A METHOD FOR ASSESSING CHANGES IN THE

A B U N D A N C E OF BUTTERFLIES

E. POLLARD

Natural Ent,ironment Research Council, Institute o]Terrestrial Ecology, Monks Wood Experimental
Station, Abbots Ripton, Huntingdon, PEI7 2LS, Great Britain

A BSTR ACT

A method based on transect counts has been developed to assess changes in abundance
of butterfliesJ~'om year to year. The method involves weekly walks around a transect
route making counts of butterflies seen within defined limits. The transects are divided
into sections related to habitat or management units. Walks are made only when
weather conditions satisjy specified minimum requirements. The method has been
testedjbr three years at Monks Wood andJbr two years at a number of other sites.
The basis jbr annual comparisons is an index of abundance which is produced for
each brood of each species, except when separation of broods is not possible. This
index is correlated with abundance, although the precise nature of the relationship will
vary from species to species. Evidence on this is presentedjor two species.
The method makes it possible to monitor the abundance of butterflies at selected
sites, using recorders, such as nature reserve wardens, who can fit in one or two hours
recording each week when the weather is suitable. Such a scheme, based on the
methods described in this paper, began in 1976. In addition to the monitoring oJ
[tuctuations of abundance, the method provides considerable injbrmation on the
phenology and ecology of butterflies. The division of the transects into sections makes
some assessment of the effects of habitat change, due to management or other factors,
possible.

INTRODUCTION

In a previous p a p e r ( P o l l a r d et al., 1975) a m e t h o d was outlined which was

c o n s i d e r e d suitable for r e c o r d i n g fluctuations in butterfly a b u n d a n c e . T h e m e t h o d is
based on transect c o u n t s which are used to calculate an index o f a b u n d a n c e . Since
ll5
Biol. Conserv. (12) 0977)--© Applied Science Publishers Ltd, England, 1977
Printed in Great Britain
116 E. POLLARD

1973 when the preliminary work was done, the study has developed as follows:
(1) A woodland transect in Monks Wood, Cambridgeshire, begun in 1973, has
been continued for a further two years.
(2) Another transect route through the two large fields in Monks Wood has been
used for recording in 1974 and 1975 to test the method in a more open habitat.
(3) At five other sites, mainly nature reserves, transect recording has been done at a
frequency of once a week by wardens or others with similar opportunities for regular
recording. This has provided a pilot trial preliminary to a proposed national
recording scheme.
(4) Population estimates have been made for two species in Monks Wood to
provide data for comparison with the index values.
This paper provides an assessment of the transect recording method and
illustrates its use with results from three years' recording in Monks Wood and two
years at five other sites in eastern England.

METHOD

The method of transect recording adopted is described here briefly with slight
modifications which have been made since the first account (Pollard et al., 1975).
Recording is from the beginning of April until the end of September. The
following criteria are observed in order to provide a degree of standardisation:
(1) Counts are started after 1045 h British Summer Time and completed before
1545 h.
(2) Counts are not made when the temperature is below i 3 °C, from 13 °C to 17 °C
counts are made in sunny conditions ( 6 0 ~o sunshine minimum); above 17°C
conditions may be sunny or cloudy.
The transect routes are divided into sections, which as far as possible coincide with
changes in the nature of the habitat being recorded. Besides simplifying counting
this enables some assessment to be made of the occurrence of butterflies in different
habitats. It also makes it possible to examine the effect of management of habitats.
The recorder walks at a uniform pace and records all butterflies seen within
prescribed limits, it is most convenient to restrict the route to rides and paths, the
boundaries of which are generally obvious. The precise width is not important
provided the boundaries are permanent, although recording may be difficult if the
width is greater than about 5 m. If no established paths are available they may be
made by using canes or other markers. In a few cases it may be necessary to gauge
distances by eye.
Butterflies are recorded up to about 5 m in front of the recorder. Individual
butterflies sometimes fly along ahead of the recorder, in which case only one entry is
made provided that there is no doubt that one butterfly is present. If there is any
doubt a further record is made. Stops may be made to resolve identification
 ASSESSING CHANGES IN ABUNDANCE OF BUTTERFLIES 117

problems, recording being resumed from the point where the walk was interrupted.
If, occasionally, a butterfly cannot be positively identified it is recorded as the
commoner of likely alternatives present in the area at that time.

