Abstract

We examined population dynamics and trophic ecology of a predator-prey system in the Simpson Desert, Australia, consisting of an assemblage of small mammals (body mass < 100 g) and 4 species of predators: the endemic letter-winged kite (Elanus scriptus), a nocturnal-hunting rodent specialist; and 3 introduced mammalian predators (dingo [Canis lupus dingo], European red fox [Vulpes vulpes], and house cat [Felis catus]). This is the 1st comprehensive study of the responses of both the kite and introduced carnivores to a rodent outbreak. The 3.5-year study period included a population outbreak of about 24 months duration involving 3 native rodent species. Mammalian predators and kites exhibited similar population responses. Kites immigrated into the area within 6 months of the outbreak commencing, and remained while rodent abundance was high; however, all birds left the area after rodent populations crashed within a 6-week period. Dingoes and foxes were more abundant than cats and both species increased during the outbreak. All carnivores were resident. The letter-winged kite fed almost entirely on rodents. Rodents were the main prey of the 3 mammalian predators during the outbreak; however, all species had intermediate niche breadths. Dietary overlap between the kite and each carnivore was high during the rodent outbreak. During a nonoutbreak period, predation on rodents by the red fox remained high, whereas that by the dingo declined. We estimated the number of average-sized rodents (body mass 32.65 g) eaten daily by a nonreproducing individual to range from 1 (letter-winged kite) to 6 (red fox). We also estimated that the 3 mammalian predators (combined) captured 11 times as many rodents per day as letter-winged kites. There is considerable potential for food-based competition between the kite and introduced mammalian predators, particularly the red fox and house cat, in arid Australia.

Small mammals in arid environments undergo population outbreaks that are characterized by both an irregular period (interval between successive density peaks) and highly variable amplitude (ratio of maximum to minimum population density—Brown and Heske 1990; Whitford 1976). This pattern contrasts with high northern latitudes where small mammal cycles are characterized by a regular period although the amplitude of cycles remains highly variable (Gilg 2002; Korpimäki and Krebs 1996). Spatial and temporal patterns in small mammal outbreaks in arid regions are determined by short- and long-term weather patterns that drive primary productivity and determine water availability (Dickman et al. 1999; Newsome and Corbett 1975). The unpredictable nature of rainfall results in food availability being highly variable through time. Peaks in abundance of small mammals result from exceptional rainfall events that stimulate pulses of primary production to which small mammals, especially rodents, display a delayed numerical increase (Corbett and Newsome 1987; Letnic et al. 2005). The interaction between population dynamics of small mammals and native predators in arid regions is poorly understood.

A significant additional factor influencing native predator–prey dynamics in arid regions, particularly Australia, is the presence of introduced carnivores. In arid and semiarid Australia, 3 species of mammalian predators occur: dingoes or wild dogs (Canis lupus dingo), house cats (Felis catus), and European red foxes (Vulpes vulpes). These species can reach high densities in periods after favorable rainfall when prey numbers peak (Letnic et al. 2005). Two of these species are recent introductions to Australia, whereas the dingo has been present for at least 3,500 years (Corbett 1995). The introduction and spread of house cats and red foxes over the past 150 years has occurred at the same time as the disintegration of the native terrestrial mammal fauna of arid and semiarid Australia (Johnson 2006; Short and Smith 1994). As a consequence, most extant terrestrial mammals in Australia's deserts occur at low abundance within limited areas of their former ranges (Kinnear et al. 2002). Evidence is available that predation by mammalian predators, together with starvation resulting from declines in food availability, contributes to the decline in numbers of rodents toward the completion of peaks in abundance in arid Australia (Letnic et al. 2005; Newsome and Corbett 1975). Such predation has the potential to disrupt native predator–prey dynamics; however, limited data on the numerical and functional response of native and introduced predators to rodent outbreaks in arid Australia have prevented further consideration of this topic.

This paper reports on a study of a predator–prey system in arid Australia. The system consisted of an assemblage of small (body mass < 100 g) rodents and dasyurid marsupials, and 4 predators: the letter-winged kite (Elanus scriptus) and 3 species of introduced mammals. The letter-winged kite is a nomadic species endemic to arid and semiarid Australia that is unique in being the only truly nocturnal raptor (Falconiformes). Kites are specialist predators on small mammals, especially rodents, and move into areas when abundance of rodents is high (del Hoyo et al. 1994). Letter-winged kites breed in colonies during such periods but do not persist at sites once rodents decline substantially (del Hoyo et al. 1994).

The goals of this study were to compare the numerical response of letter-winged kites and introduced mammalian predators to a rodent population outbreak and to assess the level of dietary overlap and potential for food competition among the predators. First, we examined the hypothesis that the numerical response of predators to rodent population outbreaks is similar for both mammalian predators and the letter-winged kite. An alternative prediction is that temporal partitioning of exploitation of rodents occurs among predators because the highly nomadic letter-winged kites rapidly move into outbreak areas and show a faster numerical response than do mammalian predators that exhibit a more limited capacity for movement and a considerably lower reproductive rate than rodents (Korpimäki et al. 2004). Second, we assessed the degree of dietary overlap between mammalian predators and letter-winged kites during the high phase of rodent population outbreaks. Specifically, we tested the hypothesis that rodents will dominate the diets of all species of predators at this time, resulting in very high levels of dietary overlap. An alternative explanation is that mammalian predators, all of which are dietary generalists, capture alternative prey that are present in greater abundance and broader variety during peaks of rodent abundance resulting in reduced dietary similarity among predators (Holmgren et al. 2006). Last, we quantified the level of predation on rodents by mammalian predators during the rodent population outbreak and contrasted this with predation levels by letter-winged kites.

