OUP user menu

Social organization and movements of desert rodents during population “booms” and “busts” in central Australia

Christopher R. Dickman, Aaron C. Greenville, Chin-Liang Beh, Bobby Tamayo, Glenda M. Wardle
DOI: http://dx.doi.org/10.1644/09-MAMM-S-205.1 798-810 First published online: 16 August 2010


We monitored populations of 2 species of desert rodents, the sandy inland mouse (Pseudomys hermannsburgensis) and spinifex hopping-mouse (Notomys alexis), over 18 years in the Simpson Desert, central Australia. Populations fluctuated synchronously from very low numbers, or “busts,” during prolonged dry periods to high numbers, or “booms,” after heavy rainfall 3 times over the study period. On the basis of observations that food resources expand after rainfall, we predicted that rodents would show increased rates of recapture, fidelity to burrows, and burrow sharing during population increase (boom) phases compared with decline or bust phases, and also reduce their movements and foraging activity in open habitats during population booms. The behavior of both species was similar but not as we had anticipated. Burrow fidelity and numbers of animals per burrow were roughly 2-fold higher during both the population increase and decrease phases as compared with the population-low phase, whereas rates of movement were reduced by about half. As revealed by giving-up density trials, animals foraged less at experimental food patches during population increase and decrease phases than during busts, and also foraged less in open than in covered habitats. Recaptures of N. alexis were similar across all population phases, whereas P. hermannsburgensis was recaptured more often when populations were decreasing than at other times. The results suggest that both species are dispersed and highly mobile during bust periods but sedentary and more social during population increases and collapses. These changes in movements and social organization appear to be unusual in desert rodents, and we propose that future studies seek to identify the roles of food and other factors in driving them.

Key words
  • Australia
  • burrows
  • desert rodents
  • long-term study
  • movements
  • Notomys
  • population dynamics
  • Pseudomys
  • social organization

Many desert environments remain dry and unproductive for long periods but experience short pulses of productivity after rainfall or flood events. Populations of small mammals in these environments often track the changing conditions, rising during the ephemeral periods of plenty and collapsing again as conditions dry out (Whitford 1976). In deserts with low but seasonally reliable rainfall, such as the Chihuahuan and Mojave of North America, rodents initiate reproduction before or during the annual wet season so that young are weaned when the availability of seeds and other food resources is high (Havstad et al. 2006; Whitford 2002). Many species are sedentary and remain in defined home ranges for much of the year and use cached stores of food to increase their chance of surviving dry periods (Randall 1993). Populations of rodents in these deserts can be influenced by climate and endogenous factors, such as competition (Lima et al. 2008), and exhibit annual or multiannual periodicities similar to those of rodents in more temperate regions (Brady and Slade 2004; Merritt et al. 2001).

By contrast, in deserts with less reliable rainfall, such as the Namib of coastal Namibia and those of central and northern Australia (van Etten 2009), rodents often breed opportunistically in response to rainfall and show lags of several weeks to months before populations increase (Griffin 1990; Plomley 1972). During prolonged drought populations may disappear locally and recover only via immigration. Many Australian desert rodents move long distances to access resources at sites that have received recent rain (Dickman et al. 1995; Letnic 2002); no evidence exists that any of the Australian desert species rely on food caching to survive dry periods. The social organization of rodents that track and exploit shifting resources is likely to be quite flexible (Dickman et al. 1995) but remains poorly known for most desert species (Anstee et al. 1997).

Long-term studies of Australian desert rodents suggest that populations fluctuate from very low densities, or “busts” (<l/ha), during long-term droughts to transient but much higher densities, or “booms” (>25/ha; exceptionally >100/ ha), after heavy rainfall (Dickman et al. 1999a; Plomley 1972). Most Australian desert rodents are omnivorous and use the flushes of green plant material, fruits, seeds, and invertebrates that are stimulated by large rainfall events (Murray and Dickman 1994a, 1994b; Murray et al. 1999). Breeding is initiated within days of rainfall, and subsequent increases in survival and recruitment of young allow populations to increase sharply 3–10 months later (Dickman et al. 1999a; Predavec 1994; Predavec and Dickman 1994). High population numbers can last 1–2 years before numbers decline. The major cause of population collapse is probably the shrinkage of the resource base as drought ensues (Letnic and Dickman, in press; Predavec 1994). However, predation from introduced carnivores such as the European red fox (Vulpes vulpes) and feral cat (Felis catus) can also have negative effects on population growth (Letnic et al. 2009; Pavey et al. 2008), especially as some desert rodents forage equally in closed and in risky, open habitats when food is scarce (Kotler et al. 1998). Social suppression of reproductive activity during population peaks (Breed 1979), depletion of nutrients (Morton and Baynes 1985), and disease (Carstairs 1974) have been suggested as additional causes of population declines, but these hypotheses remain to be tested.

