Biological Conservation 114 (2003) 453–461

www.elsevier.com/locate/biocon

Conservation implications of the distribution of genetic diversity at

different scales: a case study using the marsh fritillary butterfly
(Euphydryas aurinia)
Domino A. Joyce*, Andrew S. Pullin
School of Biosciences, The University of Birmingham, Edgbaston, Birmingham B15 2TT, UK

Received 29 November 2002; received in revised form 1 February 2003; accepted 12 February 2003

Abstract
In the UK, Euphydryas aurinia exists in fragmented habitat patches, and undergoes population fluctuations as a result of a larval
parasitoid. Its range is declining in the UK and conservation is thought to require a landscape approach since populations spread
over large areas in some years and contract to core breeding patches in others. We examined populations at a range of geographic
scales using allozyme electrophoresis to look for evidence of gene flow and differences in genetic diversity among populations.
Nationally, our FST value was 0.1542 but between population groups within the suspected colonisation range of the butterfly (ca. 20
km), FST values were not significantly different from zero. Genetic diversity in terms of number of alleles and heterozygosity was
reasonably high in natural populations (He=0.267) but low in an introduced, isolated population. We infer that migration between
closely spaced subpopulations (in a metapopulation) maintains a high genetic effective population size (large number of individuals
in a population that contribute genes to the next generation) which offsets any local reductions in population numbers due to sto-
chastic extinctions or parasitoid effects. We therefore conclude that effective conservation of the species must seek to provide net-
works of suitable habitat for groups of subpopulations, rather than maintaining habitat for isolated populations.
# 2003 Elsevier Ltd. All rights reserved.
Keywords: Bottleneck; Conservation genetics; Reintroduction; Metapopulation; FST

1. Introduction There are two schools of thought regarding the effect

Conservation goals from a genetics perspective are to
conserve as much genetic diversity and variability as
possible in order to retain adaptive capacity and evolu-
tionary potential. Frankham (1995) reviews issues in
conservation genetics, and describes habitat fragmenta-
tion and reduced migration, the loss of genetic variation
in small populations, and inbreeding depression as fac-
tors that warrant consideration. Fragmentation of
habitats can lead to the formation of a metapopulation
structure, which in turn influences the genetic structure
and diversity of a species. Genetic structure is caused by
a departure from random mating that affects allele dis-
tribution and occurs when populations are separated
ecologically or geographically.
* Corresponding author. Present address: School of Biological Sci-
ences, The University of Hull, HU6 7RX, UK.
E-mail address: d.joyce@hull.ac.uk (D.A. Joyce).
0006-3207/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0006-3207(03)00087-9
a metapopulation structure has on the genetic differ-
entiation between the subpopulations within it. Slatkin
(1985) argues that colonisation is effectively gene flow,
so subpopulations will not differentiate. Wade and
McCauley (1988) showed that this is true if colonisation
of a new patch (and therefore formation of a new sub-
population) is from a large number of individuals from
many subpopulations. On the other hand if colonisers
are low in number and come from only one or a few
source subpopulations, then subpopulation turnover
can actually enhance differentiation within the metapo-
pulation. Additionally, fluctuations in population size
lead to a low genetic effective population size (thought
to be approximately equal to the harmonic mean popu-
lation size). The rate of loss of alleles through genetic
drift is higher in a small population than in a large one
(probability of fixation of an allele=1/(2N), where
N=population size, Wright, 1931) so fluctuations in
population size that occur at a subpopulation level will
454 D.A. Joyce, A.S. Pullin / Biological Conservation 114 (2003) 453–461