Materials and Methods

Animal care and use.—We followed guidelines of the American Society of Mammalogists (Gannon et al. 2007) during animal capture and handling. Approval to conduct the research was provided by the Alice Springs Animal Experimentation Ethics Committee.

Study area.—The study area of approximately 300 km2 was located on the northwestern edge of the Simpson Desert, 230 km southeast of Alice Springs, Australia, in gibber desert (desert pavement). The climate of the region (and Australia in general) is dominated by the El Niño Southern Oscillation with an irregular periodicity and very low predictability. Rainfall is highly variable, although the majority falls from October to March. Average annual rainfall at the 3 weather stations nearest to the study area at New Crown (25°67′S, 134°83′E), Horseshoe Bend (25°21′S, 134°24′E), and Lilla Creek (25°56′S, 134°07′E) is 179.6 mm (n = 50 years), 270.1 mm (n = 25 years), and 147.3 mm (n = 10 years), respectively (Australian Bureau of Meteorology, in litt.). Seasonal temperatures are extreme, with summer maximums (December–February) often >40°C, and mean winter minimums (June–August) frequently <5°C. Humidity is low and evaporation is high. Strong southeasterly winds prevail year-round.

Abundance of small mammals.—We carried out small mammal trapping with Elliott live traps (Elliott Scientific Co., Upwey, Victoria, Australia) at 7 sites. We located trap sites across the main habitat types in the study area in proportion to their availability. Within the study area the major habitat type is gibber plains and rises with a dominant vegetation of tussock grassland. Shallow drainage areas supporting open woodlands of acacias and eucalypts occur within this landform. The remainder of the study area consists of tussock grassland on plains consisting of red earths, clay, or desert loam soils. As a consequence, 5 trap sites were located on gibber, with 1 site along a creek line, and the other on a gently undulating plain adjacent to a dune. Trap sites were a minimum of 1.0 km apart with the most widely spaced sites being >21.0 km apart.

Each site consisted of a line of either 25 or 50 traps spaced 10–15 m apart. Traps were baited with a mixture of peanut butter and rolled oats, and opened from sunset to sunrise for 3 (occasionally 2) consecutive nights per visit on 12 occasions (4,650 trap-nights) between December 1999 and March 2003. We identified, weighed, determined the sex, and checked the reproductive status of each animal captured. Because of the difference in capture effort between sites and the lack of permanent marking of individuals, we present data on population size as catch per unit effort rather than capture– mark–recapture. Numbers of captures per 100 trap-nights (1 trap-night = 1 trap open for 1 night) were calculated for each trap line. We present mean capture rate ± SE for each trapping session. We use trap lines as replicates.

Abundance of letter-winged kites.—We actively searched for letter-winged kites during each sampling period by examining potential roost sites. The letter-winged kite is a tree-roosting species that occupies communal roosts during the day. During nesting, roosts are located adjacent to nesting trees. Suitable trees were relatively rare and patchily distributed within the search area. Letter-winged kites may occupy several roosts during a given day; therefore, we used the maximum number of birds counted at a communal roost as the number of birds present during a given visit.

Abundance of mammalian predators.—We used a track-based count and spotlight count to develop an index of the abundance of dingoes, house cats, and red foxes during 6 sampling periods between May 2000 and July 2002. Given the large home ranges and low densities of mammalian predators, including the 3 target species, it was not possible to estimate absolute density.

Three 10-km track-survey transects were established in the study area, each separated by a distance of at least 5 km. During each survey period we counted tracks along each transect for 3 consecutive days. The day before the 1st count of a survey period, we cleared the transect of all previous tracks by towing a heavy drag behind a 4-wheel-drive vehicle. Tracks were counted the following morning by an observer driving an all-terrain vehicle at 10 km/h. Counts commenced approximately 30 min after sunrise and took an average of 75 min. We recorded for each track the identity of the species that produced it and the number of individuals present. The transect was prepared for the next day's count by towing a lightweight drag behind the all-terrain vehicle while counting.

For dingoes and foxes, sets of tracks were classed as new (belonging to different individuals) if separated by a minimum distance of 1 km and 0.5 km, respectively, from the previous tracks on the transect. For cats, we classed all observed tracks as new. The total number of tracks recorded across the 3 transects was summed to give the number of individuals observed per day.

Spotlighting was carried out along the track survey transects for 2 nights during each sampling period, either before or after the track surveys were conducted. Observations were made by an individual sitting on the roof of a 4-wheel-drive vehicle moving at approximately 15 km/h. We detected animals with the use of a 100-W spotlight. Surveys commenced approximately 1 h after sunset with each transect taking an average of 40 min. We used spotlight data to calculate an index of abundance for each species expressed as the number of animals observed each night.

We defined the number of individuals of each species present during each survey period as the maximum number obtained by either track counts or spotlight counts. This methodology underestimated the actual number for all species present by an unknown factor and therefore provides an index of population size. We applied the methodology in a consistent manner throughout the study and therefore it should accurately reflect changes in abundance across the study period.

Diet of letter-winged kites.—We collected fresh kite pellets below communal roosts on 4 occasions from August 2001 to September 2002. Pellets were air dried at the site and stored in plastic bags until transported to the laboratory where we prepared them for analysis by baking in an oven at 200°C for 24 h. We then randomly selected a sample of 120–140 pellets from each sample for analysis. We soaked pellets in warm soapy water to allow separation of hair, feathers, scales, and insect exoskeletons. We then placed the softened pellets through a 4-mm sieve to separate out large bones and put the remaining material through a 1-mm sieve to separate out the remaining bones. All material extracted from the pellets was stored for identification purposes.

