the vegetation surveys (Table 2). Casuarina pollen is well represented in both the surface samples and vegetation, particularly in C8 which had a high proportion of Casuarina pollen and numerous trees located nearby. Asteraceae, along with Fabaceae and Hydrocotyle, were overrepresented in the modern pollen counts, as they did not occur in significant quantities in the vegetation. However, representation of ephemeral taxa was heavily dependent on recent rainfall conditions and could therefore also be variable. A regional input into the pollen rain was seen in the counts for Eucalyptus, other Myrtaceae and Pinus, which were not recorded in the vegetation surveys, although E. camaldulensis occurred along watercourses within a few hundred metres of the sites. Poaceae pollen appeared to be mostly from extralocal or regional sources, as there was little relationship between the vegetation surveys and the surface pollen counts. The most systematic study of modern pollen rain in the context of stick-nest rat midden interpretation has been carried out by McCarthy (1999), who analyzed pollen traps at both regional and local (in cave and outside cave) scales. She demonstrated that midden pollen Radiocarbon dates for midden samples Midden sample no. Conventional/AMS Material dated Date Lab no. GH 1 (base) GH 5 (i) GH 2 (top) GH 1 (top) GH 4 (top) GH 4 (base) GH 5 (ii) GH 3 (i) GH 3 (ii) GH 8 GH 6 (b) GH 6 (a) GHE 5 (i) GHE 5 (ii) GHE 2 GHE 4 GHE 1 GHE 3 (i) GHE 3 (ii) GHE 3 (iii) C7 C8 Conventional AMS AMS Conventional Conventional Conventional Conventional AMS AMS Conventional Conventional Conventional AMS AMS Conventional Conventional Conventional Conventional AMS AMS Conventional Conventional Wood Faeces Wood Organics Organics Organics Wood Twig Wood Amberat Amberat residue Amberat Faeces Grass Grass/wood Grass/wood Grass/wood Grass/wood Faeces Casuarina cladodes Wood Amberat 220 6 70 270 6 40 280 6 40 310 6 70 460 6 90 510 6 70 530 6 80 1000 6 70 1300 6 40 2680 6 60 2960 6 100 3040 6 70 1380 6 60 3430 6 70 5240 6 90 5340 6 90 5710 6 90 6360 6 100 6470 6 60 6520 6 50 780 6 80 380 6 60 Beta-59077 Beta-86306 Beta-86305 Beta-59078 Beta-59080 Beta-59079 Beta-86307 OZA-313U OZA-312U Beta-80610 Beta-97075 Beta-89606 Beta-91811 Beta-89612 Beta-91807 Beta-91810 Beta-89609 Beta-89611 Beta-91809 Beta-91808 Beta-89607 Beta-89608 336 V. ALLEN ET AL. records represent a combination of regional and local airborne pollen, together with pollen resulting from the feeding and nesting activities of the rats. Topography, aspect and exposure of midden accumulation sites are important variables. Modern pollen rain studies of arid vegetation communities need to be applied with caution to the fossil record as these communities have experienced significant post-European change, in particular due to the impact of feral animals. Midden pollen GH 1, 2, 3, 4, 6 and 8 Asteraceae and Chenopodiaceae/Amaranthaceae dominated samples GH 1–6(a) (Fig. 5). The most important minor taxa were Casuarina and Myoporaceae. GH 6(b) and GH 8 are considerably different. GH 8 had strong Eucalyptus and very low Chenopodiaceae/Amaranthaceae counts, while GH 6(b) had strong Casuarina and Poaceae counts, though low Chenopodiaceae/Amaranthaceae values. GHE As a group the GHE samples had slightly higher levels of shrub (Myoporaceae, Myrtaceae, Sapindaceae, Solanaceae) pollen than the younger Gap Range or Coturaundee samples. Chenopodiaceae/ Amaranthaceae and Asteraceae pollens were considerably lower in these samples compared with GH 1, 2, 3, 4 and 6(a). C7 and C8 C8 was dominated by Asteraceae, Fabaceae and Poaceae. C7 was dominated by Chenopodiaceae/ Amaranthaceae and Asteraceae. Santalaceae was significant in both samples. Eucalyptus representation was low considering the