In this paper we investigate temporal changes in the social organization and movements of 2 species of desert rodents, the sandy inland mouse (Pseudomys hermannsburgensis, ∼12 g) and spinifex hopping-mouse (Notomys alexis, ∼30 g) in central Australia. We do this by drawing data from a long-term (18-year) livetrapping program that encompasses multiple boom and bust periods. We define social organization as the extent to which animals aggregate or dissociate with each other, and use live-capture data and observations of burrow use to assay and detect shifts in dispersion. This is an operational definition of social organization; we do not attempt to specify either the mating systems or modes of communication of our study species (Jones 1993). In captivity the 2 study species are socially tolerant and are able to breed year-round (Happold 1976). Family groups of N. alexis appear to be particularly cohesive, cooperating in the construction of burrows, rearing young, and repelling intruders (Stanley 1971). In the field both species nest solitarily or in groups; up to 10 adult and young N. alexis, and up to 22 nonreproductive P. hermannsburgensis have been recorded together (Finlayson 1941; Watts and Aslin 1981). Both species occupy simple, shallow (30-cm deep) burrows during winter, but N. alexis may excavate tunnels and nest chambers to 1 m deep during summer, with shafts that ascend vertically to the surface (Breed and Ford 2007; Finlayson 1941). Individuals of both species on occasion have been recorded moving 14 km over periods of several weeks, although regular nightly movements usually do not exceed several hundred meters (Dickman et al. 1995). Taken together, these observations suggest that the social organizations of both P. hermannsburgensis and N. alexis are highly variable, but the factors producing structure are not yet clear.

Our objectives are to present a brief description of the dynamics of the 2 rodent species over the period of study and test several hypotheses that predict how their social structure will change between population peaks and troughs. On the basis of the observations of Dickman et al. (1999a), Letnic et al. (2004, 2005), Newsome and Corbett (1975), Predavec (1994), and Pavey et al. (2008) that food resources are more abundant during periods of population increase than at other times, we expected that animals would be more sedentary, more aggregated, and more selective in their foraging habitats during population booms than at other times. We test 5 contingent hypotheses. We expect rates of recapture, fidelity to burrows, and numbers of animals per burrow to be greater during population booms than during the decline or bust phases. Conversely, distances moved and foraging activity in open habitats will be less during population booms than at other times. In formulating these hypotheses we define 3 population phases: increase (or boom), decrease, and low (or bust). Inspection of the population data suggested that the numbers of both species always increased rapidly to a peak and then fell, so that a high phase was never sustained. We use live-capture and mark-and-release methods to describe the dynamics of the species populations, and trapping, tracking, giving-up density (GUD) experiments, and direct observations to test the hypotheses.

Materials and Methods

Study area.—Research was carried out on Ethabuka station (now Ethabuka Reserve) in the northeastern Simpson Desert, western Queensland, Australia (23°46′S, 138°28′E). Long, parallel sand dunes are the major land form of the Simpson Desert and run about 0.6–1 km apart and rise to about 8–10 m high (Purdie 1984). The dominant vegetation is a needle-leaved hummock grass, spinifex (Triodia basedowii), which covers the floors of the dune valleys to just below the dune crests. Gidgee trees (Acacia georginae) occur on patches of heavier clay soils in the dune valleys, and shrubs such as Acacia ligulata, Eucalyptus pachyphylla, Dodonaea viscosa, Grevillea stenobotrya, and Grevillea juncifolia occur singly or in small stands throughout the study area. Annual grasses and herbs are diverse and abundant after rain but persist in the seed bank during dry periods. The regional climate is dominated by the El Niño Southern Oscillation and is highly irregular (Letnic and Dickman 2006). The long-term rainfall average is 199 mm/year (n = 94 years, recorded at Marion Downs, 120 km distant), with 74% falling over the austral summer between October and March. During the course of the present study, 1990–2008, monthly on-site rainfall varied from 0 mm in many months to 303 mm in February 1991 (Fig. 1), with annual rainfall varying from 71 mm in 2002 to 671 mm in 2000. Maximum daily temperatures in summer are 46–49°C, and winter minima fall to −6°C (Purdie 1984).

Fig. 1

Rainfall patterns in the northeastern Simpson Desert, Queensland, Australia. Actual monthly rainfall (mm) from 1989 to 2008 is shown by the vertical bars, and mean monthly rainfall is shown by the thin, consistently repeated line (_) that represents the long-term average from the 3 weather stations closest to the study area (Marion Downs, n = 94 years; Sandringham, n = 48 years; Glenormiston, n = 103 years). Rainfall during the study is shown using records from Marion Downs for 1989 to 1994 and from an on-site weather station from 1995 to 2008.

Live trapping.—Rodents were captured in pitfall traps (16 cm in diameter, 60 cm deep). To increase trap success a drift fence of aluminum wire screening (flywire) was set around the top of each trap. The fence was 30 cm high and ran for 2.5 m on each side of the pitfall opening to intercept and guide surface-active animals into the trap (Friend et al. 1989). The bottom ends of the pits were provided with a floor of wire screening to prevent captured animals from digging their way out, and all pits were capped with metal lids when not in use. Pitfall traps were arrayed in grids covering 1 ha, with each grid comprising 6 lines of 6 traps spaced 20 m apart. The top line of traps was established on the dune crest and the bottom line 100 m distant in the dune valley. Adjacent grids were spaced 0.5–2.0 km apart, with the most distant grids separated by 14 km. Six grids were operated in the 1st year of study in 1990 and up to 12 grids thereafter.

Traps were opened for 2–4 nights on each grid at sampling intervals of usually 2–3 months and checked in the mornings and sometimes afternoons. Captured animals were identified, weighed, inspected for reproductive condition, and marked uniquely by toe clipping (until 1993) or by ear notching (from 1994) before release. Co-occurring dasyurid marsupials, lizards, and frogs were also processed (see Dickman et al. 1999b, 2001; Greenville and Dickman 2009; Letnic and Dickman 2005; Predavec and Dickman 1993). Live trapping allowed us to track the population dynamics of the 2 study species, quantify movements between traps, and obtain individuals for tracking.

