BConsultants, Conservation Commission ofthe Northern Territory, P.O. Box 1046, Alice Springs, N.T.. 5750. Abstract. The effect of rabbit control methods on ...
Aust. Wildl. Res., 1985, 12, 237-47
The Response of Rabbit Populations and Vegetation to Rabbit Control on a Calcareous Shrubby Grassland in Central Australia B. D. Foran*, W. A. LowB and B. W. StrongB *Division of Wildlife and Rangelands Research, CSIRO, Central Australian Laboratory, P.O. Box 21 11, Alice Springs, N.T. 5750. BConsultants, Conservation Commission ofthe Northern Territory, P.O. Box 1046, Alice Springs, N.T. 5750.
Abstract The effect of rabbit control methods on rabbit populations and subsequent changes in vegetation were assessed over 2 years on a calcareous shrubby grassland in central Australia. Warren ripping and fumigating decreased mean population levels to four rabbits per kilometre of spotlight transect, and 1080 poisoning decreased mean levels to 9 km-1, compared to the untreated levels of 20 km-l. Release of European rabbit fleas did not have any significant effect on rabbit populations. In the untreated rabbit populations, 68% of the fluctuation in numbers was accounted for by total rainfalls in the previous summer and the previous winter, and by a forage greenness rating. Experimental control of rabbits combined with exclosures indicated that grazing by high numbers of rabbits reduced the frequency of Enneapogon spp. at one site whereas levels rose in all other sites during the two years of good rainfall. The frequency of several species, such as Chenopodium cristatum, Phyllanthus rhytidospermus and Portulaca oleracea, was highly correlated with rabbit density in the previous year, suggesting that rabbit grazing promotes their increase. Grazing'by rabbits alone depressed standing biomass in the following season by 300 kg ha-1 from a maximum 1200 kg ha-1 within the exclosures. Cattle grazing further decreased biomass by 150 kg ha-1. Seedlings of Acacia kempeana were almost absent in those areas open to rabbit grazing, but they were reasonably dense where rabbits were excluded. However, the vegetation response over the 2 years of experiment was due mainly to seasonal changes rather than rabbit control, and does not justify the expense of rabbit control at this stage. Substantial improvement in vegetation composition over the longer term may alter this conclusion.
Introduction The successful release of the European rabbit Oryctolagus cuniculus into the wild in mainland Australia in 1859 (Rolls 1969) started a series of changes to the flora and fauna which are still in train today. In addition to losses caused by simple removal of crops and pastures in agricultural areas (Myers and Poole 1963), wildlife habitat has been altered with consequent loss of animal species (Newsome and Corbett 1975), and plant communities in pastoral areas have been changed by the grazing and browsing effects of rabbits (Cochrane and McDonald 1966; Lange and Graham 1983; Friedel 1985). Introduction of myxomatosis in 1950 reduced rabbit numbers by up to 95% in much of Australia, but only persistent and diligent application of traditional methods has eradicated rabbits from rangeland areas (Parker et al. 1976). The lack of suitable insect vectors for myxomatosis, and the vast areas of suitable habitat in the arid and semiarid rangelands of Australia, allow rabbit grazing and browsing at low to moderate levels in most years. However, the release of the European rabbit flea Spzlopsyllus cuniculi has resulted in further
B. D. Foran, W. A. Low and B. W. Strong
reduction of rabbits in some semiarid areas (Cooke 1983). Rabbits may reach high numbers during suitable seasons, and in subsequent drought cause considerable damage; moreover, the few studies of vegetation that have been undertaken show that even low numbers can remove successive generations of regenerating trees and shrubs (Henzell and Lay 1981; Lange and Graham 1983). Rabbits reached the Northern Territory in the mid- 1890s (Strong 1983), a n d are now found in greatest numbers on calcareous country, fringing sand dunes, granite country and sandy stream frontage country (Low and Strong, unpublished data). Although isolated pockets of rabbits are found further north in the continent, the ratio of summer to winter rainfall, and suitability of habitats, suggest that central Australia below the Tropic of Capricorn approaches the limit of the northern distribution of large populations. Periodic rabbit plagues combined with domestic stock grazing produce demands, from land managers and government alike, that some solution be found for the 'rabbit problem'. Following a statewide enquiry into the feral animal problem in the Northern Territory (Letts et al. 1978), a program was initiated in 1980 to study the biology, impact o n vegetation, and control of rabbits in central Australia. For a program of rabbit control to be acceptable here it must be relatively cheap, effective, capable of covering large areas with moderate effort, and able to be applied when equipment and labour are available. In order t o determine the suitability of rabbit control methods, their cost a n d effectiveness, and the resultant effect on vegetation of reduced rabbit numbers, a rabbit control experiment combined with exclosure studies was initiated in 1981 near Alice Springs in central Australia.
