Parasitology (1975), 70, 103-118 With 4 figures in the text
103
Seasonal variation in numbers of the rabbit flea on the wild rabbit A. R. MEAD-BRIGGS, J. A. VAUGHAN and B. D. RENNISON* Ministry of Agriculture, Fisheries and Food, Pest Infestation Control Laboratory, Tangley Place, Worplesdon, Guildford, Surrey and * Hook Rise South, Tolworth, Surbiton, Surrey (Received 14 August 1974) SUMMARY
Samples of rabbits were obtained throughout each month over the 4-year period 1967-70. All fleas were removed, sexed and counted and the reproductive condition of the rabbits recorded. Rabbit fleas Spilopsyllus cuniculi (Dale) were present on both sexes of rabbit at all times of the year. In each year significantly high numbers of fleas were found on the rabbits in January, February, March and April and significantly low numbers in August, September and October. Mean flea counts were significantly higher on female rabbits than on males during April, May and June. During the rest of the year counts from each sex of host did not differ significantly. There were significant differences in flea numbers between years. More female than male fleas were found on both sexes of host throughout the year. The rabbit population sampled is shown to be typical of post-myxomatosis populations with regard to breeding performance and juvenile mortality. The relationship of the observed patterns of change in flea numbers to host and flea breeding and to host behaviour, population size and structure is discussed. INTRODUCTION
Allan (1956) provides an account of the numbers of rabbit fleas Spilopsyllus cuniculi (Dale) on wild rabbits Oryctolagus cuniculus (L.) on the Ballogie estate in Aberdeenshire. This study lasted for a period of 15 months, from 24 December 1953 to 4 March 1955, being terminated by the arrival of the first epizootic of myxomatosis. Allan found fleas on rabbits at all times of the year, but there were substantially higher numbers during March and April. Significantly more female than male fleas were collected. From the findings of his survey Allan suggested a tentative life-cycle for the rabbit flea. More recently other workers have provided further information both about the life-cycle of the rabbit flea and about the reproduction and social behaviour of the rabbit. With regard to the flea it has been shown that for its reproduction S. cuniculi is dependent upon the breeding of its host (Mead-Briggs & Rudge, 1960). Both sexes of flea are influenced by factors from the pre-partum female rabbit * The statistical analyses of the results of this survey were done by B.D.R.
104
A. B. MEAD-BRIGGS, J. A. VAUOHAN AND B. D. BENNISON
and/or the new-born young during their first few days of life (Mead-Briggs, 1964a; Mead-Briggs & Vaughan, 1969) and evidence has been provided that hormones from the host are involved (Rothschild & Ford, 1964, 1966; Rothschild, Ford & Hughes, 1970). With regard to the rabbit, Brambell (1942, 1944), studying highdensity populations before the advent of myxomatosis, had shown that some breeding began in January and that the period during which more than 50 % of the females were pregnant extended for 15-17 weeks between February and June. He had also found that up to 60 % of entire litters were lost in utero at the 11- to 15-day stage of gestation. Mean litter size at parturition was 5-07 and the mean number of young born to each female was 11-7 per season. In contrast, Lloyd (1963), studying a low-density post-myxomatosis population in 1957, found no evidence of intra-uterine mortality of entire litters between 11 and 15 days of gestation. He found that the duration of the breeding season was 22 weeks, the mean litter size was 5-9 and the mean number of young born per female was 29'5. Extensive studies have been made in Australia on the behaviour of wild rabbits confined in large enclosures under nearly natural conditions. These studies have provided information on the social organization and structure of rabbit populations and on the changes in territorial behaviour associated with the breeding season (Myers & Poole, 1959, 1961; Mykytowycz, 1959; Mykytowycz & Gambale, 1965). The survey reported here investigates the seasonal variation over 4 years in numbers of S. cuniculi on rabbits in a population subject to frequent epizootics of myxomatosis. The primary objective was to discover whether the seasonal variations in flea numbers recorded by Allan (1956) from a high-density, premyxomatosis population would be found in a lower density, post-myxomatosis population. The survey was also designed to collect information on host breeding and on changes in the juvenile: adult composition of the samples throughout the year. Since the reproduction of the rabbit flea is closely dependent upon that of its host this information was necessary to confirm that the Clandon population was typical of post-myxomatosis rabbit populations as described by Lloyd (1963, 1970) and contrasted with the pre-myxomatosis rabbit population investigated by Allan (1956), which most probably resembled those described by Brambell (1942, 1944). The data were also required for the interpretation of any observed fluctuations in flea numbers in relation to host and flea breeding, seasonal changes in host behaviour and changes in host numbers.