Tracking and direct observations.—Animals were radio-tracked to determine their fidelity to burrows and provide a further measure of movements. We used simple tags with single-stage circuits (i.e., with no amplifier) with 10-cm trailing whip antennas that were attached to animals using plastic cable tie collars. Tags varied in mass from 0.4 to 1.0 g, the smaller units being attached to P. hermannsburgensis and the larger to N. alexis. Individuals used for tracking were selected at random, with the proviso that their gain in load did not exceed 5% of body mass; hence, juveniles were not tracked. Animals were fitted with the collars after dusk, held for 1–2 h in a soft cotton bag to ensure correct attachment, and then released at the point of capture within 3–4 h of nightfall. No tracking was attempted on the night of release to allow animals to resume their usual activities without further disturbance. Tracking began the next day and continued for 5–11 days and nights. Animals were trapped again before batteries failed, and collars were removed. All tags operated at 150–151 MHz and were supplied by Biotrack (Wareham, United Kingdom), Holohil (Ontario, Canada), or Titley Electronics (Ballina, New South Wales, Australia) at different times during the study.

At night we located animals on foot every 1–4 h using a 3-element hand-held Yagi antenna (Titley Electronics, Ballina, New South Wales, Australia) and either a Regal 1000 (Titley Electronics, Ballina, New South Wales, Australia) or TR-2 (Telonics Inc., Mesa, Arizona) receiver. If animals were in open habitats, we attempted to establish visual contact and determine their location precisely. We approached from downwind, walking on sand to reduce noise and using red torchlight to further reduce the likelihood of being detected (Jacobs 1993). If animals could not be approached closely, locations were estimated by triangulation of 2–3 bearings taken in the direction of the peak signal. Bearings were taken using a prismatic compass either by a single observer 1–2 min apart or simultaneously by 2–3 observers. We ensured that bearings were taken from known triangulation points and that angles were >20° and <120° from each other (Kenward 2001). Pilot trials using tags placed at known locations showed that bearings were usually accurate to within ±5° up to distances of ∼110 m. By day, animals in burrows usually could be detected by walking along dune crests to pick up their signals and then walking along the line of peak signal strength. Burrows could be pinpointed by removing the Yagi antenna and following the signal using the coaxial cable and receiver. This allowed detection of animals within 1–2 m. Locations of animals and bearings were flagged in the field using marker tape and subsequently placed on fine-scale maps of the study area. We also measured linear distances between known animal locations to provide a check on the accuracy of our maps.

Direct observations were used to assess the numbers of animals using burrows. Burrows were targeted if they were known to be active (as judged by fresh tracks in the sand around burrow entrances) or used by radiotagged individuals. Just before dusk, when animals become active, 2 observers stationed themselves within 5–8 m of the burrow and counted individuals as they emerged to forage. To observe animals we used either red-light torches or a 100-W red floodlight powered by a car battery. Nightlong pilot trials, carried out on 4 occasions, showed that no additional animals emerged from burrows >4 h after darkness had fallen, so most observations were terminated after 4 h. In early trials we attempted to gain insight into the composition of burrow inhabitants by erecting temporary fences of corrugated metal around burrow systems and setting arrays of Elliott folding aluminum box-style live traps (10 × 10 × 33 cm; Elliott Scientific Equipment, Upwey, Victoria, Australia) within them. However, these attempts were abandoned as animals either remained underground or burrowed under the fence to avoid being captured.

GUD experiments.—To quantify foraging activity in open and covered habitats we established arrays of food patches at stations in open sandy sites and under the cover of hummocks of spinifex. In general, 10–24 stations were established along the midsections of sand dunes, with stations spaced >25 m apart. Two food patches were placed at each station, 1 under or at the edge of a spinifex hummock and the other >2 m away in open sand. Each food patch consisted of a plastic bowl (15 cm in diameter, 4 cm deep) that contained 20 peanut quarters mixed into 200 ml of sifted sand. The bowls were half buried in the sand and smeared on the undersurface with a mixture of Vaseline and Coopex insecticide powder (Bayer Ltd., Pymble, New South Wales, Australia) to prevent ants from accessing the food source. The food patches were prebaited for 1 night to allow animals to become accustomed to them, and then run for 3–4 successive nights. In the late afternoons, before these nights, the 20 peanut quarters were added to all food patches and the surrounding sand smoothed so that foot tracks could be read. The patches were checked at dawn after each night. Tracks in or near the bowls were identified, and the number of peanut quarters remaining per bowl were counted and taken to be the GUD (Brown 1988, 1992). The tracks of P. hermannsburgensis could be identified by their small size (pes <20 mm) and imprint of 5 hind toes, and those of N. alexis could be confirmed by their length (>30 mm), imprint of the heel, and the impression of usually just 3 toes (Triggs 1996). Although tracks could have been confused with 2 sympatric species of rodents, the desert mouse (Pseudomys desertor) and introduced house mouse (Mus musculus), we consider this unlikely because these other rodents were very rare throughout the course of the study.

