• alien flora;
  • Britain;
  • disturbed habitat;
  • ecological indication;
  • plant strategy;
  • urban flora


  1. Top of page
  2. Summary
  3. Introduction
  4. Data and methods
  5. Results
  6. Discussion and conclusions
  7. Acknowledgements
  8. References
  • 1
    Species vary according to whether they benefit from or are harmed by disturbance and intensive human activity. This variation can be quantified by indices of disturbance and unnaturalness.
  • 2
    An urban flora was characterized by comparing quadrat data from cities with several large data sets from the countryside. Existing scales of species response to disturbance and unnaturalness, ruderality (a plant's ability to survive in disturbed conditions) and hemeroby (a measure of human impact) were contrasted with derived scales based on the number of associated annuals and aliens and with ‘urbanity’, defined as the proportion of urban land in the vicinity of each quadrat.
  • 3
    Species presence data were available from 26 710 quadrats distributed through Great Britain, with urban sites only in central England. Satellite imagery was used to measure the proportion of urban land cover in the vicinity of each quadrat; 2595 quadrats were located in 1-km squares having at least 40% cover of urban land.
  • 4
    The 20 species having highest urbanity were all alien to Britain, comprising 12 neophytes and eight archaeophytes.
  • 5
    Of the 20 most frequent species in quadrats situated in 1-km squares with at least 40% urban land cover, 18 were natives. The two exceptions were Artemisia vulgaris , an archaeophyte, and Senecio squalidus , a neophyte.
  • 6
    Both ruderal and hemerobic species, as usually defined, include many non-urban arable species. The hemeroby scale of Kowarik (1990 ), designed for Berlin, does not work well in Britain.
  • 7
    The proportion of associated annuals (annuality) and the proportion of associated neophytes (alien richness or xenicity) can be developed into good indices. The annuality scale is very well defined because annuals tend to occur with other annuals. Plants with high annuality are mostly arable weeds.
  • 8
    Urban specialists in central England are, with a few exceptions, character-species of the phytosociological classes Artemisietea , Galio-Urticetea and Stellarietea . Most of them have numerous non-urban associates and they do not form a very well defined group. They have intermediate levels of annuality combined with relatively high levels of xenicity.
  • 9
    While it is possible to develop indices of hemeroby, urbanity and ruderality, these concepts are relatively complicated. Annuality and xenicity are simpler measures that can complement Ellenberg values, but definitive values for Great Britain would require additional data from southern England.


  1. Top of page
  2. Summary
  3. Introduction
  4. Data and methods
  5. Results
  6. Discussion and conclusions
  7. Acknowledgements
  8. References

The flora of urban areas has fascinated botanists for many years. The urban flora has particularly attracted attention after major incidents of destruction, for example the 1666 Great Fire of London, when Sisymbrium irio L. suddenly became abundant (Kent 1975), and the bombing of Berlin (Scholz 1960) and London (Fitter 1945) during the Second World War. Plants of great beauty appear without human assistance in the wake of smaller-scale destruction, such as clearance of old buildings (Mabey 1973), creating ‘urban commons’, which have been celebrated by Gilbert (1992) and Mabey (1996). These urban plants live in a highly unnatural environment. Certainly it is a far cry from the natural home of Chamerion angustifolium (L.) Holub in the boreal forest to the rubble of a ruined house in London.

The attribute of naturalness is much used in the evaluation of sites for nature conservation (Usher 1986). In cities, naturalness has less obvious value. Indeed, ‘nativism’ (Peretti 1998) may be quite irrelevant to the communities that are of most interest to urban ecologists. Nevertheless, the question of how natural a community is cannot be ignored. Urban people are fascinated by wild nature, which is often seen as a part of an ideal countryside (Bunce 1994); naturalness continues to be a major criterion used in site assessment.

For these practical reasons and also because the study of human impacts is of great interest in itself, applied ecologists have devised several measures of unnaturalness or human impact. Hemeroby, the best-known of these, was developed from earlier more informal concepts by Jalas (1955), who proposed a four-point scale based largely on the degree of disturbance to the soil. Jalas's (1955) definitions were subsequently extended to a 10-point scale, which has been used to categorize both plants and places in central Europe (Kowarik 1990, 1999; Sukopp 1990; Grabherr et al. 1995, 1996). Hemeroby on the 10-point scale is a measure of human impact varying from 0 (ahemerobic or completely natural) to 9 (polyhemerobic, consisting of pioneer vegetation of railways, rubbish dumps and salted motorways). Most plants in the flora of the Berlin area have been assigned a hemeroby value (Kowarik in Lindacher 1995), with the intention that these values should be used in much the same way as the indicator values of Ellenberg (1979).

As part of a study of urban floras, we wished to categorize the urban flora of central England. Although there are certain plants that are well known to be urbanophiles, many plants of cities also occur in the wider countryside. Chamerion angustifolium and Urtica dioica L. are in no sense urban specialists, although they are both frequent in British cities. An analysis of species presence in 2-km squares in Britain (Roy, Hill & Rothery 1999) showed that there was a significant effect of urban cover but, at this scale, the urban flora was not picked out as definitely as we had expected. Indeed, Phragmites australis (Cav.) Trin. and Sagina procumbens L. were indicated as species that had relatively high frequency in urban areas although neither is an urban specialist.

In order to discriminate the urban specialists more clearly, it is necessary to use a smaller unit for recording. The quadrat scale is the most obvious one to use. Numerous quadrat surveys have been made, several of which are available electronically (see below). If there is a category of urban specialists, then they should grow in quadrats together. They would be expected to have high hemeroby values, and perhaps high values on other existing scales of urbanity or disturbance, such as that of Frank & Klotz (1990).

If hemeroby is to become a clearly defined concept, then values ought to be confirmed by measurement. This is often difficult. Several different scales have been proposed (Sukopp 1969). The degree of human impact can be judged in a general way but does not necessarily correspond to a simple index like the number of human visits in a year. For example, an arable field may be passed over by machinery only a few times a year but is intensively influenced. Many visitors, on the other hand, may walk to the top of a mountain, which still retains its semi-natural character.

