Habitat and post-fire successional preferences of the Pilliga Mouse (Pseudomys pilligaensis).

By David C. Paull¹ and Sally J.Townley²

¹Zoology, School of Biological Sciences, University of New England, NSW, 2351

²19 Drummond Ave, Armidale, NSW, 2350

Abstract

Habitat preferences were assessed for P. pilligaensis from a trapping program in Pilliga East State Forest between 1993 and 1999. This species is able to tolerate a wide variety of habitat types (9 out of 14 sampled) though highest numbers were found in just 4 types; moist gullies in recently burnt scrub; regenerating Broombush Melaleuca uncinata scrub (1-2 years after fire); mature Broombush scrub; and mature Kurricabah (Acacia burrowii)-Bloodwood (Corymbia trachyphloia) scrub/woodland. The results indicate that P. pilligaensis select a variety of vegetation successional stages from recent (less than six months old), to mature scrub (>15 years old). Microhabitat factors were not quantified though all of the above mentioned types are characterised by possessing a high floristic diversity, a well developed low shrub cover (30-80%) dominated by mytaceous and heathy shrubs, a moist groundstorey layer and a sandy soil substrate of a depth greater than 20 cm. This is a species with a very low density except in areas where over-wintering congregations of this species occur.

Introduction

The Pilliga Mouse (Pseudomys pilligaensis) is a conilurine murid with a relatively restricted range from the Pilliga region of the north-west slopes of New South Wales. It was described as recently as 1980 (Fox and Briscoe 1980), its systematics were investigated by Briscoe et al. (1981) who found it to be most closely related to other Pseudomys from eastern Australia, particularly P. novaehollandiae and P. deliculatus. Apart from its systematics and reproductive behaviour (Fox and Briscoe 1980), little is known about this species. Prior to the present study (Paull, in prep.) of which this paper forms part, fewer than 30 wild individuals had been trapped (Townley 1998). In an effort to identify habitat preferences for P. pilligaensis, a trapping program targeting a range of habitats was carried out in the Pilliga State Forests between 1993 and 1999. An extensive wildfire through the study area in December 1997 enabled a study of the responses of this species to a pyric succession to be undertaken. P. pilligaensis is listed as a threatened species (Threatened Species Conservation Act 1995) due to its very restricted range, apparently confined to an area less than 500 000 ha in the Pilliga region, and because very little was known about its ecology. It is considered a rare species (Paull and Date 1999).

Methods

Study area and habitats

Pilliga East State Forest is a large forest, woodland and scrub remnant, contiguous with Pilliga West, the Pilliga Nature Reserve and other State Forests to the south. Along with several smaller State Forests in its immediate vicinity, Pilliga East State Forest covers an area of over 200 000 ha, with the whole Pilliga covering an area of approximately 500 000 ha. The study area is a relatively small section of this forest (`20 000ha) though it contains a mosaic of different habitats, mostly ironbark/cypress/oak forests, mixed eucalypt woodlands, a Broombush-dominated scrub, mallee and other Acacia-dominated scrubs. There are many different associations within each of these types, often occurring together over very small areas. Why such a level of high habitat heterogeneity is found here may be related to a combination of geomorphic, climatic and pyric factors.

The habitats of the Pilliga overlays a very old, Jurassic sandstone, much of which is exposed as low ridges and eroded gullies. The soils are derived from non-marine alluvial sediments which have covered up much of the sandstone bedrock. The soils are generally of two main types. One is a duplex clay soil with a sandy A horizon overlaying a solid clay B horizon or rocky substrate.. The depth of the sandy horizon can vary enormously from 5-100 cm and the colours from grey, brown to red and yellow. Each variation consistently supports different vegetation types. These soils are well leached and nutrient poor, the clay horizons trap much of the water for plant growth. The other main soil type found are deep earths and loamy soils, with less clay in the lower soil strata, formed by the fluvial transport from the upper Darling drainage. Their colour also varies from black to light brown.