Giving-up densities were measured separately for P. hermannsburgensis and N. alexis if the tracks of either species could be identified at each bowl. If both species had foraged and it was not clear which rodent had been the last visitor, the data for that patch were discarded. GUDs measure the tradeoff between foraging returns and the costs (energetic, predation, and missed opportunity costs of foraging—Brown 1988) of staying at a food patch and thus are low if animals spend much time foraging at a food patch and high if they do not forage for long. We expected GUDs to be higher during boom periods, when more food is available in the environment, and higher in open than in sheltered habitats owing to the higher risk of predation in the open due to the presence of predators such as the barn owl (Tyto alba), red foxes, and feral cats that occur in the study area (Dickman et al. 1991; Mahon 1999).

All procedures throughout the study were carried out under license from the Queensland National Parks and Wildlife Service (now Queensland Environmental Protection Agency) with approval from the University of Sydney Animal Ethics Committee. In addition, procedures were consistent with guidelines approved by the American Society of Mammalogists (Gannon et al. 2007).

Data processing and comparisons.—Captures were tallied for each field trip and over the duration of the study and used to generate a capture rate for each species. This was expressed as mean captures ± SE per 1,000 trap nights (1 trap night = 1 trap open for 1 night) per field trip, with grids used as replicates. Simple rate of increase (Caughley and Sinclair 1994) was calculated ajs the difference in capture rate from trip to trip and expressed using a 3-trip rolling average to visualize boom and bust periods of growth. We also show monthly rainfall for the duration of the study using records from Marion Downs until 1994 and an on-site automatic weather station (Environdata, Warwick, Queensland, Australia) from 1995. Formal analysis of the relationship between rainfall and rodents is not attempted here but has been considered elsewhere (Dickman et al. 1999a).

Recapture rate was expressed as the proportion of animals per field trip that had been captured previously on the same or on an earlier trip and was used to gauge residency. An index of movement was derived as the average distance that animals traveled between successive captures (Brant 1962), both within and between trapping grids. Animals that were recaptured successively in the same trap were excluded from movement calculations.

The positions of triangulated animals were calculated from the compass bearings and positions of the triangulation points using an iterative maximum likelihood estimator (Location of a Signal software, Ecological Software Solutions, Sacramento, California), which minimizes angular error between bearings and location of the signal (Lenth 1981). Plots of range size against number of locations failed to approach an asymptote for most individuals, suggesting that home ranges could not be estimated with reliability. As a result, distances moved between successive signal locations, expressed per unit time (m/h), were used as a surrogate measure of movement (Haythornthwaite and Dickman 2006). The distance di moved by an animal between the 1st signal location i(xi, yi) and the next (xi+1, yi+1) was calculated using the equation of White and Garrott (1990): Embedded Image

To assess fidelity to burrows, f, we used the simple index: Embedded Image where Nmax is the maximum number of visits by an individual to any 1 burrow over the period of radiotracking, and N is the total number of its visits to all burrows. Index values range from f = 1 for an individual that is faithful to a single burrow to f = 1/N for an individual that uses all burrows once. The index was calculated only for burrow usage by day; burrows seldom were used at night. The index is identical to the Berger-Parker index of species dominance (Magurran 2004), and is co-opted here for convenience.

To process GUD data we averaged the GUD for each food patch over the 3–4 nights that the food patches were set. GUDs were included in analyses if food patches had been visited by either of the study species over the study period, as judged by foot tracks in the sand matrix of the bowls, even if no food had been removed. Food patches that had been disturbed by other foragers, such as Australian ravens (Corvus coronoides), or had not been visited were omitted from analyses.

After data processing we tested each of our 5 hypotheses by comparing values between boom, decrease, and bust phases of the species’ populations. Inspection of graphs depicting capture rates and rates of increase showed 3 main periods when populations rose and then fell rapidly, with periods of low capture rate and rate of increase that preceded them (Figs. 2, 3). The increase periods occurred in 1991–1992, late 2001, and late 2007 and the decrease periods usually ≤6– 9 months later. Recapture rates and distances moved by animals were compared between phases across all 3 boom and bust periods by analysis of variance (ANOVA). Observations of burrow fidelity and numbers of animals per burrow were made only during the 1st period of population-low, increase, and decrease phases. Distances moved by radiotagged animals were recorded over the same period but with additional observations made during the 2nd population-low phase. GUD data were collected mostly during the 2nd period of population-low, increase, and decrease phases, with a few additional observations made during the 3rd population-low phase. Data from the same population phase (low, increase, or decrease) were pooled for these latter comparisons. One-factor ANOVAs were used for most tests, but 2-factor ANOVAs were used for comparisons of burrow fidelity and movements as revealed by radiotracking, with sex as a 2nd factor, and comparisons of GUDs, with habitat as a 2nd factor. Tukey's honestly significant differences test was used post hoc to locate differences between means.

Fig. 2

Population dynamics of Pseudomys hermannsburgensis in the northeastern Simpson Desert, Queensland, Australia, shown as A) captures/1,000 trap nights (mean ± SE) and B) rates of increase (positive, negative, or zero) expressed as the difference in capture rate from trip to trip, smoothed using a 3-trip rolling average.

Fig. 3

Population dynamics of Notomys alexis in the northeastern Simpson Desert, Queensland, Australia, shown as A) captures/ 1,000 trap nights (mean ± SE) and B) rates of increase (positive, negative, or zero) expressed as the difference in capture rate from trip to trip, smoothed using a 3-trip rolling average.