increase drift and reduce genetic diversity within that attention to the ‘key areas’ defined in the Species Action
subpopulation. Genetic drift could also increase differ- Plan (Butterfly Conservation, 1995). Populations are
entiation between subpopulations, and might lead to an shown in Fig. 1 with O.S. grid references and sample
overall increase in genetic diversity in the metapopula- sizes listed below. These populations were chosen to
tion. Harrison and Hastings (1996) draw the conclusion represent a range of pairwise geographic distances,
that most natural metapopulation systems with high which ranged from 5.4 km apart (within the suspected
rates of subpopulation extinction and colonisation will colonisation range of the species of 15–20 km, Warren,
show low levels of genetic differentiation among sub- 1994) between Vogwell Cottage and Broadaford Farm
populations, because such ‘high turnover’ species tend in Dartmoor, to 690 km apart, the distance between
also to have high rates of dispersal and gene flow. Ledmore on Mull, and Goss Moor in Cornwall. The
Colonies of the marsh fritillary butterfly (Euphydryas populations included an isolated, introduced population
aurinia Rott.) can be found in damp acidic or dry cal- in Lincolnshire, founded by three adult butterflies (Peter
careous grasslands. There is thought to be a regular Cawdell, personal communication).
turnover of colonies caused by high rates of extinction In each case, one larva was taken per web, in order to
and some colonisation, consistent with metapopulation avoid siblings and thus maximise the genetic diversity
structure (Warren, 1994). Fluctuations in population sampled. Larvae were taken from webs just prior to
numbers caused by larval parasitoids (Porter, 1981) hibernacula formation (September/October 1996 and
make the species extremely prone to local extinctions. A 1997) so that damage to the population should be no
review of UK colonies in 1990, compared with a survey
undertaken in 1983, shows a decline in E. aurinia inci-
dence by 62% of 10-km grid squares examined. Addi-
tionally, some habitat patches are transient, and patch
occupancy is often short, leading to a constant state of
flux in the butterfly’s distribution (Warren, 1994).
E. aurinia has declined substantially throughout Eur-
ope and is now protected under the Bern Convention
and listed on Annex II of the EC Habitats Directive. The
UK is now thought to be a major stronghold, although
here too, a substantial east–west decline has been docu-
mented and the species is listed in the UK Biodiversity
Action Plan (Asher et al., 2001). Warren (1994) suggests
that, although previously thought to be a relatively
sedentary butterfly, E. aurinia is probably fairly mobile
with a suggested colonisation range of 15–20 km. Work
done by Lewis and Hurford (1997) in Wales finds some
support for the suggestion made by Thomas (1994) that
E. aurinia comes closer to a ‘mainland-island’ popula-
tion structure than other British butterflies, but believe
that there is as yet insufficient information on dispersal
ability, or extinctions and colonisations to determine
precisely how populations are structured.
We investigated the genetic structure and diversity
among populations of E. aurinia in the UK at three dif-
ferent scales from local to national to ask the following
questions: (1) At what scale is genetic structuring of
populations evident? (2) How does recent population
history affect genetic diversity? and (3) What implications Fig. 1. Sites in the UK from which E. aurinia were sampled, as fol-
does the genetic structure have for conservation strategy? lows: (1) Ledmore, Mull NM 510 460; (2) Loch Don, Mull NM 720
320; (3) Argyll NR 800 940; (4) Islay NR 277 691; (5) Middlesceugh,
Cumbria NY 410 403; (6) Lincolnshire TF 145 743; (7) Aberbargoed
ST 162 991; (8) Pengwern Common SS 533 917; (9) Salisbury Plain SU
2. Methods 200 500; (10) Rooksmoor, Dorset ST 740 110; (11) Hog Cliff NNR,
Dorset SY632 968; (12) Cerne abbas, Dorset ST 666 018; (13) Vogwell
2.1. Sample collection Cottage, Dartmoor SX 720 821; (14) Broadaford Farm, Dartmoor SX
690 773; 15) Goss Moor, Cornwall SW 948 596; (16) Breney Common,
Cornwall SX 056 610; (17) Murlough NNR J 405 342. Solid lines
Seventeen populations were selected and sampled to group local populations, and dashed lines group regional populations.
cover the species’ range in the UK, with particular Details are given in Table 1.
 D.A. Joyce, A.S. Pullin / Biological Conservation 114 (2003) 453–461
greater than that which would occur naturally due to
overwintering mortality. Samples were frozen in liquid
nitrogen, and then stored at 70  C until use.
2.2. Gel electrophoresis
Allozyme electrophoresis was chosen so that the geo-
graphic distribution of allele frequencies could allow us
to infer the genetic structure of the population on a
spatial scale. Both alleles at any locus are fully expressed
