The process of urbanization, which so profoundly transformed the European landscape during the last century, created new environmental conditions for plants (Sukopp et al., 1973; Sukopp & Werner, 1983). Human activities and the inherent structure of cities have produced similar ecological characteristics in urban areas (such as the prevalence of artificial soils, the ‘urban heat island’ effect, comparable patterns of disturbance etc.), even in different biogeographic regions. The response of flora and vegetation to these environmental changes can be traced through the decline of elements of native and natural vegetation and the spread of alien species (Sukopp & Trepl, 1987; Kowarik, 1990), so that different cities, particularly in the inner areas, share a high proportion of spontaneous species. This might lead to the general hypothesis that urbanization causes a uniformity of urban floras. However, from studies published on the flora and vegetation of Italian settlements, the spontaneous plant cover of cities, especially in the Mediterranean basin, appears highly diversified (Anzalone, 1986; Buffa, Venturella & Raimondo, 1986; Poli Marchese, Grillo & Maugeri, 1989); Biondi et al., 1993–94; La Valva & De Natale, 1993–94; Siniscalco & Montacchini, 1993–94; Celesti Grapow & Blasi, 1996).
This paper deals with the following questions. (1) How similar are floras of settlements located in different phytoclimatic regions? (2) What is the relative importance of indigenous versus alien species? (3) What is the role played by apophytes (Table 1) in the composition of urban floras in different phytoclimatic regions? To evaluate these questions we compare the spontaneous vascular flora of five cities, selected to represent the main phytoclimatic regions of Italy (Biondi & Baldoni, 1994).
The geographic location of the study areas is shown in Fig. 1, together with their respective climate diagrams (Servizio Idrografico Nazionale, 1955–86). They are distributed across the main phytoclimatic regions of Italy: continental, temperate and Mediterranean (Biondi, Baldoni & Talamonti, 1991). In the temperate region, Ancona (103,000 inh.), and in the continental region, Milan (1,432,000 inh., the second largest Italian city), were selected. Due to its great variety, in the Mediterranean region three settlements were chosen: Rome (2,791,000 inh.) to represent the meso- mediterranean type, Cagliari (218,000 inh., the main city of the island of Sardinia) and Palermo (734,000 inh., the main city of Sicily) to represent the thermomediterranean type (Blasi, 1994).
In the relatively youthful field of urban ecology there are still notable differences in the methods used by various authors for compiling floristic inventories. Studies mainly vary in the different extent of the study area, in the choice of the species included in the lists (being only established species or including casual species), and in the research timespan (Klotz, 1990; Pyšek, 1993). Hence, in order to obtain comparable data, preference was given to data gathered in sample areas rather than to the flora of the whole city taken from the literature.
In each city, ten square plots of 1 ha were chosen, each representing (when present) areas of the following nine habitat types: parks, fallows, ruins or archaeological sites, dumping sites, recent housing developments, industrial sites, old town centres, residential areas and heavily disturbed sites (i.e. road sides). The whole urban area, from the historical centre to the urban fringe, was included in the sample. A semi-natural area located just outside the city borders was also sampled. In order to analyse comparable vegetation types, woody vegetation has not been considered in either artificial or seminatural habitats. The plots were selected randomly on the basis of a grid imposed on maps of land use, either already available or drawn expressly. The size of 1 ha was previously chosen to represent a single urban habitat type (Celesti, Nazzaro, 1993).
During one vegetative year, the spontaneous vascular flora of each plot was recorded, including native and alien species and, among the latter, both naturalized and ephemerophytes (Table 1). Only spontaneously occurring plants were taken into consideration. The distributions and frequencies of the species in the fifty plots were recorded in a data base including several features such as chorological group, taxonomic position, life form and preferential habitat (from Pignatti, 1982). Information about origin, mode of introduction, degree of naturalization (ephemerophytes v. naturalized) and time of immigration (archaeophytes v. neophytes) of alien plants (Table 1) (Viegi, Cela-Renzoni & Garbari, 1974; Celesti Grapow, 1995) was described according to the system of Schröder (1969) (see also Trepl, 1990). The nomenclature follows Greuter, Burdet & Long (1984, 1986, 1989) and Tutin et al. (1968–1976, 1993).