One way to check the values of an ecological index is to find out whether they are similar to those of associated species in large-scale quadrat data. Ellenberg indicator values could in most cases be effectively confirmed in this way, and there were generally good reasons for discrepancies (Hill et al. 2000).

In this study, we set out to answer the following questions. How distinctive is the urban flora in central England? If we can define a measure of urbanity, does it relate clearly to other measures of disturbance? How internally consistent are existing measures of disturbance? Is it even possible to measure the degree of disturbance in a satisfactory way?

Data and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Data and methods
  5. Results
  6. Discussion and conclusions
  7. Acknowledgements
  8. References

existing scales of disturbance and habitat preference

Several existing scales of measurement have already been mentioned and some additional scales can be calculated (Table 1) by methods that are explained below. All of these scales, except for the two urbanity scales, measure substantially different quantities. The existing scales are all directly numeric except for the ruderality scale of Grime (1979, 2001). We have used Grime's (1979, 2001) scale of ruderality rather than the comparable disturbance scale of Trepl (1983), because Grime's (1979, 2001) values are published and readily available. To convert established strategies to numerical values, a simple system of linear interpolation in three dimensions was used, according to the convention: C (competitor) = (10,0,0); S (stress tolerator) = (0,10,0); R (ruderal) = (0,0,10); SR (stress-tolerant ruderal) = (S + R)/2 = (0,5,5); CSR (completely intermediate) = (C + S + R)/3 = (3·3,3·3,3·3); C/CSR = (C + CSR)/2 = (6·7,1·7,1·7), etc. By definition the C + S + R-values sum to 10 (apart from irrelevant round-off error).

Table 1.  Existing scales for species, together with scales that can be calculated from data; for the definition of MA , the mean associated value, and more detailed definitions of U , V , W and A , refer to the text
Name of scaleSymbol used in this paperAuthor or sourceDefinition
1. Existing scales
HemerobyHKKowarik (1990 ); Kowarik in Lindacher (1995 ) Degree of human impact; values for West Berlin
Urbanity (central Europe)UFFrank & Klotz (1990 ) Tendency to occur in cities; values for East Germany
RuderalityRGrime, Hodgson & Hunt (1988 ) Ability to thrive where there is disturbance through partial or total destruction of plant biomass
Competitive ability and stress toleranceC , SGrime, Hodgson & Hunt (1988 ) Stress tolerance is the ability to persist and reproduce in the presence of factors that restrict photosynthetic production
Ellenberg values for BritainSymbols not usedHill et al. (1999 ) See published sources for light, moisture, reaction, fertility and salt
Annual indicator functionIAStace (1997 ), with additions from other sources = 1 for annuals, = 0 otherwise
Neophyte indicator functionIXPreston, Pearman & Dines (2002 ) = 1 for neophytes, = 0 otherwise
2. Calculated values
Urbanity (Britain)UCalculated from quadrat data and satellite imagesMean cover of urban land in vicinity of where a species occurs
Frequency in urban-and-vicinity quadratsVCalculated from quadrat data and satellite imagesFrequency of records in 1-km squares for which the cover of urban land exceeds 40%
Fidelity to urban-and-vicinity quadratsWCalculated from quadrat data and satellite imagesMean proportion of records in 1-km squares for which the cover of urban land exceeds 40%
Annuality (Britain)ACalculated from quadrat data and IAMA(IA), i.e. the proportion of annual species occurring in quadrats with the given species
Xenicity (Britain)XCalculated from quadrat data and IXMA(IX), mean proportion of aliens in same quadrat

Ellenberg indicator values, adjusted for Britain, were taken from the enumeration of Hill et al. (1999). An important existing variable that is not directly a scale of disturbance is the annual indicator function (Table 1), which is 1 for annual plants and 0 for biennials and perennials. The proportion of annuals in the vegetation was listed by Blume & Sukopp (1976) as one of several indicators of human impact (Kultureinfluss).

quadrat data

Five relevé data sets (Table 2 and Fig. 1) were combined to form a single large data matrix with 26 710 samples comprising 1244 species, of which 902 occurred in at least 10 quadrats. Species nomenclature is that of Stace (1997). Samples were available from 2508 1-km squares (c. 1% of the land area of Britain). Urban data were available for the vicinity of two cities, Sheffield in Yorkshire and Birmingham in the West Midlands.

Table 2.  Numbers of sample quadrats and British 1-km squares in relation to urban land cover; an urbanized 1-km square of the National Grid is one with more than 40% cover of urban land
SurveyDateNumber of samplesNumber of 1-km squaresMean cover percentage of urban landNumber of samples from urbanized 1-km squaresProportion (%) of samples from urbanized 1-km squares
Sample quadrats
UCPE Sheffield region surveys1965–7210 638  1 57516·3 1 39313
ITE Countryside Survey 1990199011 410    506 5·0    81 1
ITE Railway Survey1977–81 1 889    23812·1   136 7
ITE Woodland Surveys1971–91 1 737    141 4·9    32 2
West Midlands derelict land1999 1 036     4870·3   95392
Total of surveys 26 710  2 50812·6 2 59510
1-km squares
Total in Great Britain  241 726 6·510 417 4

Figure 1. Location of vegetation samples in 10-km squares of the British National Grid; 90% of urban samples were located near Sheffield and Birmingham.

Download figure to PowerPoint

The UCPE (Unit of Comparative Plant Ecology) Sheffield region surveys covered an area of 3000 km2 surrounding and including the urban centre of Sheffield (Fig. 1). Three surveys were undertaken over a 6-year period between 1965 and 1972 to record semi-natural grassland, all common herbaceous communities and rarer species and communities (Grime, Hodgson & Hunt 1988). The quadrat size was 1 × 1 m.

