Aust. Wildl.Res., 1983, 10, 105-20
Changes in the Age-Structure and Size of Populations
of Wild Rabbits in South Australia, following the
Introduction of European Rabbit Fleas,
Spilopsyllus cuniculi (Dale), as
Vectors of Myxomatosis
B. D. Cooke
Vertebrate Pests Control Authority, Box 1671, G.P.O., Adelaide, S.A. 5001.
Abstract
The age-structures of rabbit populations at four climatically different sites in South Australia have
altered significantly since the introduction of European rabbit fleas; in summer the ratio of young of
the year to older rabbits has been greatly reduced. There is evidence of decreases in size of all observed
populations and (for one population) in ability to increase rapidly in favourable years. The ratio of
young of the year to older rabbits was a convenient and sensitive index of changes in population
structure; for analytical purposes the log transformation (In [ratio] or In [ratio+ 11) was biologically
and statistically sound. Since the fleas became established, field-strain viruses may be more virulent
and morbidity rates may have increased, but the factors enabling rabbit fleas to enhance myxomatosis
as a cause of mortality are unknown. The apparent virulence may have increased because of selection
for viruses of high virulence, which are best suited for transmission by rabbit fleas, and possibly
because the dose of virus transmitted by fleas is considerably greater than that by mosquitoes or other
agents.
Introduction
There have been a number of published reports on the results of releases in Australia
of European rabbit fleas as vectors of myxomatosis. Williams and Parer (1972) concluded
that rabbit fleas made no significant difference to the epizootiology of myxomatosis at
Urana in New South Wales, except that the fleas apparently transmitted the disease in
one year when mosquitoes were scarce. The data of Dunsmore et al. (1971) and Dunsmore
and Price (1972) further suggested that at a subalpine site (Snowy Plains, N.S.W.) fleas
transmitted myxomatosis through most of the young, susceptible rabbits in summer.
Previously the disease had trickled on through the winter months when low temperatures
enhanced the virulence of the virus and resulted in the deaths of many more rabbits than
in summer epizootics. Dunsrnore and Price (1 972) therefore suggested that the introduction
of rabbit fleas might reduce the effectiveness of myxomatosis as a mortality factor in rabbit
populations.
Evidence of more favourable effects of releasing rabbit fleas has been put forward by
Sobey and Conolly (1971) and Shepherd and Edmonds (1978). They found that fleas carried
myxomatosis in advance of the normal, summer epizootics carried by mosquitoes, and
that flea-borne outbreaks resulted in the deaths of many susceptible rabbits. When fleas
were sufficiently numerous, the epizootics prevented the normal spring increase in rabbit
numbers, and the loss of young was reflected in an increase in the average age of rabbits.
Highly virulent Lausanne strain myxoma virus was introduced into the study sites of Sobey
and Conolly (1971) and Shepherd and Edmonds (1978), but attenuated field strains were
also common. The role of this Lausanne virus in the apparent changes in mortality was
unknown, and complicated the unravelling of the epizootiology.
0310-7833/83/010105$02.00
B. D. Cooke
The results of these four studies are equivocal. Because they lack adequate experimental
design, there is n o way of determining whether variations in the numbers of rabbits or
in the structure of their populations result from the introduction of fleas o r from other
factors.
The study described in this paper used a formal experimental design to provide a reliable
assessment of the effects of rabbit fleas on the epizootiology of myxomatosis. T h e
experimental design chosen was not completely rigorous, but represented an ideal
experimental design modified in the light of the resources available for a large-scale field
experiment.
Fig. 1. Southern part of South Australia, showing isohyets and location of study sites. B, Belton.
K, Keith. M, Mt Pleasant. P, Penola. W, Witchitie. The 250-mm isohyet marks the approximate
northern limit of the mediterranean-like climatic zone.
The Study Sites
South-eastern South Australia basically has a mediterranean-like climate; the winters are cool and
wet and the summers warm and dry. Between them, the sites chosen for the study have the full range
of rainfall and temperature variations found within this climatic region (Fig. 1). Two sites, Belton
and Witchitie, lie at the northern limit of the mediterranean-like environment where it abuts the
inland areas of low, erratic rainfall. Penola, on the other hand, is in a relatively cold, wet part of the
State.
Myxomatosis first became established in South Australia in 1951. Epizootics have generally occurred
in summer (December-February) and are thought to be camed by the mosquito, Anopheles annulipes,
which is common during summer (Waterhouse 1958). Lines (1952) found the epizootics were most
effective near sources of permanent water, such as the Murray River and south-eastem swamps, but
killed fewer rabbits in upland areas and drier parts of South Australia. Such an observation is consistent
with the idea that mosquitoes were the major vector of the disease.
The main use made of the land at all sites chosen for this study is the grazing of sheep and cattle.
The details of soils, vegetation and climate at each site are as follows:
Rabbit Populations after Introduction of Flea
Penola. Rabbit warrens are found along ridges of sandy, solonized soil, lightly timbered with the
remnants of native vegetation, but rabbits graze mainly on the adjoining flats of heavy, black rendzina
soil where the main pastures are introduced mediterranean annuals; these flats are often inundated
after heavy winter rains. The average annual rainfall is 655 mm. Trumble (1948) stated that the
frequency of drought (i.e. the frequency of years in which the rainfall exceeded evaporation for less
than 5 months) was 2%. Penola is not the wettest site but it has the lowest temperatures, and
evaporation from the soil is low.