Calculation of the index

As an illustration of the method, the calculation of the index of abundance for the
hedge brown (Pyronia tithonus (L.)) in Monks Wood in 1973, is shown (Table 1).
The mean count per transect is calculated each week and the index of abundance is
the sum of these means. If only one count is made each week, as suggested for routine
recording on nature reserves, then the index is simply the total number of individuals
seen. A separate index is calculated for each brood of species with more than one
generation a year, and for autumn and spring flights of species which overwinter as
adults.

TABLE 1
CALCULATIONOF THE INDEX OF ABUNDANCEFOR THE HEDGE BROWN (Pyronia tithonus (L.)) ON THE MONKS
WOOD TRANSECT IN 1973. RECORDING WEEKS ARE NUMBERED FROM l APRIL

Week 13 14 15 16 17 18 19 20 21 22

N u m b e r o f transects 4 7 5 4 3 5 4 7 5 2
M e a n c o u n t per transect 0 1,9 34.4 111.8 104.6 79-2 12.3 4.4 0-8 0
S u m o f weekly m e a n s -- I n d e x o f a b u n d a n c e = 349.4

R ES U LTS

Reliability of counts
The Monks Wood transect route (Pollard et al., 1975) has now been in use from
1973-75 at a frequency of approximately four counts a week. A number of recorders
have participated, but the three main ones have been D. O. Elias, E. Pollard and
M. J. Skelton. Counts by these recorders for the green-veined white (Pieris napi (L.))
and ringlet (Aphantopus hyperantus (L.)) illustrate the data obtained (Fig. 1). Counts
of green-veined whites are amongst the more variable, those of the ringlet very
uniform.
These counts can be examined rather more formally by calculating an index of
abundance separately for individual recorders. This has been done for two
recorders, D.O.E. hnd E.P., taking one count made by each recorder each week. If
more than one count was available, the one used was selected at random. On the
relatively few occasions when no count was available counts made by ~substitute'
recorders were used. In almost all cases (Fig. 2) the trend in index value shown by the
two recorders is the same, although for many species D. O. E. produced consistently
higher values than E.P. The most obvious discrepancy is in the small white (Pieris
rapae (L.)) data. This may be due to recorder differences in distinguishing green-
veined and small whites.
118 E. POLLARD

1,50 1973 150

oo ~o
100 100
o •
v

50 v
50 ;
o ••
'= .

0 o
i J i i

1974
'°° r e
• • v 501 oo
w~ e• o
vo

0 o v

i .... J i i i = = =

,oo[ ~o
1975

"e
o

•
5o[ ee

i i i J . I

Apr May Jun Jul Aug Sep Jul Aug

GREEN VEINED WHITE RINGLET

Fig. 1. (~ounts of green-veined whites (Pieris napi (L.)) and ringlets (Aphantopus hyperantus (L.)) m
Monks Wood by three recorders: represented as follows: O D. O. Elias, • E. Pollard, V M. J. Skehon.

Field transect
The two large fields in M o n k s W o o d National Nature Reserve, known as East
Field (4.2 ha) and West Field ( 1-7 ha), were cultivated for a period during the last
war. Since becoming part o f a nature reserve they have been maintained as grassland
by occasional mowing and scrub cutting. They are separated by only a few metres o f
woodland.
The transect route was not along a previously defined track and was marked out
with two lines o f canes to give a band 5 m wide. It was about 1 i 00 m long, and was
divided into 14 sections. The transect was interrupted when the recorder moved from
one field to the other. Counts were made during 1974 and 1975 at a frequency o f
approximately three counts a week from M a y until the end o f September. April
counts were found to be unproductive, as in early spring the species present, such as
the peacock (lnachis io (L.)) and brimstone (Gonepteo'x rhamni (k.)), fly mainly in
areas sheltered from the wind.
 ASSESSING CHANGES IN ABUNDANCE OF BUTTERFLIES ] 19

9 !

