Science of the Total Environment 360 (2006) 196 – 204

www.elsevier.com/locate/scitotenv

Biodiversity in urban habitat patches

P.G. Angold a,*, J.P. Sadler a, M.O. Hill b, A. Pullin a, S. Rushton c, K. Austin a,
E. Small a, B. Wood a, R. Wadsworth b, R. Sanderson c, K. Thompson d
a
University of Birmingham, UK
b
CEH Monks Wood, UK
c
University of Newcastle, UK
d
University of Sheffield, UK
Available online 16 November 2005

Abstract

We examined the biodiversity of urban habitats in Birmingham (England) using a combination of field surveys of plants and
carabid beetles, genetic studies of four species of butterflies, modelling the anthropochorous nature of the floral communities and
spatially explicit modelling of selected mammal species. The aim of the project was to: (i) understand the ecological characteristics
of the biota of cities model, (ii) examine the effects of habitat fragment size and connectivity upon the ecological diversity and
individual species distributions, (iii) predict biodiversity in cities, and (iv) analyse the extent to which the flora and fauna utilise the
durban greenwaysT both as wildlife corridors and as habitats in their own right. The results suggest that cities provide habitats for rich
and diverse range of plants and animals, which occur sometimes in unlikely recombinant communities. The studies on carabids and
butterflies illustrated the relative importance of habitat quality on individual sites as opposed to site location within the conurbation.
This suggests that dispersal for most of our urban species is not a limiting factor in population persistence, although elements of the
woodland carabid fauna did appear to have some geographical structuring. Theoretical models suggested that dormice and water
voles may depend on linear habitats for dispersal. The models also indicated that other groups, such as small and medium sized
mammals, may use corridors, although field-based research did not provide any evidence to suggest that plants or invertebrates use
urban greenways for dispersal. This finding indicates the importance of identifying a target species or group of species for urban
greenways intended as dispersal routeways rather than as habitat in their own right. Their importance for most groups is rather that
greenways provide a chain of different habitats permeating the urban environment. We suggest that planners can have a positive
impact on urban biodiversity by slowing the pace of redevelopment and by not hurrying to tidy up and redevelop brownfield sites.
D 2005 Elsevier B.V. All rights reserved.

Keywords: Biodiversity; Habitat corridors; Plants; Beetles; Mammals; Butterflies; Modelling; URGENT; Birmingham