Keith. This site has undulating solonized soils. On steep ridges some of the original native
vegetation, Eucalyptus and Banksia, has been retained but the flatter, cleared areas support a pasture
of introduced mediterranean pastures, e.g. Hordeurn, Erodium and Arctotheca. Rabbit warrens occur
mainly among native scrub on the edge of the pastures. The rainfall averages 455 mm annually.
Drought frequency is 22% (Trumble 1948).
Mount Pleasant. A steeply undulating site of gravelly, granitic soil. The rabbits burrow beneath
huge mounds of granite boulders. None of the original native vegetation remains and the pastures
consist of introduced annual grasses and forbs. Rainfall averages 685 mm annually and drought
frequency is 8%.
Belton. This hilly site has shallow, loamy soils and exposed, stony ridges. Vegetation largely consists
of tough, native grasses, e.g. Triodia, Stipa and Danthonia, but there is also a strong component of
introduced mediterranean grasses and medics, e.g. Hordeurn, Brornus and Medicago, in the heavier
soils of the drainage channels. Rabbit warrens are usually found on the lower hill slopes and drainage
channels. Rainfall averages 240 mm annually. The drought frequency for Johnburg, which lies close
to Belton, is 81% (Trumble 1948).
Witchitie. This site lies close to Belton and has similar soils. However, because it is more arid,
the vegetation consists largely of native chenopods, e.g. Sclerolaena (formerly included in Bassia)
and Atriplex. There are few mediterranean annuals except in the most favoured areas. Rainfall averages
170 mm annually. By Trumble's (1948) definition drought frequency exceeds 90%.
Normal rabbit control work was done by landholders at four sites during the study. At Belton,
Mount Pleasant, and Keith some warrens were destroyed by ripping with tractors, and at Keith and
Penola 1080 poison was laid each year. However, these control measures were confined to only a
small portion of each study area in any given year, and could not be regarded as causing major or
lasting reductions in the size of the rabbit populations.
Materials and Methods
Live Capture of Rabbits at Mount Pleasant
During initial trials at Mount Pleasant, rabbits were captured in a system of smeuses as described
by Southern (1940) and Copson et a/. (1981). Each warren in the study area was surrounded by a
wire-netting fence, and rabbits could leave or enter only by pushing through a smeuse fitted with
swinging gates. To trap rabbits these gates were set so that they allowed rabbits to enter the smeuses
but not leave them. The rabbits were tagged with coloured, numbered eartags to enable subsequent
identification on recapture or when seen from an observation tower. Trapping was camed out every
2 weeks except during an experimental epizootic, when it was delayed for a month to avoid spreading
virus by contact among trapped rabbits.
Shot Samples
Samples of rabbits were collected at each site at approximately 6-week intervals. The rabbits were
shot at night, with a 0.22 calibre rifle from a four-wheel-drive vehicle equipped with a spotlight.
As each rabbit was shot a blood sample was taken on a filter-paper strip for testing for myxoma
antigens or antibodies to them (Sobey et a/. 1966). The rabbits were dissected and reproductive data
recorded. For pregnant females, the numbers of corpora lutea and the number and approximate age
of embryos were noted.
An eye-lens from each rabbit was collected for estimation of age from the dried eye-lens weight
(Dudzinski and Mykytowycz 1961). During summer and autumn (December-May) the dried eyelenses of the young of the year had a weight distribution which was quite distinct from that of older
B. D. Cooke
rabbits. Consequently, it was easy to distinguish young rabbits, which were initially susceptible to
myxomatosis, from older, presumably immune ones. However, if many of the susceptible rabbits
died during an epizootic there was a rapid decrease in the percentage of young in the population.
This change in percentage, or the change in the ratio of young of a given year to older rabbits, was
a particularly sensitive indicator of juvenile mortality.
Spotlight Counts
At Belton and Witchitie, a powerful quartzhalogen spotlight was used to assist the counting of
rabbits at night along clearly marked transects. The vehicle used was driven at constant speed
(8 km hr-I) and the same observer made the count on each visit, to standardize the procedure. Each
transect was 8 km in length.
Similar counts were not made at other sites. At Mount Pleasant seasonal changes in the heigh'i
of the grasses made counts difficult to interpret, and at Keith and Penola seasonal changes in feeding
patterns of rabbits living on the margin of native vegetation would have complicated results (Cooke
1970).
Fig. 2. Decrease in the size of the rabbit
population ( 8 )at Mt Pleasant during an
epizootic of Lu myxomatosis apparently
transmitted by fleas. Bars, new cases of
myxomatosis. Arrowheads, dates of
observation.
4-
- 0
Oct.
Nov.
Dec.
Results
Initial Trial at Mount Pleasant
A small pilot trial was begun in 1973 at Mount Pleasant to test the feasibility of larger,
fully replicated experiments. Initially, it was intended to have three experimental treatment
sites: (I) no fleas, no added myxoma virus (control); (11) fleas, no added myxoma virus;
(111) fleas with added myxoma virus. The virus to be used was the virulent Lausanne
strain (Lu). It was chosen because it causes distinctive lesions in affected rabbits and because
it may have had an important role in the heavy mortality observed during the flea-borne
epizootics reported by Sobey and Conolly (1971). Each experimental site was a group of
six warrens separated from the other sites by a distance of at least 600 m.