The majority of material in the pellets was mammalian. Skulls, dentaries, and any other identifiable material were used in identification. Identification of dasyurid marsupials (Dasyuridae) was based primarily on tooth structure. We examined material under a 16 × Zeiss stereoscope (Carl Beiss, Inc., Gottingen, Germany) and compared the skull and jaw material with those held in a reference collection of central Australian mammals. The number of individuals of each prey species taken during a sampling period was determined by counting the number of skulls and left and right dentaries present and using whichever count was the highest.

Diet of mammalian predators.—We collected scats of all mammalian predators within the study area on 7 occasions between May 2000 and July 2002. Scats were searched for during the course of other activities. Five sampling periods were during a rodent population outbreak when letter-winged kites also were present in the study area. Cat scats (the most difficult to locate) also were collected at 2 other sites in the arid Northern Territory at similar latitudes and within 200 km of our study area. These cat scats were collected during the study period using identical methods and all 3 sites experienced rodent population outbreaks simultaneously (Eldridge et al. 2002). Dingo, fox, and cat scats were distinguished based on size, shape, odor, and color (Triggs 1996). A single scat was defined as 1 or more fecal pellets that appeared to have been deposited in 1 defecation event by a single animal. Only relatively fresh scats that were unbleached by age were collected to ensure that their content reflected current diet.

Scats were oven-dried at 80°C for 24 h and held in 70% ethanol solution for storage before analysis. Each scat was washed through graded sieves to break up prey remains into 2 size categories. Prey remains were initially sorted under a dissecting microscope and identified to the lowest taxonomic level possible. Plant material was not considered further. Reptiles were classified to family level based on scales and jaw bones, whereas birds could rarely be identified to familial level. Small mammals could be identified to species based on dentaries but hair of small mammals could rarely be identified below ordinal level. Therefore, we identified the majority of rodent material in predator scats only to the level of order. We defined the percentage occurrence of each prey taxon as the proportion of scats containing the taxon expressed as a percentage of the total number of scats analyzed. We calculated percentage frequency as the number of individuals of the taxon identified in scats expressed as a percentage of the total number of individuals of all prey taxa identified.

Prey selection by letter-winged kites.—We calculated Bonferroni simultaneous (95%) confidence intervals to test the null hypotheses that each category of prey was consumed in proportion to its abundance in trap samples (Alldredge and Ratti 1992; Byers et al. 1984). Bonferroni confidence intervals were calculated as follows:  

formula

where pi is the proportion of the ith species in the diet, k is the number of prey categories, and Zα/2k is the upper standard; normal table value corresponding to a probability tail area of α/2k. If the proportion of availability of a prey category fell above or below the confidence levels estimated for its occurrence in kite pellets then the null hypothesis was rejected and we concluded that the prey item was actively selected (above confidence level) or avoided (below confidence level).

Niche breadth and overlap.—We estimated dietary breadth for each of the 4 species of predators and calculated dietary overlap for each of the 6 species pairs. We used the percentage occurrence dietary data for these estimates. Levins' standardized formula for niche breadth (Krebs 1999) was used to calculate diet breadth: graphic where pi is the proportion of occurrence of each prey category in the predator's diet; and n is the number of prey categories in the predator's diet. BA ranges from 0 to 1, where a value close to 0 represents a narrow niche and one close to 1, a broad niche.

To assess the impact of high abundance of rodents on niche breadth, we calculated the niche breadth for both outbreak and nonoutbreak periods. Scats collected before May 2001 were clustered together and categorized as being from a non-outbreak period, whereas all other scats were clustered as being from the period of the rodent population outbreak.

We calculated dietary overlap between species pairs using Pianka's modification to the MacArthur and Levin measure of niche overlap (Krebs 1999): graphic where j and k are the 2 species being compared. Niche overlap ranges from 0 to 1 with 0 representing no overlap in diet and 1, complete overlap.

Predation rates.—The formula in Appendix I estimated the number of “average” rodents captured each day by a non-reproducing adult of each of the 4 species of predators during the rodent population outbreak. We calculated the body mass of an average rodent based on the abundance and mean body mass for each rodent species captured during the study (Table 1).

Table 1

Small mammal species captured during sampling in the Simpson Desert over 4,650 trap-nights during 1999–2003. Data on body mass are for the study populations except Rattus villosissimus (Menkhorst and Knight 2001).

Family and species Mean body mass (range) (g) No. captures Capture success (captures/ 100 trap-nights) 
Dasyuridae    
Sminthopsis macroura 16 (10–23) 51 1.10 
Antechinomys laniger 19 (15–24) 0.19 
Muridae    
Notomys alexis 32 (22–40) 166 3.57 
Pseudomys australis 42 (27–68) 325 6.99 
P. hermannsburgensis 13 (9–16) 136 2.92 
Rattus villosissimus 134 (60–280) 0.02 
Mus musculus 13 (11–14) 13 0.28 
Family and species Mean body mass (range) (g) No. captures Capture success (captures/ 100 trap-nights) 
Dasyuridae    
Sminthopsis macroura 16 (10–23) 51 1.10 
Antechinomys laniger 19 (15–24) 0.19 
Muridae    
Notomys alexis 32 (22–40) 166 3.57 
Pseudomys australis 42 (27–68) 325 6.99 
P. hermannsburgensis 13 (9–16) 136 2.92 
Rattus villosissimus 134 (60–280) 0.02 
Mus musculus 13 (11–14) 13 0.28 

Statistical analysis.—We performed correlation analysis (Pearson correlation coefficient) between the index of rodent abundance (captures/100 trap-nights) and the abundance indices for each of the 4 species of predators. We correlated abundance of each predator against capture rate of rodents for the current survey period and the previous 2 survey periods. For letter-winged kites, we also assessed abundance against that of each of the 3 most-abundant species of rodents. We used chi-square tests to assess variation in diet across sampling periods. We assessed adequacy of diet sample size for each species of predator by constructing a randomized cumulative curve (40 random selections of the sample order) for the number of prey taxa occurring against the number of scats sampled using the program Species Diversity and Richness 4.0 (Seaby and Henderson 2006).