proximity of trees to the midden sites. Compared with similar-aged middens from the Gap Range site, these samples had reduced Chenopodiaceae/Amaranthaceae counts, but similar shrub and herb proportions. Plant macrofossils Fig. 3. Midden GHE showing location of dated samples. Fig. 4. Modern pollen diagram. Fragmentation made many of the plant macrofossils difficult or impossible to identify. Unidentifiable fibrous S T I CK-N ES T R AT M I D D EN S Fig. 5. Table 2. Midden pollen diagram. Main features of vegetation cover at midden sites Site > 20% Ground cover (dominant taxa) GH 1 Maireana pyramidata GH 3 Maireana pyramidata GH 4 Maireana pyramidata Rhagodia spinescens Maireana pyramidata Rhagodia spinescens Maireana pyramidata Rhagodia spinescens Maireana pyramidata Rhagodia spinescens Cymbopogon ambiguus Rhagodia spinescens Rhagodia spinescens Sida calyxhymenia Eucalyptus camaldulensis Abutilon leucopetalum Sida calyxhymenia Abutilon leucopetalum Maireana pyramidata Ptilotus obovatus Rhagodia spinescens Maireana pyramidata Maireana schistocarpa Rhagodia spinescens Sida calyxhymenia Solanum ellipticum Sida calyxhymenia Poaceae spp. Capparis mitchellii Sida calyxhymenia Casuarina cristata Geijera parviflora GH 5 GH 6 Flats, 50m below a hillslope Flats, 100m below hillslope Creek, 800m from hillslope GH 8 GHE Maireana pyramidata Ptilotus obovatus Dodonaea viscosa C7 Ptilotus obovatus C8 Ptilotus obovatus a 337 5–20% Ground cover < 5% Ground cover Trees Rhagodia spinescens Sida calyxhymenia Poaceae spp. Rhagodia spinescens Cymbopogon ambiguus Alectryon oleifolius Abutilon leucopetalum Sida calyxhymenia Solanum ellipticum Cymbopogon ambiguus Rhagodia ulicina Hill slope where midden samples GH 1–6 were found. Casuarina cristata Geijera parviflora Maireana schistocarpa Pintra terranea Abutilon leucopetalum Poaceae spp. Ptilotus obovatus Sida calyxhymenia Maireana pyramidata Pintra terranea Myoporum platycarpum Maireana georgei Rhagodia spinescens Solanum ellipticum Callitris collumellaris Eucalyptus camaldulensis Myoporum acuminatum 338 V. ALLEN ET A L. Table 3. Macroscopic plant remains from middens Midden number and age GH 2 top 280 6 40 GH 2 (not dated) GH 1 top 310 6 70 GH 1 outer Type and taxon Leaf Sida calyxhymenia Seed Myoporum sp. Leaf Cassinia sp. Sida calyxhymenia Seed Myoporum sp. Leaf Casuarina sp. Burr Sclerolaena sp. C8 380 6 60 Leaf Sida calyxhymenia Casuarina sp. Cassinia sp. Epacridaceae sp. Seed Myoporum sp. Phyllode Acacia sp. C8 right (not dated) Cladodes Casuarina sp. GH 4 top 460 6 90 Cladodes Casuarina sp. GH 4 base 510 6 70 Leaf Casuarina sp. Seed Myoporum sp. Phyllode Acacia ?aneura Sida calyxhymenia Seed Brassicaceae sp. Burr Dissocarpus paradoxus Leaf Sida calyxhymenia Acacia ?aneura Seed Brassicaceae sp. Burr Dissocarpus paradoxus C7 780 6 80 C7 outer (not dated) No. 1 3 1 > 20 Midden number and date GH 3 1000 6 70 / 1300 6 40 GH 3 (not dated) Type and taxon Leaf Sida calyxhymenia Casuarina sp. Seed Myoporum sp. Seed Myoporum sp. Number 1 4 5 5 7 > 20 GH 8 2680 6 60 1 4 > 20 1 1 GH 8 (not dated) Burr Daucus glochidiatus Vittadinia sp. Calotis scapigera Sclerolaena ?patenticuspis Sclerolaena ?johnsonii Burr Daucus glochidiatus Calotis hispidula Sclerolaena sp. 1 1 > 20 1 GH 6 (a) 3040 6 70 GHE 5710 6 90 Seed pod Senna sp. 1 Cladodes Casuarina sp. 3 Burrs Calotis cuneifolia 1 Flower parts Asteraceae flower fragment 1 Seeds Calotis hispidula 5 Burrs Calotis cuneifolia 6 Calotis hispidula Cladodes Casuarina sp. Leaf Casuarina sp. > 20 Cassinia sp. 14 Sida calyxhymenia > 20 Flower remnant Asteraceae sp. head 1 Cladodes Casuarina sp. > 20 1 GH 6 (b) 2960 6 100 GHE 2 5240 6 90 Cladodes Casuarina sp. > 20 GHE 4 5340 6 90 Cladodes Casuarina sp. > 20 GHE 5 1380 6 60 / 3430 6 70 4 1 1 > 20 > 20 3 > 20 2 > 20 1 