In late August 1973, when most of the rabbits were breeding, about 2000 European
rabbit fleas were released into warrens in areas I1 and 111. Their successful establishment
Rabbit Populations after Introduction of Flea
was confirmed by trapping rabbits in the smeuses. On 25 October, when all the rabbits
in the warrens had been tagged, 1000 fleas carrying Lausanne virus were liberated into
treatment area 111. These fleas had been made infective by being shaken with a suspension
of virus, after the method of Parer et al. (1981). The subsequent epizootic was followed
closely from an observation tower erected on the site.
The first rabbits infected with Lausanne virus were seen on 5 November. By 4 December,
44 out of 49 individually known rabbits had disappeared. Twenty-nine had been seen with
myxomatosis apparently caused by busanne strain virus, and one had mild symptoms
of field-strain virus infection (Fig. 2). However, the trial did not run strictly to plan. In
early November 1973, field-strain myxomatosis broke out on area 11 where fleas had also
been established. A high proportion of the young, presumably susceptible rabbits died.
Myxomatosis did not appear on area I, the experimental control, until January 1974.
A summary of the numbers of rabbits which survived during the period of the fleaborne epizootics is given in Table 1. These data are based on the number of tagged rabbits
trapped before the epizootics in October and recaptured in late November. The rabbits
which survived the epizootic of Lausanne myxomatosis were generally old ones which
had previously been challenged with field-strain myxomatosis, and some were known to
have a high level of antibody to the disease. However, because trapping was begun well
Table 1. Results of pilot trials at Mount Pleasant to test for the effects of rabbit fleas
transmitting Lu and field-strain myxomatosis
-
Area
-
I
I1
I11
-
Treatment
-
No. of rabblts ahve
October
November
-
No fleas, no virus
Fleas, f/s %lms
Fleas, Lu vlms
Propn surviving
October to November
-
72
25
32
60
9
4'
0 8
0 4
0 1
* One rabbit died of Lu myxornatosis during trapping.
after breeding had started (August), some large young of the year were indistinguishable
from the older adults on the basis of weight, and their antibody levels were unknown.
Thus the exact number of old, immune rabbits in each area was not precisely known, but
from shot samples (Table 2; Fig. 6c) taken nearby they were at least 10% of the population.
Without replication of treatments and a precise knowledge of the proportion of older
immune rabbits in the population, it was not possible to draw any major conclusions from
this trial. However, the results did suggest that the fleas could carry an epizootic of
myxomatosis in advance of the normal summer epizootics. In addition, the high mortality
resulting from the Lausanne virus was completely in agreement with expectations, so that
it was worth testing the hypothesis that epizootics involving fleas and field-strain viruses
could also cause moderately heavy mortalities, as suggested in the trial.
Further Changes in the Epizootiology of Myxomatosis Associated with the Spread of the
Rabbit Flea
In late 1974, additional opportunities arose to test whether flea-borne epizootics were
capable of causing heavy mortality. Between 25 and 28 November of that year a large
sample of rabbits was collected by shooting over an area of 1600 km2 at Witchitie. The
main aim of this was to map the spread of the fleas since an earlier sample in 1972 (Cooke,
unpublished data), but it was immediately evident that myxomatosis, which was active
at the time, was confined to the area occupied by fleas. There was a significant association
between the presence of fleas and infection with myxoma virus, as the following contingency
table shows:
B. D. Cooke
Rabbits
with fleas
Myxomatosis
No myxomatosis
Total
Rabbits
without fleas
17
74
91
x2=12. 1 , P<O.OOl
1
70
71
Total
18
144
162
Furthermore, in the area where fleas had been well established (defined as the area they
had colonized two years previously) the age-structure of the rabbit population in late 1974
differed markedly from that of rabbits collected beyond this area. Again, these data can
be summarized as follows:
Rabbits
Young of year
Older
Fleas established
5
19
24
x 2 = 2 9 . 5 , P<O,OO1
No fleas
Total
42
7
49
47
26
73
Similar data were collected on 17-18 December 1974 from Mount Pleasant, where fleas
had spread from the initial flea release sites. Myxomatosis was strongly associated with
the presence of rabbit fleas (x2=9 3, P<O.001) and again the age-structure of the rabbit
population changed markedly across the margin of spread of the fleas (x2=6.8, P<O.01).
Thus there was strong evidence that flea-borneepizootics of myxomatosis caused marked
changes in the age structure of rabbit populations. Apart from the initial occurrence of
Lausanne strain virus at Mount Pleasant, which persisted for little more than 6 weeks,
naturally occurring myxoma viruses were apparently responsible. These data did not
indicate that flea-borne epizootics were necessarily more lethal than the usual summer
epizootics carried by mosquitoes, but simply that epizootics could powerfully affect
population structure.
It became apparent in these early trials that the ratio of young of the year to older
rabbits was a convenient index of changes in the age-structure of rabbit populations caused
by myxomatosis. Young rabbits were susceptible to the disease, whereas most adults had
acquired immunity as a result of previous infection. In addition, young rabbits can be
readily distinguished from their parents on the basis of eye-lens weight (Dudzinski and
Mykytowycz 1961). Thus, if myxomatosis killed more young rabbits than normal, the ratio
of young to old rabbits should decrease. It follows that the problem of comparing the
effects of flea-borne epizootics with the former, presumably mosquito-borne ones could
be approached by simply comparing the age structure of populations in late summer and
autumn (February-May) after the epizootics carried by both fleas and mosquitoes had run
their course.