400- hedge brown 4001 meadow brown 400 ringlet

\

300- , ' , 300- 300

200- _ 200 200

100- _ -o 100 100

0 0 0
19'73 19.74 19'75 19'73 19.74 19'75 19'73 1974 19'75

40- 80 80
green-veined white brimstone large white
30. o 60 60
/ o 1
20- , , 40~ ,o-- o

10 20"
g -
0 0
i97'3 1'974 1'97g 1'97~3 1'974 19'75 0 i973 i974 i97~
80 ¸ 40, 40-
q peacock wall small skipper
60 ' 30- 30-

/,'
40 = "'--~ 20- /'l 20
o

20 ' 10- 10

0
1'973 19~ 1'97i 0 1973 1974 1975 1973 1974 1975

40] large skipper small white red admiral

301 30 30

20.
q

10 10 o ,o, ,A', 10

0 . 0 0
1973 1974 1975 1973 1974 1975 1973 1974 1975

Fig. 2. Index values for each generation of twelve species of butterfly from the counts made by D. O.
Elias indicated by © and E. Pollard, indicated by O. Autumn and spring flights of butterflies which
overwinter as adults given separate index values. Latin names of the large white (Pieris brassicae (L.)) and
the red admiral (Vanessa atalanta (L.)) are given here as these species are not mentioned in the text.
120 E. POLLARD

It was thought that weather conditions in an open habitat might influence counts
more than in the relatively sheltered conditions of a woodland ride. However, the
consistency of counts of ringlets on the field transect (Fig. 3) compares favourably
with those on the woodland transect (Fig. 1) and this is also true of other species.

Pilot trial on nature reser~,es

Five additional sites, four of them nature reserves, were used for transect
recording, mainly by the reserve wardens. The sites were as follows:
Bevill's Wood, Cambridgeshire. A 15-year-old conifer plantation adjoining
Monks Wood. An area formerly of old deciduous woodland.
Holme Fen NNR, Cambridgeshire. Woodland, mainly of birch, established on a
peat soil after cultivation was abandoned in the late 19th century.
Castor Hanglands NNR, Cambridgeshire. Mixed woodland, heath and grassland,
mainly on calcareous soil.
Moor Farm, Lincolnshire. An 'unimproved' farmland reserve near Horncastle,
with grassland, scrub and woodland.
Gibraltar Point, Lincolnshire. Coastal sand dunes and marshes near Skegness.
In general the required frequency of one count each recording week was
maintained, but occasionally other duties made this impossible and a method of
estimating missing values was necessary. This was done by taking a mean of the
counts either side of the missing one. This will lead to underestimates if the peak
week for a species is missed. A better method may be found when more is known of
the characteristic patterns of data obtained for the different species. If two successive
weeks were missed no estimate of the index was attempted for that generation.
Examples of the data from the five transects, together with the two Monks Wood
transects (Figs. 4, 5 and 6) show that trends in index values may be common to
several sites. The hedge brown (Fig. 4) indices were more or less the same each year,
those for the large skipper (Ochlodes venata (Br. & Grey)) (Fig. 5) declined at all
sites, and those for the wall (Lasiommata megera (L.)) (Fig. 6) increased at all sites
except Gibraltar Point, where it was relatively high the previous year.

Relationship between transect counts and population estimates

The data presented suggest that the method of transects counts is 'robust' in that
the data produced by different recorders at the same site show very similar trends in
index values. It seems likely that the changes recorded reflect changes in population
levels, but without direct evidence this cannot be certain.
Some butterfly species are highly mobile and difficult to catch for studies involving
marking individuals; they may also be migratory, in which case a population unit
cannot usefully be defined. Attempts were made to estimate populations of three
species in Monks Wood, the green-veined white, small heath (Coenonympha
pamphilus (L.)) and ringlet. These were selected because they were abundant and
relatively easy to capture.
 ASSESSING CHANGES IN ABUNDANCE OF BUTTERFLIES J 21