1. Introduction valuable for human well-being as well as wildlife

Urban areas are highly modified and complex land-
scapes, within which green or open areas are seen as
* Corresponding author. School of Geography, Earth and Environ-
mental Sciences, University of Birmingham, Edgbaston, Birmingham,
B15 2TT, UK.
E-mail address: p.g.angold@bham.ac.uk (P.G. Angold).
0048-9697/$ - see front matter D 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.scitotenv.2005.08.035
(Pickett et al., 2001, 2004). The biological processes
of dispersal interact with the landscape structure in
determining the distribution of populations of species
present (Niemelä, 1999). Recent research has put urban
biodiversity into the conservation spotlight. Several
studies have focused attention on the conservation sig-
nificance of elements of the urban landscape, such as
brownfield sites (Gibson, 1998; Woodward et al., 2003)
 P.G. Angold et al. / Science of the Total Environment 360 (2006) 196–204
and gardens (Gaston et al., 2005; Thompson et al.,
2003). An essential first step to managing urban envir-
onments more effectively is a fuller understanding of
the interplay between landscape (matrix effects) and
local factors (patch effects) that affect urban biodiver-
sity. Many cities have a network of habitat fragments or
durban greenwaysT comprising areas of semi-natural
habitats, secondary succession, ruderal and pioneer
environments and open areas. These habitats may be
important features for biodiversity both as stable and as
transient habitats (McIntyre, 2000; McIntyre et al.,
2001), and may also be valuable for their possible
function as dcorridorsT and dstepping stonesT to facilitate
species dispersal (Spellerberg and Gaywood, 1993;
Kirby, 1995) and they are therefore a key part of current
ecological planning (Habitat and Species Directive,
Brussels 1993; PPG Note 9, Department of the Envi-
ronment, October 1994). In urban landscape planning,
urban greenways and wildlife corridors are increasingly
advocated to encourage animals and plants to move
around urban areas and thus to preserve or enhance
urban biodiversity.
This research aimed to: (i) understand the ecological
characteristics of the biota of cities model, (ii) examine
the effects of habitat fragment size and connectivity
upon the ecological diversity and individual species
distributions, (iii) predict biodiversity in cities, and
(iv) analyse the extent to which the flora and fauna
utilise the durban greenwaysT both as wildlife corridors
and as habitats in their own right. The project com-
prised four components involving empirical survey
work of both plants and ground beetles (Carabidae)
on a range of habitats (derelict, wetland and woodland)
(University of Birmingham), a genetic study of four
species of grassland butterflies (University of Birming-
ham), an examination of the plant communities and
hemeroby (Hill et al., 2002) (CEH, Monks Wood)
and spatially explicit modelling of mammal populations
(University of Newcastle). The research was framed
around the following hypotheses:
1. The biota of cities is drawn from local and national
species pools together with a suite of mobile heme-
robes.
2. Species richness and abundance of selected species
in habitat fragments is related to (a) patch area, (b)
habitat connectivity and (c) patch continuity over
time.
3. Genetic variation and gene flow within and between
species populations of fragmented habitats is related
to (a) patch area, (b) habitat connectivity and (c)
patch continuity over time.
197
4. Species distribution and dispersal are interrupted by
discontinuities in urban greenways.
5. The patch-dynamic structure of cities is essential for
the persistence of many species.
The paper presented here distils the results of these
components and the approach is thus cursory and aimed
at highlighting only the most significant results.
2. Methods
2.1. Study area
The West Midlands conurbation comprises the City
of Birmingham and several boroughs collectively
known as the Black Country. Birmingham City has a
population of one million, with an estimated 6 million
others living within a 50 mile radius of the city. The city
covers 27,000 hectares, 11% of the land cover is green
space in the form of parks, and the city includes ap-
proximately 4000 hectares of semi-improved neutral
grassland, pockets of ancient woodland, and 250,000
domestic gardens.
2.2. Plant data and GIS
The flora of 50 derelict sites was recorded during the
summers 1998 and 1999, by means of a total species
record and a number (between 10–30 determined by
site area) of 1 m2 quadrats (Austin, 2002). Patch size
and distance to other derelict sites were measured in the
field and from field maps transferred to GIS. Further
data were abstracted from ECORECORD (http://
www.wildlifetrust.org.uk/urbanwt/ecorecord), the Eco-
logical Database for the Black Country and Birming-
ham, provided information on over 7000 sites and has a
database with more than 280,000 species records. Stan-
dard database and GIS techniques were used to sum-
marize the data and convert them for use in a Decision
Support System (DSS).
The characterization of biota was intended to distin-
guish urban plants from those that are not urban, using
a scale of durbanityT, comparable to Ellenberg’s ecolog-
ical indicator values (Ellenberg, 1991) and the scale of
human influence called dhemerobyT (Kowarik, 1990;
Lindacher, 1995). Quadrat data were taken from Aus-
tin’s survey in Birmingham (Austin, 2002), the UCPE
quadrat database (Hodgson et al., 1999), Countryside
Survey 1990 (Barr et al., 1993), the ITE Woodland
Survey (Bunce, 1982) and the ITE Railway Survey
(Sargent, 1984). The durbanityT of a species was defined
as the mean proportion of urban land cover, according
198 P.G. Angold et al. / Science of the Total Environment 360 (2006) 196–204
to the ITE Landcover Map (Fuller et al., 1994), in the 1-
km square neighbourhood of each occurrence of the
species (Hill et al., 2002).
2.3. Carabid beetle sampling
Carabid assemblages were investigated from three
urban habitats: 28 derelict sites, 29 wetland sites and 12
woodlands (Small, 2002). In each case site selection
was stratified to select representative sites in the rural,
suburban and urban zones. In addition, all wetland sites
were located within a 5 km radius of a single burban
wildlifeQ river corridor, while the derelict sites were a
subset of those used in the flora sampling. At each site