The source of sandy sediments is almost certainly from the low sandstone ridges in the south east section of the Pilliga, deposited over geological time, these soils are high in silica and relatively nutrient-poor (Hart 1990). Today the whole of the Pilliga is drained by a complex system of intermittent or deep drainages which flow north into the Namoi River. The creek beds that cut through the Pilliga are usually dry except after rains, although much of the rainfall is retained as water in the extensive clay dome.

A steady-state in the vegetation communities appears to be regulated by water-fire interactions in a similar way to the low-nutrient savannas of South America (Medina and Silva 1990). This was borne out during the study when a large wildfire swept through the study area in December 1997, one of the most extensive wildfire on record burning out an area of around 150 000 ha. This event was followed by high levels of rainfall the following winter resulting in extensive flooding across most of the study area. What was remarkable from these stochastic events was the rate of recovery in the vegetation, which was rapid, even in the flat scrub plains. However, these two processes had different effects on the vegetation. Fire tended to promote the re-sprouters with rapid growth following the burn. It also promoted a high germination from out of the waiting seed-bank. Within several months, an enormous diversity of plants had appeared in the burnt out areas. Flooding, on the other hand, tended to kill or remove most of the groundstorey plants and low shrubs and promoted the quick growth of grasses.

Habitats were classified to type according to their floristics and a structural classification scheme as presented in AUSLIG (1990). 14 habitat types were identified, though some of these represent different pyric stages of vegetation. In terms of their extent in the study area, the five most important habitats are the Narrow-leaf Ironbark (E. crebra)-White Cypress (Callitris glaucophulla)-Bull Oak (Allocasuarina luehmanii) forest, Broad-leaf Ironbark (E. fibrosai)-Black Cypress (C. endlicheri) shrubby woodlands, Red Gum (E. blakelyi) complex and mixed eucalypt types, Broombush (Melaleuca uncinata) scrub and Bloodwood (Corymbia trachyphloia)-Kurricabah (Acacia burrowii) complex.

Trapping effort

A total of 10 040 trap-nights were conducted in 14 habitats of Pilliga East and adjacent State Forests. One survey period was undertaken between 1993-94, a second between 1997-98 and a third in 1999. All surveys used a consistent trapping design, by placing the traps in series along transects for at least 100 metres.

During the 1993-94 study period 2480 trap nights were undertaken, using only standard A type Elliott traps over 84 transects. During 1997/98, 4780 trap nights were undertaken using both standard Elliott traps and pit-fall traps. More repeated surveys were conducted over 75 transects, each containing either ten Elliott or ten pit-traps in habitats of different age since fire. Both types of trap were baited with the same oats/peanut butter/sultana mixture, the pit-traps were 40-60 cm deep by 9-12 cm wide and used mostly without drift-fences. Both types of trap were again used in 1999, when 2430 trap nights were undertaken across 27 transects treated according to a pyric succession in Broombush scrub. Four pyric stages were tested, one < 6 months old (recent), another 1-2 years old (regrowth), one 5-10 years old (intermediate age) and the other in vegetation more than 15 years old (mature), the age it normally takes for mature Broombush scrub to form based on field observations of this plant community. The high rainfall in the months following the wildfire compromised the fire treatment used in this study in the context that the results more accurately reflect the combined effect of rain and fire upon the study animals, rather than fire in isolation. Access to the study area was not possible between 6 months and 1 year after the wildfire event due to bad road access conditions, caused by widespread flooding.

When the capture rates for both methods from all trapping data were compared, it was found that the pit-traps achieved an slightly better capture rate than the Elliott traps though the difference was insignificant (Table 1). This data indicates that P. pilligaensis were trapped equally well in both sorts of traps, good capture rates seem to be entirely dependent upon the presence of local populations. A similar capture rate to that of the pit-traps was achieved by using Longworth traps, which consist of an entrance tunnel leading into a separate nest box, however trap effort using this type of trap was too low for statistical comparison.