It also became clear that controlled experiments of the type attempted at Mount Pleasant
would be difficult to complete. Trapping would need to continue over almost a year to
ascertain the ages and level of immunity of rabbits, and experimental sites would need
to be replicated. Added to this, rabbit fleas colonized areas more than 20 krn from Witchitie
between 1972 and 1974, and Shepherd and Edmonds (1976) showed that fleas appeared
20 km beyond their previous known range within a single year. In other words, there was
a high risk that the experimental controls would be overrun with fleas before the
experiments were completed.
Because of these considerations, the experimental design finally chosen represented a
compromise between an ideal experimental design and one that could be carried through
with the resources available.
Instead of experimental controls, it was planned to ascertain the age-structure of rabbit
populations for some years before the fleas were introduced, then to continue the
Rabbit Populations after Introduction of Flea
monitoring program for some years after the fleas became established, so that any changes
in the age-structure of each population could be detected. Data from Belton, Mount
Pleasant, Keith and Penola had been collected for some years before 1974, so that the
population dynamics, age-structure and epizootiology of myxomatosis in each population
were known. Fleas were introduced into all populations in late 1974 and became widely
established by the spring of 1975. Sampling of all populations was continued until at least
early 1979, although at Belton and Witchitie it continued into the 1980s.
In the absence of experimental controls, it is assumed that the effects of climate on the
age-structure of rabbit populations and epizootiology of myxomatosis were, on average,
similar in the years before fleas were introduced and after fleas became established. If this
basic assumption was not met, then the effects of the introduction of fleas would be
confounded with the effects of climatic variability.
Results of the Main Experiment
Table 2 gives the changes in the ratio of young to older rabbits as the fleas were
established at each study site. The ratios are based on samples in summer and autumn
of each year, after the annual epizootic of myxomatosis and during the non-breeding season
when no additional young were entering the population.
Table 2. The effects of introduction of rabbit fleas on the age-structure of four rabbit
populations
The variate is the ratio of young of the year to older rabbits
Treatment
Year
Site
Belton
-
No fleas
Rabb~tfleas
Mt Pleasant
Keith
Penola
-
1971
1972
1973
1974
1975
1976
1977
1978
The data were analysed by means of an analysis of variance which treated the sites as
replicates and the data from each year as observations within replicates. Initial analysis
showed that the presence of fleas was associated with a significant reduction in the ratio
of young to older rabbits. However, the residual variances about the fitted values suggested
that variances were not homogeneous; the log transformation (In [ratio+ 11) gave a more
even distribution. The results of this second analysis are given in Table 3. This analysis
showed some significant differences in the ratio of young to older rabbits between sites
but, more important, a general reduction in this ratio coincidental with the establishment
of fleas. t-Tests showed significant changes in the ratios at all four sites: Belton, t6=2.82
(O.O5>P>O .01); Mount Pleasant, t4=10.91 (P<O.001); Keith, t6=8.49 (P<O.001); Penola
t6=3.39 (0.0 1> P>0.001). The apparent effect of introducing the rabbit fleas was to reduce
the ratio of young to old rabbits from 8 . 1: 1 to 1 .68: 1.
The analysis of variance showed that there was no major component of variance within
the (sites by treatment by year) stratum, but the problem remained that if some additional
factor which influenced the age-structure of rabbit populations had changed over the period
during which fleas were introduced, its effects would be confounded with those resulting
from the release of the fleas. Weather is perhaps the most likely such factor. Rainfall and
subsequent pasture growth determines the duration of rabbit breeding each year, and so
it can influence the structure of the rabbit population in the following summer.
B. D. Cooke
Furthermore, heavy rains or drought may affect wide areas of South Australia, so it is
possible that all study sites might have been similarly influenced in any year, and that on
average one sequence of four years might differ from the next.
The relationship between weather and the age-structure of most of the experimental
rabbit populations could not be gauged because of the limited data available. However,
for Belton, data were available on the age-structure of the population for 7 years before
the fleas were introduced, and, because of the extreme variability of the climate, some of
the effects of weather were obvious.
Table 3. Analysis of variance of the changes in the ratio of young to older rabbits observed with the
introduction of rabbit fleas
The variate used is In [(ratio young: old)+ 11.
Source of variation
Sites
Sites, treatment
Treatment
Residual
Total
Sites, treatment, year
Grand total
**Significant at P<O. 01.
*** Significant at P<O. 001
Sum of
squares
Degrees of
freedom
Mean
square
Variance
ratio
P
3.0911
3
1.0304
5.719
**
11.9327
0.1003
12.0330
3 9636
19.0877
1
3
4
22
29
11,9327
0.0334
3.0082
0.1802
357.057
0.185
16.697
*t*
Injluence of Weather on the Age-Structure of Rabbits at Belton
At Belton, rabbits breed only in periods of plant growth, and females may produce a
litter each month while the growing season lasts. Productivity is therefore related directly
to the length of the growing season, except that when the growing season is more than
6 months long early-born young of the year may themselves produce young in the same
season.
The effects of climate on the age-structure of the rabbit population were analysed by
comparing the ratio of young of the year to older rabbits with the length of the growing
season. The length of each season was estimated by the soil-moisture model of Slatyer
(1962) as adapted for use at the Belton site (Cooke 1970, 1974). Table 4 shows the result;
data for 1972 differ slightly from those used in the main analysis of variance (Table 2)
because in that year there were two distinct growing seasons and two corresponding periods
of reproductive activity among the rabbits.
Because of the disproportionately high production of young in prolonged seasons, the
relationship between the length of the growing season and the ratio of young to old rabbits
was not linear. Hence, a log conversion was used to prepare the data for analysis. The
regression of In [ratio young: old rabbits] on the length of the growing seasons during the
7 years before fleas were introduced was significant (t6=4.2 1, P<O. 0 1).