in the heterozygous state, and this co-dominant infor-
mation can be used to estimate measures of genetic
diversity.
Caterpillars were decapitated, then homogenised in 80
ml of cold homogenising buffer (50 mM NaCl, 0.5 mM
EDTA, 1 mM DTT, 0.1% SDS, 1 mM b-mercaptoetha-
nol) and this mix was centrifuged at 5000 rpm for 2 min.
The supernatant was applied directly to a cellulose ace-
tate gel for horizontal electrophoresis (Helena Bios-
ciences). The entire procedure was carried out on ice
and on the same day in order to keep enzymatic activity
to a maximum. Each electrophoretic run contained
individuals from as broad a range of populations as
possible and a standard was used on each gel to facil-
itate scoring and ensure run-to-run variation did not
produce erroneous results. Enzymes with abbreviation,
number of loci and E.C. number respectively were
stained as follows: phosphoglucose isomerase (PGI, one
locus 5.3.1.9) phosphoglucomutase (PGM, one locus
2.7.5.1) aspartate aminotransferase or glutamate oxa-
loacetate transaminase (GOT, two loci 2.6.1.1) Leu-Ala
peptidase (LA, one locus 3.4.13) adenylate kinase (AK,
one locus 2.7.43). All enzymes were run in tris–gly buf-
fer (0.25 M Tris, 0.2 M glycine pH 8.6) except AK
which was run in phosphate buffer (0.03 M NaH2PO4,
0.015 M Na2HPO4 pH 6.3). Biochemical staining pro-
cedures followed Mallet et al. (1993) and Richardson et
al. (1986) with minor modifications. Protocols can be
obtained from the authors upon request. Alleles were
labelled alphabetically starting with the allele nearest
the origin.
Table 1
Population groups are described with corresponding map number (see Fig. 2)
Population group Populations FST value
Mull 1, 2 0.1027
Dorset 10, 11, 12 0.0126
Dartmoor 13, 14 0.0527
Cornwall 15, 16 0.0395
Northern group 1–4 0.0312
Wales 7, 8 0.0136
Southwest England group 7–16 0.0522
Southern group 9–16 0.0456
All populations 1–17 0.1621
FST values for population groups are with standard deviation (S.D.), chi-squared statistic (w2) degrees of freedom (d.f.) and significance. N.S.=not
significant; * significant at the P < 0.05 level; ** significant at the P <0.01 level.
455
2.3. Statistical analysis
TFPGA version 1.3 (Miller, 1997) was used for sta-
tistical analyses, except where otherwise stated. Exact
tests of Hardy–Weinberg equilibrium were used in
which the conventional Monte-Carlo method of rando-
misation gives an approximation of the true exact
probability (where more than two alleles occur at a
locus), which is then compared with the observed sam-
ple. The significance of the observed data set is found by
determining the proportion of random data sets with
probabilities less than or equal to that of the original
data set. Deviations from Hardy–Weinberg at the
P < 0.01 level were tested. Sequential Bonferroni cor-
rection (Rice, 1989) was applied both over all popula-
tions, and within populations over the different loci.
Hardy–Weinberg equilibrium at 20 km groupings was
also tested in the same way. Reynolds et al.’s coancestry
distance (1983) was calculated, since it is a measure of
genetic distance for short-term evolution when the
divergence between populations with a common ances-
try may be regarded as being solely due to drift. This is
the process most likely to have shaped E. aurinia popu-
lations in the UK. A Mantel test was performed using
NTSys version 1.7 (Rohlf, 1997).
TFPGA was also used to calculate F-statistics,
according to Weir and Cockerham (1984), and results
were jack-knifed over loci to obtain standard deviations.
F-statistics were obtained at local, regional and national
scales. National scales included the entire data set, and
the entire ‘natural population’ data set (i.e. not includ-
ing the Lincolnshire population). Populations within 20
km of one another were classed as local populations.
Regional population groupings are outlined (along with
the national and local groupings) in Table 1. Statistical
significance was tested using w2=2NFST, d.f.=number
of sites 1, where N=total number of individuals
(Workman and Niswander, 1970). Geographic distances
were calculated as the shortest distance between two
populations, found by mapping population grid refer-
ences onto a UK map in Arcview.
S.D. w2 d.f. Significance
0.0698 5.95 1 *
0.0128 2.57 2 N.S.
0.1422 2.32 1 N.S.
0.0833 2.69 1 N.S.
0.0159 4.12 3 N.S.
0.0280 11.42 1 N.S.
0.0760 23.28 9 **
0.0406 16.51 7 *
0.0371 116.06 16 **
456 D.A. Joyce, A.S. Pullin / Biological Conservation 114 (2003) 453–461
Populations that have experienced a recent reduction
of their effective population size exhibit a correlative
reduction of the allele numbers and heterozygosity
(Cornuet and Luikart, 1996). The program Bottleneck
(Cornuet and Luikart, 1996) was therefore used in an
attempt to identify populations that may recently have
been through bottlenecks.
3. Results
Twenty-one caterpillars were collected on average