MATERIALS AND METHODS
Rabbits were collected each week by shooting over areas totalling about 400 Ha on the Clandon Park Estate and an adjoining golf course. The estate, of approximately 800 Ha, is situated about 5 km east of Guildford, Surrey. The northern part of the area is level and lies at 30—45 m A.S.L. whilst in the south it rises to 165 m A.S.L., forming part of the North Downs. Land use on the estate includes arable, pasture, scrub, deciduous and coniferous woodlands and a railway cutting, in addition to the golf course. Rabbits were obtained from all such areas. A map of
Seasonal variation in numbers of rabbit
fleas
105
the study site and details of the numbers of rabbits obtained from the different parts of the estate are given in an account of the incidence of cestodes in the population (Mead-Briggs & Vaughan, 1973). Rabbit numbers were still comparatively low in the study area. It was not possible to obtain samples as large as those collected by Allan (1956) nor to obtain equal numbers of male and female rabbits in each sample. A sample size of 20 rabbits per calendar month was planned; mean sample sizes of 11-06 males and 9-08 females were obtained. The variation in monthly sample sizes evident in Table 1 reflects the difficulties encountered in obtaining rabbits during the survey period. Sampling effort was held at an approximately constant level but was increased at times of low rabbit numbers in an endeavour to obtain the planned quota. Each rabbit was placed in a flea-proof bag immediately after death and the location recorded by o.s. land parcel number. At the laboratory the bag was opened over a large enamelled tray, the fleas were removed from the carcase by vigorous combing of the fur and the bag searched for fleas which had left the body. All fleas were collected in a dry specimen tube using a suction pump and subsequently sexed and counted either under CO2 narcosis or after killing with ether. Straggler fleas (i.e. species other than S. cuniculi) were stored for subsequent identification. The unpaunched weight of the rabbit was recorded, the body cavity opened and the digestive tract removed. The sex and reproductive status were determined. For males an epididymal smear was examined and a subjective estimate made of the density of spermatozoa. Females were classed as anoestrus, pregnant or recently post-partum, and where appropriate the number of corpora lutea per ovary and of embryos per uterine horn were recorded. In pregnant animals without obvious uterine swellings the Fallopian tubes and uteri were flushed with saline and pre-implantation stages (0-8 days) aged as whole specimens. The ages of embryos from 8 to 20 days were determined by comparison with the diagrams of known age domestic rabbit embryos of Minot & Taylor (1905). Older embryos were weighed and their ages assessed by reference to the table of development weights in Brambell (1942). The state of development of the mammary glands and the presence of expressible milk were recorded. The paunched weight was obtained, after removal of the reproductive tract in the case of pregnant females. Samples of blood, lung and spleen were taken and tested for the presence of myxoma antigen and antibody. One tibia was removed, cleaned and the animal classified as juvenile or adult by examination of the degree of ossification of the proximal epiphysis, ossification being complete by 10 months of age (Watson & Tyndale-Biscoe, 1953). A small number of myxomatous rabbits'and others below 1 kg were shot but were discarded from the survey. Those less than 1 kg were rejected since they are under 4 months of age; such rabbits appear to be relatively unattractive to fleas (Allan, 1956; Rothschild & Ford, 1966).
