Urbanization, industrialization, and agriculturaliza-tion can cause natural habitats to be destroyed, fragment-ed, or severely modified. This land-use conversion is the primary factor explaining biodiversity loss (Pearce and Moran 1994, Niemela et al. 2000) because many species are restricted to small areas or certain types of habitats. They may even be forced to be separated from their pre-ferred habitats (Pyle et al. 1981, Wilcox and Murphy 1985, Desender and Turin 1989).

The urban ecosystem should have valuable green and open areas for human well-being and, thus, often con-tains gardens, parks, and woods in addition to man-made structures such as roads, commercial, and residential constructions. For animals, including arthropods, green areas in the urban landscape are important for moving around an urban area (Angold et al. 2006). However, al-though the urban green area provides an array of habi-tats for arthropods (Eversham et al. 1996, McIntyre 2000), urbanization is a leading cause of decline in biodiversity and the abundance of organisms (Pyle et al. 1981, Clark and Samways 1997, Angold et al. 2006) because a man-made environment in a city may be a dispersal barrier to less mobile arthropod species.

Ground beetles, generally consuming insects and other small arthropod species, have been extensively studied because they occur in most terrestrial habitats; their abundance and species composition can be easily monitored by pitfall traps (Lovei and Sunderland 1996). In some studies, the diversity of ground beetles is often higher in urban areas than in suburban ones (Magura et al. 2004), and forest fragmentation can lead to an increase in species richness (Halme and Niemela 1993, Niemela 2001, de Warnaffe and Lebrun 2004) because ground beetles in urban ecosystems include open-habitat spe-cies, while there is a decrease in forest specialists across the landscape, from a forest area to an urban one (Ishi-tani et al. 2003, Fujita et al. 2008). According to Niemela et al. (2000), distribution patterns of ground beetles across the urban-rural gradients could prove useful in measur-ing the urbanization effects on biota. Therefore, ground beetles have often been studied as bioindicators for frag-mentation effects along urban-rural forest gradients in many urban landscapes (Alaruikka et al. 2002, Niemela et al. 2002, Ishitani et al. 2003, Venn et al. 2003, Magura et al. 2004, 2008a, 2008b, Weller and Ganzhorn 2004, Deichsel 2006, Elek and Lovei 2007, Gaublomme et al. 2008). How-ever, Fujita et al. (2008) argued that more studies regard-ing various habitat types in urban and rural landscape are necessary.

The objective of this study was to clarify the effects of land-use types on the species richness, abundance, and composition, regarding ground beetles in urban land-scapes. We selected the potential bioindicators classifying land-use types.

The study area was located in the southwestern part of Korea, Jeonju and Iseo-myeon of Wanju-gun (35°43′-35°53′ N, 126°59′-127°14′ E) (Fig. 1). The area covered 346.7 km², where 262.3 km² of the land cover was green space in the form of parks and forests; theurban area covered 42.2 km². The border of the city included approximately 6 km² of arable land and 0.3 km² of wilderness area. The annual mean temperature and precipitation in the study area were 13.0℃ and 1,296.2 mm, respectively.

Eleven study sites were selected according to land-us-es, and these were categorized into 4 habitat types: 4 for-est areas (F_), 3 agricultural areas (A_), 2 urban roadsides (U_), and 2 riversides (R_). Table 1 shows environmental information for each sampling area, including habitat

types, surveyed altitudes (ALT), and distances from the nearest forest margin (DIST). Fujita et al. (2008) showed that ALT and DIST were significant variables in ground beetle assemblage. Proportions of each land-use type around sampling sites are shown in Table 2. The criteria for distinguishing sampling areas were the proportion of the land-use types, based on an aerial photograph within 1 km² quadrat, which was measured by the GIS database in the Environmental Geographic Information System (EGIS 2011). Choi et al. (2008) suggested that the biotope classification system consists of a 4 step system, includ-ing the biotope class (large), biotope group (medium), biotope type (small) and sub-biotope type (detail). We fol-lowed Choi et al. (2008) for the criteria of land-use types for land-use analysis in the present study. However, some of their criteria (residential, commercial, industrial, pub-lic facility and traffic facility areas) were merged into the proportion of built-up areas (PBA), because the built-up areas do not serve the habitats of ground beetles.