carabids were sampled over the course of one season
(April–October 1999 for derelict and wetland; April–
October 2000 for woodland), using standardised pitfall
trapping and hand-search techniques. The structure and
type of vegetation and the moisture, organic matter, pH,
penetrability and litter depth of the soil were also
measured. Landscape measures utilised methods used
in the analysis of the flora (above).
2.4. Butterfly sampling and genetic analysis
Four grassland butterflies were selected. Pieris napi
(Green-Veined White) is classified as having high
dispersal ability (Dennis and Shreeve, 1997), ‘migrant
or vagrant with evidence of long-distance movementT.
Maniola jurtina (Meadow Brown), Pyronia tithonus
(Gatekeeper) and Coenonympha pamphilus (Small
Heath) have moderate dispersal ability dopen popula-
tion structure with evidence of frequent movements
between habitat unitsT. However, in another classifica-
tion, based on movements in British island popula-
tions P. napi and M. jurtina were classified together as
relatively mobile and P. tithonus and C. pamphilus as
less so (Wood and Pullin, 2002). Butterfly distribution
data were available at the 1-km square scale from the
database of Butterfly Conservation (Asher et al.,
2001).
Butterflies were sampled from urban and several
outlying rural populations. Sites, and therefore popula-
tions, were sampled when at least two of the four study
species were present. If present, up to 30 adult butter-
flies were collected. Sample sizes of less than seven
were excluded from analysis. Cellulose acetate gel
electrophoresis was performed on abdominal tissue to
detect genetic variation by allozyme analysis. From 25
putative enzyme loci initially screened, nine loci were
reliably resolved for M. jurtina, ten for P. tithonus, nine
for C. pamphilus and seven for P. napi.
The departure of genotypic frequencies from those
expected under Hardy–Weinberg equilibria was tested
for statistical significance using the Fisher’s exact test.
Heterogeneity of allele frequencies among populations
was determined by calculating unbiased estimates of
the P-value of the Fisher exact test using the Markov
chain method (GENEPOP). The genetic variability of
populations for each species was expressed in terms of
observed (H o) and expected heterozygosity (H e). The
analysis of population structure, of resolved poly-
morphic loci, was determined through Wright’s F-sta-
tistics. F IS, the inbreeding coefficient, was used to
reveal the divergence of observed heterozygosity from
that expected assuming random mating, within subpo-
pulations. F IT, the overall inbreeding coefficient, pro-
vides a measure of the reduction of heterozygosity
within individuals relative to the total population, due
to non-random mating within the total population. F ST
reveals the level of geographic substructuring between
populations. Mean F IS, F IT and F ST values over all loci
are given for each species (POPGENE32). Consensus
dendrograms based on Reynold’s genetic distances
were calculated after bootstrapping with 1000 replicates
to obtain internal confidence levels for each node
(PHYLIP, Felsenstein, 1993).
2.5. Modelling mammal populations
Models were developed to analyse the impacts of
landscape structure and connectivity on the viability of
field vole Microtus agrestis, dormouse Muscardinus
avellanarius, muntjac Muntiacus reevesii and water
vole Arvicola terrestris. These species represented re-
spectively: a major food resource for predators; a con-
servation icon with restricted habitat requirements and
limited dispersal; an alien species and a declining spe-
cies associated with linear habitats. The importance of
dispersal as a factor determining population persistence
was assessed for all species. The potential effects of
varying landscape structure were determined for field
vole and muntjac.
Spatially explicit models can simulate the life-histo-
ry processes that determine where organisms are found.
These models can be used to investigate the dynamics
of populations, their response changes in landscape
structure, and their viability in the fragmented land-
scapes that typify urban environments. The models
have large data requirements both in terms of habitat
and fundamental life-history parameters, but have been
successfully used for both invertebrate and vertebrate
studies: e.g., ground-beetles (Rushton et al., 1996) and
squirrels (Rushton et al., 1997) and water voles and
 P.G. Angold et al. / Science of the Total Environment 360 (2006) 196–204
mink (Rushton et al., 2000). These techniques have
been used in developing Decision Support System
(DSS) for land use planning (Rushton et al., 1995;
Wadsworth and O’Callaghan, 1995; Watson and Wads-
worth, 1996). A generic spatially-explicit population
model, applicable to any species, was developed for
inclusion in the project’s DSS. The generic model
requires three inputs: i) a landscape habitat map pro-
vided by the GIS; ii) a classification of habitat suitabil-
ity; and iii) life history information for the selected
species. The model then undertakes a population via-
bility analysis (PVA) for the species in the given land-
scape, providing outputs on the predicted probability of
persistence at all points in the landscape and population
size.
3. Results
3.1. Patch and corridor relationships of plants: derelict
site data
Each of 50 sites was sampled by 20 quadrats of size
1 m2, so the number of species per site was a sample
but not a census of site diversity. There was no relation
between site species richness and proximity to corri-
dors; there was, however, a positive relation between
site species richness and site area, even though the total
area of land sampled was 20 m2 at each site. Of 379
species occurring in at least 5 sites, 11 were signifi-
cantly ( p b 0.05) more frequent near railway land but
only broom Cytisus scoparius and toadflax Linaria
vulgaris were strongly ( p b 0.005) associated with it.
Likewise, 6 species were significantly ( p b 0.05) more
frequent near rivers, but only redshank Persicaria
maculosa was markedly ( p b 0.005) associated with
them.
Far more influential than distance to the nearest
corridor was the effect of distance to the nearest derelict
site. 25 species were significantly ( p b 0.05) more fre-
quent in the proximity of other derelict sites, and 10 of
these were strongly so ( p b 0.005). The species that
showed this tendency were with few exceptions those