e-
O~ 0',
o

,.J

: ,o
o

0 .o
~ •
i
~0 o • "~ o •°
r "°"°
• oo ~ 0
U.I l-
° ~oo l , ~
o
OO

1 I / i I I I I I ='0
o o o o o o
~uJ
i
~0

I.U

v
o oe o ~
..~ 10

o =8

-a o
i I I i I I I i I
O O 0 O o o

I'-"
|
I.M o'~i,
...I
o o
|
Z ¢ • olO l • 3 ~
o "~ 'E
I
o o e-

|
I
o
'0
6
"q"
o
04
0
I
6 o
122 E. POLLARD

HEDGEBROWN
1974 1975

MONKS WOOD 75.2 ~ 68.8

:k
60
40
20
MONKS WOOD (FII~LDS) 0 147"2

66 J ~ 44
BEVILL'S WOOD

HOLME FEN 73 _ r ~ ~]-L.. 113

20 .
CASTOR o h 2os ~{-~l_ _ 1o6

'e L
20 -~
MOOR FARM 0 108 ~ 72

.o,
20
GIBRALTARPOINT 0 "1
Jul Aug Jul Aug
Fig. 4.
Figs. 4-6. Weekly counts (histogram columns) and index values (printed figures) from the seven
transects used for counts in 1974 and 1975.

The method used for estimating population size was based on frequency of
capture o f individuals. Essentially the sampling procedure involves walking a r o u n d
a predetermined circuit, through the habitat to be sampled, capturing butterflies
encountered and marking each with a spot o f c o l o u r e d paint. On subsequent capture
a further spot is given so that the n u m b e r o f occasions each individual has been
captured is known. After each capture the recorder moves on ten paces before
attempting a further capture, to reduce the chance o f recapturing a butterfly
 ASSESSINGCHANGESIN ABUNDANCEOF BUTTERFLIES 123

LARGE SKIPPER
1974 1975
4.7 1"0
MONKS WOOD
2.9 2"1
MONKS WOOD (FIELDS) 2I

BEVILL'S WOOD :f [~12

[1-n
HOLMEFEN . ~-I
20

15

10

5 65 0

CASTOR 0

15

10

5 45

MOOR FARM 0
%
Jun Jul Aug Jun Jul Aug
Fig. 5.

immediately after release. The circuit should be completed at least twice and
preferably ten or more times. The more recaptures obtained the greater is the
precision of the estimate. On completion of sampling a frequency distribution is
available of numbers of butterflies caught once, twice, three times and so on. A basic
assumption is that the population is closed, without immigration or emigration.
Craig (1953) developed the estimation theory applicable if each individual has an
equal chance of capture and used the distribution model appropriate to this, the
zero-truncated Poisson. Eberhardt (1969) pointed out that, more usually, some
individuals in a population are more susceptible to capture than others, and in this
124 E. POLLARD

us

E
3 I I:
,.,")