Results

Trapping commenced at the completion of a year of below-average rainfall. Mean ± SE rainfall for New Crown, Horseshoe Bend, and Lilla Creek (see “Study area” above for details) in 1999 was 118.2 ± 7.91 mm. This period was followed by 2 years with rainfall well above average: 2000 (372.87 ± 6.84 mm) and 2001 (428.2 ± 28.62 mm), which included rainfall of >200 mm in February 2000. Rainfall in 2002 was well below average (102.93 ± 9.34 mm).

Population dynamics of small mammals.—Mammal trapping produced 701 captures of 7 species of small mammals (2 dasyurid marsupials and 5 murid rodents) during 4,650 trap-nights at a rate of 15 captures/100 trap-nights (Table 1). Three rodent species, plains rats (Pseudomys australis), sandy inland mice (P. hermannsburgensis), and spinifex hopping mice (Notomys alexis), accounted for 89% of all captures.

The 3.5-year study period included a complete population cycle of rodents (Fig. 1A). The initial 12 months was characterized by very low population levels and this was followed by a period of very high abundance that featured peaks in winter–spring of 2 consecutive years. Small mammal populations crashed in spring 2002, reaching very low numbers by November 2002 (Fig. 1A). The population crash occurred between 17 October and 25 November 2002. We did not capture any rodents in March 2003 or during subsequent trapping in 2003 and 2004 (not graphed).

Fig. 1

Time series plots of capture rates of small mammals in the Simpson Desert, Australia. A) rodents <25 g and >25 g, and B) dasyurid marsupials.

Fig. 1

Time series plots of capture rates of small mammals in the Simpson Desert, Australia. A) rodents <25 g and >25 g, and B) dasyurid marsupials.

The population outbreaks of small (<25-g) and large (>25-g) rodents were generally similar, although the abundance of species >25 g was considerably higher (Fig. 1A). All but 1 capture of rodents in the >25-g class was either P. australis or N. alexis (Table 1). The abundance of species <25 g (mostly P. hermannsburgensis) increased earlier and the decline in abundance of this group commenced about 2 months earlier than the >25-g species. The abundance of dasyurid marsupials was considerably lower than the 2 classes of rodents (Fig. 1B). Overall, abundance of dasyurids did not exhibit the extreme variation of the 2 size classes of rodents. However, in October 2002 the capture rate of dasyurids was more than 3 times higher than during any other sampling period (Fig. 1B).

Abundance of letter-winged kites.—Letter-winged kites were 1st detected in the study area in May 2001 (Fig. 2A), within 6 months of small mammal abundance beginning to increase. Abundance of kites peaked in winter 2002 (July–August), when a large flock inhabited the study area for a period spanning 6 weeks. The majority of this large group only remained in the study area for a short time. We last observed birds at the site in November 2002. Abundance of letter-winged kites was not correlated with changes in abundance of all rodents or of P. australis, P. hermannsburgensis, and N. alexis during the current or previous survey periods (P > 0.05).

Fig. 2

Abundance of A) letter-winged kites and rodents, and B) dingoes, red foxes, and house cats in the Simpson Desert, Australia.

Fig. 2

Abundance of A) letter-winged kites and rodents, and B) dingoes, red foxes, and house cats in the Simpson Desert, Australia.

Abundance of mammalian predators.—The abundance index for both dingoes and red foxes was higher than that for house cats during all but the initial survey period (Fig. 2B). The abundance index for house cats was low across the study period and typically, we observed only 1 or 2 individuals during track and spotlight counts. The abundance index for both dingoes and red foxes increased during autumn 2001, when abundance of rodents began to increase, and was highest for both species in March 2002, 6 months after the abundance of rodents had peaked. However, we have no data on changes in abundance after June 2002. Abundance of red foxes was positively correlated with changes in abundance of rodents in the previous survey period (r2 = 0.94, P = 0.02). Abundance of both dingoes and cats did not show a significant correlation with abundance of rodents during the current or previous survey periods (P > 0.05).

Diet of letter-winged kites.—The diet of the letter-winged kite was composed almost entirely of small mammals, with rodents being the dominant order of prey (Table 2). Small numbers of dasyurids were consumed as well as a single reptile and 2 invertebrates. At the species level, the composition of rodents in the diet varied across sampling periods (Fig. 3). Analyses of prey consumption versus availability were carried out for the 3 most abundant species of rodents (P. australis, P. hermannsburgensis, and N. alexis), other rodents combined, and dasyurids combined (Fig. 3). P. australis was consumed in numbers similar to its availability except in September 2002, when it was selected. Kites selected the smaller P. hermanns-burgensis in August 2001, and consumed it in proportion to its abundance over the remaining sampling periods. N. alexis was consumed in a lower proportion than its availability, as measured by trap captures, during all sampling periods. The “other rodent” category was selected in November 2001 and September 2002. This category included a native rodent, Leggadina forresti (15–25 g), which, although captured by kites, did not appear in our trap samples (Table 1). The dasyurid marsupial category appeared in the diet in a proportion similar to its abundance except in September 2002, when kites avoided this category.

