Urbanization effects on carabid diversity in boreal forests

Carabid abundance, species richness and diversity were compared along an urban-rural gradient in Helsinki, Finland. Increased urbanization was found to result in significant reductions in species richness, though the reductions in abundance and diversity were not statistically significant. Forest habitat-specialist species were scarce in rural sites and virtually absent from urban and suburban sites. There was no evidence of higher diversity at intermediate disturbance levels (suburban sites), as predicted by the intermediate disturbance hypothesis. Species with flight ability and the ability to utilize open habitat were more predominant in urban and suburban sites. Flightless species were more predominant in rural and suburban sites. Carabid abundance data were suffi­ cient to reveal the negative impact of urbanization, so similar studies could be conducted in regions where carabid taxonomy is poorly known. Species composition patterns do, however, provide invaluable information. To conclude, if biodiversity is to be main­ tained in urban areas, priority must be given to the provision of those habitat features which are essential for sensitive species, such as decaying wood and wet microhabitats. These must be incorporated into urban green networks in particular, if biodiversity and species other than common generalists are to benefit from them.


INTRODUCTION
Anthropogenic modification of landscapes and associ ated landscape change which pertains to a city or town, i.e. urbanization, causes major changes to the ecosystem.Differences in a number of variables, such as tempera ture (Bornstein, 1968), acidity, soil hydrophobicity, utilizable carbon and nitrogen levels (McDonnell et al., 1997), deposition of heavy metals (Hynninen, 1986) and other pollutants (Herrmann & Hübner, 1984;Väisänen, 1986), changes in fungal biomass, bacterial flora, leaf-litter decomposition rates (McDonnell et al., 1997), fragmenta tion and edge effect (Löfström et al., 1999), all contribute to the effect of urbanization.In addition, the highly sig nificant physical covering of potential habitat with asphalt and cement causes direct loss ofhabitat.
This urbanization effect tends to result in a highly affected core at the heart of the city, surrounded by irregular rings of decreasing urban developement (Dick inson, 1996).The core area of the city often contains patches of natural land which are similar in physiognomic appearance to their rural counterparts but are generally more fragmented, polluted, warmer and drier.Further more, core areas are most affected by intense recreational use and often contain exotic species (Gilbert, 1989;Kostel-Hughes, 1995;Kostel-Hughes et al., 1996).In response to this long and probably incomplete catalogue of changes affecting urban green space, it would be sur prising if there were no concomitant effects on species composition.
On a global scale, urbanization is resulting in a vast amount of habitat loss and modification, particularly in regions where there is pressure from a rapidly growing human population.In Finland, as in much of Europe, although the human population is not increasing signifi cantly, there is an ongoing migration from rural areas to the cities (Wahlstrom et al., 1996).This results in increased construction, traffic and pollution, thus increasing the impact on natural areas within and adjacent to these cities.Intense recreational pressure also often has a major impact on such natural areas, resulting in envi ronmental changes (Lehvavirta, 1999).The aesthetic and recreational value of urban green areas is generally appre ciated by urban planners and many sites are also pre served for their conservation value.Not surprisingly though, forest patches within urbanized areas are consid erably different from their more rural counterparts.
According to Connell's (1978) intermediate disturbance hypothesis, highly disturbed sites, and undisturbed sites, have lower diversity than sites with intermediate levels of disturbance.This hypothesis has been found applicable to a number of different ecosystems (Marone, 1992;Aronson & Brecht, 1996), though other studies have shown that it is not universally applicable (Schwilk et al., 1997;Wiegand et al., 1997).
The aim of this study is to investigate the effects of urbanization upon carabid diversity along an urban-rural gradient and, in particular, to answer the following ques tions: a) are there any significant differences in the carabid assemblages of urban, suburban and rural sites, particularly regarding species richness, abundance and community composition, b) how do the different carabid Appendix 1. Features of the 12 forest sites, and their surroundings, which could affect carabid population dynamics.Distance is measured from an arbitrary point in the city centre.Width of forest indicates those sites which are most affected by edge effects, as some large patches of forest (e.g.U2 and U3) were in the form of long, narrow strips.Trampling is the total area of paths or heavily trampled ground contained in three 100 m2 squares (i.e. per 300 m2  1978) hold true for carabid com munities along an urban-rural gradient?Carabid beetles (Coleoptera: Carabidae) were selected because they are easy to collect and preserve, taxonomically and ecologically well known, include sensitive spe cialist and less sensitive generalist species and have been recommended by a number of authors (Thiele, 1979;Blake et al., 1994;Langor et al., 1994).
This study is part of the "Globenet" programme (Global Network for Monitoring Landscape Change), which is an international initiative to develop appropriate mechanisms for the global investigation of the impact of urbanization on ecosystems (see Niemela et al., 2000 and http://www.helsinki.fi/science/globenet).