Sampling was conducted from July to September in 2008. Ground beetles were collected using pitfall traps. Three pitfall traps were installed 10 m apart in each site and were emptied every month. Each pitfall trap was in-stalled at the center of a sampling site, at least 20 m from the nearest habitat edge. A pitfall trap was composed of a plastic container (10.5 cm diameter and 8 cm depth) and a lid with 6 holes (2 cm diameter in each hole), which prevented the collecting of unwanted small mammals. A plastic roof was erected in order to prevent rainfall. Traps were filled with a liquid mixture (300 mL, 95% ethyl-al-cohol:95% ethylene-glycol = 1:1) for sample preservation. Collected ground beetles were brought to the laboratory and dried, mounted, and identified to species level under a dissecting microscope. Identification was performed according to Habu (1967, 1973, 1978), Kwon and Lee (1984), Park and Paik (2001) and Sasakawa et al. (2006), and compared to voucher specimens in the laboratory. Nomenclature was confirmed by Park and Paik (2001)

and Park (2004). The habitat preferences of species were determined by general characteristics of subfamily level, and some species were confirmed by Fujita et al. (2008). Voucher specimens were deposited in the insect ecology laboratory, Seoul National University.

Abundance and species richness were measured based on the number of individuals and the number of species collected in each sampling site, respectively. Stepwise multiple linear regressions were used in order to study the relationships between the 9 variables, species richness, and abundance of ground beetle assemblages.

For analysis of the species composition of ground bee-tles among sites, a non-metric multidimensional scaling (MDS), with Bray-Curtis similarity and a two-way indica-tor species analysis (TWINSPAN) (Hill 1979), were con-ducted using the PRIMER v5.0 (Clarke and Gorley 2001, Clarke and Warwick 2001) and community analysis pack-age v2.0 (Seaby and Henderson 2002), respectively. MDS was chosen because it performs well with ecological data that do not meet the assumption of normality (McCune and Grace 2002). On the other hand, a TWINSPAN not only classifies the sites but constructs an ordered two-way table from a sites-by-species matrix (Jongman et al. 1995). TWINSPAN creates pseudo-species, and each species is subdivided into presence/absence vectors for several relative abundance levels (Dufrene and Legendre 1997). Thus, MDS and TWINSPAN are the most generally effective ordination and classification methods for eco-logical community data (Jongman et al. 1995, McCune and Grace 2002). In the MDS, stress is a measure of dis-tortion between the positions of real data points from the graphical representation. Thus, low stress represents few distortions from the real position of the data points and is associated with a graph that more accurately represents dissimilarities in species composition.

The analysis of similarity (ANOSIM) and a multiple re-sponse permutation procedure (MRPP) were conducted to further confirm significant differences in community structure among land-use types and between groups of MDS. The ANOSIM permutation test, with a maximum of 999 permutations, was used to assess significant dif-ferences among groups, and Global R value approaches 1 if differences among land-use types exist (Clarke and Warwick 2001). Similarly, the MRPP provides a measure of within-habitat homogeneity (A), which increases as the communities in different habitats deviate, to a max-imum of 1. An A value greater than 0.3 is considered to be relatively high, but statistical significance may occur for a small A value if the sample size is large (McCune and Grace 2002). The ANOSIM was conducted using the PRIMER v5.0 (Clarke and Gorley 2001, Clarke and War-wick 2001), MRPP, Pearson’s correlation, and stepwise multiple regression analyses were conducted using the statistical software package R (R Development Core Team 2010).

A total of 29 species, belonging to 18 genera of 11 sub-

families, were identified from 915 ground beetles in this study (Appendix 1). Most of the collected ground beetles were macropterous (884 individuals of 25 species) and only four brachypterous species were collected (Appen-dix 1). The species richness was relatively higher in the FE and RM, and abundance was extremely high in the AS (Fig. 2). Overall, land-use types in the urban landscape did not appear to cause a significant decrease in species richness and abundance of ground beetle assemblage. However, species richness and abundance of forest spe-cies were higher in forest areas than in open-habitat ones, but they were lower in non-forest areas.