Undated midden sections generally came from the outside of the midden and therefore may contain modern contamination. S T I CK-N ES T R AT M I D D EN S Table 4. Faunal macrofossils recovered from middens Midden number and age GH 1 (not dated) GH 2 (not dated) GH 3 1000 6 70/ 1300 6 40 GH 3 (not dated) GH 8 2680 6 60 GH 8 (not dated) GH 6 (b) 2960 6 100 GH 6 (a) 3040 6 70 GHE 5 1380 6 60/ 3430 6 70 C7 (not dated) C7 780 6 80 339 Type of macrofossil Probable identification Hair Hair Caudal vertebra Fine Dasyurid hairs Vertebra Vertebra Leg bones Ulna bones Caudal vertebra Upper incisor Hairs Hairs Rodent Macropus sp. Medium-sized mouse Sminthopsis sp. Medium-sized rodent Mouse-sized rodent Rattus sp. Rodent (not Leporillus) Rat-sized rodent Rattus villosissimus Macropus sp. Macropus rufus Hair Faecal pellet Hair Faecal pellet Trichosurus vulpecula Reptile Trichosurus vulpecula Reptile Numerous hairs Canis lupus dingo Faecal pellets Calcaneum and humerus bones Lower incisor Cervical vertebra Rodent hair Frontal bone Frontal bone Squamosal bone Bone Incisors First upper molars (x2) Upper molar Lower molars Trichosurus vulpecula Reptile Rodent Rodent Rodent Pseudomys sp., possibly P. hermannsburgensis (or P. bolami) Small gecko Pseudomys australis Similar to R. villosissimus Small Notomys sp. Medium-sized mouse Pseudomys bolami Rodent Notomys sp. or Pseudomys sp. Hair Hair Faecal pellets Canis lupus dingo Unidentified, possibly rodent though some structures are different Macropus robustus Lizard skull Amphibolurus sp. Undated midden sections generally came from the outside of the midden and therefore may contain modern contamination. material, sticks and twigs were the main midden components. Sida calyxhymenia leaves occurred in most middens, often in high numbers (Table 3). Casuarina cladodes were numerous in most samples, but absent from GH 2, GH 8, C7 and C8. Burrs from members of the Chenopodiaceae occurred in a number of samples and were usually identifiable to genus level. Acacia phyllodes (possibly A. aneura) were present in C7 and C8, but none of the Gap Range samples. Animal macrofossils Eleven of the midden samples contained mammalian or reptilian remains (Table 4), and at least one part of every midden contained insect fragments (Table 5). Most of the midden samples contained Leporillus or rodent faecal pellets. Macropus sp. remains were recovered from GH 1, GH 3 and C7. Hairs belonging to Sminthopsis sp. were recovered from GH 3. Hair and faecal pellets of Trichosurus sp. (probably T. vulpecula based on known distribution and habitat preference) were recorded in GH 8 and GH 6, respectively. Hairs belonging to the dingo (Canis lupus dingo) were recorded in GH 6(b), and GHE 5. Reptilian faecal pellets were recorded in GH 8 and GH 6(b). Many of the teeth and bone fragments were only identifiable as ‘rodent’. GH 6(a) contained 340 V. ALLEN ET AL. Table 5. Insect remains from middens Midden sample Type Taxa GH 1 GH 4 Beetle fragments Beetle fragments Beetle fragments Pupal case Beetle fragments Beetle fragments Mandible Order: Coleoptera Order: Coleoptera, possibly family Buprestidae (jewel beetles) Possibly families Buprestidae, Cerambycidae Order: Diptera Families: Dermestidae,Curculionidae (weevils), Buprestidae and colleagues Family Cerambycidae Family Gryllacrididae (raspy crickets) C7 GH 8 GH 6 bones of the long-haired rat (Rattus villosissimus), the plains rat (Pseudomys australis), one of the smaller hopping-mice (Notomys sp.) and two upper first molars of Bolam’s mouse (Pseudomys bolami). Hair, possibly from P. hermannsburgensis (or P. bolami) was also found here. An upper incisor consistent with the long-haired rat (Rattus villosissimus) was recorded from GH 3. A complete lizard skull of the genus Amphibolurus