Results

Nine habitat types were identified as being used by P. pilligaensis. Habitat preferences were determined by comparing the pooled capture rates found in each type. The study resulted in a total capture of 176 individuals (Table 2). The habitats with the highest capture rates were (Figs 1-4): moist gullies in areas recently burnt by wildfire (15%); 1-2 year-old Broombush regrowth (8.1%); mature Broombush (Melaleuca uncinata) scrub and associated heathland (3.1%); and Kurricabah/Bloodwood (Acacia burrowii/Corymbia trachyphloia) scrub/woodland (3.4%). Consistent features of the latter three habitats were: a relatively high plant species richness; a 30-80% shrub cover at a height of <100 cm; a moist groundcover of plants, litter and fungi; and a sandy upper horizon of greater than 20 cm depth. The more common understorey species found in the mature Broombush associations include Micromyrtus spp., Calytrix tetragona, Acacia triptera, Dampieria lanceolata, Hibbertia sericea, Eriostemon ericifolius, Westringia cheelii with M. uncinata, Eucalyptus dwyeri and E. viridis as overstorey species. The mature Bloodwood-Kurricabah association understorey was dominated by Philotheca salsifolia, Prostanthera leichardtii, Phebalium glandulosum, Cassinia spp., Hibbertia riparia with, Acacia burrowii, Corymbia trachyphloia and Eucalyptus fibrosa as overstorey species.

Low to moderate abundances of P. pilligaensis were found in: mixed eucalypt/Callitris spp. shrubby woodlands; thick Acacia tindaleae scrub with Narrow-leaf Ironbark (Eucalyptus crebra) overstorey; intermediate age (5-10 years old) Broombush scrub; pure stands of Green Mallee (E. viridis) scrub; and flat scrub areas which had been burnt less than 6 months ago, with little or no vegetation cover.

P. pilligaensis was not trapped in Narrow-leafed Ironbark /Cypress Pine/Oak (E. crebra-Callitris-Allocasuarina leuhmanii) forest and Crowned Wattle (Acacia tindaleae) scrub. These habitat were characterised as having a poor low shrub layer with high levels of grass, hard clay substrate or a low plant species richness. They were also not trapped in Broad-leaf Ironbark (E. fibrosa) woodland or rocky outcrops, perhaps partly due to a lower sampling effort in these types, or in mature riparian or gully woodland.

The preference of P. pilligaensis for different pyric stages of Broombush were assessed by pooling results from 1998 and 1999 (Fig 5). Each age-class represented vastly different structural stages of the vegetation. Based on capture rates, to account for the differences in trap effort in the four stages, the results show a clear preference for the early and later stages of vegetation, with a significant avoidance for recent post-fire and intermediate age Broombush.

The pre- and post-fire responses of P. pilligaensis at two sites are shown in Figure 6. One is from a gully and the other is from a community of near-mature Broombush scrub burnt by the December 1997 wildfire. P. pilligaensis were absent from both sites prior to the fire, the gully site showed a quick colonisation after the fire, by 3 months a congregation of mostly females had gathered in and around the shallow creek-line. No trapping was undertaken 6 to 12 months after the fire, though when trapping commenced again in December 1998, no P. pilligaensis were trapped.

The other site showed a different pattern of colonisation, with only Mus domesticus present at the site in the six months following the fire. In December 1998, no Mus were trapped, but one adult male Pseudomys was found. Trapping during the Autumn and Winter, saw the number of animals caught at the site (approximately 3 hectares) had climbed to 22 adults and juveniles with a fairly even sex ratio. Subsequent trapping at the site saw a spring decline with only two breeding females caught, along with a breeding female Mus.

17 Mus domesticus were trapped in the same areas which were trapped for P. pilligaensis, eight of these were caught straight after the fire in burnt scrub, three were caught in 1-2 year old Broombush regrowth, three were caught in over-wintering congregations of Pseudomys and six were caught in spring Pseudomys breeding areas.