such as wormwood Artemisia absinthium and fennel
Foeniculum vulgare that had been found to be mark-
edly urban according to the land-cover criterion.
Table 1
Numbers of plant species recorded per patch in ECORECORD; results for most habitats have been not been listed individually
Number of patches Broad-leaf wood
Off corridor 1817 31.8
On corridor 2702 19.7
Total 4519 26.0
199
3.2. Indices for urban flora
Species presence data were available from 26,710
quadrats, of which 2595 were located in 1-km squares
with more than 40% cover of urban land. The concept of
hemeroby, (or degree of human influence), is too diffuse
to be the basis of an index; there are too many kinds of
disturbance (Grubb, 1988) and the most intensely dis-
turbed land is arable land, which is rural. High propor-
tions of annual plants characterize arable land and many
of these agrestal species are absent or relatively uncom-
mon in cities. This result shows up clearly when species
are plotted in a two-dimensional space comparing
durbanityT (the mean proportion of nearby urban land
cover) with dannualityT (the mean proportion of annuals
found in the same quadrat) (Fig. 1).
Of the 902 species found in at least 10 quadrats,
the large majority were in perennial communities not
close to urban cover. These appear at the bottom left
of the diagram. Plants that were mainly restricted to
urban sites are at top, and arable species are at the
right.
3.3. Patch and corridor relationships of plants: ECOR-
ECORD data
From the ECORECORD data, it appears that patches
on corridors are for the most part less rich in plant
species than those which are off them (Table 1). Out
of 18 habitat types recognized in the ECORECORD
survey, only open water (for which there were effec-
tively no patches off corridors) and swamps were richer
on corridors.
Out of 433 species having at least 10 occurrences, 7
were positively associated with corridors ( p b 0.005) if
habitat is taken into account, and of these only Indian
balsam Impatiens glandulifera, reed canary-grass Pha-
laris arundinacea and butterbur Petasites hybridus
were at least 25% more frequent than would be
expected by chance. These are all associated with
watercourses. Of the remaining four species, reed
sweet-grass Glyceria maxima and sallow Salix cinerea
are also characteristic of moist habitats, but lupin Lupi-
nus polyphyllus and weld Reseda luteola are plants of
disturbed habitats.
Neutral grass Open water Swamp All habitats
15.5 – 6.0 17.1
12.7 14.8 12.5 12.3
13.8 14.8 10.8 14.2
200 P.G. Angold et al. / Science of the Total Environment 360 (2006) 196–204
80
70
Mean urban land cover
60
50
40
30
20
10
0 Sedum acre Potentilla argentea
Fig. 1. Mean urban land cover in 1-km squares where species occurred in relation to the proportion of associated annuals in quadrats. Longevity
classes are distinguished by symbols: circle = annual, triangle = biennial, dot = perennial.
In the other direction, there were 113 species neg-
atively associated with corridors ( p b 0.005). Among
species occurring in less than half the expected number
of corridor sites were woody plants such as holly Ilex
aquifolium, rhododendron Rhododendron ponticum,
and yew Taxus baccata and woodland herbs such as
wood sorrel Oxalis acetosella and wood dock Rumex
sanguineus. A few grassland and wayside plants had a
significantly negative association with corridors, but
the shortfall was much less, 15% fewer than expected
for ryegrass Lolium perenne, 11% for cock’s-foot Dac-
tylis glomerata and 36% for daisy Bellis perennis.
Likewise a few garden plants had negative associations
with corridors, notably ground-elder Aegopodium
podagraria with 39% fewer occurrences than
expected, 62% fewer for hairy bittercress Cardamine
hirsuta and 31% fewer for Chinese privet Ligustrum
ovalifolium.
3.4. Factors influencing the carabid beetle fauna of
urban habitat patches
Habitat quality was the most important factor influ-
encing beetle diversity in the urban habitat patches
surveyed (Small et al., 2006-this volume). Diversity
of carabids on derelict land was significantly related
to the vegetation structure and was highest amongst
early successional tall herb plants (Small et al., 2003).
In wetland patches the beetle assemblages were also
closely determined by the nature of the vegetation with
a clear distinction between the communities trapped on
wetter sites, where the vegetation was dominated by
Buddleja davidii Lactuca serriola
Melilotus alba
Aster novi-belgii Conyza canadensis
Senecio squalidus Apera interrupta
Solidago canadensis
Solanum tuberosum
Brassica oleracea
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Proportion of annual associates
reeds (predominantly Glyceria spp., P. arundinacea,
Phragmites australis), and the drier sites where the
vegetation was dominated by rushes (Juncaceae)
(Small, 2002). The level of disturbance from recrea-
tional users heavily influenced woodland carabid com-
munities. Woods in the urban zone are typically
subjected to much higher usage by people and conse-
quently a clear divide in beetle assemblage species
richness (df = 2,21; F = 7.018; p = 0.005) and diversity
(df = 2,21; F = 14.329; p = b0.001) was shown across an
rural–urban gradient. This was most clearly illustrated
by increasing dominance of a few ubiquitous urban
generalists (e.g., Pterostichus madidus) (df = 2,21;
F = 12.345; p = 0.001) and the decline of the woodland
(e.g., Cychrus carabioides) (df = 2,21; F = 8.842;
p = 0.002), woodland associate species (e.g., Abax par-
allepidus (df = 2,21; F = 8.842; p = 0.002), and other
large-bodied species (e.g., Cararbus spp.) (df = 2,21;
F = 19.858; p = b0.001) (Sadler et al., in press).
The issue of spatial location of a patch in the land-
scape was typically of secondary and minor signifi-
cance. Habitat specialists of wetland and derelict sites
showed little sensitivity to isolation measured as dis-
tance from river and railway corridors (Fig. 2). The
number of wetland specialists at sites located on corri-
dors was no greater than the number found at off
corridor sites (Fig. 3, t = 0.52, df = 20, n.s.), although
corridor sites were shown to have significantly higher
numbers of generalist species (t = 4.64, df = 29,
p b 0.001). However, landscape effects such as site
size, the amount of built up and wooded land within
5 km of the sample sites were significantly ( p b 0.05)
 P.G. Angold et al. / Science of the Total Environment 360 (2006) 196–204 201