Table 2

Diet composition of scats or pellets of letter-winged kites, dingoes, red foxes, and house cats in the Simpson Desert, 2000–2002. Data for mammalian predators includes both rodent outbreak and nonoutbreak periods. A, percentage frequency = number of individuals of the taxon in scats/total number of individuals of all prey in scats; B, percentage occurrence = proportion of scats that contained the taxon.

 Letter-winged kite (0.2–0.4 kg) Dingo (12–24 kg) Red fox (3.5–8 kg) House cat (2.5–6.5 kg) 
Prey taxon 
Invertebrates (total) 0.28 0.43 10.27 26.58 12.17 22.22 12.22 25.00 
Vertebrates         
Reptilia (total) 0.14 0.21 10.15 22.78 1.74 3.17 4.44 9.09 
Aves (total) — — 3.67 9.49 6.09 11.11 10.00 20.45 
Mammalia (total) 99.59 98.93 75.92 97.15 80.00 96.83 73.33 100.00 
Dasyuromorphia         
Dasyuridae 3.08 4.49 — — — — 1.11 2.27 
Diprotodontia         
Macropodidae — — 6.97 18.04 4.35 7.94 2.22 4.55 
Rodentia         
Muridae 96.51 97.01 50.98 70.25 71.30 95.24 61.11 93.18 
Lagomorpha         
Rabbit — — 9.05 23.42 0.87 1.59 6.67 13.64 
Carnívora         
House cat — — 1.47 3.80 — — — — 
Artiodactyla         
Cattle — — 7.46 19.30 3.48 6.35 2.22 4.55 
No. scats or pellets 468  316  63  44  
No. prey items 715  818  115  90  
 Letter-winged kite (0.2–0.4 kg) Dingo (12–24 kg) Red fox (3.5–8 kg) House cat (2.5–6.5 kg) 
Prey taxon 
Invertebrates (total) 0.28 0.43 10.27 26.58 12.17 22.22 12.22 25.00 
Vertebrates         
Reptilia (total) 0.14 0.21 10.15 22.78 1.74 3.17 4.44 9.09 
Aves (total) — — 3.67 9.49 6.09 11.11 10.00 20.45 
Mammalia (total) 99.59 98.93 75.92 97.15 80.00 96.83 73.33 100.00 
Dasyuromorphia         
Dasyuridae 3.08 4.49 — — — — 1.11 2.27 
Diprotodontia         
Macropodidae — — 6.97 18.04 4.35 7.94 2.22 4.55 
Rodentia         
Muridae 96.51 97.01 50.98 70.25 71.30 95.24 61.11 93.18 
Lagomorpha         
Rabbit — — 9.05 23.42 0.87 1.59 6.67 13.64 
Carnívora         
House cat — — 1.47 3.80 — — — — 
Artiodactyla         
Cattle — — 7.46 19.30 3.48 6.35 2.22 4.55 
No. scats or pellets 468  316  63  44  
No. prey items 715  818  115  90  
Fig. 3

Changes in proportion of use (open circles) ± 95% confidence level and abundance (filled circles) of 5 categories of mammalian prey of letter-winged kites across 4 sampling periods in the Simpson Desert, Australia.

Fig. 3

Changes in proportion of use (open circles) ± 95% confidence level and abundance (filled circles) of 5 categories of mammalian prey of letter-winged kites across 4 sampling periods in the Simpson Desert, Australia.

Diets of mammalian predators.—Although the sample sizes for red foxes and house cats are smaller than that for dingoes (Table 2), the randomized cumulative curve of prey taxa reached an asymptote for both species, indicating sample sizes were adequate. Five of the 7 sampling periods for mammal scats were during the rodent population outbreak (May 2001–July 2002). Relatively few scats were collected during the period of low rodent abundance (65, 10, and 1 scats for dingoes, foxes, and cats, respectively).

Rodents were the main prey of all 3 mammalian predators, being found in >90% of scats of red foxes and house cats and in 70% of dingo scats (Table 2). Dingoes also regularly consumed rabbits, cattle, macropods, reptiles, and invertebrates. Secondary prey of red foxes included invertebrates and birds, whereas cats also preyed on invertebrates, birds, and rabbits.

Dietary niche breadth and overlap.—We calculated niche breadth for all 4 species of predators using occurrence data across 9 food categories: invertebrates, birds, reptiles, and 6 groups of mammals (dasyurids, macropods, rodents, rabbits, cats, and cattle). During the rodent outbreak, letter-winged kites had a very narrow niche (BA = 0.035), whereas the broadest niche was that of dingoes (BA = 0.46). House cats (BA = 0.28) and red foxes (BA = 0.14) had intermediate niche-breadth values.

Both dingoes and red foxes had a much broader dietary niche during the nonoutbreak period (dingoes, BA = 0.85; foxes, BA 0.56). Percentage occurrence of rodents in the diet of dingoes was markedly higher during the rodent outbreak, whereas occurrence of all other prey categories (except cats and macropods) declined (Fig. 4). Percentage occurrence of rodents in the diet of foxes showed little change across the 2 periods, whereas occurrence of cattle, birds, reptiles, and invertebrates was higher in the nonoutbreak period. The difference in occurrence of prey in the diet between rodent outbreak and nonoutbreak periods was significant for both dingoes (χ2 = 50.42, d.f. = 7, P < 0.001; based on all prey categories except dasyurids) and red foxes (χ2 = 13.76, d.f. = 5, P < 0.01; based on all prey categories except rabbits, dasyurids, and cats).

Fig. 4

Comparison of the percentage occurrence of prey categories in the diet of A) dingoes and B) red foxes during outbreak and nonoutbreak conditions in the Simpson Desert, Australia.

Fig. 4

Comparison of the percentage occurrence of prey categories in the diet of A) dingoes and B) red foxes during outbreak and nonoutbreak conditions in the Simpson Desert, Australia.

