Journal of Biogeography, 28, 129±138 Effects of habitat fragmentation on carabids in forest patches Tibor Magura1*, Viktor KoÈdoÈboÈcz2 and BeÂla ToÂthmeÂreÂsz3 1HortobaÂgy National Park Directorate, Debrecen, Sumen u. 2, H-4024, Hungary 2Zoological Department of Debrecen University, Debrecen, PO Box 3, H-4010, Hungary 3Ecological Institute of Debrecen University, Debrecen, PO Box 71, H-4010, Hungary Abstract Aim The aim of this study was to test some of the classical rules of island biogeography for the carabid beetles (Coleoptera: Carabidae) in 15 forest patches during 1995±99. Location The 15 forest patches studied are located on the Bereg Plain. The Bereg Plain is at the foot of the Carpathians, partly in Hungary and partly in the Ukraine. Even in recent times, the area was covered by continuous woodland of deciduous trees, and the species of the closed canopy deciduous forest of the hills and mountains were able to disperse from the Carpathians to these lowland forests. But now, because of agricultural activities and forest management, this woodland is fragmented into forest patches. Methods The species±area and the number of species±distance to mainland relationships, and the in¯uence of other factors like shape, isolation, and altitude above the see level, on the number of species were studied. We have used three categories of species during the analyses: (i) total number of species; (ii) the number of species of the closed canopy deciduous forest of the hills and mountains (ForHim species); and (iii) the number of widely distributed generalist species (WidGe species). Results There were positive, but statistically insigni®cant correlations between the distance to the Carpathians and the total number of species, and also between the distance to the Carpathians and the number of WidGe species. The correlation was negative for the number of ForHim species, and it was also not signi®cant. There were signi®cant negative correlation between both the total number of species and the number of WidGe species and the size of the forest patches, while there were signi®cant positive correlation between the size of forest patch and the number of ForHim species. The number of ForHim species decreased signi®cantly by the increase of isolation, while the number of WidGe species increased by the isolation. Shape of the forest patches, and their altitude above the sea level had no in¯uence on the total species richness, the ForHim species richness, and the WidGe species richness. We have also proved that each of the total species richness, the ForHim species richness, and the WidGe species richness, is higher for many small forest patches than for one large patch of the same total area. Main conclusions Our results suggest that historical reasons have a vital in¯uence on the present species patterns. Moreover, in biogeographical studies we must distinguish between species which recognize the habitat as a patch or island and those that can survive in the neighbouring habitats as well. An ignorance of these two categories may disguise basic biogeographic rules. Keywords Carabid beetles, island biogeography, fragmentation, isolation, species±area relationship, edge-effect, habitat heterogeneity, subpopulations. *Correspondence: Zoological Department of Debrecen University, Debrecen, PO Box 3, H-4010, Hungary. E-mail: magura@tigris.klte.hu Ó 2001 Blackwell Science Ltd 130 T. Magura, V. KoÈdoÈboÈcz and B. ToÂthmeÂreÂsz INTRODUCTION There have been vital changes in forest and agricultural management of landscapes during the twentieth century. The new forms of agricultural activities and forest management have contributed largely to the fragmentation of natural habitats (Mader, 1980; Forman & Godron, 1986). There are two vital effects of fragmentation on living organisms. Firstly, the total area of the habitats sustaining the populations decreases. Secondly, these habitats tend to be more isolated (Saunders et al., 1991; Halme & NiemelaÈ, 1993). The Bereg Plain is at the foot of the Carpathians, partly in Hungary and partly in the Ukraine, and it was covered by continuous woodlands of deciduous trees even during the 18±19th centuries. Because of the clear-felling of forests and agricultural activity, the former large forested areas are reduced to small isolated forest fragments separated by agricultural areas and by open seminatural assemblages. This particular situation provides excellent possibilities to test island biogeographical models and/or their predictions. When the Bereg Plain was covered by