(a) (b)
Carabid species richness Wetland carabids
30
25
R2 = 0.0077
No. of species

No. of species
20
15
20
10

R2 = 0.005 5

10 0
0 0.2 0.4 0.6 0 2 4 6
Log (Distance to railway) Distance to R. Cole (km)
Fig. 2. Relationships between carabid species richness and distance to (a) the nearest railway habitat corridor, (b) the wetland corridor (River Cole).

related to the woodland beetle assemblages indicating (v 2 = 23.0, df = 9, p b 0.01, Table 2) were significant.
some structuring of assemblages as a result of wood- Wright categorised mean F ST values of 0–0.05 as indi-
land fragmentation and isolation. Woodland specialists cating no structuring (P. napi and M. jurtina) and
and species were all found to be less likely to occur in values between 0.05 and 0.15 as moderate structuring
the more isolated urban woodland patches (Sadler et al., (P. tithonus and C. pamphilus).
in press). Testing the isolation-by-distance model of geograph-
ic structuring revealed no significant relationship be-
3.5. Distribution and genetic structure of butterfly tween geographic and genetic distance between
populations populations at this spatial scale for any of the four
species. The Mantel test revealed no matrix association
M. jurtina was recorded in 20.4% of 1 km squares, for P. napi (matrix correlation: r = 0.34, approximate
P. tithonus in 18.6%, P. napi in 20.1% and C. pamphi- Mantel t-test: t = 1.54, p = 0.06), M. jurtina (r = 0.13,
lus in 6.5%. These apparently low values for fairly t = 0.75, p = 0.77), P. tithonus (r = 0.18, t = 1.08,
common species are put into context when it is realised p = 0.14) or C. pamphilus (r = 0.15, t = 0.84,
that the percentage of squares where at least one record p = 0.2). No significant relationships were found when
has been submitted is only 36.8%. geographic and genetic distances were logged. Dendro-
Little evidence of geographic structuring was found grams, based on genetic distances, revealed little con-
for P. napi and M. jurtina populations, but moderate gruence with geographic proximity of populations nor
structuring was revealed for both C. pamphilus and P. on whether populations were urban or rural.
tithonus. The overall mean F ST values across all re- The relatively low mean F ST values, indicating no
solved polymorphic loci for the four species ranged significant structuring, and the lack of evidence linking
from 0.042 to 0.075. Mean F ST values were not signif- genetic similarity and geographic proximity of popula-
icant for P. napi (mean F ST = 0.042, v 2 = 8.9, df = 7, tions, suggest that dispersal ability is not the key factor
p N 0.05) and M. jurtina (mean F ST = 0.048, v 2 = 13.2, determining distributions of P. napi and M. jurtina at
df = 9, p N 0.05), but F ST values of 0.068 for P. tithonus this spatial scale. We conclude that they are restricted
(v 2 = 21.9, df = 10, p b 0.05) and 0.075 for C. pamphilus by habitat availability. Despite the local distribution of
C. pamphilus, the genetic data do not suggest that
20 populations are isolated nor that the species is not
Species richness

15
capable of dispersing to other habitat patches within
the conurbation.
10

5 3.6. Corridors and butterfly dispersal

0 C. pamphilus butterflies were sampled from eight

Wet Wet Other Other
specialists: specialists: species: species: urban and two outlying rural populations. Up to 30
ON OFF ON OFF adult butterflies were collected. F ST values (Wright,
Fig. 3. Comparison of the number of specialist and generalist carabid
1978) revealed the level of geographic structuring be-
beetle species between wetland survey sites located on or off the river tween populations. A test of differences in allele fre-
Cole corridor. quencies revealed a highly significant result for C.
202 P.G. Angold et al. / Science of the Total Environment 360 (2006) 196–204