The largest set of quadrat data was from the Institute of Terrestrial Ecology (ITE) Countryside Survey 1990 (Barr et al. 1993). It was based on a stratified random sample of 508 1-km squares throughout Britain. Urban areas were excluded. The quadrats were of variable size, including large plots (14 × 14 m) placed at random throughout each 1-km square, linear plots (10 × 1 m) and habitat plots (2 × 2 m) targeted in areas of semi-natural vegetation. The ITE Woodland and Railway Surveys also excluded urban land. The ITE Railway Survey was based on a stratified random sample of verges and cuttings throughout the track network of Britain (Sargent 1984). Vegetation was recorded in 2 × 2-m quadrats. The ITE Woodland Surveys used 14 × 14-m quadrats. In 1971, R. G. H. Bunce recorded vegetation in 103 cartographically defined woods in localities widely distributed across Great Britain (Bunce 1982). From 1989 to 1991, 38 farm woods were surveyed using the same methodology; these were mostly small woods, many of which had been established less than 100 years previously.

In the West Midlands, K. C. Austin sampled the vegetation of derelict sites using quadrats of size 1 × 1 m.

urban land cover

The Land Cover Map of Britain is a classification of the land cover of Britain derived from satellite images (Barr et al. 1993; Fuller, Groom & Jones 1994). The map assigns each 25 × 25-m pixel of land to one of 25 land cover categories. Two land cover categories are urban, one applying to pixels where buildings and hard surfaces occupy most of the area and the other to pixels containing a mixture of buildings and permanent vegetation. The proportion of these two categories taken together was used as a measure of urban land within each 1-km square cell of the British National Grid.

calculation of urbanity and frequency in highly urbanized squares

Species occurring in less than 10 samples were omitted from the data set and not considered further. The quadrat data were originally all quantitative but were reduced to presence and absence for analysis. They can be represented by a matrix of m samples and n species:

  • A  = [ aij ] ( i  = 1, … , m ; j  = 1, … , n )

where aij= 1 if species j is present in sample i and aij= 0 otherwise. Let ui be the proportion of urban land cover in the 1-km grid square containing sample i. Then the urbanity Uj of species j is defined as the mean proportion of urban land in its vicinity:

  • image

One-kilometre squares with at least 40% urban land cover were defined as ‘high-urban’ squares. Quadrat samples from high-urban squares were called ‘urban-and-vicinity’ quadrats. Another measure of urbanity can be defined using the indicator function of these squares. Specifically, let v(i) = 1 if sample i is in a 1-km square with > 40% urban land cover and v(i) = 0 otherwise.

Then, for a species j, the frequency in urban-and-vicinity quadrats is defined in the obvious way:

  • image

The proportion of occurrences in urban-and-vicinity quadrats is defined similarly:

  • image

calculation of mean associated values derived from field data

For several attributes, values were available for species j but not for sites i. Given a species attribute:

  • Y  = [ yj ] ( j  = 1, … , n )

such as urbanity or the annual indicator function, then the mean associated value inline image for species j is defined as the mean value of yk for those species k that are found in the same quadrats as species j. MA(Y), the vector of mean associated values, is then calculated as follows. Stage 1: calculate row (sample) totals and weights:

  • image
  • wi  = 1/ ai+
  • image

Stage 2: calculate mean values for associates, not including the species itself:

  • image
  • image

Note that this method of calculating mean associated values gives approximately equal overall weight to each sample. An associate of a given species has less influence if it is one among many associates in a sample than if it is the only associate.


  1. Top of page
  2. Summary
  3. Introduction
  4. Data and methods
  5. Results
  6. Discussion and conclusions
  7. Acknowledgements
  8. References

Of the 20 most urban species (Table 3) none was unequivocally a British native, although the eight European species that were first found before 1680 are now classified as archaeophytes, presumed to have been established in Britain before 1500 (Preston, Pearman & Dines 2002). The great majority of the most urban species were in fact relatively uncommon in our sample; only Artemisia absinthium and Senecio squalidus were present in more than 0·5% of the quadrats. While this pattern in part reflected the mainly rural position of the sample quadrats, for which the mean urban land cover was 13%, it also emphasized the fact that most urban specialists are relatively rare. The proportion of annual associates varied widely from Acer platanoides, Aster novi-belgii, Foeniculum vulgare, Lupinus × regalis and Solidago canadensis, which had fewer annual associates than the average (12%) for all quadrats, to Anthemis cotula, Apera interrupta, Conyza canadensis, Diplotaxis tenuifolia, Lactuca serriola, Reseda luteola, Senecio squalidus and Vulpia myuros, which had more than twice the average proportion of annual associates.

Table 3.  The 20 most urban species, ordered by the proportion of urban land cover in 1-km squares where they occurred. None of these species is native to Britain; those marked (a) are archaeophytes, introduced before 1500
NameLongevity (a-annual, b = biennial, p = perennial)Number of samplesU : mean urban land cover (%) A : mean proportion of annual associates (%) Continent of originDate of first British record in wildDate of introduction to Great Britain for known garden plants
Buddleja davidiip    407519Asia19271890s
Lactuca serriola (a) b    677431Europe1632
Melilotus albusa, b    416422Europe1822
Melilotus officinalisb    955921Europe1848
Lupinus  ×  regalisp    2858 9North America19551937
Conyza canadensisa    755832North America1690
Solidago canadensis s.l. p   1045610North America18881648
Aster novi-belgiip    585611North America18601710
Artemisia absinthium (a) p   1715621Europe15511200
Oenothera glaziovianab    165421North America18601778
Senecio squalidusa, b, p   5855031Europe17921620s
Apera interruptaa    234946Europe1848
Diplotaxis tenuifolia (a) p    244827Europe1597
Acer platanoidesp    2647 8Europe19051683
Saponaria officinalis (a) p    254614Europe1597Medieval
Calystegia pulchrap    234613Asia18841823
Vulpia myuros (a) a   1144528Europe1633
Foeniculum vulgare (a) p    1644 8Europe1677Roman
Reseda luteola (a) b   1064435Europe1570Iron Age
Anthemis cotula (a) a    384437Europe1523
All samples 26 7101312   

By contrast, all but three of the 20 most common species in highly urbanized 1-km squares had no particular association with urban land but were merely common in both town and country (Table 4). Artemisia vulgaris and Senecio squalidus were strongly concentrated in urban areas and Chamerion angustifolium was nearly twice as common there as in the countryside at large. In the other direction, Agrostis capillaris, Dactylis glomerata, Poa trivialis, Rubus fruticosus and Urtica dioica were less frequent in urban areas than in the countryside at large but were still present in at least 14% of quadrats located in highly urbanized 1-km squares. Much the same pattern was shown by the mean cover of urban land in 1-km squares containing each species. By this criterion, Agrostis capillaris and Urtica dioica were the least urban of the common species; Artemisia vulgaris and Senecio squalidus were the most urban.