Study sites
The urban-rural gradient comprised of a transect from central Helsinki (60°10'N, 24°56'E) to northern Espoo (60°17'N, 24°38'E) in southern Finland (hemiboreal vegetation zone, Ahti et al., 1968), covering a distance of approximately 20 km.Twelve sites along the gradient were selected for this study: Four urban, four suburban and four rural.Urban sites were located within the core of the city of Helsinki, with busy roads, densely populated and industrial sites adjacent.Rural sites were in areas with minimal human population, only light traffic and no nearby industrial sites, and suburban sites were intermediate between these two extremes (Appendix 1).The forest stands in which the study sites were located were spruce dominated (Picea abies L.), with Myrtillus -type vegetation (Cajander, 1949).The age of the dominant spruce trees was estimated to be in excess of 80 years.Other site characteristics are shown in Appendix 1.

Pitfall trapping
Whilst there are complications to be considered, pitfall trapping is the best means currently available for sampling carabid communities (Greenslade, 1964;Thiele, 1979;Spence & Niemela, 1994).We placed ten pitfall traps (mouth diameter = 65 mm) at least 10 m apart in an irregular line at each site.The traps were sunk into the ground, their mouths flush with the surface.Approximately 40 ml of a propylene-glycol solution (1:1 aqueous) was used in order to kill and preserve the trapped arthropods.Plastic covers were placed a few centimeters over the traps to avoid dilution of the glycol by rainfall.
The traps were emptied on a monthly basis during one growing season (May-September 1998).Carabids were identi fied to species using Lindroth (1985Lindroth ( , 1986) ) and spiders (Araneae) were reserved for possible use in subsequent studies.

Environmental factors
We assessed the amount of trampling at each site as a crude measure of disturbance on the forest floor by taking three, 100 m2 squares along each transect.The area of paths and any other eroded patches which lie within each of these squares, was recorded.
Other environmental factors such as size of forest patch, amount of edge, distance from extensive forest, area of adjacent fields/open land and extent of urbanized land were assessed using the Greater Helsinki recreational map (see Appendix 1).Fig. 1.Carabid species richness and abundance along the urban-rural gradient.Urban (U1-4), suburban (S1-4) and rural (R1-4) sites are indicated along the horizontal axis.