Table 2
Summary of F-statistics at the resolvable polymorphic loci for the four butterfly species; columns labeled No. gives the sample size
Butterfly species
Pieris napi Pyronia tithonus Maniola jurtina Coenonympha pamphilus
Locus No. F IS F IT F ST No. F IS F IT F ST No. F IS F IT F ST No. F IS F IT F ST
IDH1 – – – – – – – – 272 0.12 0.171 0.05 – – – –
IDH2 – – – – 330 0.06 0.01 0.05 278 0.06 0.03 0.03 258 0.38 0.41 0.05
G6PD – – – – 304 0.12 0.17 0.06 – – – – 216 0.56 0.63 0.16
GOT1 208 0.05 0.01 0.06 – – – – – – – – – – – –
GOT2 208 0.07 0.04 0.03 – – – – 302 0.10 0.07 0.03 304 0.07 0.15 0.09
AK2 – – – – – – – – – – – – 298 0.04 0.02 0.05
PEP-LA 204 0.39 0.43 0.06 290 0.03 0.13 0.10 296 0.06 0.11 0.05 214 0.24 0.30 0.09
MPI – – – – 232 0.22 0.30 0.09 254 0.03 0.08 0.05 – – – –
PGI 208 0.05 0.01 0.04 328 0.01 0.03 0.04 280 0.09 0.12 0.03 304 0.11 0.17 0.07
PGM 198 0.08 0.10 0.02 318 0.01 0.06 0.05 300 0.03 0.03 0.06 304 0.03 0.04 0.07
Mean 206 0.08 0.12 0.04 310 0.08 0.14 0.07 286 0.03 0.07 0.05 275 0.06 0.13 0.08

pamphilus, indicating heterogeneity between popula-
tions. Consensus dendrograms, based upon Reynold’s
genetic distances, revealed little congruence of popula-
tions according to geographic proximity, with one of
the three River Cole corridor populations acting as an
outgroup to all other populations. Confidence limits on
the nodes of this tree were low, however. These data
suggest that putative wildlife corridors may not be
acting as conduits to gene flow any more effectively
than the rest of the landscape.
3.7. Mammal populations
fecundity within their habitat rather than dispersal be-
tween habitats.
In conclusion, dispersal distance was a key life
history parameter for water vole and dormouse, but
not for muntjac and field vole. Habitat availability is
low for both water vole and dormouse but high for the
other two species.
4. Discussion
4.1. The character and diversity of urban habitat
patches