Table 4.  Most common species in quadrats located in 1-km squares with more than 40% cover of urban land (urban-and-vicinity quadrats); all species are British natives except for those marked (a) and (n), which are archaeophytes and neophytes, respectively
NameTotal number of occurrences in all quadratsNumber of occurrences in urban-and-vicinity quadratsW : proportion (%) of records in urban-and-vicinity quadrats U : proportion (%) of urban land near quadrats where species present
Agrostis stolonifera 7 483106914·316
Holcus lanatus 8 732 95911·013
Arrhenatherum elatius 6 962 82711·915
Festuca rubra 6 918 71410·313
Dactylis glomerata 7 677 651 8·512
Cirsium arvense 4 527 60413·316
Plantago lanceolata 4 337 59813·815
Lolium perenne 5 232 59711·414
Taraxacum officinale 4 714 57712·214
Rubus fruticosus 6 723 567 8·412
Trifolium repens 4 863 50310·312
Poa pratensis 4 226 49611·714
Agrostis capillaris 5 945 464 7·8 9
Elytrigia repens 4 546 46310·215
Poa trivialis 5 979 456 7·612
Poa annua 4 009 43710·913
Chamerion angustifolium 2 198 41919·120
Artemisia vulgaris (a)    796 39249·241
Urtica dioica 6 484 370 5·710
Senecio squalidus (n)    585 36662·650
All quadrats26 7102595 9·713

Although many urban plants grow in disturbed habitats, there was no very marked tendency for the characteristic plants of urban areas to be associated with annuals (Fig. 2). Indeed, the characteristically urban plants, i.e. those with at least 40% mean urban cover near to where they were found, had an intermediate proportion of associated annuals (A), mostly in the range 10–40%. Above this range, the great majority of species were annuals that occur frequently if not always on arable land. Two of the three perennials for which A > 40%, Brassica oleracea and Solanum tuberosum, are grown as annual field crops. The third perennial, Potentilla argentea, was more surprising. In fact, although it was found in open communities with many annuals, its most frequent associates were Agrostis capillaris, Plantago lanceolata, Poa pratensis and Veronica arvensis, only one of which is an annual. Three species, Apera interrupta, Sisymbrium altissimum and Sisymbrium orientale, had A > 40% and U > 40%.


Figure 2. 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.

Download figure to PowerPoint

When species with high numbers of annual associates were compared with urban species, their mean Ellenberg values were generally similar (Table 5). The plants with numerous annual associates were almost exclusively arable weeds. They and urban species had a mean signature that indicated drier, less acid and more fertile sites than the average for all species. The largest difference from the overall average was in the nitrogen values, which are broadly an indication of fertility. In terms of Grime's (1979, 2001) established strategies, the annual arable species were almost entirely ruderals. Of the species characterized by high cover of nearby urban land, only three, Artemisia absinthium, Saponaria officinalis and Senecio squalidus, had been assigned values for their established strategy. As a result, the mean was not significantly different from that of the species that were frequent in urbanized 1-km squares. The proportion of associated annuals was, however, significantly higher, 0·22 as opposed to 0·15, suggesting that the mean ruderality value would have been genuinely higher if more species had been scored for it.

Table 5.  Mean attributes of species in categories defined by annuality ( A ), urban land cover ( U ) and frequency in urban-and-vicinity quadrats ( V ); means are arithmetic means except for plant heights, which are medians. Values in parentheses are standard errors
Ecological attributeSpecies with annuality > 0·5Species with high urban land cover (Table 2)Species most frequent in urban squares (Table 3)All species
Ellenberg indicator value
Light7·2 (0·13)7·6 (0·28)7·0 (0·15)6·8 (0·04)
Moisture4·4 (0·10)4·6 (0·18)5·2 (0·14)6·0 (0·07)
Soil reaction6·6 (0·13)6·9 (0·15)6·4 (0·18)6·1 (0·05)
Nitrogen6·1 (0·21)5·8 (0·29)6·0 (0·23)4·8 (0·06)
Salt0·2 (0·14)0·4 (0·22)0·4 (0·15)0·2 (0·03)
Number of species with values291920830
Mean (SE) established strategy
Competitive ability0·8 (0·31)3·6 (0·99)5·1 (0·67)3·0 (0·12)
Stress tolerance0·0 (0·00)1·1 (0·55)1·6 (0·35)3·5 (0·14)
Ruderality9·2 (0·31)5·3 (0·71)3·3 (0·60)3·5 (0·13)
Number of species with values15319488
Mean (SE) values of other attributes
Plant height (cm)60 (9)150 (25)95 (14)66 (1)
Number of species with values302020726
Hemeroby (HK)6·8 (0·12)6·8 (0·48)5·2 (0·56)3·9 (0·08)
Number of species with values2168466
Associations with annual plants (A)0·59 (0·01)0·22 (0·02)0·15 (0·01)0·15 (0·00)
Total number of species302020902

Species whose associates were mostly annuals and species found in places with high urban land cover both had a mean score for hemeroby of 6·8 (Table 5). This was significantly higher than the mean for frequent species in highly urban areas, but for the urban specialist category the significance was only borderline at 5%. In comparison with the average for all species, it therefore appeared that the urban specialists were more ruderal and hemerobic, and that the frequent species were more competitive and of intermediate hemeroby. These patterns are clearly apparent in Fig. 3. This shows that the highly urban species included a few highly hemerobic species, notably Buddleja davidii (several urban specialists did not have a value for hemeroby). From a British perspective, some of the highly hemerobic species towards the bottom left of the diagram are surprising, for example Ajuga reptans, Digitalis purpurea and Veronica serpyllifolia, which have a hemeroby value of 6 but are not confined to situations with high human impact.