was recovered from midden C7. DISCUSSION Timing and nature of midden accumulation Where it has been possible to date the top and base of single deposits, our results indicate that stick-nest accumulation was relatively rapid. Both GH 1 and GH 4, whose radiocarbon dates overlap within their error margins, could have accumulated in a single episode of nest-building. Although plant macrofossils indicated that GH 3 formed over a period of up to 300 years, this could be much less if older twigs were incorporated into a nest structure. This contrasts with studies of middens with much higher proportions of amberat, for which sporadic accumulation can occur over periods of about 600 years (Pearson & Dodson 1993) and 1000 years (McCarthy 1996). Such studies, using detailed dating of different components, have also shown that stratigraphic integrity cannot be assumed, even in freestanding Leporillus middens with distinct layering (Head et al. 1998). Our results from GHE, albeit not a vertically layered midden, illustrate this point. GHE fills the full height (about 40 cm) of the back of a well-protected rock ledge, extending 180 cm along the ledge (Fig. 2). Of the eight dated samples, six date between 5200 BP and 6500 BP, suggesting sporadic accumulation of the midden, not necessarily vertically formed, through a 1300-year period. The three dated subsamples of GHE 3 (Table 1) all overlap within their error ranges, indicating the integrity of this sample. Sample GHE 5 is more problematic. It not only has a dating discrepancy of 2000 years between faeces (1380 6 60 BP) and grass (3430 6 70 BP) components, but is apparently overlain by sample GHE 1 (5710 6 90 BP). Comparison of dates from paired leaf and faeces samples from Flinders Ranges sites shows that faecal pellet dates are most likely to be stratigraphically inconsistent (Head et al. 1998). This is consistent with historical descriptions of the rats themselves moving through the structure they have built without fundamentally disturbing it. We therefore argue that the pollen and macrofossil evidence from GHE 5 is more likely to be around 3400 years old, but this is a tentative interpretation. GHE suggests also that complex erosion, reworking and/or reoccupation of deposits should always be considered. Extrapolation from dated to undated parts of middens is potentially inaccurate, and pollen and macrofossil evidence should be interpreted in ‘snapshots’ of time, rather than as a continuous record. Holocene environmental change In keeping with our interpretation of this evidence providing palaeo-ecological ‘snapshots’ rather than a continuous record, we discuss environmental change in three temporal windows. 5200–6500 BP Non-chenopod shrubs, particularly Myrtaceae, Sapindaceae and Solanaceae, were more important in the vegetation than they are today, while tree cover and species composition appears to have been similar to today. Herbs including Asteraceae, while important, appear to have been less abundant than today. Casuarina cladodes are the most abundant plant macrofossil in these samples, and Casuarina still occurs near the site. It is difficult to separate spatial and temporal variability on the basis of these samples, which all come from one site, GHE, which has no young samples. For example, amounts of Chenopodiaceae/ Amaranthaceae pollen are lower in all these samples than in those less than 1300 years old, suggesting a possible expansion of the Chenopod shrublands in the late Holocene. However, the GHE surface sample has the lowest level of Chenopodiaceae/Amaranthaceae S T I CK-N ES T R AT M I D D EN S representation of all GHE samples, so low levels of this pollen may partly be due to the fact that that it is a more protected site, with more influence from surrounding trees of other families. 