Dietary overlap between letter-winged kites and house cats and red foxes was very high (>0.90), and overlap with dingoes was moderately high during the rodent outbreak (Table 3). Dietary overlap among the 3 mammalian predators also was very high at this time (Table 3). Overlap between dingoes and foxes decreased during the nonoutbreak period.

Table 3

Dietary overlap indices for the 4 species of predators in the Simpson Desert. Diets of letter-winged kites and house cats were not assessed during the nonoutbreak period.

 Rodent outbreak (May 2001–September 2002) Nonoutbreak (June 2000–May 2001) 
Dingo Red fox House cat Red fox 
Letter-winged kite 0.88 0.98 0.93  
Dingo  0.93 0.96 0.76 
Red fox   0.98  
 Rodent outbreak (May 2001–September 2002) Nonoutbreak (June 2000–May 2001) 
Dingo Red fox House cat Red fox 
Letter-winged kite 0.88 0.98 0.93  
Dingo  0.93 0.96 0.76 
Red fox   0.98  

Predation rates.—The daily biomass (g) of rodents captured by a nonreproducing adult of each species of predator was estimated based on the formula in Appendix I as: letter-winged kite, 39 g; dingo, 75 g; house cat, 111 g; and red fox, 191 g. This formula adjusted the biomass contribution of large mammalian prey to the maximum daily food consumption of the species of predator. For example, if a dingo captured a macropod with body mass of 2,000 g, we considered that this contributed only 1,000 g to the diet because this is the daily food intake of the dingo (DC in Appendix I). Based on the biomass values given above, the estimated daily number of average-sized rodents (body mass of 32.65 g; Appendix I) consumed by an individual of each species (rounded to the nearest whole number) was letter-winged kite, 1; dingo, 2; house cat, 3; and red fox, 6.

Discussion

The 3.5-year study period included a rodent population outbreak in response to 2 consecutive years of rainfall more than double the annual average. The duration of the outbreak from initial increase to completion of the decline phase was almost 24 months. This duration is similar to findings for the study species elsewhere in arid Australia (Brandie and Moseby 1999; Dickman et al. 1999; Letnic et al. 2005).

Our study is the 1st to assess population responses of both introduced mammalian predators and a native rodent specialist to a population outbreak of native rodents. Letter-winged kites moved into the study area in May 2001, within 6 months of the beginning of the rodent population outbreak, and remained for 18 months. All birds left the area after rodent populations crashed within a 6-week period. The 3 species of mammalian predators were resident in the study area. The index of abundance of foxes and dingoes began increasing in autumn 2001 (Fig. 2B) at the same time that numbers of rodents were increasing and as the kites migrated into the study area (Fig. 2A). Abundance of foxes and dingoes was highest in early 2002, following the peak in abundance of rodents in winter 2001. Although the amplitude of the change varied across the species of predators, our findings support the hypothesis of a similar numerical response to rodent population outbreaks by both the kites and mammalian predators. Little potential for temporal partitioning of prey exploitation appears possible in this system.

The rodent outbreak in our study was driven primarily by 3 species with mean body mass <50 g. These species differed in the timing of their response to the increased availability of food in 2000–2001 with N. alexis and P. hermannsburgensis peaking in winter-spring 2001 and P. australis in autumn-winter 2002. The trajectory and amplitude of the population outbreaks of rodents in this study are in line with other studies of native rodents, including these 3 species, in the Simpson Desert and elsewhere in arid Australia (Brandie and Moseby 1999; Dickman et al. 1999; Letnic et al. 2005). In particular, the peaks in abundance over successive winters shown in our study (Fig. 1A) also were exhibited by P. hermannsburgensis and N. alexis in the Simpson and Great Sandy Deserts (Dickman et al. 1999).

Dietary overlap between letter-winged kites and each of the mammalian predators was high to very high during the rodent population outbreak (Table 3). Rodents were the major component of the diet for all 4 species, suggesting that predators were concentrating on the most abundant prey during the rodent outbreaks. Such optimal foraging is characteristic of dingoes in arid Australia, where they specialize on the most abundant prey at any given time, switching prey species as abundance changes over time (Newsome 1990). Red foxes and house cats at sites in the eastern Simpson Desert also concentrate on rodents when these are abundant but include other prey in the diet, including reptiles, birds, invertebrates, and carrion, when rodent numbers are low (Mahon 1999).

Letter-winged kites captured the 2 species of Pseudomys in greatest abundance, taking them in proportion to availability during most of the sampling periods. In contrast, kites took N. alexis in lower numbers than available during all sampling periods (Fig. 3). N. alexis is the only bipedal rodent in the study area and it is likely to be similar to other bipedal rodents in possessing superior sprint speed and unpredictable escape behavior in open habitats compared to sympatric quadrapedal rodents (Kotler et al. 1994). This increased predator avoidance ability in open space, which is typical of our study area, may increase the probability of it avoiding kite predation. All 3 species are readily captured in the aluminum traps used in our study (C. Pavey, in litt.), therefore, it is unlikely that N. alexis was overrepresented in our trap samples.

The letter-winged kite is a highly nomadic species that erupts in areas after high rainfall and associated increases in primary productivity that trigger population outbreaks and large-scale dispersal events of rodents. Breeding by kites apparently occurs only during rodent population outbreaks and letter-winged kites have a series of adaptations that enables rapid reproductive output (summarized in Marchant and Higgins [1993]). Once densities of prey decrease in an area, kites leave and search for food elsewhere (Marchant and Higgins 1993; J. D. Pettigrew, in litt.), or undergo mass mortalities resulting from starvation (Marchant and Higgins 1993), or both. In our study, kites were no longer present once rodents had disappeared from the trapping plots, although dasyurid marsupials persisted. Therefore, during nonoutbreak periods, the nomadic nature of the kites ensures that competition for food with mammalian predators does not occur.