The model for field voles was only sensitive to one
parameter, the carrying capacity of the habitat, and this
has a major impact on predicted population size. Of the
four species in this study, field voles had the most
catholic tastes with regard to breeding and dispersal
habitat. Essentially, in the West Midlands landscape,
field voles live in a single, well-connected habitat that
is (and likely to remain) largely unfragmented.
The muntjac deer model, dealing with a large mam-
mal with high dispersal potential, was most sensitive to
juvenile mortality, with fecundity and adult mortality
also being significant. Although their habitat is largely
fragmented, they were able to disperse between suitable
habitats with relative ease.
Water voles rely on bank side vegetation. They
require a contiguous habitat, thus dispersal distance
was a significant life history parameter in the model.
However, carrying capacity of the environment and
brood size were the most important factors.
The dormouse model was sensitive to all life history
parameters. Not only are dormice small mammals with
little dispersal ability, but they are highly restricted to
breeding habitat and are very much dependent on high
There is a considerable diversity of flora on urban
wasteland sites, with a total of 378 species found on the
50 sites surveyed. The highest diversity both within and
between derelict sites occurs while they are still young,
with convergent succession evident over a 20-year
period leading to domination by Arrhenatheretum
grassland and subsequently a Rubus fruticosus–Holcus
lanatus understorey. The more diverse pioneer and tall
herb phases of the succession will however persist
longer on infertile substrate or under continual and
sporadic physical disturbance of the site. The amount
of urban cover that surrounded a site was not related to
the plant community present on the site, and there was
no evidence of an urban–rural gradient in the commu-
nities of derelict sites. However, more urban sites did
include a greater proportion of neophyte aliens than
rural sites. There was no difference in the relative
proportions of archeophyte alien species between
urban and rural sites.
The derived scales of annuality and urbanity provide
a means of environmental assessment additional to
Ellenberg values for characterising the urban flora
(Hill et al., 1999). It is interesting to note, however,
 P.G. Angold et al. / Science of the Total Environment 360 (2006) 196–204
that whereas annuals are strongly associated with other
annuals, urban plants have many other associates. In-
deed, the ten commonest plants found in the plant
quadrat survey are all British natives that occur in
lawns and by roads, and are frequent in hedges and
pastures in eutrophic lowland countryside. The defining
urban plants are those associated with waste places. Out
of the 20 with highest urbanity, none is a British native;
8 are archaeophyte aliens that reached Britain before
1500; 12 are neophyte aliens that were introduced more
recently.
Butterflies, however, did not show an effect of prox-
imity to other sites. Our results are consistent with the
hypothesis that the four species are either able to move
freely around the city or have established colonies from
relatively distant source populations in separate coloni-
zation events.
Similarly, carabids illustrated the importance of hab-
itat (site-based) variables over landscape variables in
defining assemblage type and species distributions
(Small et al., 2006-this volume). Only in the case of
woodlands were variables such as isolation, position on
the urban–rural gradient and site size shown to be
important for a limited number of large bodied, flight-
less woodland associated species (Sadler et al., in
press).
It maybe therefore that for many species of inverte-
brates in the urban environment, the maintenance and
even restoration/creation of good quality habitat is the
key to their continued survival rather than the more
difficult task of increasing habitat connectivity.
4.2. Role of corridors in the urban environment
Green corridors, which have become a recognized
feature of urban planning and conservation for 10
years, may make little difference to the diversity of
plants and beetles found in our towns and cities by
virtue of their function as corridors. They do often
provide valuable habitat, especially on river corridors
and railway land. However, the plant communities of
derelict sites show no greater similarity to each other
if adjacent or close to designated urban corridors,
suggesting that the corridors do not increase connec-
tivity. Nevertheless, sites within areas with a high local
density of derelict sites were more similar, suggesting
greater inter-site dispersal of propagules when sites are
less isolated. We found no evidence that wetland spe-
cialist beetle diversity is greater on or near the green
corridors, and no evidence that corridors are necessary
for dispersal of butterflies. For the invertebrates in
particular, habitat quality appears to be the significant
203
factor, and the habitat of designated dgreen corridorsT
in the Black Country is rather a chain of habitats of
different type and/or quality rather than a linear con-
tinuous habitat.
Theoretical models suggest that dormice and water
voles may depend on linear habitats for dispersal (but
dormice are nowadays more likely to find new habitat
by deliberate human introduction than by dispersal).
However, the models do indicate that other groups,
such as small and medium sized mammals, may use
corridors although plants and invertebrates may not.
This finding indicates the importance of identifying a
target species or group of species for urban greenways
intended as dispersal routeways rather than as habitat in
their own right.
Our research has not found any evidence that plants
or invertebrates use urban greenways for dispersal.
Their importance for these groups is rather as a chain
of different habitats permeating the urban environment.
We suggest that planners can have a positive impact on
urban biodiversity by slowing the pace of redevelop-
ment and by not hurrying to tidy up and redevelop
brownfield sites. We found that derelict sites had a
more distinctive flora when they were close to other
such sites, suggesting that dispersal between sites may