Data analyses
A nested ANOVA was used to analyze data of carabid abun dance and species richness per trap and per site.The hypothesis that urbanization has an effect upon carabid abundance and/or species richness was tested across the three levels of urbaniza tion: urban, suburban and rural.Data were log (X + 1) trans formed to achieve approximate normality for both abundance and species richness.Pearson's correlation coefficient was used to examine the degree of correlation between abundance and species richness.In addition, we calculated the Brillouin diver sity index for each site.
Table 1.Nested ANOVA table showing the effect of urbani zation on (a) carabid abundance, (b) species richness and (c) abundance with site R1 removed.Abundance and species rich ness data were log (x + 1) transformed to comply with para metric test assumptions.The analyses were performed at the trap level with the 12 sites (11 in 'c') nested in the gradient fac tor.Var -estimated variance component, % -variance compo nent expressed as a percentage (see Sokal & Rohlf, 1995).Ordination by Principal Component Analysis (PCA) was used to compare the similarity of the carabid assemblage at each site and to examine the significance of different species characteris tics.Species that were collected from one site only and only one individual, were excluded from this analysis.
The species characteristics used were flight ability and the ability to utilize open habitat, which were taken from Lindroth (1985Lindroth ( , 1986Lindroth ( and 1992)), Thiele (1977) and Kinnunen (1999) (see Appendix 3).Species were considered to be capable of dispersal by flight if either of the former authorities had confirmed, or strongly suggested, ability to fly and, in the case of dimorphic or polymorphic species, if part of the population in this region was known to be capable of flight (Lindroth, 1992).Some macropterous species are therefore considered as flightless.RESULTS

Response of carabid abundance and species richness to urbanization intensity
A total of 2203 individuals and 25 species were col lected (Appendix 2).Most of the abundant species, in particular Pterostichus melanarius, P. niger, P. oblongopunctatus and Carabus hortensis, were markedly less abundant in the urban sites than the rural sites.One of the most abundant species, Calathus micropterus, however, appeared to be equally abundant across the three regions.
The nested ANOVA revealed a highly significant effect of urbanization along the gradient for species richness (Table 1).For abundance the effect was just outside the 5% confidence interval and abundance was highly corre lated with species richness (Pearson correlation coeffi cient: r = 0.848).Both abundance and species richness were low in the urban sites and high in the rural sites, par ticularly site R1 (Fig. 1).These ANOVA models explained a high proportion of the variance in the data.Because rural site R1 had a conspicuously high value for abundance (641 individuals), the ANOVA was repeated excluding this site, though this had a negligible effect on the result.
Diversity was lowest at urban site U4 and highest at rural site R3.Mean diversity was highest in the rural and lowest in the urban area (Table 2), and a Kruskal-Wallis ANOVA revealed a marginal trend of decreasing diver sity with increasing urbanization intensity (Kruskal-Wallis test statistic = 4.769, p = 0.092).
The PCA of the carabid data (Fig. 2) showed distinct clustering of the sites, with clear separation according to urbanization intensity.The cumulative percentage of vari ance explained by the first two axes of the ordination were 39.3% and 56.1% respectively.Separation of the sites occurred in three different directions within two dimensional space.The urban sites separated negatively along the X-axis, the suburban sites positively along the Y-axis and the rural sites towards positive X and negative Y.
The incorporation of relevant species characteristics into the PCA (Fig. 2) showed that species which are able to fly and utilize open habitat (e.g.Amara brunnea, Leistus ferrugineus and Pterostichus strenuus), were rela tively more abundant in urban and suburban sites, whereas flightless species (e.g.Pterostichus niger, Carabus nemoralis, C. hortensis), were restricted to rural and suburban sites.Open habitat species were predomi nant in the urban sites.

Effects of urbanization upon carabid communities
Our results indicate that urbanization has a negative effect upon carabid species richness and abundance.Both species richness and abundance were markedly reduced in the urban sites compared to the other two levels, although the abundance effect along the gradient was not statisti cally significant at the 5% risk level.Mean diversity also decreased with increasing urbanization (rural > suburban > urban).Our hypothesis that urbanization has a strong, negative effect upon carabid communities in boreal for ests is thus supported.
There are several possible explanations for the lower carabid diversity in the urban sites.McDonnell et al. (l997) have shown that leaf-litter decomposition pro ceeded significantly faster at urban sites and that carbon and nitrogen processing proceeded to very stable forms, from which they were unavailable for re-utilization.This result implies that the cycling of the two most important nutrients in ecosystems, upon which whole food chains depend, could have been replaced by sinks which, whilst not destroying them, effectively remove these nutrients from the food chain.This could account for the dramatic reduction in carabid species richness and abundance in urban sites reported in this study.