Figure 3. Hemeroby in relation to the proportion of annuals and percentage urban land cover near where species were found; large dots signify species with high hemeroby (6–9), small dots denote the remainder.

Download figure to PowerPoint

The pattern shown by highly ruderal species was somewhat more concentrated but the ruderals (in the sense of Grime 1979, 2001) were nevertheless spread widely along the annuality axis (Fig. 4). Few urban specialists were highly ruderal; only Senecio squalidus and Sisymbrium altissimum fell into this category. Highly ruderal species with very few annual associates were Cardamine flexuosa, Cardamine pratense, Geranium robertianum, Rhinanthus minor and Senecio aquaticus.


Figure 4. Ruderality in relation to the proportion of annuals and percentage urban land cover near where species were found; large dots signify species with high ruderality (i.e. 6·6–10·0, equivalent to R, R/CR, R/SR and R/CSR), small dots denote less ruderal species.

Download figure to PowerPoint

When levels of urbanity were plotted against annuality and xenicity, the nature of the urban specialists was clear (Fig. 5). They had medium levels of annuality. Relative to their annuality, the proportion of associated neophytes was high. Arable species had a medium to high number of neophyte associates. Both cities and arable fields experience high human impact. However, in the British samples studied, arable specialists had in general higher xenicity than urban specialists.


Figure 5. Urbanity in relation to the proportions of annuals and neophytes; species are represented by differing symbols according to their urbanity.

Download figure to PowerPoint

The internal consistency of the scales was measured by correlating species scores with mean associated values. For annuality the correlation was 0·954; for xenicity it was 0·888; for urban cover it was 0·815. The relation (not shown here) between A and MA(A) was curvilinear over only a small part of its range, as was the relation between X and MA(X). The relation between U and MA(U) was so curvilinear that MA(U) was shaped more like U3 than U. This was because much of southern Britain is urbanized, so that for U < 25%, U often did not measure whether the species occurred in cities but merely whether it occurred in southern Britain where urban land cover is greater. To eliminate the influence of these irrelevant species, a new variable:

  • U25  = max (( U  − 25),0)

was defined and used to derive a new associated-value variable MA(U25). This variable had a slightly higher correlation, 0·821, with U and lacked the tendency shown by MA(U) to give relatively high values to agrestal and aquatic plants.

The 20 most urban plants according to the original variables U and W, and the 20 with most urban associates, as measured by MA(U) and MA(U25) were, with only six exceptions, species that are indicative of the phytosociological classes Artemisietea vulgaris, Stellarietea mediae and Galio–Urticeteae (Table 6). The exceptions were Centranthus ruber, normally a plant of walls, Cerastium tomentosum and Symphoricarpos albus, which are garden plants usually found close to where they were planted, Vulpia myuros, which in central Europe grows in dry grasslands of the Koelerio–Corynephoretea, Cichorium intybus, found on waysides in Molinio–Arrhenatheretea grassland, and Acer platanoides, a large tree whose seedlings and saplings are found in urban areas near to where it is planted.

Table 6.  Species that either had high urbanity, as measured by mean urban cover or frequency in highly urban 1-km squares, or which were strongly associated with species having high urbanity, together with their main phytosociological class. The Stellarietea have been divided into two orders, Polygono-Chenopodietalia (P) and Sisymbrietalia (S). The urban code specifies the presence (1) or absence (0) of the species in the list of 20 most urban plants according to the urban variables urbanity ( U ), fidelity to highly urban squares ( W ), associated urbanity ( MA ( U )) and associated urbanity above the 25% threshold ( MA ( U25 ))
SpeciesNumber of occurrencesUrban codeClass
(a) Urban specialists with many urban associates
Apera interrupta 231111Stellarietea mediae (P)
Artemisia absinthium1711111Artemisietea vulgaris
Buddleja davidii 401111Galio-Urticetea
Conyza canadensis 751111Stellarietea mediae
Diplotaxis tenuifolia 241111Artemisietea vulgaris
Foeniculum vulgare 161001Artemisietea vulgaris
Lactuca serriola 671111Stellarietea mediae (S)
Melilotus albus 411111Artemisietea vulgaris
Melilotus officinalis 951111Artemisietea vulgaris
Oenothera glazioviana 161111Artemisietea vulgaris
Reseda luteola1061011Artemisietea vulgaris
Senecio squalidus5851111Artemisietea vulgaris
Vulpia myuros1141111Koelerio-Corynephoretea
(b) Urban specialists with fewer urban associates
Acer platanoides 261100Querco-Fagetea
Anthemis cotula 381100Stellarietea mediae (P)
Aster novi-belgii 581100Galio-Urticetea
Calystegia pulchra 231100Galio-Urticetea
Centranthus ruber 160100Asplenietea trichomanis
Cerastium tomentosum 150100(garden throw-out)
Lupinus  ×  regalis 281100Artemisietea vulgaris
Saponaria officinalis 251000Galio-Urticetea
Solidago canadensis s.l. 1041100Galio-Urticetea
Symphoricarpos albus 100100(usually planted)
(c) Less urban species with many urban associates
Cichorium intybus 120010Molinio-Arrhenatheretea
Crepis vesicaria 330001Stellarietea mediae (S)
Diplotaxis muralis 280011Stellarietea mediae (S)
Lepidium campestre 110010Stellarietea mediae (S)
Melilotus altissimus 460011Artemisietea vulgaris
Oenotheria biennis 220011Artemisietea vulgaris
Papaver dubium1160010Stellarietea mediae (P)
Sisymbrium altissimum 350011Stellarietea mediae (S)
Sisymbrium orientale 380011Stellarietea mediae (S)
Tanacetum vulgare 960001Galio-Urticetea

The major classes Artemisietea vulgaris (order Onopordietalia acanthii) and Galio–Urticeteae (order Lamio albi–Chenopodietalia) correspond, respectively, to plants that are characteristic of rather open habitats and plants that form dense, often clonal, vegetation. The assignment of Buddleja davidii, which colonizes open habitats but later forms thickets, to Galio–Urticeteae follows Mucina (1993). Plants corresponding to the Stellarietea mediae can be divided into two orders, Polygono–Chenopodietalia (basically agrestal) and Sisymbrietalia (annual-dominated communities of waysides and waste places). Of the normally agrestal species, Anthemis cotula, Apera interrupta and Papaver dubium occupied mainly or entirely non-agrestal habitats in our sample.