The high levels of dietary overlap between letter-winged kites and mammalian predators revealed in this study indicate the possibility that competition for food may occur during rodent outbreaks, particularly between kites and red foxes and house cats. Examination of our data indicates that letter-winged kites have a very narrow dietary niche and, therefore, have limited alternative food sources, whereas mammalian predators in arid Australia all have relatively broad niches and can exploit alternative prey (Dickman 1996; Mahon 1999; Paltridge 2002). Competition between kites and mammalian predators, if it does take place, is most likely to occur at the completion of population peaks when numbers of rodents are declining because of starvation but numbers of mammalian predators are high, as is predation on rodents (Newsome and Corbett 1975; Mahon 1999). Although kites can disperse during this stage of rodent outbreaks, individuals persisted until rodent populations collapsed (Fig. 2A), probably because alternative food resources are limited at a landscape scale. A consequence of any competition for food at this stage of rodent outbreaks will be a reduction in survivorship of kites during nonoutbreak periods.

Although high levels of dietary overlap indicate the possibility of food-based competition between kites and mammalian predators, this topic requires further research. Direct manipulative experiments that enable long-term reductions of mammalian predators and measure the impact on population dynamics of rodents (e.g., Banks 1999) are required. A manipulative field experiment in the Simpson Desert has established that predation by foxes and cats limits the postoutbreak density of P. hermannsburgensis, but that it does not dampen population responses of rodents to subsequent rainfall (Mahon 1999). However, this experiment could not continue over a long time span because of the advent of unpredictable rainfall (Mahon 1999). We cannot overemphasize the logistic challenges to carrying out long-term manipulative experiments in arid Australia where rainfall is highly unpredictable, as are the resultant population outbreaks of rodents.

Based on the dietary data from our study and daily food consumption rates of the predator species, the number of average-sized rodents eaten daily by a nonreproducing individual ranged from 1 (letter-winged kite) to 6 (red fox). The estimated daily rodent-capture rate for the 3 mammalian predators (combined) was 11 times that of the letter-winged kite. Of the 3 mammalian predators, dingoes exerted relatively low predation pressure on rodents. Dingoes had the broadest dietary niche during the period of low density of rodents, when occurrence of rodents in scats was much lower (Fig. 4A). In comparison, although red foxes and house cats both have comparatively broad dietary niches, rodents are an important component of the diet of each species in arid Australia during outbreak and nonoutbreak periods (Fig. 4B; Mahon 1999; Paltridge 2002). This finding supports previous research that has identified red foxes and house cats as significant predators of remnant populations of small mammals in Australia (Dickman 1996; Kinnear et al. 2002).

Acknowledgments

CRP thanks F. Geiser and staff at Zoology, University of New England, for support and encouragement during the write-up of this study. R. Paltridge identified prey in all the mammalian predator scats and J. Gorman assisted with identification of prey in letter-winged kite pellets. B. Shakeshaft and A. Nano provided invaluable support with fieldwork. J. Hughes and J. Cole assisted with preparation and identification of prey in letter-winged kite pellets. G. Medlin of the South Australian Museum in Adelaide provided expert advice in the identification of mammalian prey in letter-winged kite pellets. Rangers from the Parks and Wildlife Service provided field assistance. We thank C. Krebs, C. Dickman, and A. Glen for comments and critiques of previous drafts of the manuscript. Funding for the research was provided by the Bureau of Rural Sciences. We thank the Bloomfield family for allowing us to work on their property.

Literature Cited

Alldredge
J. R.
Ratti
J. T.
.
1992
.
Further comparison of some statistical techniques for analysis of resource selection
.
Journal of Wildlife Management
 
56
:
1
9
.
Banks
P. B.
1999
.
Predation by introduced foxes on native bush rats in Australia: do foxes take the doomed surplus?
Journal of Applied Ecology
 
36
:
1063
1071
.
Brandle
R.
Moseby
K. E.
.
1999
.
Comparative ecology of two populations of Pseudomys australis in northern South Australia
.
Wildlife Research
 
26
:
541
564
.
Brown
J. H.
Heske
E. J.
.
1990
.
Temporal changes in a Chihuahuan Desert rodent community
.
Oikos
 
59
:
290
302
.
Byers
C. R.
Steinhorst
R. K.
Krausman
P. R.
.
1984
.
Clarification of a technique for analysis of utilization–availability data
.
Journal of Wildlife Management
 
48
:
1050
1053
.
Corbett
L. K.
1995
.
The dingo in Australia and Asia
 .
University of New South Wales Press
,
Sydney, New South Wales, Australia
.
Corbett
L. K.
Newsome
A. E.
.
1987
.
The feeding ecology of the dingo III. Dietary relationships with widely fluctuating prey populations in arid Australia: an hypothesis of alternation of predation
.
Oecologia
 
74
:
215
227
.
del Hoyo
J.
Elliott
A.
Sargatal
J.
(eds.).
1994
.
Handbook of the birds of the world
 . Vol.
2
.
New World vultures to guineafowl
.
Lynx Edicions
,
Barcelona, Spain
.
Dickman
C. R.
1996
.
Overview of the impact of feral cats on Australian native fauna
 .
Australian Nature Conservation Agency
,
Canberra, Australia
.
Dickman
C. R.
Mahon
P. S.
Masters
P.
Gibson
D. F.
.
1999
.
Long-term dynamics of rodent populations in arid Australia: the influence of rainfall
.
Wildlife Research
 