Materials and Methods Study Area
The study area is approximately 40 km to the south-west of Alice Springs and receives approximately 263 mm average annual rainfall. On an area of Muller Land System (Perry et al. 1962) characterized by undulating limestone country, discrete areas of calcareous shrubby grassland were surveyed and found to have warren densities ranging from 0.5 to 1.5 warrens per hectare. The plant community was composed of a sparse shrub layer dominated by Acacia kempeana with minor A. aneura, A, estrophiolata, Atalaya hemiglauca and Cassia nemophila, and a herbage layer dominated by Enneapogon spp. and Sclerolaena spp. together with a wide range of annual and perennial grasses and herbs. The soil was a Gcl. 12 (Northcote 1971)with an average depth of 50 cm. The area has been grazed by domestic stock for 100 years and by rabbits for perhaps 80 years. Degradation is evident in many areas, with degraded herbage composition, tree death and extensive soil erosion. Study Sites
Eight similar sites of approximately 1 km2 were selected over a distance of 30 km within the larger surveyed area. At six of these sites, three plots of 1 ha each were chosen within 50 m of large or multiple warren complexes, for erection of exclosures. One plot was netted to exclude rabbit and cattle grazing (no grazing), another fenced to exclude cattle but allow rabbit grazing (rabbit grazed) and the other left open to all grazing (rabbit plus cattle grazed). Two additional sites were chosen on the same land system 20 km distant, on which to test the effectiveness of European rabbit fleas in controlling rabbits. Rabbit Control Treatments
Sites were selected at random for the application of control treatments as follows: two as untreated experimental controls (control for ripping-fumigating and poisoning treatments); two for warren ripping and fumigating; two for 1080 poisoning; two for poisoning, ripping and fumigating; one for release of the European rabbit flea; one with no treatment (control for the European rabbit flea trial). The grazing exclosures were erected on the first six of these sites: the untreated experimental control, ripping and fumigating, and 1080 poisoning. The ripping and poisoning treatments were applied in March-April of 1981. Multiple passes with a single-tined ripper were made every 60-100 cm across each
Effects of Rabbit Control
warren to a depth of 40 cm. Holes reopened 1 week later were fumigated with 'Gastoxin' (V.E.B. Delicia) tablets and closed by hand. The poison bait was applied as 1080 one-shot oats, being a mixture of poisoned and ordinary oats in the ratio of 1: 100, one poisoned grain being sufficient to kill a rabbit. The bait was laid in trails around each warren at a rate of 6 kg km-1, each warren receiving an average of 85 m of trail. A total of 3500 rabbit fleas from CSIRO, Canberra, were released into flea treatment warrens in December 1980, February 1981 and July 1981.
Rabbit Survey Spotlight counts along 1-km standard transects were made in each replicate from a vehicle driven at 15 km h-1 with two observers. Counts began at dusk under standard conditions of moonlight and were conducted on two successive nights, at bimonthly intervals, from March 1981 to July 1983.
Vegetation Four permanent transects 50 m long were allocated at random within each of the 18 (three rabbit control X two replicates X three grazing exclosure) vegetation study plots. A total of 100 nested quadrats of 0.1, 0.25 and 1.00 m2 were systematically placed on the transects to record species' 'presence and absence' frequency data. Aerial plant cover was measured by 500 points of a wheel point apparatus within each plot. In 1983, grass, herb and total green standing biomass (dry weight) was collected from 10 randomly placed quadrats 1 m square, in each of the 18 vegetation study plots. The density and size class of shrub and tree species was collected on belt transects (10 by 50 m) centred on the four frequency transects in each plot. Data collection took place during autumn (post summer rainfall) in 1981 (pre-treatment) 1982 and 1983 (post-treatment).
Ancillary Information Rainfall was collected in storage gauges and measured every 2 months. An index of cattle grazing pressure was obtained by noting, at the time of rabbit survey, the presence of fresh cattle hoofprints along permanent vegetation transects in the 'rabbit plus cattle grazed' treatment at each of the six sites. Presence of myxomatosis in the rabbit population was noted during rabbit survey. A work activity log was kept during the application of rabbit control treatments to enable cost of control to be computed.