be important in the development of the characteristic
urban derelict flora.
Acknowledgements
This research was the responsibility of a consortium
including researchers from the University of Birming-
ham, CEH Monks Wood, the University of Newcastle
and the University of Sheffield. It was funded under the
NERC URGENT programme (GST/03/1979). The
duser communityT including the local authorities gave
valuable support and a special note of thanks is offered
to EcoRecord, who donated the use of their database to
the project.
References
Asher J, et al. The millenium atlas of butterflies in Britain and Ireland.
Oxford University Press.
Austin KC. Botanical processes in urban derelict spaces. Unpublished
PhD thesis, School of Geography and Environmental Sciences,
The University of Birmingham, 2002.
Barr CJ, Bunce RGH, Clarke RT, Fuller RM, Furse MT, Gillespie
MK, et al. Countryside survey 1990: main report. London7
HMSO; 1993.
Bunce RGH. A field key for classifying British woodland vegetation
Part 1. London7 Institute of Terrestrial Ecology.
Dennis RLH, Shreeve TG. Diversity of butterflies on British
islands: ecological influences underlying the roles of area,
204 P.G. Angold et al. / Science of the Total Environment 360 (2006) 196–204
isolation and the size of the faunal source. Biol J Linn Soc
1997;60:257 – 75.
Ellenberg H. Zeigerwerte von Gefässpflanzen Mitteleuropas. Scr
Geobot 1991;18:1 – 248.
Felsenstein J. PHYLIP (phylogeny inference package) V 35c. Seattle,
Washington7 Department of Genetics, University of Washington;
1993.
Fuller RM, Groom GB, Jones AR, 1994. Photogramm Eng Remote
Sensing 1994;60:553 – 62.
Gaston KJ, Smith RM, Thompson K, Warren PH. Urban domestic
gardens (II): experimental tests of methods for increasing biodi-
versity. Biodivers Conserv 2005;14:395 – 413.
Gibson CWD. Brownfield: red data The values artificial habitats have
for uncommon invertebrates. Peterborough7 English Nature; 1998.
Grubb PJ. The uncoupling of disturbance and recruitment, 2 kinds of
seed bank, and persistence of plant-populations at the regional and
local scales. Ann Zool Fenn 1988;25:23 – 36.
Hill MO, Mountford JO, Roy DB, Bunce RGH. Ellenberg’s indicator
values for British plants. ECOFACT Volume 2 technical annex.
Institute of Terrestrial Ecology.
Hill MO, Roy DB, Thompson K. Hemeroby, urbanity and ruderality:
bioindicators of disturbance and human impact. J Appl Ecol
2002;39:708 – 20.
Hodgson JG, Wilson PJ, Hunt R, Grime JP, Thompson K. Allocating
C–S–R plant functional types: a soft approach to a hard problem.
Oikos 1999;85:282 – 94.
Kirby K. Rebuilding the English countryside: habitat fragmentation
and wildlife corridors as issues in practical conservation. Engl Nat
Sci 1995;10 1995 [Peterborough].
Kowarik I. Some responses of flora and vegetation to urbanization in
Central Europe. In: Sukopp H, Hejny S, Kowarik I, editors. Urban
ecology: plants and plant communities in urban environments.
Amsterdam7 SPB Academic Publishing; 1990. p. 45 – 74.
Lindacher R, 1995. Veröff Geobot Inst Eidgen Tech Hochsch Stift
Rübel Zür 1995;125:1 – 436.
McIntyre NE. Ecology of urban arthropods: a review and a call to
action. Ann Entomol Soc Am 2000;93:825 – 35.
McIntyre NE, Rango J, Fagan WF, Faeth SH. Ground arthropod
community structure in a heterogeneous urban environment.
Landsc Urban Plan 2001;52:257 – 74.
Niemelä J. Ecology and urban planning. Biodivers Conserv
1999;8:119 – 31.
Pickett STA, Cadenasso ML, Grove JM, Nilon CH, Pouyat RV,
Zipperer WC, et al. Urban ecological systems: linking terrestrial
ecological, physical, and socioeconomic components of metropol-
itan areas. Ann Rev Ecolog Syst 2001;32:127 – 57.
Pickett STA, Cadenasso ML, Grove JM. Resilient cities: meaning,
models, and metaphor for integrating the ecological, socio-eco-
nomic, and planning realms. Landsc Urban Plan 2004;69:369 – 84.
Rushton SP, Cherrill AJ, Tucker K, O’Callaghan JR. The ecological
modelling system of NELUP. J Environ Plan Manag 1995;
38:35 – 52.
Rushton SP, Sanderson RA, Luff ML, Fuller R. Modelling the spatial
dynamics of ground beetles (Carabidae) within landscapes. Ann
Zool Fenn 1996;33:233 – 41.
Rushton SP, Lurz PWW, Fuller R, Garson PJ. Modelling the distri-
bution of the red and grey squirrel at the landscape scale: a
combined GIS and population dynamics approach. J Appl Ecol
1997;34:1137 – 54.
Rushton SP, Barreto GW, Cormack RM, Macdonald DW, Fuller R.
Modelling the effects of mink and habitat fragmentation on the
water vole. J Appl Ecol 2000;37:475 – 90.
Sadler JP, Small EC, Fiszpan HA, Telfer M. Investigating environmental
variation and landscape characteristics of an urban–rural gradient
using woodland carabid assemblages. J Biogeogr in press.
Sargent C. Britain’s railway vegetation. Institute of Terrestrial
Ecology.
Small EC. Biodiversity and persistence of carabid beetles (Coleoptera:
Carabidae) in fragmented urban spaces. Unpubl. PhD thesis,
School of Geography and Environmental Sciences, The Univer-
sity of Birmingham, 2002.
Small EC, Sadler JP, Telfer MG. Carabid beetle assemblages on
urban derelict sites in Birmingham, UK. J Insect Conserv 2003;
6:233 – 46.
Small EC, Sadler JP, Telfer MG. Do landscape factors affect brown-
field carabid assemblages?. Sci Total Environ 2006;360:205 – 22.
doi:10.1016/j.scitotenv.2005.08.051 [this volume].
Spellerberg IF, Gaywood M. Linear features: linear habitats and
wildlife corridors. English Nature Research Report. Peterborough7
English Nature; 1993.
Thompson K, Austin KC, Smith RM, Warren PH, Angold PG, Gaston
KJ. Urban domestic gardens (I): Putting small-scale plant diver-
sity in context. J Veg Sci 2003;14:71 – 8.
Wright S. Evolution and the genetics of populations Volume 4 Var-
iability within and among natural populations. University of
Chicago Press.
Wadsworth RA, O’Callaghan JR. Empirical searches of the NELUP
land use database. J Environ Plan Manag 1995;38:107 – 16.
Watson PM, Wadsworth RA. A computerised decision support
system for rural policy formulation. Int J Geogr Inf Syst 1996;
10:425 – 40.
Wood BC, Pullin AS. Persistence of species in a fragmented
urban landscape: the importance of dispersal ability and habitat
availability for grassland butterflies. Biodivers Conserv 2002;
11:1451 – 68.
Woodward JC, Eyre MD, Luff ML. Beetles (Coleoptera) on brown-
field sites in England: an important conservation resource? J
Insect Conserv 2003;7:223 – 31.