from each of 17 UK populations. A total of 20 alleles
were found across six polymorphic loci, with a mini-
mum of nine in Lincolnshire and a maximum of 17 in
Islay (Table 2); no population contained every allele,
and only one population contained a private allele at
very low frequency (Cerne abbas, PGM-e at 0.027).
Individual populations did not deviate significantly
from Hardy–Weinberg equilibrium after the application
of a Bonferroni correction, suggesting random mating
occurs within populations. The same was true for
population groups, but the entire data set was found to
deviate significantly from Hardy–Weinberg equilibrium
at all loci except GOT2. This was true whether Bonfer-
roni correction was applied over the entire dataset or
over different loci at individual populations.
3.1. Genetic structure and geographic distance
The FST (Weir and Cockerham’s y) over all popula-
tions was 0.1621, S.D. 0.0408, and without Lincoln-
shire, 0.1542, S.D. 0.0408, both figures being significant
(P < 0.01). An FST value that differs significantly from
zero among a group of populations implies that they
have undergone genetic differentiation, and are there-
fore subdivided, as would be expected for populations
over such a large scale. An FST value which does not
differ significantly from zero, on the other hand, shows
no subdivision or genetic differentiation, and implies
that gene flow between populations is high, most likely
a result of migration between them. FST tests were car-
ried out on various regional subsets of populations
(Table 1). The southern group of populations, which are
located a maximum of 244 km apart (between Salisbury
and Goss Moor), had an FST value of 0.0522 (P < 0.01).
This value represents a small but significant amount of
population subdivision, leading to genetic differentia-
tion. This southern group was then further partitioned
into Welsh and English populations. The two Welsh
populations (63 km apart) appeared not to show sig-
nificant genetic differentiation but the south east popu-
lation group did so with FST=0.0456 (P < 0.05). The
populations in Scotland, separated by a maximum of 81
km, also showed no sign of population subdivision
(FST=0.0312, P > 0.05). The finest scale examined con-
sisted of those populations within the suspected coloni-
sation range of the butterfly (20 km). The Dorset,
Cornwall and Dartmoor populations all produced non
significant FST values (0.0126, 0.0395, and 0.0527
respectively, P > 0.05 in all cases). Mull’s populations
are slightly further apart (25 km) and the FST for this
subgroup was surprisingly high at 0.1027, which was
significant at the P < 0.05 level.
Reynolds et al.’s coancestry distance (1983) was plot-
ted against the pairwise geographic distance between
populations (Fig. 2). Using a Mantel test, no correlation
between genetic and geographic distance was found,
either with or without the Lincolnshire population
(Z=0.248, P > 0.1 and Z=0.264 and P > 0.1, respec-
tively).
3.2. Genetic diversity and population history
Fig. 3 shows the distribution of genetic diversity. Two
measures are shown, Nei’s (1978) unbiased hetero-
zygosity (termed here He), and the number of alleles per
population (Na). For both measures, the introduced
Lincolnshire population was lowest but populations
containing the highest values varied according to the
measure chosen.
The program ‘Bottleneck’ (Cornuet and Luikart,
1996) uses a sign test for probability of heterozygosity
excess as a result of a recent population bottleneck.
Each population was tested and none was found to have
a significant excess of heterozygotes (P > 0.05), thus
these data do not provide evidence that any population
has been through a recent genetic bottleneck.
4. Discussion
4.1. Genetic structure and geographic distance
At the national scale, the FST value is indicative of
subdivision of populations, coupled with genetic differ-
entiation. This contrasts with the FST values obtained
for populations within the local scale of ca. 20 km, the
suspected colonisation distance for the species. Over the
entire data set, significant deviations from Hardy–
Weinberg equilibrium are found, whereas these devia-
tions are not present within population groups. This
may indicate a threshold distance at which random
mating between individuals does not occur and there-
fore populations start to become genetically distinct.
Populations in the Scottish group do not show signs of
subdivision, which is surprising given that they are
island populations separated by sea. It should be noted,
however, that local scale sample sizes may have been
too small to test for differences reliably, because of the
problems associated with sampling from small habitat
patches containing small subpopulations of rare species.
Table 2
Allele frequencies in E. aurinia populations at 6 allozyme loci. N=sample size, He=expected heterozygosity, Ho=observed heterozygosity