2600–3400 BP GHE 5, with its dating uncertainty, has a very similar pollen signal to the other older GHE samples. However, the two GH 6 samples and GH 8 indicate the highest tree representation of any of the samples. The combination of pollen and animal macrofossil evidence suggests the possibility that Eucalyptus trees were closer to the Gap Range than they occur today. GH 8 and GH 6(b), particularly the former, have the highest Eucalyptus pollen values of any samples, including modern ones. Although the pollen data alone do not provide clear evidence, since Eucalyptus pollen representation in this context is highly variable (McCarthy 1999), these are also the two samples in which T. vulpecula hairs and faecal pellets were found. T. vulpecula does not require Eucalyptus species for nesting hollows, and can use termite mounds and rock-holes for den sites if available, but it generally prefers a Eucalyptus component in its diet (Kerle 1984) providing some evidence for Eucalyptus spp. occurring closer to the range than they do today, probably by expanding out from the creeks several hundred metres away. < 1300 BP High levels of Chenopodiaceae/Amaranthaceae and Asteraceae counts in middens younger than 1300 years and surface samples suggest structurally similar vegetation to present, with chenopod and other shrublands and desert herbfield the most important regional vegetation communities. The differences between the Gap Range and Coturaundee samples are generally consistent for both fossil and modern pollen counts, reflecting greater exposure of the Gap Range to the regional Chenopodiaceae/Amaranthaceae pollen input from the surrounding plains (McCarthy 1999, chapter 5), notwithstanding some contribution to the Chenopodiaceae/Amaranthaceae and Asteraceae components from the rats themselves. 341 found in several of the midden samples, indicating either some form of utilisation of the middens by these taxa or collection of the material by Leporillus spp. for the nest building process. If other species were using the middens, this may explain some of the discrepancies relating to contamination and inconsistencies with the stratigraphy. Of the middens which were dated and strongly indurated (types 2 and 3), bone material, hair and faecal pellets were found in GH 6(a), GH 6(b), GH 8 and GHE 5. Since the bone occurred within the amberat, it is reasonable to assume that the enclosed faunal remains predate European occupation and were not there as a result of recent contamination. The results from Table 4 indicate at least Pseudomys australis, P. bolami, Rattus villosissimus, Notomys sp. (smaller species), Trichosurus vulpecula and Canis lupus dingo were contemporaneous with Leporillus spp. The lizard skull found in C7 is more difficult to place in the timescale, since it came from a type 5 midden where there is some potential for younger material to fall into the mass of sticks, despite the age for the midden (780 6 80 BP) derived from radiocarbon dating. The rodents Rattus villosissimus, Pseudomys australis and P. bolami (as P. hermannsburgensis) have been recorded only once before at Mutawintji National Park as subfossil evidence (Dickman 1993; Ellis 1995). In New South Wales, P. australis is considered extinct while P. hermannsburgensis and P. bolami are considered rare and R. villosissimus is considered sparse (Dickman 1993). Trichosurus vulpecula is a relatively rare animal in far western New South Wales and Dickman et al. (1993) list it as being rare and possibly endangered in western New South Wales. Trichosurus vulpecula has not been recorded before at Mutawintji National Park; however, anecdotal evidence suggests that it was once relatively common in the area. Further evidence for a