26
:
389
403
.
Eldridge
S. R.
Shakeshaft
B. J.
Nano
T. J.
.
2002
.
The impact of wild dog control on cattle, native and introduced herbivores and introduced predators in central Australia
.
Final Report to Bureau of Rural Sciences
 .
Northern Territory Parks and Wildlife Commission, Alice Springs
,
Northern Territory, Australia
.
Gannon
W. L.
Sikes
R. S.
the Animal Care and Use Committee of the American Society of Mammalogists
.
2007
.
Guidelines of the American Society of Mammalogists for the use of wild mammals in research
.
Journal of Mammalogy
 
88
:
809
823
.
Gilg
O.
2002
.
The summer decline of the collared lemming, Dicro-stonyx groenlandicus, in high arctic Greenland
.
Oikos
 
99
:
499
510
.
Holmgren
M.
et al
.
2006
.
A synthesis of ENSO effects on drylands in Australia, North America and South America
.
Advances in Geo-sciences
 
6
:
69
72
.
Johnson
C.
2006
.
Australia's mammal extinctions: a 50 000 year history
 .
Cambridge University Press
,
Cambridge, United Kingdom
.
Kinnear
J. E.
Sumner
N. R.
Onus
M. L.
.
2002
.
The red fox in Australia—an exotic predator turned biocontrol agent
.
Biological Conservation
 
108
:
335
359
.
Korpimäki
E.
Brown
P.
Jacob
J.
Pech
R. P.
.
2004
.
The puzzles of population cycles and outbreaks of small mammals solved?
BioScience
 
54
:
1071
1079
.
Korpimäki
E.
Krebs
C. J.
.
1996
.
Predation and population cycles of small mammals: a reassessment of the predation hypothesis
.
BioScience
 
46
:
754
764
.
Kotler
B. P.
Brown
J. S.
Mitchell
W. A.
.
1994
.
The role of predation in shaping the behaviour, morphology and community organisation of desert rodents
.
Australian Journal of Zoology
 
42
:
449
466
.
Krebs
C. J.
1999
.
Ecological methodology
 .
Benjamin-Cummings
,
Menlo Park, California
.
Letnic
M.
Tamayo
B.
Dickman
C. R.
.
2005
.
The responses of mammals to La Niña (El Niño Southern Oscillation)–associated rainfall, predation, and wildfire in central Australia
.
Journal of Mammalogy
 
86
:
689
703
.
Mahon
P. S.
1999
.
Predation by feral cats and red foxes and the dynamics of small mammal populations in arid Australia
.
Ph.D. dissertation
 ,
University of Sydney
,
Sydney, New South Wales, Australia
.
Marchant
S.
Higgins
P. J.
(eds.).
1993
.
Handbook of Australian, New Zealand and Antarctic birds
. Vol.
2
.
Raptors to lapwings
 .
Oxford University Press
,
Melbourne, Victoria, Australia
.
Menkhorst
P. W.
Knight
F.
.
2001
.
A field guide to the mammals of Australia
 .
Oxford University Press
,
Melbourne, Victoria, Australia
.
Newsome
A.
1990
.
The control of vertebrate pests by vertebrate predators
.
Trends in Ecology and Evolution
 
5
:
187
191
.
Newsome
A. E.
Corbett
L. K.
.
1975
.
VI. Outbreaks of rodents in semi-arid and arid Australia: causes, preventions, and evolutionary considerations
. Pp.
117
153
in
Rodents in desert environments
  (
Prakash
I.
Gosh
P. K.
, eds.).
Dr. W. Junk
,
The Hague, The Netherlands
.
Paltridge
R.
2002
.
The diets of cats, foxes and dingoes in relation to prey availability in the Tanami Desert, Northern Territory
.
Wildlife Research
 
29
:
389
403
.
Seaby
R. M.
Henderson
P. A.
.
2006
.
Species diversity and richness. Version 4
 .
Pisces Conservation International
,
Lymington, United Kingdom
.
Short
J.
Smith
A.
.
1994
.
Mammal decline and recovery in Australia
.
Journal of Mammalogy
 
75
:
288
297
.
Triggs
B.
1996
.
Tracks, scats and other traces
.
A field guide to Australian mammals
 .
Oxford University Press
,
Melbourne, Victoria, Australia
.
Whitford
W. G.
1976
.
Temporal fluctuations in density and diversity of desert rodent populations
.
Journal of Mammalogy
 
57
:
351
369
.

Appendix I

Calculation of the daily predation rate on rodents by predators during the rodent population outbreak. The total number of “average-sized” rodents eaten by a nonreproducing adult of each predator species each day was calculated as follows:  

formula

where RP is the number of rodents eaten during a 24-h period; DC is the daily food consumption of predator species; PR is the proportion biomass of rodents in the diet during rodent outbreak; and RBM is the index of body mass of an average rodent (32.65 g). Daily food consumption (DC) was estimated for each species of predator as: letter-winged kite, 40 g (based on proportion of body mass of other raptors and owls); dingo, 1,000 g; house cat, 500 g; and red fox, 500 g (estimates for carnivores based on Paltridge [2002]). Proportion biomass of rodents (PR) in the diet of each species of predator was based on data collected during the rodent outbreak: letter-winged kite, 0.98; dingo, 0.08; house cat, 0.22; and red fox, 0.38. Biomass contribution of mammals considered too large to be consumed by a predator during a 24-h period was adjusted on the basis of daily food consumption data given above.

Index of body mass of an average rodent (RBM) was calculated from data on abundance and mean body mass given in Table 1.

Author notes

Associate Editor was Gerardo Ceballos.