Discussion

Like other Australian Pseudomys, P. pilligaensis may be regarded as an early post-fire colonist (reviewed by Friend 1993). This species is well adapted to survive even the hottest fires, with its habit of burrowing at depths around 20 cm below the soil (Paull, unpbl. thesis). However the need to find vegetation cover and food seems to have driven P. pilligaensis out from the recently burnt flat scrub plains, with only one capture out of 490 trap nights recorded from trap-lines placed in burnt scrub from 6 weeks to 6 months after the fire, despite the good rains which also fell during this period.

When access was gained again to the study area in December 1998-Februaury 1999, only two mature P. pilligaensis were captured at these transects. However by April, 1999, populations at these same transects had then become established. P. pilligaensis capture rates then peaked in regrowth Broombush when it was around 14-16 months old, with numbers staying high through the winter. Some trap nights resulted in 50% capture rates and recapture rates were at 12% (Paull, unpbl. thesis).

Like the closely related species, P. novaehollandiae (Fox and McKay 1981), P. pilligaensis responds well to early regenerating vegetation, though with an apparent earlier peak in numbers than the 2-3 years described for P. novaehollandiae. In these over-wintering grounds, congregations of P. pilligaensis consisted of both non-reproductive adults and the cohort of animals probably born the previous spring. Subsequent trapping during the following Spring revealed a substantial decline in numbers with a dispersal of most males and occupancy by reproductive females from the previous years juvenile cohort (Paull, unpbl. thesis). So like P. novaehollandiae, breeding did not commence in the newly burnt scrub lands until nearly two years after the wildfire event.

P. pilligaensis did decline significantly in the intermediate-age class, with a capture rate of only 0.6%. This may in part be attributable to a decrease in the low shrub cover in this age class where the dominant Broombush shrubs have gained a height of two metres where it forms a dense shrub layer. Understorey shrubs may be deprived of light or may not be able to compete with the dense growth of Broom. The few animals caught in these areas were never recaptured and were nearly all juveniles or males (Paull unpbl. thesis).

It must be remembered that these capture rates are the result of the combined effect of fire and the subsequent high level of rainfall. Rainfall has been shown to be important for population growth in other arid Pseudomys, particularly P. deliculatus (Braithwaite and Brady 1993) and P. hermannsburgensis (Masters 1993; Southgate and Masters 1996). Rain seems to be very important for P. pilligaensis as it speeded the regrowth in the burnt scrub, and provided the groundstorey with increased humid conditions, which persisted in the study area until the time of this writing, two years later. Moist ground and subterranean conditions were a consistent factor at all sites where P. pilligaensis was caught. Moist gullies with a lush grassy growth may offer important refuges for P. pilligaensis within areas recently burnt by fire. This was the only habitat selected by this species not covered with an extensive shrubby understorey, though straight after the wildfire, there were few areas in the burnt out sections of the study which possessed any kind of shrub layer or ground moisture. No P. pilligaensis were trapped in the same gully before the fire or one year later. As well, no P. pilligaensis were caught at any of the mature gully sites.

High capture rates were also encountered in mature scrub. Both over-wintering congregations and breeding females were trapped in mature Broombush scrub, while only over-wintering congregations were found in Bloodwood-Kurricabah scrub. Both types are probably important refuge and resource rich habitat for P. pilligaensis with extensive litter cover, a mesic groundstorey of ferns, moss and lichens, a deep sandy substrates of at least 50 cm, a diverse low shrub layer with a cover of about 50-80% and an overstorey of tall shrubs with a low cover and clumps of low eucalypts. This selection for a high density of understorey of heathy and myrtaceous shrubs with a high species richness has also been identified in a number of other Australian Pseudomys (Posamentier 1976; Fox 1978; Cockburn 1978; Cockburn et al. 1981a, b; Fox and McKay 1981; Fox 1982; Fox 1990; Wilson 1991; Higgins and Fox 1993).