Statistical Analysis Analysis of variance and covariance were used to compare the effects of rabbit control treatment on vegetation measures and rabbit population.
Results
Control T h e mechanical and chemical rabbit control treatments were applied in March of 198 1. As a result of these treatments, together with myxomatosis which reappeared in the population immediately after they were applied, rabbit numbers decreased from 13-22 km-I to less than 5 km-1 (Fig. la). For the 'untreated control' areas, numbers also fell, due most probably to myxomatosis exacerbated by dry conditions. Over the experimental period of 2 years, the rabbit control treatments restricted increase in numbers and kept the population below that of the untreated areas. A n analysis of covariance which adjusted mean population levels for pre-treatment levels showed that 'untreated control', with 19.9 rabbits km-1, was always greater than '1080 poisoning', with 8.5 rabbits km-1, 'poison, rip a n d fumigate', with 3.4 rabbits km-1, and 'rip and fumigate', with 3.0 rabbits = 3 . 3 rabbits km-1). However, the '1080 poisoning' treatment was effective km-I ( L S D ~ % only in the short term, because in March 1983 population numbers were climbing steadily towards the levels of the 'untreated control' areas a n d a further poisoning was required. The release of European rabbit fleas does not appear to have had a n effect o n populations o n the 'flea' sites during the experimental period (Fig. lb). Repeated sampling of the rabbit population from these areas indicated that fleas did not survive.
B. D. Foran, W. A. Low and B. W. Strong
240
Rabbit populations in the 'untreated control' areas varied according to antecedent rainfall. For the mean of populations of the two replicates, 68% of the variation was accounted for by the rainfall of the previous winter, that of the previous summer, and a greenness index: y = 39.1 - 0 . 2 5 6 ~ 0~ . 0 2 2 ~ ~1-.35x3;where y is mean rabbit population, xl is the previous winter (April-September) rainfall in millimetres, x2 is the previous summer (October-March) rainfall in millimetres, and x3 is forage greenness rated on a scale 1-3, two months previously. 30-
25
-
20
-
15
-
10
-
5
-
(a)
A
J
P
O
1980
1981
1982
1983
Fig. 1. Density, in numbers per spotlight kilometre, of rabbits in the mechanical and chemical control No treatment. 0 0 1080 poisoning. treatments. .-. . - - -. Rip and fumigate. xx Rip, poison and fumigate. Myxomatosis present. o Date of treatment. Follow-up 0 No treatment. treatment. (b) Density of rabbits in the rabbit flea treatments. 0 -x x Flea treatment. o Date oftreatment. (c) Monthly rainfall, from February 1980 to April 1983, at the experimental site near Alice Springs in central Australia.
*
The cost ofrabbit control treatments was $3.57, $5.71, $8.83 and $0.48 per hectare for the '1080 poison', 'rip and fumigate', 'poison, rip and fumigate', and 'flea release' treatments respectively (Table I).
Vegetation Differences in species frequency between the years were related to amount and timing of
Effects of Rabbit Control
rainfall, rather than to the rabbit control and exclosure treatments. From 1981 to 1983 frequency of species such as Enneapogon spp., Sclerolaena patenticuspis, Helipterum jloribundum, Chenopodium cristatum, Ptilotus atriplicifolius and Boerhavia dzffusa increased, while that of Tribulus terrestris, Phyllanthus rhytidospermus, Portulaca oleracea and Citrullus colocynthis decreased (Table 2). Some of these species usually dominate in a pioneer or overgrazed situation, but this data suggests that their reaction to the wetter years of 1982 and 1983 is as important as their response to grazing intensity. Rabbit control treatment, alone, did not hate a significant effect on the frequency of any species. After adjustment for pre-treatment levels, the 'rabbit and cattle grazed' treatments had higher frequencies