Argyll Breney Goss Cumbria Vogwell Broadaford Rooksmoor Cerne Hog Cliff Murlough Islay Ledmore Loch Don Pengwern Aberbargoed Salisbury Lincolnshire
(4) Common (16) Moor (15) (5) Cottage (13) Farm (14) (10) abbas (12) NNR (11) NNR (17) (4) (1) (2) (8) (7) Plain (9) (6)

AK* a 0.056 0.059 0.033 0.000 0.039 0.000 0.000 0.016 0.089 0.000 0.000 0.000 0.033 0.000 0.000 0.000 0.000
b 0.944 0.941 0.967 1.000 0.962 1.000 1.000 0.984 0.911 0.515 0.960 1.000 0.967 1.000 1.000 1.000 1.000
c 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.485 0.040 0.000 0.000 0.000 0.000 0.000 0.000
N 9 17 15 14 13 7 40 31 28 34 25 12 15 19 17 23 17
He 0.105 0.111 0.064 0.000 0.074 0.000 0.000 0.032 0.163 0.500 0.077 0.000 0.064 0.000 0.000 0.000 0.000
Ho 0.111 0.118 0.067 0.000 0.077 0.000 0.000 0.032 0.179 0.382 0.000 0.000 0.067 0.000 0.000 0.000 0.000