1967 1968 1969 1970 Mean
1967 1968 1969 1970 Mean
Year
6 8 9 5
—
6 12 13 12
X
X
10 8 12
7
478 673
12 13 8 17 —
1
450 995 769
1368 1982 3246 1814
661
y
A
968
y
2114 1857 1489 1533
671
2713 1005 2262 1737
Feb.
2207 2476 1426
266 753
856 4 10 12 8
—
5 5 8 12
1 X
A
y
915
1668
699
1164
130
143 220 691 340 348
May y
1
X
Sex of rabbit: female 4 7 535 9 8 1986 11 13 1027 10 17 1546 — 1274
Sex of rabbit: male 18 6 129 16 2 68 13 8 118 11 10 155 — 118 —
x
June A
528
498 421
51
395 642 688 731 614
y
1142
July
16 16 6
7
—
11 12 11 15
X
Aug. *
357
119 696 399 215
236
343
168 340 91
y
1
7 10
8
7
—
12 12 6 11
X
A
Sept.
N.s. Not significant.
00689 1-3347 0-0302 0-491 0-631 0-042 0-421 0-050 0-148
Mean square
•»* Significant at P < 0-001.
33 1 32 3 11 33 1 3 11
D.F.
Analysis of covariance of the logarithms of the monthly flea counts on rabbit numbers
5 5 11 11
346 946 329
y
1805
Source of variance
3306 2189 1702
659 652
A
Apr.
Residual error in y Regression of y on x in error Deviations from regression in error Between years (adjusted) Between months (adjusted) Months x years (adjusted) Between sexes of rabbit (adjusted) Sexes x years (adjusted) Sexes x months (adjusted)
11 9 11 14
—
756
1037
—
X
1653 1252
f
8 15 11 12
487
y
11 18 12 4
X
Mar.
227
190 241 222 254
161 162 137 321 195
y
11 12 10
2
—
10 9 12 14
X
1-4 N.s. 14.0*** 1-7 N.s. 4.9***
20-9»«*
16-2**»
1
A
y
51 419 229 274 243
182 346 278 351 289
Oct.
6 8
7
13
19 14 11 13 —
X
y
346 517 178 344 346
899 870 755 702 806
Nov.
6 11 4 14
7 —
13 13 8
X
•
.°H
GO
©
H
w
O
g
A. I
Jan. *
Table 1. The numbers of male and female rabbits (x) caught per month from January 1967 to December 1970 and the total numbers of fleas (y) found on them
06
Seasonal variation in numbers of rabbit fleas
107
60
5-6
5-2
E "o 4-8
. 40
3-6
3-2
2-8 20
2-2
2-4
2-6
2-8
30
32
Log10 mean flea count
Fig. 1. Showing the positive linear relationship between the logarithms of the means in Table 1 and the logarithms of the variances. RESULTS AND INFERENCES
Monthly rabbit samples and flea counts Table 1 shows the numbers of male and female rabbits in each monthly sample and the total numbers of fleas collected from them. The variability of the flea counts in the Table suggested that the flea numbers obtained per month from the male and female rabbit samples were distributed in a Poisson or log-normal manner and, therefore, that they should be appropriately transformed before being subjected to analysis. The method described by Jeffers (1960) was used to test which, if any, transformation should be used. For the purposes of the test the logarithms of the means shown in Table 1 and of their variances were plotted as in Fig. 1. The slope of the regression line expressing the relationship between sample variance and sample mean indicated by its value, 2-5, that a logarithmic transformation of the counts was necessary.
108
A. E. MEAD-BRIGGS, J. A. VATJGHAN AND B. D. RENNISON
1600
J F M A M I J A S O N D
J F M A M J J A S O N D
Fig. 2. Geometric mean monthly flea counts, adjusted to the standard sample size of 10-07 animals, for (A) male and (B) female rabbits. Counts derived from the data in Table 1. (Pairs of values that differ by a factor of 1-80 are significantly different at P = 0-05.) Black bars: counts significantly above average (P < 0-05). Shaded bars: counts not significantly different from average. White bars: counts significantly below average (P < 0-05). a,b,c: months in which counts on male and female rabbits were significantly different (P < 0-05).