The floristic similarity of the five cities was investigated by means of classification and ordination. Three data sets were examined. In the first set, the cities are described by the presence/absence of the 684 species (set A; 5×684); in the second (set B; 5×14), they are described by the following fourteen quantitative variables, chosen from the floristic features in the data base: number of species, percentage of native species, neophytes, therophytes, geophytes, chamaephytes, hemicryptophytes, phanerophytes, endemics, Eurasiatics, stenomediterranean, eurimediterranean, Atlantics and wide-distributed species, a heterogeneous group which includes cosmopolites, subcosmopolites and aliens (Pignatti, 1982). In the third data set, the fifty plots are described by the presence/absence of the species (set C; 50×684).
Each set was analysed by numerical classification and ordination using the SYNTAX package (Podani, 1994). The classification of the data was performed by Group Average Clustering, a hierarchical agglomerative procedure in which cluster to cluster distance (in this case chord distance) is the arithmetic average of all distances of objects (five cities, fifty plots) which do not occur together in the two clusters (Podani, 1994). In order to clarify the data structure a Minimum Spanning Tree was built from the distance matrix (chord distance). A Minimum Spanning Tree is a tree graph connecting all objects in such a way that there are no loops and the sum of the edges is minimal (Podani, 1994). The ordination of the data was performed by applying Principal Co-ordinates Analysis (PCoA, metric multidimensional scaling) a method designed to obtain coordinates of objects (fifty plots) from their distance matrices (in this case from chord distance matrix).
Frequency of the species
Of the 684 plant species found in the fifty plots, forty-two occur in all five cities (Table 2) yielding a similarity of only 6.1% of the total flora. Considering the proportion of species shared by four cities, the highest value (10.8%) is obtained between the floras of Ancona, Rome, Cagliari and Palermo, that is by excluding Milan, the furthest removed biogeographically. These percentages are lower than in Central Europe, where for instance Kunick (1982), in comparing the flora of nine cities, found a floristic similarity of about 15%.
The forty-two species listed in Table 2 are mostly annuals (twenty-eight therophytes) and generally belong to the native flora (thirty-three indigenous); aliens are mainly neophytes, all naturalized, and confined to man-made habitats. Ephemeral species rarely showed up in more than one city: being mostly escapees from cultivation and their presence being accounted for by historical and cultural rather than ecological factors. The most frequently occurring species in the fifty plots are also native: Parietaria judaica, Chenopodium album, Malva sylvestris and Plantago lanceolata. The most frequent aliens in the fifty plots were Aster squamatus, Conyza albida and Ailanthus altissima.
Species richness and alien flora
The total number of species found in each city, the number of species in each plot, and the proportion of the alien flora, are illustrated in Fig. 2. The lowest values of species richness (both in total and for each habitat type) were found in Milan, and the highest in Rome. Generally, the highest values were found in the archaeological sites, in the open fields on the outskirts, and in the parks, where plants from semi-natural vegetation find shelter. The lowest values were mainly found on compact and poor soils of residential areas and generally on heavily disturbed sites. Although the impact of human actvity is usually intense, old town centres generally show a high number of species, mainly growing on ancient walls.
The proportion of the alien flora varies between 11.9% (Ancona) and 25.6% (Milan). The higher proportion measured in Milan is due to the presence of several neophytes, mainly from North-America. Among the latter the most frequent are: Acer negundo, Ailanthus altissima, Ambrosia artemisiifolia, Bidens frondosa, Erigeron annuus, Lepidium virginicum, Oxalis fontana and Solidago gigantea. However, to this list should be added thirty-seven thermophilous species native to Italy but of a more southern origin than Milan, such as Inula graveolens, Inula viscosa, Lagurus ovatus, Lobularia maritima and Silybum marianum. If these plants are treated as aliens, the total percentage of the introduced flora reaches 42.8% in the continental city of Milan. The proportion of native and introduced species is not influenced by the presence of such ‘Italian aliens’ in the other four cities.
The relative percentage of the alien species recorded in the different habitat types (Fig. 2) varies between 4% (Palermo, semi-natural area) and 30% (Milan, old town centre). A general increase can be observed along the gradient of increasing urbanisation from semi-natural to disturbed plots (moving from left to right in the scheme), although it will be noticed that in the Mediterranean settlements (Rome, Cagliari and Palermo), even in the most heavily disturbed sites, aliens only reach 22.5% (Cagliari, old town centre).
Several neophytes native to the Americas are found among the alien flora (Table 3, A–B), the most numerous being Amaranthus, Conyza and Datura species, followed by species introduced from Asia and from other European regions. They are mostly naturalized, generally limited to man-made sites, and only a few invade natural vegetation. In Palermo and Cagliari are found a few species from Africa, such as Acacia karoo, Boherhaavia repens, Oxalis pes-caprae and Pennisetum setaceum. Of these, only Oxalis pes-caprae, an invader from the Mediterranean-type region of the Cape, is widespread and well established in both cities, although it is still confined to man-made sites.