Discussion and conclusions

  1. Top of page
  2. Summary
  3. Introduction
  4. Data and methods
  5. Results
  6. Discussion and conclusions
  7. Acknowledgements
  8. References

suitability of the sample for distinguishing attributes of the urban flora

The quadrat samples do not represent a random or even a stratified random sample of the vegetation of Britain. Many of them were, however, located in this way. Many quadrats, including some urban quadrats, were selected to show the range of variation in urban vegetation. The scaling of the variable U is therefore inevitably somewhat arbitrary. In principle, a true scale for Britain could be calculated by taking random quadrats across the entire country, including both urban and rural areas. In practice, we have opted to use available data, deliberately including a higher proportion of urban quadrats than the average for the whole country. This resulted in the mean urban cover of our quadrats being 12·6% whereas that of the country as a whole was 6·5%.

The choice of 1-km squares as the units within which urban cover was measured was dictated more by practicality than optimality. For the ITE surveys, more detailed localization was not available in machine-readable form. The regions of both Sheffield and the West Midlands include large cities, so that broadly similar results would be expected if urban land cover had been measured over a larger area. The sensitivity of the decision to use 1-km squares has not been tested, but is likely not to be great.

The quadrats varied in size from 1-m square (Sheffield and Birmingham surveys) to 14-m square (Countryside Survey and Woodland Survey). The size difference could in principle have biased the results. In the Countryside Survey, arable fields were normally sampled near the edge. The large quadrat size may therefore have resulted in more perennials being included as associates of annuals than if a small quadrat had been used. Fortunately, this bias should have little effect on the order of species. An index such as annuality is merely an index. Its calculated values depend to some extent on details of quadrat size and position, but whatever the quadrat size, arable plants would have had the highest values. The common and characteristically arable species Fallopia convolvulus, Urtica urens and Viola arvensis all had annuality values exceeding 0·6. No characteristically urban plant had A > 0·4.

Different results would undoubtedly have been obtained if we had chosen different cities. London, in particular, has a larger urban flora than central England, including warmth-loving plants such as Ailanthus altissima and Galinsoga parviflora, which were scarcely recorded in our cities. Likewise, several of the characteristic urban species of central England are uncommon or rare further to the north, for example Artemisia absinthium, Conyza canadensis and Vulpia myuros, which are rare in Glasgow (Dickson, Macpherson & Watson 2000). Cities, like many other biotopes, show a strong gradient of biodiversity from north to south in Europe. Only in the Mediterranean region does this gradient cease. Grapow & Blasi (1998), using a standardized survey technique in Italian cities, found 372 species in Rome, compared with 215 in Milan and 273 in Palermo. Rome, appropriately for its age and area, had the richest urban flora but is in the centre of Italy.

urban and rural flora

Although there are a few characteristically urban species such as Artemisia absinthium, Buddleja davidii, Conyza canadensis and Senecio squalidus, the urban flora of central England is mainly a mixture of common plants in the wider countryside, together with species that occur generally on waysides and in waste places. Most of the commonest species in highly urbanized 1-km squares are merely those that are well represented in grassy vegetation outside towns, not necessarily where there is any strong disturbance. However, Artemisia vulgaris and Chamerion angustifolium are plants of disturbed ground in the wider countryside, especially by roads. Only Senecio squalidus is both common and distinctively urban. Note the close parallels here with the results of Grime (1986), who listed both the commonest plants of spoil in the Sheffield region, and the plants most restricted to spoil. Of the common species, 16 are also found in Table 4 in the present study. There are clearly a handful of species, most of them grasses, that are common in herbaceous vegetation throughout England. As in this study, Senecio squalidus was one of the few spoil specialists that was also common on spoil.

The fact that the majority of the commonest urban species were also common in ordinary lowland countryside raises the possibility that the 40% threshold for highly urban squares was too low and that many of the samples from these 1-km squares were in fact from countryside close to towns. If the threshold for urbanized squares is raised to 80% cover of urban land, a very similar list of commonest species results. Four natives, Elytrigia repens, Poa pratensis, Poa trivialis and Urtica dioica, drop out of the list, to be replaced by four other natives, Crepis capillaris, Epilobium montanum, Medicago lupulina and Rumex obtusifolius. Of the 594 samples satisfying the 80% criterion, 434 were from derelict sites in Birmingham, so they are less representative of central England as a whole. We have also (M.O. Hill, D.B. Roy & K. Thompson, unpublished results) found a strong preponderance of natives in the eastern English city of Peterborough. The preponderance of natives is not an artefact of sampling.

Because only one of the urban specialists is common, the effect of urban aliens on the native flora is almost certainly small. This is reflected in the fact that they are characteristic of phytosociological classes whose communities occupy waysides and waste places. Only when the aliens form dense thickets are they likely to have a relatively large effect. Dense stands of Aster novi-belgii, Buddleja davidii, Calystegia pulchra and Solidago canadensis s.l. are occasionally found, but only in places where there is a strong human influence. Native plants almost always have alternative habitats where there is less human influence. There is, however, one neophyte, Calystegia silvatica, that may have displaced a native in towns. This was not distinguished from the native Calystegia sepium in some of the surveys in our database. We therefore lack information on how highly urban it is.

scales of disturbance

Hemeroby has hitherto been mainly used a scale for site assessment, rather than as a scale for characterizing species. The definitions have been flexible and designed to meet the needs of particular surveys. For the assessment of hemeroby in Austrian forests (Grabherr et al. 1995), not only floristic criteria but also the naturalness of the restocking process, the amount of dead wood, size of felling coupes and stand structure were taken into account. Our aim in seeking a hemeroby index for central England was to develop a floristic criterion that would be sufficiently simple and objective to be used directly in site assessment. Site floristic values for hemeroby could then feed into more complex evaluations.