The total monthly flea counts in Table 1 were accordingly transformed to common logarithms before being used as dependent (y) variables, with the numbers of rabbits as independent (x) co-variables, in an analysis of covariance. The analysis was carried out in the manner described by Snedecor & Cochran (1967; section 14-4). The covariance method was adopted because, as shown by Quenouille (1948), it was a satisfactory, labour-saving alternative to the least-squares method of analysing, factorially, non-orthogonal data of this type. For the same reason covariance analyses were applied to the data in Tables 3 and 4. The analysis of the data in Table 1 shows that after adjustment by covariance to an average sample size of 10-07 rabbits per month, highly significant differences occurred between the average numbers of fleas obtained in each year, in different months of the year, on male and female rabbits and between the proportions of the monthly flea counts that were derived from male and female rabbits (P < 0-001 in each case). Considering firstly the adjusted mean transformed counts for the four years 1967-70, which were respectively 2-49, 2-82, 2-71 and 2-79; that for 1967 was very significantly lower than average (P < 0-001), while those for 1968 and 1970 were significantly higher (respectively P < 0-01 and P < 0-05). The mean for 1967 was, at the same time, very significantly lower than each of the means for the three subsequent years (P < 0-001) and the mean for 1969 was just significantly lower than that for 1968 (P < 0*05). Thus, the highly significant difference between the four means arose mainly because many fewer fleas were collected in 1967 than in the next 3 years.
Seasonal variation in numbers of rabbit
fleas
109
Table 2. The percentages of male fleas in the monthly flea counts from male and female rabbits caught each month from January 1967 to December 1970 Total Jan. Feb. Mar. Apr. May June July Aug. Sept. Oct. Nov. Dec. sample Sex of rabbit: male %(J fleas 42-1 43-9 49-3 47-0 49-3 45-1 44-5 41-3 45-4 41-2 42-2 44-2 44-6 S.E.± 2-3 0-6 1-6 2-3 2-9 2-9 0-6 2-5 2-3 1-3 2-2 0-6 — Sex of rabbit: female %
103
Seasonal variation in numbers of the rabbit flea on the wild rabbit A. R. MEAD-BRIGGS, J. A. VAUGHAN and B. D. RENNISON* Ministry of Agriculture, Fisheries and Food, Pest Infestation Control Laboratory, Tangley Place, Worplesdon, Guildford, Surrey and * Hook Rise South, Tolworth, Surbiton, Surrey (Received 14 August 1974) SUMMARY
Samples of rabbits were obtained throughout each month over the 4-year period 1967-70. All fleas were removed, sexed and counted and the reproductive condition of the rabbits recorded. Rabbit fleas Spilopsyllus cuniculi (Dale) were present on both sexes of rabbit at all times of the year. In each year significantly high numbers of fleas were found on the rabbits in January, February, March and April and significantly low numbers in August, September and October. Mean flea counts were significantly higher on female rabbits than on males during April, May and June. During the rest of the year counts from each sex of host did not differ significantly. There were significant differences in flea numbers between years. More female than male fleas were found on both sexes of host throughout the year. The rabbit population sampled is shown to be typical of post-myxomatosis populations with regard to breeding performance and juvenile mortality. The relationship of the observed patterns of change in flea numbers to host and flea breeding and to host behaviour, population size and structure is discussed. INTRODUCTION
Allan (1956) provides an account of the numbers of rabbit fleas Spilopsyllus cuniculi (Dale) on wild rabbits Oryctolagus cuniculus (L.) on the Ballogie estate in Aberdeenshire. This study lasted for a period of 15 months, from 24 December 1953 to 4 March 1955, being terminated by the arrival of the first epizootic of myxomatosis. Allan found fleas on rabbits at all times of the year, but there were substantially higher numbers during March and April. Significantly more female than male fleas were collected. From the findings of his survey Allan suggested a tentative life-cycle for the rabbit flea. More recently other workers have provided further information both about the life-cycle of the rabbit flea and about the reproduction and social behaviour of the rabbit. With regard to the flea it has been shown that for its reproduction S. cuniculi is dependent upon the breeding of its host (Mead-Briggs & Rudge, 1960). Both sexes of flea are influenced by factors from the pre-partum female rabbit * The statistical analyses of the results of this survey were done by B.D.R.