The habitats most selected by this species appears to be at both ends of the successional spectrum with an avoidance of taller regenerating vegetation, making Broombush scrub probably sub-optimal for P. pilligaensis when it is aged between about 5-15 years old. It can be concluded that in mature vegetation, understorey conditions improve again for colonisation by this species. This is not a pattern that has been recorded in other Pseudomys, only two other Pseudomys are known to inhabit mature or old associations, P. shortridgei (Baynes et al. 1987) and P. occidentalis (Kitchener 1995) both from Western Australia. The selection pattern shown by P. pilligaensis suggests that it is making its habitat selection based upon density of groundcover and understorey cover which may become suitable at different times in the succession. The presence of logs is not a prerequisite for this species as it is able to use habitats without trees or logs by constructing burrows into the sandy substrate (Paull unpbl. thesis), usually under low bushes.

It shares these habitats with sympatric resident populations of Antechinus flavipes and Sminthopsis murina. The larger A. flavipes was not common in areas which are selected by P. pilligaensis, except for the occasional adult male. This species was more common wherever there was rock cover, where females congregate or in forest and woodlands with logs. The Common Dunnart S. murina was seems to have an even wider habitat selection than P. pilligaensis, it is common in both burnt areas, in intermediate age and mature habitats, and in woodlands, scrub and heath.

The other species caught Mus domesticus has only been caught in the burnt scrub habitats surveyed, where it is sympatric with P. pilligaensis. The highest numbers of this species were recorded straight after the wildfire (Fig. 3), they declined at the time P. pilligaensis numbers had increased 1-2 years after the fire though some remained in areas which were inhabited by over-wintering congregations of P. pilligaensis. M. domesticus is not a common species in the study area though is probably widespread. Evidence to date suggest that it only becomes more common immediately after fire.

Previous study on the habitat selection of P. pilligaensis was undertaken by Lim (1992 unpbl.) in the Pilliga Nature Reserve, who found a similar selection for habitat with a high shrub cover, high plant species richness and a sandy substrate. The habitats where P. pilligaensis were detected were sandstone ridge with open woodland or mallee and in E. fibrosa/C. trachyphloia woodlands with a dense heath layer. This species is still being caught at some of these same locations, as well as in rocky outcrops with a tall heath and woodlands with a very sparse understorey (Stuart Green, pers. comm., 1999) which is not entirely consistent with the results of this study. Fox and Briscoe (1980) reported this species to be in low densities in heathy scrub and in woodlands of various associations with a sparse understorey.

Conclusion

Given the wide habitat selection by this species, P. pilligaensis cannot be regarded as a habitat or even a pyric stage specialist. It is more likely to be limited in its habitat selection by other microhabitat factors, such as a gradient in the cover of low shrubs and depth of the sandy substrate, perhaps selecting a particular suite of shrubs. However, Broombush, heath and Kurricabah/Bloodwood scrub support populations of this species in both early and late successional stages of vegetation and may be important habitat for P. pilligaensis. The extent of these scrub habitat types in Pilliga East is about 20 000 ha out of a total of over 200 000 ha for the whole state forest. This scrub occurs in only small patches outside Pilliga East. Bearing in mind that this species’ density is very sparse except where over-wintering congregations occur, this is a very restricted distribution for a small rodent. However its scattered presence in shrubby woodlands of Pilliga East and the adjacent Pilliga Nature Reserve indicate that it is capable of occupying a range habitats within the Pilliga region as long as a favourable set of habitat conditions exist.

Acknowledgments

We would like to thank Fritz Geiser, Gerhardt Körtner and Dean Metcalfe for their assistance with this note and Stuart Green for his personal communication.