Species characteristics and the urban habitat
Consideration of individual species trends supports the suggestions that: 1. Urbanization has a negative impact upon the abun dance of even the most abundant species, with the single exception of Calathus micropterus 2. Forest specialist species do not thrive in areas which are affected even by quite moderate levels of urbaniza tion, and 3.The species which are more abundant in urban sites are generalist species and species of open habitat (Leistus ferrugineus and Patrobus atrorufus), capitalizing on the altered conditions of this disturbed habitat.A major component of anthropogenic impact upon boreal forests is habitat fragmentation (Andrén, 1997;Hanski, 1999).The forest cover of the study region has become highly fragmented, particularly during the latter Appendix 3. Carabid species characteristics which could affect dispersal ability and recolonization prospects.Details of wing mor phology are shown, as definite knowledge of flight ability is unavailable for many species.Fields refers to the ability of species to utilize agricultural land, dry/wet indicates hygrophilia.U/R shows preference for rural or urban areas, with (U) indicating possible preference for the urban environment (Thiele, 1977, Lindroth, 1985, 1986, 1992;Kinnunen, 1999) half of the 20th century (Wuorenrinne, 1978).The urban and suburban areas of the gradient contain a mosaic of mostly small patches of forest, parks, wasteland and asphalt, the proportion of asphalt and the small size of the forest patches being more marked in more urban areas.The size, width and connectivity data in Appendix 1 give an indication of how much each site is affected by frag mentation and edge effects.Although sites U2 and U3, for example, are situated in a 440 ha patch of forest, the portion of it which lies within the urban region and con tains these sites, is in the form of a narrow strip, c. 300-500 m in width.These sites were actually quite spe cies rich, though this was predictably due to the presence of open habitat generalist species and species which are atypical of the spruce forest habitat (Halme & Niemela, 1993).Edge effects and intense recreational pressure thus appear to counteract the advantages conferred upon these sites by their large size and connectivity with extensive forest.
Communities generally contain a few very abundant species and a comparatively large number of scarce spe cies (Magurran, 1988;Niemela, 1993).Generalist species usually fall into the former of these groups and specialists mostly into the latter.Thus, even if several specialist spe cies were lost from a community, this would be unlikely to have any discernable effect upon carabid abundance, though species richness should decrease.Despite this, however, a loss of abundance was also evident, although not significantly so.Most of the highly abundant gener alist species, e.g.Pterostichus melanarius, P. niger, P. oblongopunctatus and Carabus hortensis, which are readily able to utilize even highly disturbed habitat, were less abundant in the urban sites.Thus we can surmise that either urbanization is resulting in a significant loss of resources available or that it is otherwise making the environment more harsh.
Wet forest habitats, decaying wood and extensive forest cover are scarce in urban areas (Karjalainen, 1991), though can still be found in the rural region.The rarest habitat specialist species caught in this study, Agonum mannerheimii, as well as the scarce A. fuliginosum, A. thoreyi and Trechus rubens, are all dependent on such features and were only found from the rural region.A. mannerheimii, in particular, is in decline within the Fennoscandian region (Niemela et al., 1987;Lindroth, 1992, p. 611) and is clearly unlikely to benefit from the protec tion of urban or suburban sites, unless they possess such features as are required by sensitive habitat specialists.Even less sensitive forest specialist species, such as Carabus glabratus, are also quite rare in the more urban ized sites.This supports the suggestion that urban forests are inhospitable places for the forest carabids of the region.
One of the questions which this study set out to answer was whether Connell's (1978) hypothesis of greater diversity at intermediate levels of disturbance applies to the case of carabid diversity along an urbanization gradi ent.In this study, neither the Brillouin diversity index nor the species richness values support the suggestion of higher diversity in the intermediately disturbed suburban sites.This could be due to the intensity of the negative effect of urbanization, which has been demonstrated in this study.
The urban and suburban areas also contained a couple of species for which there were equally few recordings, Amara eurynota and Synuchus vivalis.These, however, are not regionally threatened species but are species which are atypical of the hemi-boreal, spruce-forest habitat (Lindroth, 1985(Lindroth, , 1986)).They are species which generally prefer drier, more open habitat (Appendix 3), apparently taking advantage of the altered conditions of the more urban forest sites.Such areas are thus providing an opportunity for some species to gain access to a new kind of habitat (Eversham et al., 1996) but this is always likely to be at the cost of the most sensitive species, such as Agonum mannerheimii and a large number of similarly sensitive species which were not captured in this study (Halme &Niemela, 1993).
The ability of species to utilize open countryside and to fly might significantly affect different species' persistence prospects in the more urbanized areas.Indeed, the PCA shows that flight-capable species which were able to utilize open habitat (e.g.Amara brunnea, Leistus ferrugineus) were predominant in urban and suburban areas.These traits are obviously of benefit to species inhabiting areas where the forest cover is more fragmented and interspersed with parks, streets and buildings.Flightless species, on the other hand, were more predominant in rural and suburban sites, as their dispersal between the more isolated urban habitat patches will be handicapped.Accordingly, all of the species which were predominant in urban sites were also those which utilize open habitat.Thus, consideration of species characteristics suggests that flight ability and the ability to utilize open habitat, which are commonly traits of generalist species, confer an important advantage in more urbanized habitats.Flight less species in particular are thus more common in non urban sites.