of Dactyloctenium radulans, P. oleracea, Swainsona burkei and C. cristatum. None of these species were more frequent in the 'rabbit grazed' than the 'no grazing' treatments, and so increase in these less desirable species may be due to the combined grazing effects rather than to rabbits alone. Table 1. Cost of four methods of rabbit control on a calcareous shrubby grassland in central Australia Warrens per hectare are from the total numbers counted in 1 km* around each study site. Mean numbers of rabbits per spotlight kilometre are adjusted by covariance for pre-treatment levels
Warrens per hectare Mean No. rabbits per spotlight km Cost of initial treatment per ha: total ($) Labour ($) Equipment and materials ($) Cost of initial treatment per warren ($) Cost of follow-up treatment per ha: total ($) Labour ($) Equipment and materials ($) Cost of follow-up treatment per warren ($) Total cost per hectare ($) Total labour costs ($) Total equipment and materials ($)
Untreated control
1080 poisoning
Rip and fumigate
1080, rip and fumigate
Flea treatment
No flea treatment
1,07 19.9t7.0
0.81 8.5k5.6
0.96 3.4+ 1.2
1.13 3.7k2.5
0.71 5.4k2.1
0.81 5,4i2,3
-
2.21 0.42 1.79
5.71 0.97 4.74
8.41 0.97 7.44
-
-
0.48 0.08 0.40
-
2.71
5.94
7.43
-
1.37 0.30 1.07
-
0.43 0.35 0.08
-
-
1.69 3.58 0.72 2.86
-
-
-
5.71 0.97 4.74
0.38 8.83 1.38 7.52
-
-
-
0.59
-
-
-
-
0.48 0.08 0.40
-
However, the mean frequency on both grazed treatments in 1983 of C. cristatum (r = O.9l), D. radulans (r= O.88), P. oleracea (r = O.82), P. rhytidospermus (r = O.92), Eragrostis barrelieri (r=0.91) and Malvastrum americanum (r=0.93) was positively correlated to the mean rabbit counts of the previous year, and so a direct causative relationship may be indicated. The dominant biennial grasses Enneapogon spp. are the most valuable components of the herbaceous layer, both for rangeland stability and domestic animal forage. Mean treatment values for the frequency ofEnneapogon spp. suggest that it responds positively to wet years and grazing influences, but its frequency is negatively correlated with rabbit counts (r= -0.70). Data from the individual treatment replicates indicates that both grazed and ungrazed areas show a stable or upward trend in the frequency ofEnneapogon spp. over the period (Fig. 2). However, at the 'untreated control' replicate which received the heaviest rabbit and cattle grazing, the frequency of Enneapogon spp. in the 'rabbit grazed' area decreased, that in the 'rabbit and cattle grazed' area remained stable while that in the 'no grazing' area increased. The decrease corresponded with mean rabbit coutns of 2 1 km-1 and
Sclerolueno patenticuspis Tribulus rerrestris Phyllanthus rhylidospermus Sida spp. Heliptrmm jlorihundum Portulaca oleracra Euphorbia drummondii Swainsona hurker Solanum spp. Chenopodrum cristatum Ptrlofus alriplicifolrus Salsola kali C'ifmllis coiocynlhis Malvustrum americanum Bocrhavra d~@sa
Forbs
Enneapogon spp. Eragrostrs barrelieri Dactyloctenium radulans T r a p auslra/ianus
Grasses
Species
15.6 33.6 37-6 51.6 11.4 26.7 79-4 12.5 25.8 19.1 10.1 18.1 5.8 6-1 7-8
49.2 3-8 8.0 10.2
1981
14-6 28.2 5.8 47.0 21-9 18.2 54.8 14.0 33.2 23.1 13.2 17.3 1.3 5.7 9.0
56.1 1.0 5-0 2.5
1982
28.5 16-8 6.2 53.6 78.5 17-1 83-1 13.8 29.3 38.2 18.4 19.1 2.1 3-7 32-2
63.1 3.9 12.2 12-5
13.6** 3-4~s 19-O*** 9.6**
F
2.5 5.0
-
8-0
k
LSD
3.9 34.8*** 7.9 16.9*** 78.7*** 6.1 7.3:' 3.7 152.9*** 8.7 4.1* 7.7 17.8*** 10-8 1-4~s 2.8 15.4*** 10.P 9.1 2.9 19.0a** 0 . 1 ~ ~ 2.3 10.0'* 2.1~s 5.9 48.6***
Between years 1983
23.6 14.9 4.6 52-8 78-3 9.7 76.1 14.1 29-3 78-5 19-5 7.9 1.5 10.3 21-6
75.0 1.2 18.7 22.4
Untreated control
38.5 19.5 5-0 52.1 79.8 21.0 89.1 13-6 34-9 14.2 19.7 32-7 2-3 1.7 55.1
50.8 5-8 7.3 7.9 23-3 15.9 9.1 55-8 77.4 20-5 84-2 13.6 23.7 21.9 15.9 16.6 2-3 0 20.0
63.5 4.6 10.6 7.2
Rabbit control treatment 1080 Rip poisonand ing fumigate
Significance: *P