0
z

¢',1

0,. .,o
c~
U.I

:D
,6

0
z
Q..
U') '9 0
0

i~ (~ i i i I i i i i J L J i r i ,
u') 0 u'~ 0 u') 0 0 ") 0 0 u') 0 u') 0 ") 0

" ~ " Z ~
0 o_., 0 ~ ,,,

0 ,,, 0 "<

v 0 ~
o
z
0
0
,<
"x.
oZ ~0 ~.~ :~ ,,,<
,." @
Z
0
 ASSESSING CHANGES IN ABUNDANCE OF BUTTERFLIES 125

case a truncated geometric model is likely to fit the data better. In both cases the
basic principle is the same; a frequency distribution is available with one value
missing, that is the number of individuals caught on zero occasions. Assuming the
zero term to be part of the same frequency distribution the total population can be
estimated and standard errors given. In nearly all cases the observed data fitted the
geometric distribution (using the X 2 goodness of fit statistic for comparison) better
than the Poisson model and the former was therefore used for estimates (a worked
example is given in the appendix). The geometric model results in higher estimates
than the Poisson model but the two are very closely correlated and either would have
given similar population trends.
Population estimates were made for the small heath on nine occasions in 1974 and
1975 in East Field in Monks Wood. The small heath is essentially an open grassland
or heathland species. In Monks Wood the field transect had much higher counts
than the woodland transect and therefore the transect counts made in East Field in
the same week as the population estimates were used for comparison with them (Fig.
7). The two were significantly correlated (r = 0.548 p <0.001 using square root
transformations).
Sampling of the ringlet population was on five occasions, all in 1975 in East Field.
Correlations with field and wood transects were significant (field 0-755 p < 0.001,
wood 0.763 p < 0.001). The data are presented (Fig. 8) to show the passage of a
generation of butterflies. This was not possible with the small heath because different
broods overlap considerably.
Population estimates were also attempted for the green-veined white using the
same sampling method but with the sampling circuit around a system of rides in the
wood. The data proved unsuitable for estimates, however, because the assumption
of a closed population was apparently false. There was no increase in the proportion
of previously captured individuals as a day's sampling progressed. This is considered
to be due to the movement of individuals through the area as suggested by Baker
(1969), so that marked butterflies only remained in the locality a short time before
being replaced by new unmarked ones. Because of this the population estimate
increases consistently with the number of captures made and is clearly quite
unreliable. Thus for the green-veined white, although transect counts may be well
correlated with population levels in a large area this could not be demonstrated by
the sampling method adopted.
Even if transect counts are perfectly correlated with population on the day of the
count there remains the possibility that large differences in adult mortality in
different seasons could make comparison of the indices misleading. Thus many
individuals dying quickly could give a lower index than fewer present for a long
period. Inspection of the data so far obtained suggests that this is unlikely to be a
serious problem, but it is one which should be borne in mind.

Phenological injormation
The data obtained by the transect counts show, for most species, the duration of
126 E. POLLARD

,.-.O_

¢..--

I
" E

I t
o

~-J'5

I o i

r o

o o I

u
k-~--I

o~

.o"~

I •
c~

I I
o 0 0 O
8 0
¢,3
0 0
o
r ~
se~,ow!~,s~ uo!~,olndo d Lz~
 ASSESSING CHANGES IN ABUNDANCE OF BUTTERFLIES 127

1000-

800-

600-

FREQUENCY OF CAPTURE
ESTIMATE
400-

200 -
-f-
t
0

2°t
10 FIELD TRANSECT COUNTS

30-

20-
WOOD TRANSECT COUNTS
10-

0
Jul Aug
Fig. 8. Relationship between population estimate (with standard errors) and means of transect counts
made in the week of each estimate on the field and woodland transects in M o n k s Wood. Data for the
ringlet (Aphantopus hyperantus (L.)).
128 E. POLLARD

each adult generation and so give information on the number of generations and
their timing. For example the small white (Fig. 2) had three broods in 1973 and 1975
but only two in 1974 in this part of England. Some species, however, have rather
complex life cycles; for example, the speckled wood (Pararge aegeria (L.)) has three
or sometimes four broods a year with overwintering in larval and pupal stages
(Goddard, 1967). Often the different broods overlap.
Another species with an apparently complex life history is the small heath.
information derived from a number of sources (South, 1906; Frowhawk, 1934; Lees,
t962, 1965, pers. comm.) has been combined to produce the following 'probable' life
cycle in southern England. An early May-June brood of adults produces two types
of larvae; rapid developers which become adult the same season and fly in
August-September, and slow developers which overwinter as large larvae and
become adult in May-June the following year. The August-September adults give
rise to larvae which overwinter at an early stage, mature the following spring and
become adult from late June-early August. The larvae from these overwinter as large
larvae and are added to the following May-June brood. This sequence is considered
to be the typical one but in unusual seasons more or fewer broods may occur. The
data obtained from transect counts in the Monks Wood fields, at Gibraltar Point
and at Moor Farm (Fig. 9), where the species has been fairly abundant, support the
suggested life-cycle in that three broods are usually indicated. Rearing of progeny
from wild-caught females may be necessary for full confirmation.