GOT1* a 0.000 0.000 0.000 0.063 0.042 0.000 0.000 0.000 0.000 0.000 0.019 0.000 0.000 0.000 0.000 0.000 0.000

D.A. Joyce, A.S. Pullin / Biological Conservation 114 (2003) 453–461

b 1.000 1.000 1.000 0.625 0.958 1.000 0.988 0.984 1.000 0.819 0.815 1.000 1.000 0.957 1.000 0.978 1.000
c 0.000 0.000 0.000 0.313 0.000 0.000 0.012 0.016 0.000 0.181 0.074 0.000 0.000 0.044 0.000 0.022 0.000
d 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.093 0.000 0.000 0.000 0.000 0.000 0.000
N 10 17 17 16 12 9 42 31 28 36 27 13 16 23 19 23 17
He 0.000 0.000 0.000 0.508 0.080 0.000 0.024 0.032 0.000 0.296 0.322 0.000 0.000 0.083 0.000 0.043 0.000
Ho 0.000 0.000 0.000 0.375 0.083 0.000 0.024 0.032 0.000 0.250 0.296 0.000 0.000 0.085 0.000 0.044 0.000

GOT2 a 1.000 1.000 1.000 1.000 1.000 1.000 0.987 0.983 1.000 0.879 0.981 1.000 1.000 1.000 1.000 1.000 1.000
b 0.000 0.000 0.000 0.000 0.000 0.000 0.013 0.014 0.000 0.121 0.019 0.000 0.000 0.000 0.000 0.000 0.000
N 10 17 17 15 12 9 0 30 29 33 26 12 16 22 17 23 15
He 0.000 0.000 0.000 0.000 0.000 0.000 0.025 0.033 0.000 0.213 0.038 0.000 0.000 0.000 0.000 0.000 0.000
Ho 0.000 0.000 0.000 0.000 0.000 0.000 0.026 0.033 0.000 0.242 0.039 0.000 0.000 0.000 0.000 0.000 0.000

LA* a 0.200 0.059 0.100 0.071 0.000 0.000 0.167 0.083 0.069 0.047 0.091 0.227 0.033 0.405 0.235 0.196 0.000
b 0.800 0.941 0.900 0.929 1.000 1.000 0.833 0.917 0.931 0.953 0.909 0.773 0.967 0.595 0.765 0.804 1.000
N 10 17 15 14 12 9 36 24 29 32 22 11 15 21 17 23 16
He 0.320 0.111 0.180 0.133 0.000 0.000 0.278 0.153 0.128 0.089 0.165 0.351 0.064 0.482 0.360 0.315 0.000
Ho 0.400 0.118 0.200 0.143 0.000 0.000 0.167 0.167 0.069 0.094 0.182 0.273 0.067 0.429 0.235 0.391 0.000

PGI* a 0.000 0.000 0.000 0.000 0.000 0.000 0.103 0.033 0.155 0.000 0.000 0.000 0.000 0.000 0.000 0.044 0.206
b 0.350 0.765 0.529 0.333 0.583 0.667 0.641 0.633 0.621 0.015 0.352 0.208 0.313 0.674 0.842 0.609 0.706
c 0.650 0.235 0.471 0.667 0.417 0.333 0.256 0.317 0.224 0.909 0.630 0.792 0.688 0.326 0.158 0.348 0.088
d 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.017 0.000 0.076 0.019 0.000 0.000 0.000 0.000 0.000 0.000
N 10 17 17 15 12 9 39 30 29 33 27 12 16 23 19 23 17
He 0.455 0.360 0.498 0.444 0.486 0.444 0.513 0.497 0.540 0.168 0.479 0.330 0.427 0.440 0.266 0.507 0.452
Ho 0.700 0.471 0.471 0.267 0.167 0.444 0.590 0.400 0.552 0.182 0.444 0.417 0.500 0.391 0.316 0.435 0.529

PGM* a 0.000 0.033 0.083 0.067 0.091 0.071 0.135 0.036 0.000 0.076 0.037 0.100 0.192 0.132 0.039 0.109 0.094
b 0.625 0.633 0.667 0.533 0.909 0.571 0.351 0.464 0.345 0.879 0.500 0.650 0.231 0.632 0.615 0.674 0.906
c 0.375 0.300 0.250 0.300 0.000 0.286 0.473 0.357 0.586 0.046 0.389 0.250 0.577 0.211 0.308 0.174 0.000
d 0.000 0.033 0.000 0.100 0.000 0.071 0.014 0.143 0.069 0.000 0.074 0.000 0.000 0.026 0.039 0.044 0.000
e 0.000 0.000 0.000 0.000 0.000 0.000 0.027 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
N 8 15 12 15 11 7 37 28 29 33 27 10 13 19 13 23 16
He 0.469 0.507 0.486 0.611 0.165 0.582 0.634 0.635 0.533 0.220 0.592 0.505 0.577 0.539 0.524 0.502 0.170
Ho 0.250 0.600 0.167 0.667 0.173 0.857 0.622 0.679 0.483 0.152 0.482 0.400 0.385 0.316 0.462 0.435 0.188