Correctingfor the Effects of Weather
It was now possible to correct for the effect of seasonal conditions, in the comparison
of data collected before and after fleas were introduced. Regression analysis showed that
slopes fitted to the data sets collected before and after did not differ (tlo=O.525, NS), but
there was a significant separation of the two data sets, as shown by difference between
their intercepts if a common slope was assumed (tl,=2.773, P<O. 05).
It is worth noting that the data collected for the 1977 cohort of young lie within the
range of those before the establishment of the fleas. This may be significant because
1977 was a poor year for rabbit fleas; only 70% of the rabbit population at Belton was
infested with fleas, and fleas actually died out over large parts of Witchitie where they
had previously been well established. Myxomatosis was observed at Belton in October
Rabbit Populations after Introduction of Flea
and November 1977, but from the small numbers of sick rabbits seen it seemed that the
epizootic may have been less intense or differed in some way from those when fleas are
numerous and infest nearly all of the rabbits.
After correction for seasonal differences it was still possible to recognize significant effects
of the fleas among the rabbits at Belton. On average there were considerably fewer young
in the rabbit population in the years following the release of fleas than before. For given
seasonal conditions, the ratio of young to older rabbits had been reduced by a factor of
approximately 3.2.
Table 4. The age-structure of the rabbit population at Belton in relation
to the length of each growing season
European rabbit fleas were introduced in late 1974
Year
1968
1969
1970
1971
1972 (a)
1972 (b)
1973
1974
1975
1976
1977A
1978
1979
1980
-\
Ratio of young
to old
11.36
5.00
1.34
12.43
0.17
1 29
9 29
9 30
0.24
0.49
2 69
0 96
1.43
1.02
In [ratio]
2.43
1.61
0.29
2.52
-1.77
0.25
2.23
2.23
-1.43
-0 71
0.99
-0.04
0.36
0-02
Length of
season (weeks)
38
25
25
37
7
17
48
47
25
24
24
33
34
30
Drought year in which fleas survived poorly.
Effects of Rabbit Fleas on the Size of Rabbit Populations
Although rabbit fleas may change the age-structure of rabbit populations, the question
of greatest practical importance for landholders who see rabbits as competing with their
livestock for food is whether or not the presence of fleas is associated with a fall in numbers
of rabbits.
To answer this it is necessary to draw mostly on data from Belton and Witchitie.
Although these sites are 20 km apart and differ markedly in their annual average rainfall,
it is worth considering whether Belton can be used as an experimental control for the
Witchitie study site during 1970-74, when fleas were established at Witchitie but had not
spread to Belton.
It can be argued that if population changes were influenced by weather, then the numbers
of rabbits seen at Belton and Witchitie should have been highly correlated, because both
sites were influenced by the same major falls of rain although to a different degree. In
fact, there was a strong correlation in 1968-69, before the establishment of fleas. However,
this correlation broke down once fleas became established at Witchitie, and despite the
extremely favourable seasons of 1973 and 1974 the numbers of rabbits at Witchitie did
not increase, even though large numbers were counted at Belton. Since the establishment
of fleas at Belton in 1975 the rabbit populations at both sites have remained low. The
numbers seen at Belton have fallen to about 30% of the average number before the
introduction of fleas. In recent years, 1979-8 1, there has been a slight but steady climb
in the numbers seen at Witchitie.
Fig. 3 shows the numbers of rabbits counted at Witchitie and Belton on each visit to
the study sites between 1968 and 1981, together with the proportion of rabbits infested
B. D. Cooke
Fig. 3. Numbers of rabbits counted on transects at: (a) Belton; (b) Witchitie. Arrowheads indicate
when fleas were released at Witchitie and first noticed at Belton. Bars between graphs indicate the
presence of rabbits with myxomatosis: solid, at Belton; hatched, at Witchitie. Shaded areas, the
proportions of rabbits carrying fleas.
Fig. 4. Relationship between the number of
rabbits counted at Belton and Witchitie
before and after the release of fleas.
1968-69; R2=0.789.
0 197 1-74; R2=0.106.
Rabbits counted at Belton
with fleas and the incidence of myxomatosis. It should be noted that the general correlation
between myxomatosis and the decline in population in 1969 does not necessarily imply
a causal relationship. In the summer of 1968-69, when rabbits had abundant food,
myxomatosis did not cause a marked decline in their numbers. Furthermore, the massive
population crash at Witchitie in 1969 was associated with a shortage of succulent vegetation
(Cooke 1982) and actually preceded the appearance of myxomatosis.
Rabbit Populations after Introduction of Flea
Fig. 4 shows how spotlight counts at Witchitie and Belton were correlated before the
introduction of the fleas (1968-69) and how that correlation was lost when fleas were present
at Witchitie but not at Belton (1971-74).
There were no comparable data on rabbit numbers from the other study sites. However,
from the effort expended in obtaining samples it was judged that at Keith and Penola
numbers fell in 4 years to 30% of their original level. The population decrease at Mount
Pleasant was even more extreme. Before the release of the rabbit fleas it was possible to
shoot 20-30 rabbits in an area of less than 50 ha. Three years after the release of the fleas
it was necessary to search hard, over 200-300 ha, to obtain a sample, and by late 1978
rabbits were so scarce that it was no longer feasible to continue shooting samples.