Implications for the management of urban forests
It is quite evident that altered environmental conditions are resulting in assemblage changes in urban forests, as exemplified here for carabid beetles.There is also a growing awareness of the need for appropriate strategies for the management of urban green areas, with considera tion of specific objectives such as aesthetic, recreational and economic interests, as well as sustainable developement criteria, such as biodiversity and the amelioration of pollution (Gilbert, 1989).This study illustrates the need for the implementation of, and evaluation of, potential measures to improve the biodiversity of urban green space.Measures to improve the quality and connectivity of urban green spaces must be planned in accordance with the needs of the most sensitive species, rather than being content with connectivity elements in the form of strips of lawn or trees.Many of the habitat features of which these sensitive species require in forested regions are well known, such as dead and decaying trees and woody material, as well as wet forest, mature deciduous trees and natural water-channels (Karjalainen, 1991).Attempts are being made to include these into urban green networks but they must also be preserved in, and introduced into, the corridors which connect urban green spaces (Bueno et al., 1995).Without urban green corri dors which are planned for and function effectively for sensitive and specialist species, there is little possibility of improving the biodiversity of urban green space.How ever, as Niemela (2001) points out, the functioning of ecological corridors is a question of considerable contro versy and it is evident that more research is needed on the usefulness of ecological corridors in the urban setting.

Fig. 2 .
Fig. 2. PCA ordination of carabid species distributions with regard to site.Pterostichus niger favours rural, Amara brunnea sub urban and Leistus ferrugineus urban sites.Species which are close to the origin are less affected by urbanization intensity.The carabid species are marked to indicate those which are capable (fly) or incapable (flightless) of flight and those which are (open) or are not (close) able to utilize open habitat.Species which are able to fly and utilize open habitat favour the suburban and urban sites, whereas most flightless species, open habitat species in particular, preferred rural and suburban sites.All of the species which were predominant in urban sites were open habitat species.The key to the abbreviations used for the scientific names is in Appendix 2. Urban, suburban and rural (U, S and R) sites are marked with a cross (X).
). Connectivity indicates if and how the sites are connected to extensive areas of forest cover.Adjacent open vegetation indicates whether there are significant, large patches of open land adjacent which may be advantageous to those species which also thrive in such habitat.Urban open land is heavily disturbed open land, which provides a similar resource for just a few of the hardiest generalist species.Nuuksio is an area of c. 70 000 ha of forest, part of which is a national park.U, S and R indicate urban, suburban and rural sites 1-4.

Table 2 .
Brillouin diversity index for each site, including the mean values for each area. .