104
A. B. MEAD-BRIGGS, J. A. VAUOHAN AND B. D. BENNISON
and/or the new-born young during their first few days of life (Mead-Briggs, 1964a; Mead-Briggs & Vaughan, 1969) and evidence has been provided that hormones from the host are involved (Rothschild & Ford, 1964, 1966; Rothschild, Ford & Hughes, 1970). With regard to the rabbit, Brambell (1942, 1944), studying highdensity populations before the advent of myxomatosis, had shown that some breeding began in January and that the period during which more than 50 % of the females were pregnant extended for 15-17 weeks between February and June. He had also found that up to 60 % of entire litters were lost in utero at the 11- to 15-day stage of gestation. Mean litter size at parturition was 5-07 and the mean number of young born to each female was 11-7 per season. In contrast, Lloyd (1963), studying a low-density post-myxomatosis population in 1957, found no evidence of intra-uterine mortality of entire litters between 11 and 15 days of gestation. He found that the duration of the breeding season was 22 weeks, the mean litter size was 5-9 and the mean number of young born per female was 29'5. Extensive studies have been made in Australia on the behaviour of wild rabbits confined in large enclosures under nearly natural conditions. These studies have provided information on the social organization and structure of rabbit populations and on the changes in territorial behaviour associated with the breeding season (Myers & Poole, 1959, 1961; Mykytowycz, 1959; Mykytowycz & Gambale, 1965). The survey reported here investigates the seasonal variation over 4 years in numbers of S. cuniculi on rabbits in a population subject to frequent epizootics of myxomatosis. The primary objective was to discover whether the seasonal variations in flea numbers recorded by Allan (1956) from a high-density, premyxomatosis population would be found in a lower density, post-myxomatosis population. The survey was also designed to collect information on host breeding and on changes in the juvenile: adult composition of the samples throughout the year. Since the reproduction of the rabbit flea is closely dependent upon that of its host this information was necessary to confirm that the Clandon population was typical of post-myxomatosis rabbit populations as described by Lloyd (1963, 1970) and contrasted with the pre-myxomatosis rabbit population investigated by Allan (1956), which most probably resembled those described by Brambell (1942, 1944). The data were also required for the interpretation of any observed fluctuations in flea numbers in relation to host and flea breeding, seasonal changes in host behaviour and changes in host numbers.
MATERIALS AND METHODS
Rabbits were collected each week by shooting over areas totalling about 400 Ha on the Clandon Park Estate and an adjoining golf course. The estate, of approximately 800 Ha, is situated about 5 km east of Guildford, Surrey. The northern part of the area is level and lies at 30—45 m A.S.L. whilst in the south it rises to 165 m A.S.L., forming part of the North Downs. Land use on the estate includes arable, pasture, scrub, deciduous and coniferous woodlands and a railway cutting, in addition to the golf course. Rabbits were obtained from all such areas. A map of
Seasonal variation in numbers of rabbit
fleas
105
the study site and details of the numbers of rabbits obtained from the different parts of the estate are given in an account of the incidence of cestodes in the population (Mead-Briggs & Vaughan, 1973). Rabbit numbers were still comparatively low in the study area. It was not possible to obtain samples as large as those collected by Allan (1956) nor to obtain equal numbers of male and female rabbits in each sample. A sample size of 20 rabbits per calendar month was planned; mean sample sizes of 11-06 males and 9-08 females were obtained. The variation in monthly sample sizes evident in Table 1 reflects the difficulties