Fig. 5. Changes in the effect of summer epizootics of myxomatosis on the rabbit population at Belton
before (a) and after (b) the introduction of rabbit fleas. Before, myxomatosis passed through the
population with no change to the age structure, but subsequent epizootics resulted in major decreases
Percentage of young in population. @---a Percentage of young
in the percentage of young. A---A
with myxomatosis.
How Does the Rabbit Flea Enhance the Effectiveness of Myxomatosis?
Although no attempt was made to look closely at all the factors which might have
accounted for the changes in epizootiology of myxomatosis, several general observations
are relevant.
During the initial studies at Mount Pleasant and Witchitie, the flea-borne epizootics
obviously occurred before the normal mosquito-borne ones. It seemed possible that the
mortality of young rabbits associated with the presence of fleas might have been higher
because rabbits were infected at an earlier age (Fenner and Ratcliffe 1965) or at a lower
ambient temperature (Marshall 1959).
More recently, however, epizootics have occurred in summer when ambient
temperatures were high and most young of the year were subadult, i.e. the epizootics were
directly comparable to those before the introduction of the fleas. For example, Fig. 5 shows
the percentage of young in the rabbit population at Belton during summer epizootics both
before and after the establishment of fleas. It can be seen that the age-structure of the
rabbit population changed greatly with the passage of epizootics in which fleas were
involved, even though the epizootics of previous years, which were presumably carried
by mosquitoes, caused no such changes. This implies that the myxoma virus was effectively
more virulent in the presence of the fleas than in their absence.
B. D. Cooke
In addition, there was evidence that fleas enhanced the proportion of young rabbits
which became infected with myxomatosis-at least at some sites. If the ages of rabbits
suffering from myxomatosis before and after the introduction of fleas are compared (Fig. 6),
it is apparent that before the establishment of fleas about 80-90% of the sick rabbits were
less than a year old, and 10-20% did not contract myxomatosis until they were in their
second year. Since fleas have been present it has been rare to shoot a rabbit more than
1 year old with symptoms of myxomatosis. This implies that the annual morbidity rate
is higher than previously.
Age (months)
Fig. 6. Cumulative frequency of rabbits with symptoms of myxomatosis in each 3-month age class:
(a) Penola; (b) Keith; (c) Mt Pleasant; (d) Belton. 0-0
Before introduction of fleas. e---• After
introduction of fleas. The data imply that since the introduction of fleas most rabbits contracted
myxomatosis in their first year of life, i.e. morbidity rates are higher than previously.
Discussion
In this study, the early observations and field trials supported some of the suggestions
made by Sobey and Conolly (1971) and Shepherd and Edmonds (1976). First, it was
apparent that rabbit fleas transmitted myxomatosis in the field, because at times the disease
was confined to those areas where fleas were established. Second, there was evidence of
heavy mortality among young rabbits as a result of myxomatosis spread by the fleas. The
age-structure of rabbit populations changed as a result of these epizootics and young rabbits
formed a smaller proportion of the summer population than previously. Third, there was
a general decrease in numbers of rabbits and a conspicuous failure to increase during
favourable periods.
Rabbit Populations after Introduction of Flea
The work described in this paper took a stronger experimental approach to field studies,
and an analysis of variance of data collected from four experimental sites confirmed a
significant and consistent effect of rabbit fleas on the age-structure of rabbit populations
at all sites.
The major experiment did not include experimental controls, but instead relied upon
data collected for a number of years before the release of the rabbit fleas for comparison
with similar data collected after the fleas were established. Therefore it lacks the rigour
of a properly controlled experiment and there is a chance that some factor, such as weather,
might also have had different effects on the structure of the rabbit populations before and
after the release of the fleas, and that its effects are confounded with those of the fleas.
However, in view of the wide separation of the study sites at Belton, Mount Pleasant,
Keith and Penola and the small variability in weather patterns at the more southerly sites,
the chance that widespread heavy rains or drought would influence all sites to the same
extent is small. In addition, there is no a priori reason why a period of heavy rain which
prolonged reproduction at Belton should have such marked effects at other sites. At Penola,
for instance, heavy rains cause inundation of pastures and low-lying rabbit warrens, and
this could offset the benefits of a longer growing season to some extent. It was further
shown that at Belton, where there was great variation in weather each year, the seasonal
correction of data enabled the confounding effects of weather to be separated from those
of the rabbit fleas.
These corrected data showed that in the presence of fleas the ratio of young to old
rabbits in the population at Belton was lowered by a factor of 3.2. However, this estimate
might be a little to:, low in view of the fact that the epizootic of 1977 may have been
abnormal. In that year, the ratio of young to old rabbits remained much as expected on
the basis of the length of the growing season alone, despite a minor epizootic of
myxomatosis. Fleas survived poorly in 1977 and only 70% of the rabbits were infested.
It is relevant that Sobey and Conolly (1971) showed that, in their study areas, epizootics
did not occur unless about 70% of the rabbits carried fleas.
If the result for 1977 is omitted from the analysis of data from Belton it can be calculated
that the ratio of young to old rabbits in the population is reduced by a factor of 4.2 during
normal epizootics where fleas are involved in the transmission of disease. This factor is
not too different from the 4.8-fold reduction obtained from the analysis of data from all
four sites (Tables 2, 3) when the effects of weather were assumed to be negligible.
Theoretically, data from Mount Pleasant, Keith and Penola might be corrected for the
variability in the annual growing season in much the same way as data from Belton. This
would enable corrected data to be used in an analysis of variance (Table 3) and so remove
the major objection that the experimental results are confounded with climatic variation.