Chorology and life forms
The chorological spectra of the total flora recorded in each city are shown in Table 3 (C). They are generally dominated by species with wide distribution (cosmopolites, subcosmopolites and alien species), varying from 24.5% in Palermo to 43.7% in Milan. Nonetheless, the flora of each city remains distinct and reflects the chorological features of the surrounding semi-natural landscape (Pignatti & Sauli, 1976). In the samples gathered in Cagliari and Palermo, stenomediterranean and southern-Mediterranean species predominate, such as Asparagus albus, Emex spinosa, Euphorbia dendroides and Thapsia garganica. In Rome, Ancona and Milan, the importance of the Eurasiatics, palaeotemperate and europeo-caucasics (e.g. Alliaria petiolata, Anthriscus sylvestris, Calystegia sepium, Clematis vitalba, and Sambucus nigra) increase progressively. In Milan, a conspicuous group of species of Mediterranean origin was found, mainly stenomediterranean and Mediterraneo-turanian.
The percentage distribution of the life forms also reflects the biogeographic gradient moving south from Milan to Palermo, as shown by the increase of the ratio of therophytes to hemicryptophytes (Table 3 D). Compared to the percentage values of the life forms measured in the surrounding regions (Pignatti, 1982), the cities analysed show a general increase of phanerophytes, due to several woody ornamental aliens escaped from cultivation, and of therophytes which are well-adapted to those features of the urban habitats such as drought, disturbance and instability.
The two classifications of the five cities based on the presence/absence of the species (set A) and on the fourteen quantitative characters (set B), produced the same result: the first division separates Cagliari and Palermo from the other cities, the second separates Milan, while the highest similarity is measured between Rome and Ancona. The dendrogram of the five cities obtained for the quantitative characters (set B) is shown in Fig. 3.
The classification of the fifty plots described by the presence/absence of the 684 species (set C), clearly shows that the floristic similarity between those plots located in the same city is higher than between those from similar habitat types (Fig. 4). Even the most heavily disturbed sites show a higher affinity with their respective semi-natural areas rather than with similar habitats (such as road sides, walls, trampled areas, and dumping sites) in other cities.
The plots from Ancona are divided into two groups, showing the steep phytoclimatic gradient measured inland from the coast: sites with a distinctive temperate phytoclimatic type (cluster on the left side) are separated from those close to the sea, which tend towards the mesomediterranean type (cluster on the right side).
Both the classification and the ordination (Fig. 5) suggest a clear separation into three geographic zones: the continental zone including Milan, the thermomediterranean zone including Cagliari and Palermo and, between them, the temperate and the meso-mediterranean zones including Ancona and Rome. The first axis of the PCoA ordination represents a latitudinal gradient from Cagliari and Palermo to Milan, while the different degrees of urbanization and the impact of human activities do not play an important role in the ordination of the plots.
Although the analyses generally show a separation of the three zones and of the five cities, there is evidence of similarity among the floras of historical centres, where plant communities growing on ancient walls and ruins have several species in common. The Minimum Spanning Tree of the fifty plots (Fig. 6) shows the close relation between the flora of the historical centres of Ancona, Rome and Milan (those plots highlighted with larger letters in Fig. 6). Nevertheless, few aliens are present, but many native species from natural vegetation of rocky habitats (e.g. Capparis spinosa and Sedum spp.), which constitute a peculiar characteristic of the plant cover of historical towns in Italy (Fig. 7). A high degree of similarity amongst the floras of old town centres in Europe has previously been reported by several authors (e.g. Saarisalo-Taubert, 1963; Aey, 1990; Brandes, 1995).
Although in the study areas human impact dates back thousands of years, geographical location still plays the major role in determining the constitution of the flora. The plant cover of each city, especially in the Mediterranean settlements, reflects the characteristics of the surrounding region (i.e. chorology, life forms, and biodiversity) even in the most heavily disturbed sites, and the proportion of species shared between different cities is relatively low.
A similar trend of species richness, decreasing from fields and archaeological sites to residential and heavily disturbed sites, characterizes each of the different habitat types (Fig. 2), and is mainly due to the depth and maturity of the soil and the length of time vegetation has been able to develop undisturbed. Nevertheless, it can generally be observed that even the species richness (highest in Rome, lowest in Milan) tends to be related to the floristic characteristics of the surrounding region (Pignatti, 1982) rather than to urban features such as the size of the city. The high number of species recorded in Rome, for instance, has been explained by reference to the environmental diversity of the Roman countryside (Blasi et al., 1995), together with the permanence of corridors, allowing the high connectivity of semi-natural plant communities and extended green areas within the city (such as archaeological sites), with the vegetation of the outskirts (Celesti Grapow, 1995).