457
* Indicates a locus that exhibited significant deviations from Hardy–Weinberg expectations when entire dataset was tested. Numbers in parentheses correspond to population numbers in Fig. 1.
458 D.A. Joyce, A.S. Pullin / Biological Conservation 114 (2003) 453–461

Fig. 2. Relationship between the geographic and genetic distances between all pairs of populations, including the introduced population in
Lincolnshire shown as open circles.

Fig. 3. Distribution of genetic diversity in each of the 17 populations of E. aurinia in the UK. He=Nei’s 1978 unbiased heterozygosity; Na=number
of alleles.

No correlation between genetic and geographic dis-
tance was found at the national scale. As the distance
between populations increases, one would expect that a
relationship between genetic and geographic distance
might become evident, but this might only be detectable
if distance were the only important variable. Our study
includes populations so well separated that the effects of
random genetic drift will most likely counteract the
limited migration which may occur. Genetic drift there-
fore affects the genetic distance between populations
rather more than geographic distance, since exchange of
individuals at large distances will not occur.
4.2. Genetic diversity and population history
Packer et al., (1988) compared estimates of hetero-
zygosity across 56 species of Lepidoptera and found the
minimum value to be 0.001 (Ecdytolopha mana) and the
maximum 0.324 (Pectinophora gossypiella) with a mean
value of 0.105. The mean for UK E. aurinia populations
(0.267) is thus well above average, and suggests no cause
for concern regarding the genetic diversity in this spe-
cies. However, there are a number of problems asso-
ciated with this kind of comparison. The fact that
estimates depend on the loci used in each individual
 D.A. Joyce, A.S. Pullin / Biological Conservation 114 (2003) 453–461
study makes them somewhat unreliable since there are
variations in mean heterozygosity of specific loci among
species and taxa (Ward et al., 1992). One way round this
problem is to use the method of correction described in
Ward et al. (1992) which uses a weighting factor calcu-
lated from the mean heterozygosity per enzyme for
insects, and the mean heterozygosity per locus. As one
example, data from the pest species Heliothis veriscens
(Mallet et al., 1993) was subjected to this correction
factor, and the mean adjusted heterozygosity was 0.117.
The mean adjusted heterozygosity for E. aurinia is
0.222. This still indicates that the species compares
favourably with at least one pest species which pre-
sumably maintains relatively large population sizes. We
do not have any data on what genetic diversity levels
have been historically for E. aurinia; these data would
be invaluable. This survey can perhaps be thought of as
a snapshot of current diversity and may provide a mar-
ker with which to assess future changes in population
genetic diversity.
In conservation terms, although changes in allelic
composition are informative about changes in popula-
tion dynamics, irreversible loss of alleles is more
important. Unfortunately, investigating allelic loss risks
confusing sampling variation due to genetic drift with
sampling variation due to sample size (Brookes et al.,
1997). In this case, we have a population known to
have been founded from three individuals (Lincoln-
shire), where we would expect genetic diversity to be
low, and are able to use such a population as a com-
parison. Allelic diversity is reduced faster than hetero-
zygosity during a bottleneck, and a reduction in
numbers of alleles is therefore to be expected. The Lin-
colnshire population contained the smallest number of
alleles and was lowest in terms of both diversity mea-
sures used. This indicates that either (1) natural popu-
lations do not undergo such severe bottlenecks, or (2)
the bottlenecks have been over-estimated (although
larval web counts are thought to provide reasonably
accurate estimates of population size), or (3) there is
enough migration to immediately counteract any loss of
diversity associated with bottlenecks and founder
events. The isolation of the Lincolnshire population
means that a lack of diversity could not be offset by
immigration from other populations. However, perhaps
diversity is maintained by migration in the other, nat-
ural populations investigated. Further evidence for (3)
could be inferred from the population sampled from
Murlough NNR in Northern Ireland. This remains one
of the more diverse populations, despite the fact that
the population was reduced to just one larval web in
1993 (Whatmough, personal communication) prior to
sampling. Although there are differences in allele fre-
quencies in this population compared to other popula-
tions, diversity is higher than in the Lincolnshire
population.
459
Distribution of genetic diversity in the UK does not
seem to show a clear pattern, as no areas are ‘hotspots’
of diversity, and there is no obvious geographical cline
north–south or east–west. The Northern Ireland and
Islay populations, as well as Rooksmoor and Cerne
abbas in Dorset, seem to be the most diverse for both
chosen measures. Both Dartmoor populations stand out
as being relatively poor in genetic diversity compared to
other populations, and other southern populations in
particular.
No populations were found to contain an excess of
heterozygotes, which indicates no recent genetic bottle-
neck in any population. It is possible that the genetic
effective population size (the number of individuals in a
population that contribute genes to the next generation)
was not reduced in these populations, despite data to
the contrary on population size from transect walks and
larval web counts. This may also imply that genetic
effective population sizes are larger than suspected due
to immigration from (and emmigration to) nearby
colonies. The nonparametric test used is suitable for use
with a small number of loci, although the statistical
power of the bottleneck tests can be increased by
screening more loci (Cornuet and Luikart, 1996). For a
number of populations the percentage of polymorphic
loci is low, especially where a loss of genetic diversity
has occurred (e.g. Lincolnshire). This probably explains
why the test failed to identify the population as having
been through a bottleneck.
The theory that movement between populations leads
to a large genetic effective population size and therefore
high genetic diversity is corroborated by studies on
similar species such as Euphydryas phaeton in North
America (Brussard and Vawter, 1975). In a study to
investigate the long-distance dispersal in Euphydryas
editha bayensis, Harrison (1989) concluded that this
species made episodic but successful colonisations of
new sites that were important to the butterfly’s regional
distribution and persistence in the long term. Debinski
(1994) found higher heterozygosity levels than one
would expect in Euphydryas gillettii, which she sus-
pected to be indicative either of overdominance or dis-
persal among populations.
5. Conservation recommendations
Populations in the UK fall into two main geo-
graphically separate populations, one containing Scot-
tish populations, and one containing southern English
and Welsh populations. Mitochondrial DNA studies
(Joyce and Pullin, 2001) show that these two regions