In the event, the hemeroby scale of Kowarik in Lindacher (1995) was not sufficiently consistent with the unnaturalness of habitats in Britain to provide a viable basis for quantifying the degree of human impact. This does not rule out the possibility that another, mainly new, scale of hemeroby could be constructed for central England. There are, however, two substantial difficulties. One is that there are many types of human influence, just as there are many types of disturbance (Grubb 1988). The other is that hemerobic plants often occur in quadrats together with plants that are not particularly characteristic of sites with strong human influence. In other words, the hemerobic signal may be rather weak.

Grime's (1979, 2001 ) ruderality was not originally designed as a scale of disturbance but as a strategy predisposing plants to thrive in environments subject to biomass destruction. In practice, the published ruderality scale generally rates annuals high and perennials low. This relationship between life history and the R strategy was formally recognized by Hodgson et al. (1999 ), who used association with monocarpic species (and vernals) as the objective ‘gold standard’ of ruderality.

As a scale for environmental assessment, Grime's (1979, 2001) ruderality has the relatively unsatisfactory feature that some species with high ruderality, such as Cardamine pratensis, Rhinanthus minor and Senecio aquaticus, are mainly associated with perennials that are not ruderal. Although the present study uses the R-values published in Grime, Hodgson & Hunt (1988), it is interesting to note that application of the more recent protocol in Hodgson et al. (1999) makes the first two of these species significantly less ruderal. For basic understanding of plant communities, it is very interesting to know that plants with contrasting strategies can live together in a community. However, for bioindication of environmental conditions, this fact is less relevant.

Associated annuality A was highly correlated with its mean associated value MA(A), making it suitable as a measure of environmental condition. In northern Europe, high annuality is generally an indication of human disturbance but it can also result from there being an unfavourable season. This applies particularly to communities of winter annuals, which sometimes occur on little-disturbed sand dunes in Britain and are widespread in the Mediterranean region where open conditions are maintained by summer drought. Annuality therefore is not specifically a measure of human disturbance of the soil surface. Drought, if severe enough, destroys biomass just as surely as ploughing. Annuality does, however, appear to be a good community attribute that has been unjustly overlooked by most authors as a measure of environmental condition. High annuality is not the same as being an annual. Impatiens glandulifera is an annual that mostly occurs with perennials and Sedum acre is a perennial that is normally associated with annuals.

Urbanity U was less well correlated with its mean associated values MA(U) than annuality. It indicates a different kind of disturbance, longer-term and less regular. In terms of Grubb's (1988) classification of the means of persistence at a regional scale, urban specialists fall mainly into category 8 ‘Occupancy of transient but not very short-lived sites, with newly available sites often immediately adjacent’. By contrast, many of the common species of urban 1-km squares belong to category 3 ‘Long-term occupancy of an appreciable part of the landscape as adults and juveniles’.

Because urbanity picks out species that are specialists of irregularly disturbed sites, it does not necessarily provide a measure of the ‘ruderalization of the countryside’ (Scholz 1960), which is apparent in the loss of stress-tolerant species from the landscape (Thompson 1994; Grime 2001). For measuring this and other changes, annuality and xenicity may be more useful than urbanity. They are not the same as ruderality and hemeroby but are closely related to them. They are simply defined and, given a good database of quadrats, easily measured.


The urban flora of central England is not sufficiently distinctive to be characterized satisfactorily by an index. Hemeroby and ruderality have to be carefully defined in order to give them a comprehensive operational definition. In practice, the most effective indices were not derived from environmental information but from attributes of associated species. We recommend that indices of annuality and xenicity should be developed for the vascular plants of Great Britain and other European countries, to complement the Ellenberg values already available. For Great Britain, a wider range of samples from urban sites and waste places in the south of England would be required, as well as from Sheffield and Birmingham.


  1. Top of page
  2. Summary
  3. Introduction
  4. Data and methods
  5. Results
  6. Discussion and conclusions
  7. Acknowledgements
  8. References

This work was funded by the UK Natural Environment Research Council through its programme URGENT, award GST/03/1979. We are grateful to Kevin Austin, Bob Bunce, Mike Le Duc, Owen Mountford and, especially, John Hodgson, who made their quadrat data available for this analysis. We thank the land owners who permitted botanical recording on their land. Chris Preston gave us up-to-date information on the native status of species in Britain. Jon Marshall and three anonymous referees made very helpful comments on an earlier draft.