The dominant species was Pheropsophus jessoensis, comprising 32.7% of all ground beetles captured. Harpa-lus (Psedoophonus) sp. was the most second dominant species, comprising 16.7%. Synuchus nitidus and Synu-chus sp.1 made up 11.0% each. Thus, these four dominant species accounted for 71.2% of total abundance.

The Pearson’s correlation matrix among variables is il-lustrated in Table 3. The PBA was significantly correlated with other variables except on the paddy field area. Step-wise multiple regression analyses between ground bee-tle assemblages and variables are presented in Table 4.

Overall, many variables significantly affected the species richness (r² = 0.9654, F = 47.57, P = 0.001) and abundance (r² = 0.9737, F = 62.80, P < 0.001) of forest species, but not the total ground beetles and open-habitat species.

In the MDS, eleven study sites were clustered into 2 major groups (i.e., forest areas vs. non-forest areas) and 5 subgroups at species level (Fig. 3a). On the other hand, eleven sites were clustered into 2 major groups and 3 sub-groups at subfamily level (Fig. 3b). The axis 1 in both the MDS of species and subfamily level may represent the gradient from forest to non-forest areas. The ANOSIM and MRPP showed that similarity among land-use types and

between 2 major groups, forest and non-forest areas, in MDS were significantly different at both the species and subfamily level (Table 5).

According to the TWINSPAN, all study sites were di-vided into 2 groups at species level (Fig. 4a), forest and non-forest groups by the first indicator species, Doli-chus halensis halensis for non-forest. The second indica-tor species of forest and non-forest areas were Harpalus discrepans and Chlaenius costiger, respectively. However, classification at subfamily level showed relatively unclear rather than analysis at species level (Fig. 4b).

Our results showed that land-use types in urban land-scapes did not appear to cause significant decrease in overall species richness and abundance of ground beetle assemblage. In the gradient of land-use disturbance, pre-vious studies (Vanbergen et al. 2005, da Silva et al. 2008) revealed that more disturbed areas, such as agricultural areas, had high species richness and a greater abundance of ground beetles. In our study, most non-forest areas were highly disturbed by human activity, such as agricul-ture, construction, or habitat management. These distur-bances may have lead to relatively low species richness and abundance of ground beetles in non-forest areas, compared to the results of Vanbergen et al. (2005) and da Silva et al. (2008). Park (2010), studying the urban park, noted that the management strategy of the inhabitants can also affect insect diversity, including ground beetles.

Instead of the taxonomic approach, however, the use of habitat preferences of ground beetles might become a more comprehensive tool with which to assess and monitor biodiversity (Niemela 2000, da Silva et al. 2008). Our results can also be explained by the predominance of some forest and open-habitat species. Forest species, such as large (Aulonocarabus semiopacus and Coptola-brus jankowskii jankowskii), middle, or small species (Synuchus spp.) are generally found in natural forest areas but not in non-forest ones. This is consistent with find-ings from recent studies (Fujita et al. 2008, Gaublomme et al. 2008). This may explain that the abundance of these species tended to decrease from forest areas to urban ones. In addition, ground beetles in a forest environ-ment are more influenced by habitat complexity due to the restriction of flight capability (Darlington 1943, Kava-naugh 1985, Gobbi et al. 2006), and many forest species are predominantly composed of brachypterous species (Darlington 1943). In particular, Fujita et al. (2008) dis-cussed that large and/or flightless species are vulnerable to natural and human disturbances. Unlike forest species, most open-habitat species, such as species in subfamilies Harpalinae, Zabrinae, Brachininae and Callistinae, were abundant in non-forest areas. In the Korean agricultural landscapes, subfamilies Harpalinae, Brachininae, and Callistinae are known to be dominant groups (Choi et al. 2004, Kang et al. 2009), and our results also showed a similar species composition in non-forest areas where D. h. halensis, Harpalus (Psedoophonus) sp., Pheropsophus javanus, and P. jessoensis were dominant. Gray (1989) hypothesized that opportunistic species, in other words, generalist species or open-habitat species, should gain dominance in disturbed habitats, although overall diver-sity should decrease. Therefore, differences between for-est and non-forest areas may be explained by the habitat preferences of ground beetles. Consequently, these eco-logical characteristics of ground beetles may be impor-tant to understanding their biodiversity patterns and spe-cies compositions.