Effects oJ management on counts

One of the objectives of the transect recording method is to assess the effects of
habitat management on butterfly populations, and the transects are divided into
sections with this in mind. However, there are a number of factors which make such
an assessment difficult. Many species are very mobile and the counts may largely
reflect the attraction of the adult butterflies to flowering plants. As an example, the
peacock butterfly counts increased markedly in section l of the Monks Wood
transect (Fig. 10) after the clearing of trees to widen the ride in winter 1973-74. This
was, at least in part, a response to a flush of teasels (Dipsaeus fullonum) which
themselves respond to sunny conditions and flower vigorously after clearance of
shading vegetation.
Three sections of ride on the Monks Wood transect have been widened during the
period that counts have been made. In addition to section 1, already mentioned,
section 2 was widened in winter 1973-74, and section 5 in 1974-75 (routine cutting of
ride edge vegetation is also practised, on a rotational basis). The data for the green-
veined white (Fig. 10) indicate that it has benefited from the widening of rides, as has
the meadow brown (Maniolajurtina (L.)) (Fig. 11). Again there can be no certainty
that the distribution of breeding populations has changed.
Parts of Monks Wood are managed in other ways. Existing clearings are mown,
occasionally new ones are created, and some woodland compartments are managed
 ASSESSINGCHANGESIN ABUNDANCEOF BUTTERFLIES 129

~° o
• :..." . 0
E
• 00 • 080 •
c-

,.. : . E~
0

~o
e~ • °

~Q

• E

0
~:'~
0e
..'nee. . . ole
~ ." <
%0 • •
oee e •
O00~O0 • • • "~
• el O ~

°° Z
8
• @00 •
eeo o •
•
e-

i | l l i l l a o i i i i i i
0 0 0 0 0 0 0
130 E. POLLARD

G R E E N V E I N E D WHITE PEACOCK

~J

___r~.__ I'°o I
40 Z

io .... 0
rrN
U

L\]~-~----r-~ L~ I I I !~.~ ~ 0 0 ,--

80 x

i.o
-20
II
0 v- , ,. . . . . 0

I '°
1 2 3 4 5 6 7 8 9 1011 12 13 14 1 2 3 4 5 6 7 8 9 1 11 1213 14
SECTION SECTION

Fig. 10. Effect of management on index values for the green-veined white (Pieris napi(L.)) and peacock
(lnachis io (L.)) in the sections of the Monks Wood transect. Histogram columns hatched to indicate the
seasons following winter ride widening in sections 1,2 and 5. The two broods of these species indicated by
I and II.

as coppice, or in other ways. The c o m p a r t m e n t to the south of section 10 of the

woodland transect was coppiced in winter 1974-75, but in the first year since
coppicing there has been no immediate response in the butterfly counts in section 10.
Results are given for the ringlet (Fig. 11) one of the more abundant species in this
part of the wood.
Change is not restricted to managed sections; in the absence of management
habitats are subject to gradual change. In view of this and the other problems
discussed it is probably advisable to attempt no formal analysis of management
effects at present. Information is needed on range of movement from existing
breeding areas, on rates of colonisation of new breeding areas and on other basic
 ASSESSING CHANGES IN ABUNDANCE OF BUTTERFLIES 131

RINGLET MEADOW BROWN 60

40 ~ f 40
1975 ~ r , , . 20 20
~vV-~ H l__ - - 0 ~ 0

f20
f4°
20
Z
o_
t,-

0 0 <

100 Z

80

I
1973
40 40

[~ 20 20

~ -0 0
1 2 3 4 5 7 8 9 1011 12 13 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14
SECTION SECTION
Fig. I 1. Effect of management on index values for the ringlet (Aphantopus hyperantus(L.)) and meadow
brown ( Maniol.jurtina (L.)) in the sections of the M o n k s Wood transect. Histogram columns hatched to
indicate the season following winter ride widening in sections 1, 2 and 5.

aspects of the ecology of even our most common butterflies, before management
effects can be interpreted adequately, in spite of these inherent difficulties in
interpretation, the changing patterns of occurrence of butterflies over a period of
years do illustrate the effects of management in terms which are of considerable
interest to the reserve manager, that is the effects on numbers of butterflies seen in
flight.