may be the product of two separate colonisation events
after the last ice-age, and allozyme data support their
treatment as separate Management Units. This means
that if either Unit declines, conserving the only remain-
460 D.A. Joyce, A.S. Pullin / Biological Conservation 114 (2003) 453–461
ing Unit will not leave us with a representative sample
of today’s genetic diversity in E. aurinia.
Gene flow between local populations seems to be
responsible for maintaining genetic diversity, which
would otherwise be eroded by repeated bottlenecks. We
therefore suggest that populations must remain con-
nected, i.e. at distances apart approximating the coloni-
sation range of the butterfly. Relevant landscape
restoration projects are already underway in some areas
such as the Blackmore Vale area of Dorset, one of four
trial areas in English Nature’s Habitat Restoration
Project (English Nature, 1999). The programme should
benefit E. aurinia in the Dorset region, by creating more
flower-rich grassland as suitable breeding habitat.
Isolated populations, such as those in Cumbria, may
be of particular conservation concern as these are unli-
kely to recover from any stochastic extinction events
that might occur. Work carried out on the extinction
and minimum viable metapopulation size of the marsh
fritillary by Bulman (2001) shows that isolation of
populations is a major factor in their extinction.
Since the population in Lincolnshire is isolated and
therefore has no way of increasing its genetic diversity
through immigration, its progress will be extremely
interesting to monitor from a scientific point of view.
The population may also be free from the larval para-
sitoid that is thought to cause population cycles else-
where. This should be examined, and the population
monitored to see whether cycles in population size occur
irrespective of this.
The management of species increasingly embraces
landscape scale processes. This study suggests that
measuring the distribution of genetic diversity at a vari-
ety of scales provides useful information on the appro-
priate scale at which to focus conservation resources.
Acknowledgements
The University of Birmingham wishes to acknowledge
the financial support of English Nature. This work was
funded by grant number 03/GNT/125. Thanks also to
the following organisations and individuals from them
who greatly aided larval collection: Butterfly Conserva-
tion, the National Trust, English Nature, Scottish
National Heritage, the Countryside Council for Wales,
Dartmoor National Park Authority, Dorset Wildlife
Trust and the Defence Estate Organisation, MoD.
References
Asher, J., Warren, M., Fox, R., Harding, P., Jeffcoate, G., Jeffcoate,
S., 2001. The millenium atlas of butterflies in Britain and Ireland.
Oxford University Press, Oxford.
Brookes, M.I., Graneau, Y.A., King, P., Rose, O.C., Thomas, C.D.,
Mallet, J.L.B., 1997. Genetic analysis of founder bottlenecks in the
rare British butterfly Plebejus argus. Conservation Biology 11, 648–
661.
Brussard, P.F., Vawter, A.T., 1975. Population structure, gene flow
and atural selection in populations of Euphydryas phaeton. Heredity
34, 407–415.
Bulman, C.R., 2001. Conservation Biology of the Marsh Fritillary
Butterfly Euphydryas aurinia. PhD Thesis, University of Leeds.
Butterfly Conservation, 1995. Species Action Plan: Marsh Fritillary
Eurodryas aurinia. Wareham, Dorset.
Cornuet, J.M., Luikart, G., 1996. Description and power analysis of
two tests for detecting recent population bottlenecks from allele
frequency data. Genetics 144, 2001–2014.
Debinski, D.M., 1994. Genetic diversity assessment in a metapopula-
tion of the butterfly Euphydryas gillettii. Biological Conservation 70,
25–31.
English Nature, 1999. English Nature’s Habitat Restoration Project.
English Nature, Peterborough.
Frankham, R., 1995. Conservation genetics. Annual Review of
Genetics 29, 305–327.
Harrison, S., Hastings, A., 1996. Genetic and evolutionary con-
sequences of metapopulation structure. Trends in Ecology and
Evolution 11, 180–183.
Harrison, S., 1989. Long distance dispersal and colonization in the bay
checkerspot butterfly Euphydryas editha bayensis. Ecology 70, 1236–
1243.
Joyce, D.A., Pullin, A.S., 2001. Phylogeography of the marsh fritillary
Euphydryas aurinia (Lepidoptera: Nymphalidae) in the UK. Biolo-
gical Journal of the Linnean Society 72, 129–141.
Lewis, O.T., Hurford, C., 1997. Assessing the status of the marsh fri-
tillary butterfly Eurodryas aurinia: an example from Glamorgan,
UK. Journal of Insect Conservation 1, 159–166.
Mallet, J., Korman, A., Heckel, D.G., King, P., 1993. Biochemical
genetics of Heliothis and Helicoverpa (Lepidoptera: Noctuidae) and
evidence for a founder event in Helicoverpa zea. Annals of the
Entomological Society of America 86, 189–197.
Miller, M.P., 1997. Tools for population genetic analyses (TFPGA)
1.3: A Windows program for the analysis of allozyme and molecular
population genetic data. Computer software distributed by the
author.
Nei, M., 1978. Estimation of average heterozygosity and genetic dis-
tance from a small number of individuals. Genetics 89, 583–590.
Packer, L., Taylor, J.S., Savignano, D.A., Bleser, A., Lane, C.P.,
Sommers, L.A., 1998. Population biology of an endangered butter-
fly Lycaeides melissa samuelis (Lepidoptera: Lycaenidae): genetic
variation, gene flow, and taxanomic status. Canadian Journal of
Zoology 76, 320–329.
Porter, K., 1981. The Population Dynamics of Small Colonies of the
Butterfly Euphydryas aurinia. PhD Thesis, Oxford University.
Reynolds, J., Weir, B.S., Cockerham, C.C., 1983. Estimation for the
coancestry coefficient: basis for short term genetic distance. Genetics
105, 767–779.
Rice, W.R., 1989. Analyzing tables of statistics. Evolution 43, 223–
225.
Richardson, B.J., Baverstock, P.R., Adams, M., 1986. Allozyme Elec-
trophoresis. A Handbook for Animal Systematics and Population
Structure. Academic Press, Sydney.
Rohlf, F.J., 1997. NTSYS-pc v1.7 Numerical Taxonomy and Multi-
variate Analysis System. Exeter Software, New York.
Slatkin, M., 1985. Gene flow in natural populations. Annual Review
of Ecology and Systematics 16, 393–430.
Thomas, C.D., 1994. Local extinctions, colonisations and distribu-
tions: habitat tracking by British butterflies. In: Leather, S.R., Watt,
A.D., Mills, N.J., Walters, K.F.A. (Eds.), Individuals, Populations
and Patterns in Ecology. Intercept, Andover, UK, pp. 317–337.
Wade, M.J., McCauley, D.E., 1988. Extinction and colonisation: their
effects on the genetic differentiation of local populations. Evolution
42, 995–1005.
 D.A. Joyce, A.S. Pullin / Biological Conservation 114 (2003) 453–461 461

Ward, R.D., Skibinski, D.O.F., Woodwark, M., 1992. Protein het-
erozygosity, protein structure and taxonomic differentiation. Evo-
lutionary Biology 26, 73–159.
Warren, M.S., 1994. The UK status and suspected metapopula-
tion structure of a threatened European butterfly, the Marsh
Fritillary Eurodryas aurinia. Biological Conservation 67, 239–
249.
Weir, B.S., Cockerham, C.C., 1984. Estimating F statistics for the
analysis of population structure. Evolution 38, 1358–1370.
Workman, P.L., Niswander, J.D., 1970. Population studies on south-
western Indian Tribes II Local genetic differentiation in the Papago.
American Society of Human Genetics 22, 24–49.
Wright, S., 1931. Evolution in Mendelian populations. Genetics 16,
97–159.