One possible explanation for the lack of uniformity amongst the cities results from the low contribution of alien species, especially of neophytes, to the plant cover of the settlements analysed: in the total flora of the cities they number less than 20%. The most frequent species found in all cities are mainly local, and in the Mediterranean settlements, even in the heavily disturbed sites, the highest percentage of aliens is still under 23%. These values are much lower than those reported for Central European cities, where aliens make up about 40–50% of the total flora (PysË/link>;ek, 1989; Kowarik, 1995). Only in the continental city of Milan has such a high proportion been observed.
It might be argued that the higher representation of aliens in Central European urban floras reported in the literature is based on long term floristic research, leading to a higher number of aliens, mainly occurring only once or ephemerally, being added to the lists. Moreover, since a high proportion of alien species occur only infrequently in cities (Kowarik, 1995), the minimum area required to register all of them might be over the 10 ha sample used in this study. Nevertheless, the percentages obtained within our sample are very close to those obtained for the total flora of each of the cities (Hruska, 1989; Celesti Grapow, 1995). Besides, a lower than expected proportion of neophytes had already been noted in studies of Italian cities (Anzalone, 1951; Brandes, 1985; Arrigoni & Rizzotto, 1993–94; Celesti Grapow, 1995; Hruska, 1993–94).
The minor role played by alien species in the Mediterranean urban ecosystem may be explained by the pre-adaptation of some species of the Mediterranean basin to urban sites and more generally to man-induced disturbance (di Castri, 1991; Groves, 1991; Naveh & Vernet, 1991). In fact, some of these commonly occurring species (e.g. Cardaria draba, Carduus pycnocephalus, Chondrilla juncea, Echium plantagineum and Medicago spp.) are endowed with high invasive potential and are often invaders of other territories (Groves, 1986; di Castri, 1 990; Fox, 1990).
Beside the high level of disturbance, which is one of the principal factors responsible for the increase of alien species (not only in urban environments), in this paper we briefly consider two other features typical of the urban ecosystem which, according to Kowarik (1995), promote the enhancement of alien species in Central European cities: the availability of specific ‘urban’ niches, and the minor isolation of seed sources.
Many Mediterranean species are well-adapted to the conditions of those niches created inside cities, particularly related to environmental stresses such as limited water supply, high temperatures and high luminosity. High temperature requirements and an ability to grow in dry sites are in fact among the most common features of successful invaders in man-made sites (PysË/link>;ek, Prach & SËmilauer, 1995). Analysis of the distribution of the vascular flora in the city of Rome showed that there is no group of thermophilous species restricted to the city centre, as happens in several Central European cities (Gödde & Wittig, 1982–83; Wittig, Diesing & Gödde, 1985). Although Milan shows a conspicuous number of species with thermophilous requirements, whose actual expansion is related to the ‘urban heat island’ (Frattini, 1993–94), a consistent entry of plants of more southern origin, such as African species, was not found in the Mediterranean cities. Hence, it may be hypothesized that the effect of the ‘urban heat island’ does not play such an important role in the distribution of vascular plant species in the Mediterranean region as in the continental region and in Central Europe.
The constant arrival of seeds of exotic species in cities allows the maintenance of established populations, and more generally the lack of isolation from seed sources is considered one of the most important factors which conditions the distribution of plant species in urban environments (Trepl, 1995). A common characteristic of the Mediterranean cities analysed is the structure of the urban mosaic, which allows connections between the green areas inside the cities and the plant communities of the surrounding areas. In Rome, for instance, the distribution of some species (e.g. Capparis spinosa, Centranthus ruber, and Quercus suber) shows the presence of such connections by the numerous remnants of semi-natural vegetation (Celesti Grapow, 1995). The minor isolation from seed sources may contribute to the establishment of this unusual spatial pattern, with local plant populations occurring even in the city centres and in heavily disturbed sites.
The results of the comparison of five Italian cities suggest that the ‘uniformity hypothesis’ is not valid for the Italian urban ecosystem, where several phytoclimatic regions are represented and the flora is mainly composed of local species. From this study it may be concluded that in the Mediterranean urban ecosystem, at least as far as Italian territory is concerned, alien species do not play an important role, while the apophytes are the most successful group of the spontaneous flora.