CONCLUSIONS

The studies in M o n k s W o o d , using several recorders, suggest that the counts are
reliable in the sense that they are repeatable. Different recorders show similar trends
in counts and the data indicate that variation in weather conditions within the
recommended limits is not o f great importance.
132 E. POLLARD

It remains to be demonstrated that the index of abundance obtained is closely

related to population size, but studies on the small heath and ringlet strongly suggest
that this will prove to be true in many cases.
The initial trial in a woodland area, the transect in the fields at Monks Wood and
the coastal transect at Gibraltar Point produced data that suggest that the recording
method is suitable for a wide range of habitats.
The primary aim is to detect year-to-year trends. In addition it is expected that
information will be acquired on many aspects of the biology and ecology of
butterflies, including life-cycles, habitat requirements and the effects of
management on populations. The interpretation of some of this information,
especially on the effects of management, can only be tentative as little is known of the
range of flight of adults of most species. In turn, however, the information acquired
in transect recording should contribute to this understanding and stimulate further
lines of research. The very mobile species present the greatest problems and for these
species the results from a number of sites in a region may have to be combined to
examine trends in numbers.
In 1976 a national recording scheme, based initially on National Nature Reserves,
was begun. These reserves should provide the necessary continuity of recording but
as they are protected areas they are not typical of our changing countryside.
Recording in unprotected areas will be equally important for comparison with the
nucleus of nature reserves.

ACKNOWLEDGEMENTS

D. O. Elias, M. J. Skelton and J. A. Thomas have been associated with this study
since its beginning and it would have been impossible without their contributions. 1
am also grateful to P. Burnham, R. V. Collier, L. Farrell, N. R. Payne, N. Pitts-
Drake, J. Robinson and S. J. Hitchman, who have tested the recording method
under a variety of conditions and to K. Lakhani for statistical advice including the
provision of a formula for calculating the standard errors used in population
estimates. Mrs J. M. Welch and Mrs M. L. King have both provided valuable
technical help. The final year of the work described was financed by the Nature
Conservancy Council.

REFERENCES

BAKER, R. R. (1969). The evolution of the migratory habit of butterflies. J. Anim. Ecol., 38, 70346.
CRAIG,C. C. (1953). On the utilisation of marked specimensin estimating populations of flying insects.
Biometrika, 40, 170-6.
EBERHARDT,L. L. (1969). Population estimates from recapture frequencies. J. Wildl. Mgmt, 33, 28-39.
FROWHAWK,F. W. (1934). The complete book of British butterflies. London, Ward Lock.
 ASSESSING CHANGES IN ABUNDANCE OF BUTTERFLIES 133

GODDARD, M. J. (1967). Broods of the speckled wood (Pararge aegeria aegerides) (Lep. Satyridae).
Entomologist, 100, 241-54.
LEES, E. (1962). On the voltinism of Coenonymphapitmphilus (L.) (Lep. Satyridae). Entomologist, 95,
5-6.
LEES, E. (1965). Further observations on the voltinism of Coenonymphapamphilus (L.) (Lep. Satyridae).
Entomologist, 98, 43-5.
POLLARD, E., ELIAS, D. O., SKELTON, M. J. & THOMAS, J. A. (1975). A method of assessing the
abundance of butterflies in Monks Wood National Nature Reserve in 1973. Entomologist's Gaz., 26,
79-88.
SOUTH, R. (1906). The butterflies of' the British Isles (lst edition). London, Warne.

APPENDIX

'Frequency of capture' population estimate using

the geometric model (Eberhardt, 1969)
Data for the small heath ( C o e n o n y m p h a p a m p h i l u s (L.))

Number of occasions captured

l 2 3 4 5 6
N u m b e r of individuals 20 7 6 2 1 1
No. of captures 20 14 18 8 5 6
Then total number of individuals r = 37
total number of captures s = 71
estimated geometric parameter
s - r 34
s- I 70 0.486
/~= 1 - q = 0 . 5 1 4

The fitted geometric distribution is found by

E ( n x ) = rt)(l x - l x = O, 1 . . . .

where n:, is the number caught on x occasions.

Thus
Number of occasions captured (x)
1 2 3 4 5 6 7
Observed 20 7 6 2 1 1 0
Expected (n.,) 19.0 9.2 4-5 2-2 1.1 0.5 0.3
The fit is clearly good, but can be compared with the expected Poisson
distribution (Craig, 1953) by calculating the X z goodness of fit statistic for each.
134 L. POLLARD

The p o p u l a t i o n estimate for the g e o m e t r i c d i s t r i b u t i o n is

_ r(s - 1) _ 76'2
S -- Y

a n d the s t a n d a r d e r r o r

= = 12.9