decline in T. vulpecula populations in arid environments (Evans 1992; Pearson & Dodson 1993) indicates that the range of this generally common species may have declined significantly through large areas of Australia since European settlement. Post-European vegetation change The macrofossil faunal record Most of the bone of small mammals in the middens probably came from pellets regurgitated by barn owls (Tyto alba) sharing the same caves and rock shelters with Leporillus spp. Since T. alba can forage over distances of 5–10 km or more in the arid zone, the source area of this bone material can be extensive. Evidence (hair and faecal pellets) of the brushtail possum (Trichosurus vulpecula), macropods and reptiles was Semi-arid and arid rangelands have suffered considerably since the introduction of feral and domestic grazing animals. Working from geomorphological evidence, Fanning (in press) argues that removal of landcover has been the principal cause of the current phase of valley floor gullying and associated land degradation in the arid rangelands of western New South Wales. It is difficult to relate our very general pollen evidence to this argument for two reasons. First, it is possible that degradation of the biocrust, a change which is not 342 V. ALLEN ET AL. identified in the pollen record, was more significant in the process of erosion than, for example, removal of grasses. Second, the broad classification of pollen types may be masking considerable vegetation change at lower taxonomic levels. The fundamental changes in vegetation may have occurred at the genus and species level rather than the broader family or structural levels (e.g. trees, shrubs and herbs). Without more refined pollen identification and better macrofossil records, it is impossible to assess the impact of grazing by feral and domestic animals on the vegetation at Mutawintji. CONCLUSIONS These midden samples provide a sensitive local pollen record, picking up some minor taxa that would not normally be represented palynologically. The result is a spatially variable picture of past vegetation. Shrubby vegetation in samples dated between 5200 and 6500 years is consistent with that period falling within the wettest part of the Holocene, although this wet period affected vegetation differently in different parts of the arid zone (Singh 1981; McCarthy et al. 1996). Our finding that vegetation since 1300 BP was similar to present is also consistent with the results of McCarthy et al. (1996) from the Flinders Ranges . The possible increase in tree cover around 2600–3400 BP needs to be explored in further research. Continued careful attention to taphonomic issues, and taking full account of spatial and temporal variability, are crucial to palaeo-ecological interpretations of Australia’s expanding Leporillus midden data set. ACKNOWLEDGEMENTS This research was funded by the Australian Research Council, the Quaternary Environments Research Centre, University of Wollongong and the Australian Institute of Nuclear Science and Engineering. We thank the NSW National Parks and Wildlife Service and the Mutawintji Land Council for permission to undertake the project and providing support in the field. J. Atchison, L. McCarthy, F. Bates, W. and M. Riley, G. O’Donnell, B. Bates, R. McKinnon and J. Collins assisted in the field. Barbara Triggs, Linda Gibson and employees of the Sydney Herbarium assisted with identifications. REFERENCES Berry S. L. (1991) The potential of fossil mammal middens as indicators of vegetation history in Central Australia. Aust. Bot. 39, 305–13. Betancourt J. L., Van Devender T. R. & Martin P. S. (1990) Introduction. In: Packrat Middens: the Last 40 000 Years of Biotic Change, (eds J. L. Betancourt T. R. Van Devender & P. S. Martin) pp. 2–11. The University of Arizona Press, Tuscon. Boyd W. E. (1990) Quaternary pollen analysis in the arid zone of Australia: Dalhousie Springs, Central Australia. Rev. Palaeobot. Palynol. 