From these findings, we speculate that the species richness and abundance of ground beetles according to habitat preferences, especially forest species, would be explained by specific patterns, with several variables such as PBA, DIST, crop, and natural forests. These find-ings mean that forest species should be considered in ac-curately detecting the diversity pattern along the urban-rural gradient, but open-habitat species or total ground beetles may not appropriate to detecting the diversity pattern along the urban-rural gradient. In many previous studies, such as the GLOBENET project (Alaruikka et al. 2002, Ishitani et al. 2003, Magura et al. 2004, 2008a, 2008b, Deichsel 2006, Elek and Lovei 2007, Gaublomme et al. 2008), species richness and the abundance of forest spe-cialists showed a negative relationship along the urban-rural forest gradient. Unlike the GLOBENET project, we focused on the effects of change in ground beetle assem-blage according to various land-use types and variables; we did not focus on the urban-rural forest gradient. Thus, species richness and abundance of forest species in our studies were significantly decreased with variables for ur-banization or environmental status. However, many vari-ables in our studies generally represented the proportion of the disturbed area. In addition, variables in our studies do not directly represent habitat conditions. For example, PBA and DIST were not surrogates for other environ-mental variables, such as temperature, humidity, canopy covers, and leaf litter. Therefore, further studies on the relationships between ground beetles and environmental variables are necessary.

The ordination and classification at species level showed that several land-use types are clustered into 2 major groups of forest and non-forest areas, while analy-ses at subfamily levels are relatively less appropriate for classifying land-use types. By TWINSPAN, some species (D. h. halensis, H. discrepans, C. costiger, and Chlaenius naeviger) and subfamilies (Carabinae, Licininae, Scariti-nae, Harpalinae) can be used as candidates for indicators. In general, D. h. halensis, H. discrepans, C. costiger and C. naeviger are known to be open-habitat species. Carabi-nae and Licininae species are generally forest specialists, while Scaritinae and Harpalinae species are generalists. However, there is little information about their ecologi-cal roles in ecosystems because only a few references can be applied to Korean ground beetles. Thus, we still need more information in order to determine the indicators for classifying land-use types. For example, Magura (2002) studied the spatial distribution of ground beetles and the effects of edge on their diversity; the study indicated the habitat preferences of each ground beetle species.

In general, an urban landscape can provide an impor-tant habitat for the maintenance of biodiversity due to mosaic landscapes. Recent studies regarding urban land-scape, using ground beetles, have focused on examining the connectivity and change of diversity of ground beetles among habitat patches along urban-rural forest gradients (Alaruikka et al. 2002, Niemela et al. 2002, Ishitani et al. 2003, Venn et al. 2003, Magura et al. 2004, 2008a, 2008b, Weller and Ganzhorn 2004, Deichsel 2006, Elek and Lovei 2007, Gaublomme et al. 2008). However, Sadler et al. (2006) and Fujita et al. (2008) suggested that continuous forests do not necessary serve as a “mainland” for forest specialists. However, Fujita et al. (2008) concluded that every urban habitat acts as a temporary reservoir of spe-cies possibly. In the present study, although most forest specialists were not found in non-forest areas, some for-est species, including C. j. jankowskii, Pterostichus (Nia-loe) sp., Pterostichus sulcitarsis, and Synuchus sp.1, were collected in non-forest areas such as AU, RM and UU. In-deed, AU is located within mountain areas, but RM and UU are located in urban environments, which are man-made green areas or habitat patches. This finding indi-cates that several habitat patches in the urban landscape may operate as “stepping-stones” for small-scale disper-sion. However, interpretation of connectivity among vari-ous habitat types regarding results from anthropogenic disturbance is still limited because characterizing various forms of anthropogenic landscape modification effects is a difficult task (Csorba and Szabo 2009) and mechanisms in habitat use of ground beetles is uncertain.

In conclusion, environmental change by anthropo-genic disturbance can cause different effects on ground beetle assemblages; examples of consequences are nega-tive effects on forest specialists. However, we needed a greater understanding of the response mechanisms of ground beetles in changing the environment. In addition, studying the ground beetles in Korea has been mainly fo-cused on the diversity of mountainous areas rather than relationships between ground beetles and habitat condi-tions. Therefore, an examination of the characteristics of ground beetles will be required prior to evaluating an-thropogenic disturbance.