encountered in obtaining rabbits during the survey period. Sampling effort was held at an approximately constant level but was increased at times of low rabbit numbers in an endeavour to obtain the planned quota. Each rabbit was placed in a flea-proof bag immediately after death and the location recorded by o.s. land parcel number. At the laboratory the bag was opened over a large enamelled tray, the fleas were removed from the carcase by vigorous combing of the fur and the bag searched for fleas which had left the body. All fleas were collected in a dry specimen tube using a suction pump and subsequently sexed and counted either under CO2 narcosis or after killing with ether. Straggler fleas (i.e. species other than S. cuniculi) were stored for subsequent identification. The unpaunched weight of the rabbit was recorded, the body cavity opened and the digestive tract removed. The sex and reproductive status were determined. For males an epididymal smear was examined and a subjective estimate made of the density of spermatozoa. Females were classed as anoestrus, pregnant or recently post-partum, and where appropriate the number of corpora lutea per ovary and of embryos per uterine horn were recorded. In pregnant animals without obvious uterine swellings the Fallopian tubes and uteri were flushed with saline and pre-implantation stages (0-8 days) aged as whole specimens. The ages of embryos from 8 to 20 days were determined by comparison with the diagrams of known age domestic rabbit embryos of Minot & Taylor (1905). Older embryos were weighed and their ages assessed by reference to the table of development weights in Brambell (1942). The state of development of the mammary glands and the presence of expressible milk were recorded. The paunched weight was obtained, after removal of the reproductive tract in the case of pregnant females. Samples of blood, lung and spleen were taken and tested for the presence of myxoma antigen and antibody. One tibia was removed, cleaned and the animal classified as juvenile or adult by examination of the degree of ossification of the proximal epiphysis, ossification being complete by 10 months of age (Watson & Tyndale-Biscoe, 1953). A small number of myxomatous rabbits'and others below 1 kg were shot but were discarded from the survey. Those less than 1 kg were rejected since they are under 4 months of age; such rabbits appear to be relatively unattractive to fleas (Allan, 1956; Rothschild & Ford, 1966).
1967 1968 1969 1970 Mean
1967 1968 1969 1970 Mean
Year
6 8 9 5
—
6 12 13 12
X
X
10 8 12
7
478 673
12 13 8 17 —
1
450 995 769
1368 1982 3246 1814
661
y
A
968
y
2114 1857 1489 1533
671
2713 1005 2262 1737
Feb.
2207 2476 1426
266 753
856 4 10 12 8
—
5 5 8 12
1 X
A
y
915
1668
699
1164
130
143 220 691 340 348
May y
1
X
Sex of rabbit: female 4 7 535 9 8 1986 11 13 1027 10 17 1546 — 1274
Sex of rabbit: male 18 6 129 16 2 68 13 8 118 11 10 155 — 118 —
x
June A
528
498 421
51
395 642 688 731 614
y
1142
July
16 16 6
7
—
11 12 11 15
X
Aug. *
357
119 696 399 215
236
343
168 340 91
y
1
7 10
8
7
—
12 12 6 11
X
A
Sept.
N.s. Not significant.
00689 1-3347 0-0302 0-491 0-631 0-042 0-421 0-050 0-148
Mean square
•»* Significant at P < 0-001.
33 1 32 3 11 33 1 3 11
D.F.
Analysis of covariance of the logarithms of the monthly flea counts on rabbit numbers
5 5 11 11
346 946 329
y
1805
Source of variance
3306 2189 1702
659 652
A
Apr.
Residual error in y Regression of y on x in error Deviations from regression in error Between years (adjusted) Between months (adjusted) Months x years (adjusted) Between sexes of rabbit (adjusted) Sexes x years (adjusted) Sexes x months (adjusted)
11 9 11 14
—
756
1037
—
X
1653 1252
f
8 15 11 12
487
y
11 18 12 4
X
Mar.
227
190 241 222 254
161 162 137 321 195
y
11 12 10
2
—
10 9 12 14
X
1-4 N.s. 14.0*** 1-7 N.s. 4.9***
20-9»«*
16-2**»
1
A
y
51 419 229 274 243
182 346 278 351 289
Oct.
6 8
7
13
19 14 11 13 —
X
y
346 517 178 344 346
899 870 755 702 806
Nov.