64, 331–41. Copley P. (1988) The Stick-Nest Rats of Australia: a Final Report to the World Wildlife Fund (Australia). Department of Environment and Planning, Adelaide. Dickman C. R. (1993) The Biology and Management of Native Rodents of the Arid Zone in New South Wales. NSW National Parks and Wildlife Service, Hurstville. Dickman C. R., Pressey R. L. & Lim L. (1993) Mammals of particular conservation concern in the Western Division of New South Wales. Biol. Conserv. 65, 219–48. Ellis M. (1995) A discussion of the large extinct rodents of Mootwingee National Park, western New South Wales. Aust. Zool. 30, 39–41. Evans M. (1992) Diet of the Brushtail Possum Trichosurus vulpecula (Marsupialia: Phalangeridae) in Central Australia. Aust. Mammal 15, 25–30. Fanning P. C. (2000) Recent landscape history in arid western new South Wales, Australia: a model for regional change. Geomorphology (in press). Gerritsen J. (1977) Mootwingee – the Rockholes. T. Nicholls & Sons, Broken Hill. Green N., Caldwell J., Hope J. & Luly J. (1983) Pollen from an 1800 year old stick-nest rat (Leporillus spp.) midden from Gnalta, western New South Wales. Quaternary Australasia 1, 31–41. Head L., McCarthy L., Quade J., Witter D., Allen V. & Lawson E. (1998) Classification and radiocarbon dating of Leporillus nests in semi-arid Australia and palaeoclimatic implications. Palaeoclimates 3, 161–77. Kerle J. A. (1984) Variation in the ecology of Trichosurus: Its adaptive significance. In: Possums and Gliders (eds A. Smith & I. Hume) pp. 115–28. Surrey Beatty & Sons, Sydney. Luly J. G. (1993) Holocene palaeoenvironments near Lake Tyrrell, semi-arid north-western Victoria, Australia. J. Biogeogr. 20, 587–98. Martin H. A. (1973) Palynology and historical ecology of some cave excavations in the Australian Nullarbor. Aust. J. Bot. 21, 283–316. McCarthy L. (1996) Stick-nest rat (Leporillus spp.) middens from the Flinders Ranges, South Australia: do they stand the test of time? Ninth International Palynological Congress Abstract p. 101. Publisher?, Houston, Texas. McCarthy L. (1999) A Holocene vegetation history of the Flinders Ranges South Australia: evidence from Leporillus spp. (Stick-nest rat) middens. PhD Thesis, School of Geosciences, University of Wollongong, Wollongong. McCarthy L., Head L. & Quade J. (1996) Holocene palaeoecology of the northern Flinders Ranges, South Australia, based on stick-nest rat (Leporillus spp.) middens: a preliminary overview. Palaeogeog. Palaeoclimatol. Palaeoecol. 123, 205–18. Moore P. D., Webb J. A. & Collinson M. E. (1991) An Illustrated Guide to Pollen Analysis. Hodder and Stoughton, London. New South Wales National Parks and Wildlife Service (1989) Mootwingee National Park, Mootwingee Historic Site, S T I CK-N ES T R AT M I D D EN S Coturaundee Nature Reserve: Plan of Management. National Parks and Wildlife Service, Hurstville. Pearson S. (1999) Late Holocene biological records from the middens of stick-nest rats in the central Australian arid zone. Quaternary Int. 59, 39–46. Pearson S. & Dodson J. R. (1993) Stick-nest rat middens as sources of palaeoecological data in Australian Deserts. Quat. Res. 39, 347–54. Rose G. (1974) 1: 250 000 Geological Series, explanatory notes 343 WHITE CLIFFS (sheet SH 54–12), Department of Mines, Sydney. Singh G. (1981) Late Quaternary pollen records and seasonal palaeoclimates of Lake Frome, South Australia. Hydrobiologia 82, 419–30. Singh G. & Luly J. (1991) Changes in vegetation and seasonal climate since the last full glacial at Lake Frome, South Australia. Palaeogeog. Palaeoclimatol. Palaeoecol. 84, 75–86.