Before analyses we used Levene's test to check for equality of variances. As a result, data on movements of N. alexis, as revealed by radiotracking, were log(x + l)-transformed, and recapture data for both species were transformed by arcsine (Quinn and Keough 2002; Zar 1974). Equal variances could not be obtained for the GUD results, even when data were transformed. We therefore used the untransformed data, accepting significance at α = 0.01 in the GUDs comparisons, but αa = 0.05 in all other tests (Underwood 1997). All comparisons were made separately for each of the 2 study species using SPSS 15.0 (SPSS 2006).


Population dynamics and rainfall.—The study area received rainfall above the annual average in February 1991, December 2000, and January 2007 and rainfall exceeding 100 mm in January 1991, February 1992, January 1995, February 1997, April 1998, and April 2000 (Fig. 1). Periods of rainfall deficit prevailed in 1990, 1994, 1996, and from 2002–2006 (Fig. 1).

A total of 4,723 captures was made of the 2 study species between March 1990 and June 2008 in 105,192 trap nights, yielding an overall trap success of 4.5%. P. hermannsburgensis was captured most commonly, with 2,988 captures and a recapture success of 12.2% (n = 365 recaptures). N. alexis was captured 1,735 times with a recapture success of 17.0% (n = 295 recaptures).

Peak capture rates of P. hermannsburgensis were recorded in July–August 1991, July–August 1992, March 2001, and February 2008, with rates in each month exceeding 170 captures/1,000 trap nights. Captures fell to <7 captures/1,000 trap nights for much of the period from 1994 to mid-1997 and to zero in 2002 (Fig. 2A), with corresponding changes in the rate of increase (Fig. 2B). Capture rates of N. alexis showed a similar pattern, peaking at 213 captures/1,000 trap nights in June 2001, with more modest peaks in July–August 1991, July–August 1992, and February 2008. No N. alexis were captured between November 1993 and September 1997; rates were <3 animals/1,000 trap nights from 2002 to 2006 (Fig. 3A) with concomitantly low rates of increase (Fig. 3B). In general, the capture rates of both species peaked 3–6 months after heavy rainfall, although the peak in February 2008 arrived a year after drought–breaking rains.

Rates of recapture.— Recapture rates differed for P. hermannsburgensis between population phases (F8.17 = 5.94, P = 0.001), with recaptures during the 2nd and 3rd decrease phases being greater than at other times (Fig. 4A). Although recaptures of N. alexis appeared also to be greater during population decreases (Fig. 4B), differences between phases were not significant (F8,17 = 2.02, P = 0.11).

Fig. 4

Recapture rates, expressed as proportions of animals per field trip that had been captured previously on the same, or on earlier, trips of A) Pseudomys hermannsburgensis and B) Notomys alexis in the northeastern Simpson Desert, Queensland, Australia, shown during 3 low phases (L1–L3), 3 increase phases (I1–I3), and 3 decrease phases (D1–D3) of populations of each species. Means are shown ± SE, with n = 3 sampling sessions per phase.

Fidelity to burrows.—Animals were tracked to a total of 297 different burrows on 663 occasions, with individuals of both P. hermannsburgensis and N. alexis using either the same burrow or different burrows for up to 6–8 consecutive days. Burrow fidelity was less in the population-low phase than in either the increase or decrease phases in both species (Fig. 5). In P. hermannsburgensis fidelity to burrows was roughly double during the increase and decrease phases compared with that during the low phase (F2,45 = 16.56, P < 0.001); we found no difference between the sexes (F1,45 = 1.72, P = 0.20) nor any sex × phase interaction (F2,45 = 0.206, P = 0.82). Patterns of burrow fidelity were similar in N. alexis (phase: F2,36 = 17.02, P <0.001; sex: F1,36 = 0.08, P = 0.78; interaction: F2,36 = 0.004, P = 0.99).

Fig. 5

Burrow fidelity of A) Pseudomys hermannsburgensis and B) Notomys alexis in the northeastern Simpson Desert, Queensland, Australia, shown for females and males during low, increase, and decrease phases of the populations. Means are shown ± SE, with samples sizes above the bars. Fidelity is expressed as the maximum number of visits by an individual to any 1 burrow during radiotracking divided by the total number of its visits to all burrows.

Numbers of animals per burrow.—Animals were observed emerging from burrows on 43 nights over the study period. Numbers per burrow varied from 1 to 11 for P. hermannsburgensis and from 1 to 7 for N. alexis. Fewer animals were seen emerging from burrows during the population-low phase in both species than at other times (Fig. 6). The phase difference was less marked, and not significant, in P. hermannsburgensis where numbers per burrow increased from 2 to ∼3.5 during population booms (F2,20 = 1–05, P = 0.37; Fig. 6A), compared with the >2-fold increase in numbers per burrow between phases in N. alexis (F2,17 = 4.61, P = 0.025; Fig. 6B).

Fig. 6

Numbers of A) Pseudomys hermannsburgensis and B) Notomys alexis observed using the same burrow at any one time during low, increase, and decrease phases of the populations in the northeastern Simpson Desert, Queensland, Australia. Means are shown ± SE, with samples sizes above the bars.