However, in the absence of a detailed knowledge of the interaction between weather and
productivity of rabbits at each site, such an approach was not feasible.
The Meaning of the Change in Ratios
In the main experiment the effect of introducing fleas was to change the ratio of young
to old rabbits from 8 . 1: 1 to 1 .9: 1. A population change of this size, without changes in
adult mortality, assumes a loss of 80% of the young which would normally have survived
the usual mosquito-borne epizootics. However, there are other factors involved in this
change in ratio apart from additional mortality of young. This is best shown by means
of a simple example.
Fig. 7a shows the age-specific mortality curve for a stable rabbit population. For
simplicity, the survival rate of young rabbits does not differ from that of rabbits over a
year old. The yearlings and older rabbits have survived myxomatosis of low virulence,
B. D. Cooke
and the ratio of yearlings (n,) to older rabbits (n2+n3)has been set at 8: 1, which is similar
to that observed in the experimental populations before the release of the fleas.
Now consider what would happen if the rabbit fleas enhanced the effects of myxomatosis
and increased juvenile mortality to such an extent that the adult population declined and
the production of young was reduced by 25% annually, while adult survival remains as
before at 12%per year. The family of survival curves given in Fig. 7b shows the mortality
curve from each successive age class; the ratio of yearlings to rabbits in the older ageclasses has been reduced to 6: 1, simply because of the relatively large number of older
rabbits in a declining population.
It could further be assumed that as the rabbit population fell adult survival may have
increased slightly: some minor increase simply because of the lower incidence of
myxomatosis among older rabbits (Fig. 6), but also some increases if older rabbits survived
better at low than high densities.
Age (years)
Fig. 7. Theoretical curves showing changes in age structure which may result from decreasing size
of population and from changes in adult mortality, in addition to changes in mortality caused by
changes in apparent virulence of myxomatosis. (a) Adult survival 12% per year, mortality from
myxomatosis negligible; ratio n , : n2+n3,8: 1. (b)Adult survival 12% per year, 40% mortality of young
from myxomatosis; ratio n , :n2+n3, 6 : 1. (c) Adult survival 25% per year, 40% mortality of young
from myxomatosis; ratio nl:n2+n3,2: 1.
Whether or not survival among older rabbits improved is not known; however, Fig. 7c
shows that if adult survival was increased to 25% annually, substantial changes in the ratio
of young to older rabbits would follow. Given the combined effects of the decreased survival
of young and the enhanced survival of adults, the ratio of young to adults would change
from 8: 1 to 2: 1, even if flea-borne myxomatosis accounted for only 40% of the susceptible
rabbits which might otherwise have survived. This is considerably less than the mortality
rate of 80% implied earlier, and it illustrates how carefully data of this type should be
interpreted to avoid the wrong conclusions about the virulence of myxomatosis.
Factors Underlying the Apparent Changes in Virulence of Myxomatosis
The data collected during summer epizootics before and after the establishment of rabbit
fleas rule out the possibility that apparent changes in virulence can be attributed solely
to the age of rabbits at infection (Fenner and Ratcliffe 1965) or low temperatures (Marshall
1959). In addition, only a small component of the change in mortality could be attributed
to changes in the morbidity rate of the disease. At most, morbidity rates might have been
increased from about 80% to something approaching 100%. For a virus causing a mortality
of 10%in a rabbit population, such an increase in morbidity would only increase mortality
to a little more than 12%.
There is a possibility that rabbit fleas would favour the persistence of strains of myxoma
virus which are more virulent than those favoured in transmission by mosquitoes. In
Rabbit Populations after Introduction of Flea
Britain, where fleas are the dominant vector of myxoma virus, grade I1 and grade IIIa
viruses are common. In Australia, where mosquitoes are the principal vector, less virulent
grade IIIb viruses are most common (Vaughan and Vaughan 1968). For South Australia,
a few viruses tested in 1964 were largely in grades lIIb or IV (J. Bromell, unpublished
data).
It is, therefore, of interest that in a rabbit population in which a low grade IIIb virus
caused 10% mortality a virus of high grade IIIa or low grade I1 would cause about 40%
mortality (Rendel 1971). Such a change in virulence might be sufficient to cause the changes
in age-structure observed in populations of wild rabbits, as suggested earlier. The rapidity
of these apparent changes in virulence (i.e. within a single year) is no greater than the rate
of change in virulence observed when myxoma virus attenuated rapidly following its initial
introduction into Australia (Fenner and Ratcliffe 1965).
Another possible cause of apparent changes in virulence is the difference in mode of
transmission of virus by fleas and mosquitoes (Mykytowycz 1956). It might be that fleas
transmit virus particles in greater numbers, at different times or to different sites on the
host than mosquitoes. Such an effect should not be ruled out as influencing apparent
virulence. Fenner and Ratcliffe (1965) showed that the initial dose of virus can affect
survival of rabbits, although their experimental results show the effect to be rather small
over the range 50 IDSOto 500000 I D ~ O
myxoma virus. Nevertheless, it would be useful to
know the effects of dosage in the range which might be more common in the field, where
vectors are relatively scarce.
If the mode of transmission of virus or a dosage effect caused a change in virulence,
then the change in virulence when fleas were first introduced would be easily understood.
It might also explain the low mortality of rabbits during the epizootic at Belton in 1977,
when rabbit fleas were low in numbers and found on only 70% of rabbits.
For the future there are some important consequences of the release of rabbit fleas.