References

1. AUSLIG, 1990. Vegetation. Atlas of Australian Resources. 3^rd Series. Vol. 6. Auslig, Canberra.
2. Baynes, A., Chapman, A. and Lynam, A.J, 1987. The rediscovery, after 56 years, of the heath rat Pseudomys shortridgei (Thomas1907) (Rodentia: Muridae) in Western Australia . Rec. West. Aust. Mus. 13: 319-22.
3. Braithwaite, R. W. and Brady, P., 1993. The Delicate Mouse, Pseudomys delicatulus: a continous breeder waiting for the good times. Aust. Mammal. 16 (1): 94-8.
4. Briscoe, D. A., Fox, B. J. and Ingleby, S., 1981. Genetic differentiation between Pseudomys pillagaensis and related Pseudomys (Rodentia: Muridae). Aust. Mammal. 4: 89-92.
5. Cockburn, A., 1978. The distribution of Pseudomys shortridgei (Muridae : Rodentia) and its relevance to that of other heathland Pseudomys. Aust. Wildl. Res. 5 : 213-219.
6. Cockburn, A., Braithwaite, R. W. and Lee, A. K., 1981a. The response of the heath rat, Pseudomys shortridgei, to pyric succession: a temporally dynamic life-history strategy. J. Animal Ecol. 50: 649-666.
7. Cockburn, A., 1981b. Population regulation and dispersion of the Smoky Mouse, Pseudomys fumeus 1. Dietary determinants of microhabitat preference. Aust. J. Ecol. 6: 231-254.
8. Fox, B.J., 1978. Temporal changes in a small mammal community on coastal heath regenerating after fire. Bull. Aust. Mamm. Soc. 5: 30.
9. Fox, B.J., 1982. Fire and mammalian secondary succession in an Australian coastal heath. Ecology 63: 1332-41.
10. Fox, B.J., 1990. Changes in the structure of mammal communities over successional time scales. Oikos 59: 321-29.
11. Fox, B. J. and Briscoe, D. A., 1980. Pseudomys pilligaensis, a new species of murid rodent from the Pilliga Scrub, northern New South Wales. Aust. Mammal. 3: 109-26.
12. Fox, B.J. and McKay, G.M., 1981. Small mammal responses to pyric successional changes in eucalypt forest. Aust. J. Ecol. 6: 29-42.
13. Friend, G.R., 1993. Impact of fire on small vertebrates in mallee woodlands and heathlands of temperate Australia: a review. Biol. Cons. 65: 99-114.
14. Hart, D., 1990. Afield appraisal of the role of plant opal in the Australian environment. Unpbl. PhD thesis, Macquarie University.
15. Higgins, P. and Fox, B.J., 1993. Interspecific competition: A mechanism for rodent succession after fire in wet heathland. Aust. J. Ecol. 18: 193-201.
16. Lim, L., 1992. Conservation and biology of the rare Pilliga Mouse Pseudomys pilligaensis Fox and Briscoe 1980 (Muridae: Conilurini). Unpublished report to the Zoological Parks Board, Mossman, NSW.
17. Masters, P., 1993. The effects of fire driven succession and rainfall on small mammals in spinifex grassland at Uluru National Park, Northern Territory. Wildl. Res. 20: 803-13.
18. Medina, E. and Silva, J.F., 1990. Savannas of northern South America: a steady state regulated by water-fire interactions on a background of low nutrient availability. J. Biogeog. 17: 403-14.
19. Paull, D.C., 1999. Patterns of decline in the native mammal fauna of the north-west slopes of New South Wales. Aust. Zool. 31(1): 210-24.
20. Posamentier, H.G., 1976. Habitat requirements of small mammals in coastal heathland of New South Wales. MSc Thesis, University of Sydney.
21. Southgate, R. and Masters, P., 1996. Fluctuations of Rodent populations in response to rainfall and fire in a Central Australian hummock grassland dominated by Plectrache schinzii. Wildl. Res. 23: 289-303.
22. Townley, S.J., 1998. Collation of current knowledge of the Pilliga Mouse Pseudomys pilligaensis prior to the preparation of a recovery plan. Unpublished report to NSW NPWS Western Zone, Dubbo.
23. Wilson, B. A., 1991. The ecology of Pseudomys novaehollandiae (Waterhouse, 1843) in the eastern Otway Ranges, Victoria. Wildl. Res. 18: 233-47.

Table 1. Pooled P. pilligaensis capture rates for different trapping techniques used

Table 2. Pooled trapping results, 1993-99.

* number of P. pilligaensis captures during each survey period
** habitats with logs

| National Parks Association of NSW | NPA Western Woodlands Index |