Distances moved.—Too few rodents were recaptured during 2 of the 3 population-low phases to include in comparisons. However, distances moved within grids differed little between phases for either P. hermannsburgensis (F6,43 = 1.17, P = 0.34) or N. alexis (F6,50 = 1.35, P = 0.25), perhaps reflecting low sample sizes and heterogeneity in distances moved. Relatively few animals moved between grids, but distances covered by those doing so varied greatly. Individuals of both species shifted to neighboring grids 500 m away and between the most distantly separated grids 14,000 m apart. Only 3 individuals shifted between grids during population-low phases, thus precluding formal comparisons; an adult male P. hermannsburgensis moved 12,000 m, an adult male N. alexis moved 750 m, and an adult female moved 5,500 m. Pooled over the period of study, movements between grids comprised 11.9–19.2% of all recapture-based movements between population phases for P. hermannsburgensis and 11.1–22.2% for N. alexis. We found no evidence that proportions of within-grid to between-grid movements differed between population phases for either species (P. hermannsburgensis: χ22 = 0.68, P = 0.70; N. alexis: χ22 = 1.41, P = 0.49).

During radiotracking we obtained 874 location fixes for P. hermannsburgensis and 726 for N. alexis and followed the 2 species for total distances of 47.6 km and 49.0 km, respectively. In general, both species moved at similar rates (m/h, grand means: N. alexis = 63.05 ± 5.93 SE; P. hermannsburgensis = 57.94 ± 4.54 SE), but the maximum speed recorded was considerably greater for N. alexis (1,018 m/h) than for P. hermannsburgensis (422 m/h) between any 2 consecutive locations. Movements were greater during the population-low phase than during the increase or decrease phases (P. hermannsburgensis: F2,45 = 8.88, P < 0.001; N. alexis: F2,36 = 13.44, P < 0.001) and also were greater for males than for females (P. hermannsburgensis: F1,45 = 15.95, P < 0.001; N. alexis: F1,36 = 9.30, P = 0.004). No phase × sex interaction existed for either P. hermannsburgensis (F2,45 = 2.17, P = 0.13) or N. alexis (F2,36 = 0.39, P = 0.72). Patterns of movement were similar for both species (Fig. 7).

Fig. 7

Movements (m/h) of radiotracked A) Pseudomys hermannsburgensis and B) Notomys alexis in the northeastern Simpson Desert, Queensland, Australia, shown for females and males during low, increase, and decrease phases of populations of each species. Means are shown ± SE, with samples sizes above the bars.

Foraging activity in open habitats.—For P. hermannsburgensis GUDs were lower in the population-low phase than in either the increase or decrease phases (F2,110 = 11.66, P < 0.001) and lower also in spinifex compared with open habitats (F1,110 = 8.41, P = 0.005). The phase × habitat interaction approached significance (F2,110 = 3.08, P = 0.05) owing to the similarity in GUDs between habitats in the population-low phase but not in the other 2 phases (Fig. 8A), but did not meet the more stringent criterion of a = 0.01 in this test. Patterns in the GUD results were similar for N. alexis (Fig. 8B), with strong main effects (phase: F2,132 = 45.37, P < 0.001; habitat: F1,132 = 20.67, P < 0.001) and no phase × habitat interaction (F2,132 = 1.02, P = 0.37).

Fig. 8

Giving-up densities (GUDs) of A) Pseudomys hermannsburgensis and B) Notomys alexis in the northeastern Simpson Desert, Queensland, Australia, shown for open and spinifex-covered habitats during low, increase, and decrease phases of populations of each species. GUDs represent the numbers of peanut quarters left by rodents in food patches from 20 quarters that had been provided. Means are shown ± SE, with samples sizes above the bars.


Population dynamics and rainfall.—The 3 population booms over the study period followed exceptionally heavy summer rainfalls, whereas bust periods generally coincided with prolonged periods of rainfall deficit. Broadly similar patterns of eruption and subsequent collapse have been described in several previous studies in both central Australia (Brandie and Moseby 1999; Newsome and Corbett 1975; Pavey et al. 2008; Southgate and Masters 1996) and in arid environments elsewhere (Beatley 1969; Ernest et al. 2000; Lima et al. 2008; Whitford 1976).

Although the capture rates of both study species shared similar trajectories over time, they differed in the magnitude of response to rainfall. Except during June 2001, when the capture rate of N. alexis peaked at 213 animals/1,000 trap nights, this species usually achieved lower capture rates during boom periods than did P. hermannsburgensis. As litter sizes, gestation, and weaning periods are similar in both species (Breed and Ford 2007), and similar proportions of females breed after rain (Predavec 1994), higher population peaks in P. hermannsburgensis might reflect higher juvenile survival or larger initial population sizes before eruptions. The relative survival rates of juveniles of the 2 species could not be ascertained. However, comparison of Figs. 2 and 3 shows that P. hermannsburgensis consistently achieved capture rates of >9 animals/1,000 trap nights in the months preceding eruptions, 3-fold higher than those for N. alexis. Capture rates for N. alexis were particularly low before the 3rd population eruption in late 2007, which was the most muted of the 3 booms exhibited by this species. N. alexis also may experience stronger social suppression at high density than P. hermannsburgensis (Breed 1979). This too could reduce population maxima in this species.

Two aspects of the results were unexpected. First, the capture rates of both species decreased very rapidly after peaking, even if further rain fell during the decline phase.

Second, both species disappeared from the trapping record for periods of months or years before returning. As trappability varies little over time or population phase in either N. alexis or P. hermannsburgensis (Dickman et al. 1999a), we assume that these differences in capture rate reflect true changes in population size or dispersion and hence that factors other than rain and trappability are influential. Resource availability declines as conditions dry (Predavec 1994), but the rate of decline appears to be slower for food resources of the rodents (seeds, invertebrates, and green plant material) than for the rodents themselves (Ricci 2003). In addition, as sufficient invertebrates (the preferred prey of both study species— Murray and Dickman 1994b) are available at all times to sustain populations of dasyurid marsupials (Dickman et al. 2001), it seems unlikely that the rodents disappear due to lack of food.