The initial effects for South Australia indicate a general reduction in rabbit numbers, but
the apparent increase in the virulence of the disease coupled with the increase in morbidity
rates imply a high rate of selection for genetic resistance to myxomatosis (Rendel 1971,
p. 91). It is therefore interesting that the population of rabbits at Witchitie has increased
slowly but steadily over the three years 1979-81, after a decade in which rabbit numbers
were low following the initial release of the fleas in 1969.
Acknowledgments
I should like to thank Messrs B. Sutton and D. Chinner for assistance in the field and
laboratory, and Mr J. E. Bromell and Dr R. P. Henzell for useful discussions and advice.
Dr W. R. Sobey, CSIRO, provided rabbit fleas and myxoma virus. I am indebted to
numerous landholders for assistance and permission to use their properties as experimental
sites. Mr B. V. Fennessy, CSIRO, provided helpful criticism of the manuscript, and Mr
P. L. Bird prepared the figures. This study was supported in part by funds from the
Australian Wool Corporation.
References
Cooke, B. D. (1970). Factors which influence the numbers of rabbits in natural populations. M.Sc.
Thesis, University of Adelaide.
Cooke, B. D. (1974). Food and other resources of the wild rabbit Oryctolagus cuniculus (L.). Ph.D.
Thesis, University of Adelaide.
Cooke, B. D. (1982). A shortage of water in natural pastures as a factor limiting a population of
rabbits, Oryctolagus cuniculus (L.), in arid north-eastem South Australia. Aust. Wildl. Res. 9,465-76.
Copson, G. R., Brothers, N. P., and Skira, I. J. (1981). Distribution and abundance of the rabbit,
Oryctolagus cuniculus (L.), at subantarctic Macquarie Island. Aust. Wildl. Res. 8, 597-61 1.
Dudzinski, M. L., and Mykytowycz, R. (1961). The eye lens as an indicator of age in the wild rabbit
in Australia. CSIRO Wildl. Res. 6, 156-9.
B. D. Cooke
Dunsmore, J. D., and Price, W. J. (1972). A non-winter epizootic of myxomatosis in subalpine southeastern Australia. Aust. J. 2001. 20, 405-9.
Dunsmore, J. D., Williams, R. T., and Price, W. J. (1971). A winter epizootic of myxomatosis in
subalpine south-eastem Australia. Aust. J. 2001. 19, 275-86.
Fenner, F., and Ratcliffe, F. N. (1965). 'Myxomatosis.' (Cambridge University Press.)
Lines, E. W. L. (1952). History of epizootics of myxomatosis in rabbits in South Australia in 1951-52.
J. Agric. S. Aust. 56, 232-8.
Marshall, I. D. (1959). The influence of ambient temperature on the course of myxomatosis in wild
rabbits. J. Hyg. 57, 484-97.
Mykytowycz, R. (1956). The effects of season and mode of transmission on the severity of myxomatosis
due to an attenuated strain of virus. Aust. J. Exp. Biol. 1(34), 121-32.
Parer, I., Conolly, D., and Sobey, W. R. (1981). Myxomatosis: the introduction of a highly virulent
strain of myxomatosis into a wild rabbit population at Urana in New South Wales. Aust. Wildl. Res.
8, 6 13-25.
Rendel, J. M. (1971). Myxomatosis in the Australian rabbit population. Search (Syd.) 2, 89-94.
Shepherd, R. C. H., and Edmonds, J. W. (1976). The establishment and spread of Spilopsyllus cuniculi
(Dale) and its location on the host Oryclolagus cuniculus (L.), in the Mallee region of Victoria. Aust.
Wildl. Res. 3, 29-44.
Shepherd, R. C. H., and Edmonds, J. W. (1978). Myxomatosis: changes in the epidemiology of
myxomatosis coincident with the establishment of the European rabbit flea SpilopsyNus cuniculi (Dale)
in the Mallee region of Victoria. J. Hyg. 81, 399-403.
Slatyer, R. 0. (1962). Climate of the Alice Springs Area. In 'Lands of the Alice Springs Area, Northern
Temtory, 1956-7'. pp. 109-28. CSIRO Land Res. Ser. No. 5.
Sobey, W. R., and Conolly, D. (1971). Myxomatosis: the introduction of the European rabbit flea
Spilopsyllus cuniculi (Dale) into wild rabbit populations in Australia. J. Hyg. 69, 331-46.
Sobey, W. R., Conolly, D., and Adams, K. M. (1966). Myxomatosis: a simple method of sampling
blood and testing for soluble antigens or antibodies to them. Aust. J. Sci. 28, 354-5.
Southern, H. N. (1940). The ecology and population dynamics of the wild rabbit (Oryctolagus
cuniculus). Ann. Appl. Biol. 27, 509-26.
Trumble, H. C. (1948). Rainfall, evaporation and drought frequency in South Australia. S. Aust. J.
Agric. 52, 55-64.
Vaughan, H. E. N., and Vaughan, J. A. (1968). Some aspects of the epizootiology of myxomatosis.
Symp. Zool. Soc. Lond. 24, 289-308.
Waterhouse, E. J. (1959). A survey of the mosquitoes of Coonalpyn Downs, South Australia. Trans.
R. Soc. S. Aust. 82, 259-64.
Williams, R. T., and Parer, I. (1972). The status of myxomatosis at Urana, New South Wales, from
1968 until 1971. Aust. J. 2001. 20, 391-404.
Manuscript received 15 June 1982; accepted 2 August 1982