continuous forest, the species of the closed canopy deciduous forest of the hills and mountains were able to disperse from the Carpathians to these lowland forests. For these species the Carpathians were a potential colonization source (Varga, 1995; Magura et al., 1997, 1999; KoÈdoÈboÈcz & Magura, 1999). Now, however, most of the species are widely distributed generalist species. Studies on less mobile species, such as non¯ying arthropods living on the soil surface, provide a useful measure of the impact of human-caused disturbances. Ground-dwelling carabid beetles (Coleoptera: Carabidae) are exceptionally useful study organisms for examining biogeographical topics because they are diverse and abundant, their ecology and systematics are relatively well known, and they seem to be highly sensitive to the effects of landscape changes such as fragmentation (NiemelaÈ et al., 2000). According to the predictions of island biogeography, the number of species supported by an island decreases with the distance from the continent (mainland) and it increases with the area of the island (MacArthur & Wilson, 1963, 1967). Therefore, it is negatively correlated with the distance and positively with the size of the island. Isolates, which are surrounded by different types of area/communities, can also be regarded as islands. We have tested these classical rules of island biogeography for the carabids of the Bereg Plain. Other authors have stressed the in¯uence of other factors like shape (Laurence & Yensen, 1991), isolation (Vos & Stumpel, 1995), and altitude above the sea level (BaÂldi & Kisbenedek, 1999), on the number of species. We have also tested these factors. MATERIALS AND METHODS Study area and sampling Carabid assemblages of 15 forests of the Bereg Plain in Hungary and in the Ukraine were studied during the period of 1995±99 (Table 1 & Fig. 1). The Bereg Plain is the most humid (630±660 mm annual precipitation) and the coolest area of the Great Hungarian Plain. Originally Salicetum triandrae, Salicetum albae-fragilis and Fraxino pannonicae± Ulmetum forest associations were dominant along the rivers. Oak-hornbeam forest (Querco robori-Carpinetum) is the most extensive forest type of the area. It is a relatively undisturbed, forested marginal area of the Great Hungarian Plain, which provides refuge for several endangered species Table 1 Characteristics of the studied forest patches. Forest Area (ha) Distance to the Carpathians (km) Shape -index Isolation -index Height above the sea level (m) Trapping year Collected carabid species Bockerek-forest DeÂda-forest in Hungary LoÂnya-forest DeÂda-forest in Ukraine Dobrony-forest Peres-forest Rafajna-forest TeÂglaÂs-forest Dobrony-forest GuÂt-forest AlsoÂremete-forest BereguÂjfalu-forest PuskinoÂ-forest Rafajna-forest PuskinoÂ-forest BereguÂjfalu-forest MunkaÂcs-forest AlsoÂkerepec-forest GaÂt-forest 1249.66 197.47 1047.80 76.80 1191.67 249.97 1609.09 41.19 1191.67 870.78 463.23 3995.46 522.74 1609.09 522.74 3995.46 179.68 1520.61 436.72 36.91 29.43 28.07 29.15 9.94 12.08 19.48 15.20 9.94 20.71 11.22 3.57 13.15 19.48 13.15 3.57 9.99 7.58 15.04 3.66 1.51 1.91 1.14 1.85 1.26 1.16 1.14 1.85 1.84 1.65 2.18 1.47 1.16 1.47 2.18 1.29 1.55 1.74 0 76.80 0 197.47 0 0 870.78 0 0 1609.09 3995.46 463.23 0 870.78 0 463.23 1520.61 616.40 1520.61 101 109 101 109 103 101 107 102 103 108 136 136 121 107 121 136 109 109 109 1995 1995 1995 1996 1996 1996 1996 1996 1997 1997 1998 1998 1998 1998 1999 1999 1999 1999 1999 13 18 17 17 13 29 18 20 13 10 11 12 14 16 12 12 14 11 12 Ó Blackwell Science Ltd 2001, Journal of Biogeography, 28, 129±138 1 Effects of habitat fragmentation on carabids 131 Figure 1 The studied forest patches in the Bereg Plain. 1: Bockerekforest, 2: DeÂda-forest on the Hungarian side, 3: LoÂnya-forest, 4: DeÂda forest on the Ukrainian side, 5: Dobrony-forest, 6: Peresforest, 7: Rafajna-forest, 8: TeÂglaÂs-forest, 9: GuÂt-forest, 10: AlsoÂremete-forest, 11: BereguÂjfalu-forest, 12: PuskinoÂ-forest, 13: MunkaÂcs-forest, 14: AlsoÂkerepec-forest, 15: GaÂt-forest and 16: Forest belt of the Carpathians. of the fauna. There are also remarkable features of the carabid fauna of the studied region. Most of the species are generalist with wide distributions, characteristic both of forest and open areas of lowlands and hills. However, there are species which are characteristic of the closed canopy deciduous forest of the hills and mountains and usually do not occur in plain areas. The potential colonization source of these latter species is the Carpathians (Varga, 1995; Magura