6 11 4 14
7 —
13 13 8
X
•
.°H
GO
©
H
w
O
g
A. I
Jan. *
Table 1. The numbers of male and female rabbits (x) caught per month from January 1967 to December 1970 and the total numbers of fleas (y) found on them
06
Seasonal variation in numbers of rabbit fleas
107
60
5-6
5-2
E "o 4-8
. 40
3-6
3-2
2-8 20
2-2
2-4
2-6
2-8
30
32
Log10 mean flea count
Fig. 1. Showing the positive linear relationship between the logarithms of the means in Table 1 and the logarithms of the variances. RESULTS AND INFERENCES
Monthly rabbit samples and flea counts Table 1 shows the numbers of male and female rabbits in each monthly sample and the total numbers of fleas collected from them. The variability of the flea counts in the Table suggested that the flea numbers obtained per month from the male and female rabbit samples were distributed in a Poisson or log-normal manner and, therefore, that they should be appropriately transformed before being subjected to analysis. The method described by Jeffers (1960) was used to test which, if any, transformation should be used. For the purposes of the test the logarithms of the means shown in Table 1 and of their variances were plotted as in Fig. 1. The slope of the regression line expressing the relationship between sample variance and sample mean indicated by its value, 2-5, that a logarithmic transformation of the counts was necessary.
108
A. E. MEAD-BRIGGS, J. A. VATJGHAN AND B. D. RENNISON
1600
J F M A M I J A S O N D
J F M A M J J A S O N D
Fig. 2. Geometric mean monthly flea counts, adjusted to the standard sample size of 10-07 animals, for (A) male and (B) female rabbits. Counts derived from the data in Table 1. (Pairs of values that differ by a factor of 1-80 are significantly different at P = 0-05.) Black bars: counts significantly above average (P < 0-05). Shaded bars: counts not significantly different from average. White bars: counts significantly below average (P < 0-05). a,b,c: months in which counts on male and female rabbits were significantly different (P < 0-05).
The total monthly flea counts in Table 1 were accordingly transformed to common logarithms before being used as dependent (y) variables, with the numbers of rabbits as independent (x) co-variables, in an analysis of covariance. The analysis was carried out in the manner described by Snedecor & Cochran (1967; section 14-4). The covariance method was adopted because, as shown by Quenouille (1948), it was a satisfactory, labour-saving alternative to the least-squares method of analysing, factorially, non-orthogonal data of this type. For the same reason covariance analyses were applied to the data in Tables 3 and 4. The analysis of the data in Table 1 shows that after adjustment by covariance to an average sample size of 10-07 rabbits per month, highly significant differences occurred between the average numbers of fleas obtained in each year, in different months of the year, on male and female rabbits and between the proportions of the monthly flea counts that were derived from male and female rabbits (P < 0-001 in each case). Considering firstly the adjusted mean transformed counts for the four years 1967-70, which were respectively 2-49, 2-82, 2-71 and 2-79; that for 1967 was very significantly lower than average (P < 0-001), while those for 1968 and 1970 were significantly higher (respectively P < 0-01 and P < 0-05). The mean for 1967 was, at the same time, very significantly lower than each of the means for the three subsequent years (P < 0-001) and the mean for 1969 was just significantly lower than that for 1968 (P < 0*05). Thus, the highly significant difference between the four means arose mainly because many fewer fleas were collected in 1967 than in the next 3 years.
Seasonal variation in numbers of rabbit
fleas
109
Table 2. The percentages of male fleas in the monthly flea counts from male and female rabbits caught each month from January 1967 to December 1970 Total Jan. Feb. Mar. Apr. May June July Aug. Sept. Oct. Nov. Dec. sample Sex of rabbit: male %(J fleas 42-1 43-9 49-3 47-0 49-3 45-1 44-5 41-3 45-4 41-2 42-2 44-2 44-6 S.E.± 2-3 0-6 1-6 2-3 2-9 2-9 0-6 2-5 2-3 1-3 2-2 0-6 — Sex of rabbit: female %