During declines of some species of desert rodents, such as the long-haired rat (Rattus villosissimus), animals appear to become stressed as their numbers fall, levels of aggression also increase, and individuals frequently are captured with missing fur and scabs on the body (Predavec and Dickman 1994). Carstairs (1974) suggested that disease might play a role in driving population declines and noted further that individuals lose body mass during this phase (Carstairs 1976). However, we saw no obvious indication that our study animals were stressed during the decline phases. No signs of individuals in poor condition were evident, and internal examination of small samples of both species by Ricci (2003) found very low incidence of disease or parasitic infection at any time. Some evidence suggests, instead, that red foxes and feral cats hasten the declines of both N. alexis and P. hermannsburgensis (Pavey et al. 2008) and, in the study area, can drive populations to local extinction in the sand dunes (Mahon 1999). Both species of rodents probably persist in small numbers in patches of gidgee woodland in the dune valleys (Gregory 2008). The trees, with their sparse understory of associated shrubs, may provide rodents with both food and shelter, and perhaps act as loci from which animals disperse when conditions improve in the surrounding environment.

Social organization and movements.—Our initial hypotheses received very limited support, with animals responding similarly during the increase and decrease phases and very differently during population troughs. There were few obvious differences in responses between the 2 study species. Recapture rates of P. hermannsburgensis differed more between population phases than did those of N. alexis, and numbers of animals cohabiting burrows differed between phases only in the latter species. Although our sample sizes were sometimes small, these results suggest that the 2 species responded in broadly similar ways between population phases.

In formulating hypotheses we followed previous work showing that post–rain pulses of food resources facilitate increases in reproduction and juvenile survival in the study species (Dickman et al. 1999a; Letnic et al. 2004, 2005; Newsome and Corbett 1975; Predavec 1994) and assumed that these events would be associated with changes in movements and social organization. We assumed further that, with the depletion of food resources, populations would fall concomitantly and exhibit similar patterns of social organization during both the decline and subsequent low phases. However, populations declined faster than their food resources, and factors such as predation may have exerted stronger depressive effects on populations than we had expected. This could have resulted in per capita availability of resources being similar for individuals during both the increase and decrease population phases and hence account for the observed similarities in social organization during these times. That individuals were not resource-limited during the decline phases is supported by observations that both species continued to reproduce as capture rates fell (Predavec 1994; Ricci 2003) and that experimentally added food retarded, but did not prevent, population declines (Predavec 2000; C.-L. Beh, pers. obs). We use this interpretation to focus our discussion of the results below.

Both species exhibited low recapture rates throughout the study, thus providing little obvious indication that increased residency or amount of time in the study area contributed to population increases. As populations increased rapidly during the boom periods, however, the influx of new individuals at each census would have masked any tendency for increased residency. Predavec (1994) also found little difference in recapture rates of the 2 species during different population phases but noted that lactating females contributed disproportionately to overall recapture rates as populations increased. With the influx of new animals declining during the decline phases, increased residency by adults could account for the observed increases in recapture rates at these times, at least in P. hermannsburgensis. Krebs et al. (1994) showed that low recapture rates of house mice were due to low trappability during the breeding season and to nomadic movements at other times. We have little evidence that trappability varies over time in either N. alexis or P. hermannsburgensis (Dickman et al. 1999a) and thus suggest that differences in residency between population phases account best for our results.

Two other sets of results support the interpretation that residency was relatively high during the population increase and decline phases. First, fidelity to burrows was high during these phases, with all tracked individuals returning to the same burrow at least once and several using the same burrow for periods of 6–8 days. By contrast, burrow fidelity was low during bust periods, with no individuals using the same burrow for runs of more than 3 consecutive days and many not returning to a previously used burrow over the tracking period. Second, rates of movement of tracked animals were approximately 2-fold greater during the population-low phase than during the increase and decrease phases. In support of the observations by Predavec (1994) that breeding females were relatively sedentary, we also found that females moved less than males in both our study species but that this sex difference in movements was consistent during all phases of the population cycle. Movements based on recaptures revealed no clear patterns. However, as radiotracking showed that the 2 study species moved at an overall average speed of ∼60 m/h, they could traverse a trapping grid readily within 1–2 h. This suggests that the grids were simply too small to provide a reliable index of nightly movements.

The reduction in movements and increased burrow fidelity during population booms was accompanied by an increase in the numbers of animals using the same burrows. As reproduction and juvenile survival increase during eruptions, and some reproduction continues during the decline phases (Predavec 1994; Ricci 2003), adults and their young probably contributed most to the increases in burrow occupancy. On 1 occasion a female P. hermannsburgensis was observed leaving a burrow with 3 pups attached to her nipples, and adult N. alexis were observed on 4 occasions either retrieving pups that had strayed from the burrow entrance or shepherding them during brief foraging bouts within a radius of 4–5 m of the burrow. As up to 7 N. alexis and 11P. hermannsburgensis were observed exiting from burrows during the increase or decrease phases, and usual litter sizes do not exceed 4–5 (Breed and Ford 2007; Jackson 2003), P. hermannsburgensis, at least, appears to tolerate additional animals within burrows beyond the parent–offspring unit. We do not know whether additional animals represent extended family members or nonrelatives. However, our single observation of 7 N. alexis emerging from a burrow suggested that the group contained 2 adults and 5 young, hence allowing the possibility that this large group was a single family unit.