  1. Top of page
  2. Summary
  3. Introduction
  4. Data and methods
  5. Results
  6. Discussion and conclusions
  7. Acknowledgements
  8. References
  • Barr, C.J., Bunce, R.G.H., Clarke, R.T., Fuller, R.M., Furse, M.T., Gillespie, M.K., Groom, G.B., Hallam, C.J., Hornung, M., Howard, D.C. & Ness, M.J. (1993) Countryside Survey 1990: Main Report. HMSO, London, UK.
  • Blume, H.-P. & Sukopp, H. (1976) Ökologische Bedeutung anthropogener Bodenveränderungen. Schriftenreihe für Vegetationskunde, 10, 7589.
  • Bunce, R.G.H. (1982) A Field Key for Classifying British Woodland Vegetation, Part 1. Institute of Terrestrial Ecology, Cambridge, UK.
  • Bunce, M. (1994) The Countryside Ideal: Anglo-American Images of Landscape. Routledge, London, UK.
  • Dickson, J.H., Macpherson, P. & Watson, K. (2000) The Changing Flora of Glasgow. Edinburgh University Press, Edinburgh, UK.
  • Ellenberg, H. (1979) Zeigerwerte von Gefässpflanzen Mitteleuropas. Scripta Geobotanica, 9, 1122.
  • Fitter, R.S.R. (1945) London's Natural History. Collins, London, UK.
  • Frank, D. & Klotz, S. (1990) Biologisch-Ökologische Daten Zur Flora der DDR, 2 edn. Martin-Luther-Universität, Halle-Wittenberg, Germany.
  • Fuller, R.M., Groom, G.B. & Jones, A.R. (1994) The Land Cover Map of Great Britain: an automated classification of Landsat Thematic Mapper data. Photogrammetric Engineering and Remote Sensing, 60, 553562.
  • Gilbert, O. (1992) The Flowering of Cities: The Natural Flora of ‘Urban Commons’. English Nature, Peterborough, UK.
  • Grabherr, G., Koch, G., Kirchmeir, H. & Reiter, K. (1995) Hemerobie österreichischer Waldökosysteme – Vorstellung eines Forschungsvorhabens im Rahmen des österreicheschen Beitrages zum MAB-Programm der UNESCO. Zeitschrift für Ökologie und Naturschutz, 4, 105110.
  • Grabherr, G., Koch, G., Kirchmeir, H. & Reiter, K. (1996) Wie Natürlich Is der Österreichische Wald? Symposiumsmappe. Institut für Pflanzenphysiolgie, Vienna, Austria.
  • Grapow, L.C. & Blasi, C. (1998) A comparison of the urban flora of different phytoclimatic regions in Italy. Global Ecology and Biogeography, 7, 367378.
  • Grime, J.P. (1979) Plant Strategies and Vegetation Processes. John Wiley, Chichester, UK.
  • Grime, J.P. (1986) The circumstances and characteristics of spoil colonization within a local flora. Philosophical Transactions of the Royal Society of London, B314, 637654.
  • Grime, J.P. (2001) Plant Strategies, Vegetation Processes, and Ecosystem Properties. Wiley, Chichester, UK.
  • Grime, J.P., Hodgson, J.G. & Hunt, R. (1988) Comparative Plant Ecology: A Functional Approach to Common British Species. Unwin-Hyman, London, UK.
  • Grubb, P.J. (1988) The uncoupling of disturbance and recruitment, two kinds of seed bank, and persistence of plant populations at the regional and local scales. Annales Zoologici Fennici, 25, 2336.
  • Hill, M.O., Mountford, J.O., Roy, D.B. & Bunce, R.G.H. (1999) Ellenberg's Indicator Values for British Plants. ECOFACT, Vol. 2Technical Annex. Institute of Terrestrial Ecology, Huntingdon, UK.
  • Hill, M.O., Roy, D.B., Mountford, J.O. & Bunce, R.G.H. (2000) Extending Ellenberg's indicator values to a new area: an algorithmic approach. Journal of Applied Ecology, 37, 315.
  • Hodgson, J.G., Wilson, P.J., Hunt, R., Grime, J.P. & Thompson, K. (1999) Allocating C-S-R plant functional types: a soft approach to a hard problem. Oikos, 85, 282294.
  • Jalas, J. (1955) Hemerobe und hemerochore Pflanzenarten. Ein terminologischer Reformversuch. Acta Societatia Pro Fauna et Flora Fennica, 72, 115.
  • Kent, D.H. (1975) The Historical Flora of Middlesex. Ray Society, London, UK.
  • Kowarik, I. (1990) Some responses of flora and vegetation to urbanization in Central Europe. Urban Ecology: Plants and Plant Communities in Urban Environments (eds H.Sukopp, S.Hejný & I.Kowarik), pp. 4574. SPB Academic Publishing, Amsterdam, the Netherlands.
  • Kowarik, I. (1999) Natürlichkeit, Naturnähe und Hemerobie als Bewertungskriterien. Handbuch für Naturschutz und Landschaftspflege (eds W.Konold, R.Böcker & U.Hampicke), V-2·1, pp. 118. Ecomed, Landsberg, Germany.
  • Lindacher, R. (1995) PHANART: Datenbank der Gefässpflanzen Mitteleuropas. Veröffentlichungen Des Geobotanischen Institutes der ETH, Stiftung Rübel, Zürich, 125, 1436.
  • Mabey, R. (1973) The Unofficial Countryside. Collins, London, UK.
  • Mabey, R. (1996) Flora Britannica. Sinclair-Stevenson, London, UK.
  • Mucina, L. (1993) Galio-Urcitetea. Die Pflanzengesellschaften Österreichs. 1. Anthropogene Vegetation (eds L.Mucina, G.Grabherr & T.Ellmauer), pp. 203251. Gustav Fischer, Jena, Germany.
  • Peretti, J.H. (1998) Nativism and nature: rethinking biological invasion. Environmental Values, 7, 183192.
  • Preston, C.D., Pearman, D.A. & Dines, T.D. (2002) New Atlas of the British and Irish Flora. Oxford University Press, Oxford, UK.
  • Roy, D.B., Hill, M.O. & Rothery, P. (1999) Effects of urban land cover on the local species pool in Britain. Ecography, 22, 507515.
  • Sargent, C. (1984) Britain's Railway Vegetation. Institute of Terrestrial Ecology, Huntingdon, UK.
  • Scholz, H. (1960) Die Veränderungen in der Ruderalflora Berlins. Willdenowia, 2, 379397.
  • Stace, C.A. (1997) New Flora of the British Isles, 2 edn. Cambridge University Press, Cambridge, UK.
  • Sukopp, H. (1969) Der Einfluss des Menschen auf die Vegetation. Vegetatio, 17, 360371.
  • Sukopp, H. (1990) Stadtökologie: das Beispiel Berlin. Dietrich Reimer, Berlin, Germany.
  • Thompson, K. (1994) Predicting the fate of temperate species in response to human disturbance and global change. Biodiversity, Temperate Ecosystems and Global Change (eds T.J.B.Boyle & C.E.B.Boyle), pp. 6176. Springer Verlag, Berlin, Germany.
  • Trepl, L. (1983) Zum Gebrauch von Pflanzenarten als Indikatoren der Umwelddynamik. Sitzungsberichte der Gesellschaft Naturforschender Freunde Zu Berlin, 23, 151171.
  • Usher, M.B. (1986) Wildlife conservation evaluation: attributes, criteria and values. Wildlife Conservation Evaluation (ed. M.B.Usher), pp. 344. Chapman & Hall, London, UK.