Although both rodent species can be maintained communally in captivity (Happold 1976), the greater variance in numbers of P. hermannsburgensis that shared burrows, and the lack of a significant difference in this variable between population phases, suggest that this species shows more flexibility in this aspect of its social behavior than N. alexis. The mating system of P. hermannsburgensis appears to be promiscuous, and the propensity of females during estrus to mate with multiple males suggests further that sperm competition is likely (Breed and Ford 2007; Jackson 2003). In contrast, N. alexis has relatively small testes (0.1–0.2% of body weight, an order of magnitude less massive than in P. hermannsburgensis), a low rate of sperm production, and is thought not to exhibit either sperm competition or multiple paternity within litters (Breed 1997; Breed and Washington 1991). Although our observations provide little insight into the mating systems of the 2 species, they are consistent with the notion that N. alexis exhibits a cohesive family structure that is less extended than in P. hermannsburgensis (Stanley 1971). If this is correct, variance in the numbers of N. alexis in burrows may be reduced within population phases if burrows are occupied largely by pairs with or without their dependent young.

The GUD results indicate that animals spent more time at the food patches during the population-low phases than during the eruptions and probably reflect the reduced availability of food in the broader environment during the bust periods. The similarity of the GUDs during the increase and decrease phases, and their high values, provide further evidence that animals were not strongly resource-limited during these phases. In earlier work in the same area Kotler et al. (1998) reported similarly high GUD values for P. hermannsburgensis and suggested that this species either did not like peanuts or took few of them because of the richness of food in the background environment. As both species of rodents readily ate peanuts at times, the latter explanation appears most plausible.

Despite spending more time at the food patches during bust phases, N. alexis, at least, returned lower GUDs while foraging under spinifex than while foraging in the open. Introduced terrestrial carnivores such as the red fox and feral cat occur in the study area, as do predatory marsupials such as the mulgara (Dasycercus blythi; Woolley 2005, 2006). Barn owls occur also, albeit uncommonly. Encounters with such predators are more likely to occur in open habitats than under the protective cover of spinifex (Dickman 2009; Dickman et al. 1991; Mahon 1999), suggesting that the lower GUDs under spinifex reflect risk-aversive behavior or more time spent foraging where it is safe to do so. P. hermannsburgensis returned similar GUDs in open and spinifex habitats during the bust phase, suggesting that it showed little sensitivity to predation risk at this time. In contrast to the fast–moving and saltatorial N. alexis, P. hermannsburgensis is quadrupedal and could be at particularly high risk in open habitats. Pocket mice (Perognathus) and other quadrupedal heteromyids, for example, usually forage preferentially under cover and reduce their risk of predation by doing so (Kotler and Brown 1988; Price and Waser 1985). However, similar results have been obtained previously for P. hermannsburgensis (Kotler et al. 1998) and for other small mammal species in the same area (Haythornthwaite and Dickman 2000). We speculate that as predator activity is reduced during prolonged dry periods (Mahon 1999) it is possible that P. hermannsburgensis is sensitive to this. Certainly, both rodents returned lower GUDs in spinifex than in the open habitats during the increase and decrease phases when predator activity is higher, indicating that they are risk-aversive at these times.

In conclusion, our results show that both P. hermannsburgensis and N. alexis can persist for months or years at very low densities and then increase their rates of capture by 1–2 orders of magnitude in the wake of drought–breaking rains. During the bust periods both species are dispersed, do not occupy obvious ranges, and exhibit high levels of mobility that may allow them to harvest food resources that are scarce and distributed in patches. In these respects the Australian desert rodents appear to differ profoundly from the more sedentary heteromyids of the North American deserts (Jones 1993; Reichman and Price 1993) and the murids of the northern (Fichet-Calvet et al. 1999) and southern African deserts (White et al. 1997). After rain, rodents in the study area show a reduction in their movements, become more sedentary, and show increased burrow-sharing with conspecifics. These changes in movements and social organization remain throughout the population boom periods and prevail even as populations are falling again to low levels. We suggest that future studies should seek to disentangle the relative roles of food and other factors in driving declines (Ford and Pitelka 1984; Krebs et al. 2001) and also identify how these factors contribute to the temporal shifts in movements and social organization that we have described.


We are indebted to D. and P. Smith and family for support and access to Ethabuka in the early years of this study and to Bush Heritage Australia for support, logistical assistance, and access to the reserve since 2004. We also are indebted to hundreds of volunteers, students, and assistants for help in the field, especially P. Banks, A. Frank, A. Glen, A. Haythornthwaite, P. Higgs, F. Koch, M. Letnic, P. Mahon, G. McNaught, T. Parratt, L. Pastro, M. Predavec, F. Quails, S. Ricci, and M. Tischler, and to the Australian Research Council for funding, the American Society of Mammalogists (ASM) for sponsoring the symposium in Albuquerque, New Mexico, and B. Blake, W. Breed, and C. Pavey for their constructive review comments on an earlier version of the manuscript. C. R. Dickman is especially grateful to P. Stapp and the ASM for supporting his participation in the symposium and for encouragement during the production of this manuscript.


  • Special Feature Editor was Barbara H. Blake.

Literature Cited

View Abstract