et al., 1997, 1999; KoÈdoÈboÈcz & Magura, 1999). Based on these features of the fauna, we have used three categories of species during the analysis: (1) total number of species; (2) the number of species of the closed canopy deciduous forest of the hills and mountains (ForHim species in the followings); and (3) the number of widely distributed generalist species (WidGe species). This categorization is based on the paper of SzeÂl (1996), using the bigeographical distribution of carabids in Hungary. Beetles were collected using unbaited pitfall traps consisting of plastic cups (diameter 100 mm, volume 500 mL) containing ethylene glycol as a killing preserving solution. There were nine to 18 traps in a forest, scattered randomly in a typical part of the forest. There was at least 10 m distance between the traps. Trapped individuals were collected monthly from April to October. All carabids taken in pitfall traps were identi®ed to species using standard keys (Freude et al., 1976). package on a digitized 1 : 25,000 map. Distance was calculated by taking the shortest possible distance between the Carpathians and the forest patch. The shape of forests was characterized by the shape index (Patton, 1975). It is de®ned as P/(200 ´ (p á A))1/2), where P is the perimeter of the forest patch in metre, and A is the area of it in hectares. Its value is 1 for a round shape forest, while values > 1 represent deviation from circularity (Laurence & Yensen, 2 1991). Isolation of a habitat island is most often measured as the distance to the nearest patch. Isolation of a habitat patch depends not only on the distance to the nearest patch, but on the area of the nearest patch, too. Namely, a larger habitat patch is more likely to have greater number of species that can colonize in the neighbouring patch. However, the distance between the adjacent patches is also relevant. In our study isolation of the forest patches was measured by the inverse isolation measure, as proposed by Vos & Stumpel (1995). It was de®ned as the total forest area within a radius of 600 m around the studied forest patch. This measure was used as an inverse of the isolation, because its value decreases as the isolation of the forest increases. The radius was chosen as 600 m, because even poor colonist forest specialists can cover this distance through inhospitable habitats. Cumulative number of species richness and the number of species in the categories in ranked sequences of patches from large-to-small and small-to-large were calculated to explore whether many small forest patches contain more species than one large patch of the same total area (Quinn & Harrison, 1988; Douglas & Lake, 1994; Honnay et al., 1999). Linear regression analysis was used to study the relationships between the studied variables (distance to the Carpathians, size of the forest island, shape index, inverse isolation index, altitude above the sea level) and the total number of carabid species of the island, the number of ForHim species, and the number of WidGe species. The dissimilarity of the species composition of forest patches was measured by the Hellinger distance, d(p,q), in the following way: d p; q† ˆ 1 ÿ S X p pi
qi iˆ1 where pi is the relative frequency of the species i in the ®rst compared forest patch (p), and qi is the relative frequency of the species i in the second compared forest patch (q), and S is the total number of species. Heterogeneity of the carabid fauna of the forest patches was characterized by the Whittaker's b-diversity (Whittaker, 1960), which is de®ned as the ratio of the total species pool and the average number of species of the patches and minus one. Its minimum is 0, and a large beta-diversity value means a large heterogeneity. RESULTS Data analyses Distances of the forest islands from the Carpathians, and their area, were measured by the ArcView GIS program Ó Blackwell Science Ltd 2001, Journal of Biogeography, 28, 129±138 During the period of 1995±99, we have collected 10,983 individuals of 56 carabid species (see Appendix). There were four species identi®ed at the Bereg Plain, which are S = 33.71 ) 0.19 (Height) r = 0.4333 F = 3.9300 d.f. = 1,17 P = 0.0638 S = 5.64 + 5.79 Log (Distance) r = 0.3021 F = 1.7067 d.f. = 1,17 P = 0.2088 S = 26.85 ) 5.23 Log (Area) r = 0.5192 F = 6.2742 d.f. = 1,17 P = 0.0227 Widely distributed generalist species (WidGe species) S = 15.81 ) 2.18 (Shape) S = 15.33 ) 1.89 Log (Isolation) r = 0.2338 r = 0.5266 F = 0.9832 F = 6.5231 d.f. = 1,17 d.f. = 1,17 P = 0.3353 P = 0.0205 S = )0.49 + 0.03 (Height) r = 0.1867 F = 0.6140 d.f. = 1,17 P = 0.4441 S = 1.28 + 0.79 Log (Isolation) r = 0.6621 F = 13.2667 d.f. = 1,17 P = 0.0020 S = 4.47 ) 1.67 Log (Distance) r = 0.2620 F = 1.2531 d.f. = 1,17 P = 0.2785 S = )1.72 + 1.54 Log (Area) r = 0.4599 F = 4.5612 d.f. = 1,17 P = 0.0475 Forest species of the hills and mountains (ForHim species) S = 3.44 ) 0.52 (Shape) r = 0.1686 F = 0.4972 d.f. = 1,17 P = 0.4903 S = 33.40 ) 0.17 (Height) r = 0.4508 F = 4.3367 d.f. = 1,17 P = 0.0527 S = 9.89 + 4.36 Log (Distance) r = 0.2742 F = 1.3822 d.f. = 1,17 P = 0.2559 S = 19.44 ) 2.78 (Shape) S = 16.64 ) 1.09 Log (Isolation) r = 0.3602 r = 0.3647 F = 2.5350 F = 2.6087 d.f. = 1,17 d.f. = 1,17 P = 0.1298 P = 0.1247 Height above the sea level Log (Isolation) Shape S = 25.68 ) 3.87 Log (Area) r = 0.4636 F = 4.6539 d.f. = 1,17 P = 0.0456 There were no signi®cant correlations between the total number of species, the number of WidGe species and the distance to the Carpathians. There was negative correlation between the ForHim species and the distance to the Carpathians, although it was also not signi®cant. This may be explained by the fact that the forest of the Carpathians are the source of the species characteristic of the closed canopy deciduous forest of hills and mountains, and they spread from there to the forests of Bereg Plain (Varga, 1995; Magura et al., 1997, 1999; KoÈdoÈboÈcz & Magura, 1999). Therefore, the observed negative species number±distance to Total species Species±distance relations Log (Distance) DISCUSSION Log (Area) usually characteristic to the closed forests of the hill and mountains. They are as follows: Carabus intricatus Linnaeus 1761; Cychrus caraboides (Linnaeus 1758); Abax parallelus (Duftschmid 1812); and Cymindis cingulata Dejean 1825. These are the ®rst published occurrence of these species in a plain (lowland) situation. We have collected three more species which were recorded just once at the Hungarian Great Plain: Carabus arcensis carpathus Born 1902; Pterostichus melas (Creutzer 1799) and Molops piceus (Panzer 1793). There were positive, but statistically insigni®cant correlations, between the distance to the Carpathians and the total number of species, and also between the distance to the Carpathians and the number of WidGe species. The relationship was negative for the number of ForHim species, and it also was not signi®cant (Table 2). Both the total number of species and the number of WidGe species are negatively correlated with the size of the forest patches. These relationships were signi®cant. There was a signi®cant positive correlation between the size of forest patch and the number of ForHim species (Table 2 & Fig. 2). There was also a signi®cant positive correlation between the number of ForHim species and the inverse isolation index. Therefore, the number of ForHim species decreased by isolation. Moreover, we have found a signi®cant negative correlation between the number of WidGe species and the inverse isolation index. There was no signi®cant correlation between the other factors (Table 2). We have further found a signi®cant positive correlation between the distances of the patches and the dissimilarity of their species composition measured by the Hellinger distance (r ˆ 0.4654, F ˆ 46.7232, d.f. ˆ 1,169, P < 0.0001; Fig. 3), which also stresses the effect of isolation. A plot of the cumulative species richness of forest patches against the ranking of the size of the patches proved for each of the total species richness, the ForHim species richness, and the WidGe species richness, that many small forest patches contain more species than one large patch of the same total area (Fig. 4). Whittaker's beta-diversity of the carabid fauna was 2.8, suggesting high heterogeneity of the local fauna of the forest patches. Table 2 Relationships between the total species number, the number of species of the distribution categories (ForHim and WidGe species) and the studied variables (size of the forest patches, distance to the Carpathians, shape index, inverse isolation index, altitude above the sea level). The base of the logarithm is 10. 132 T. Magura, V. KoÈdoÈboÈcz and B. ToÂthmeÂreÂsz Ó Blackwell Science Ltd 2001, Journal of Biogeography, 28, 129±138 1 Effects of habitat fragmentation on carabids 133 (a) (b) (c) Figure 2 Species±area relationships for the three categories of species. Figure 3 The correlation between the distance of forest patches and the dissimilarity of their species composition measured by the Hellinger distance. the Carpathians relationship is consistent with the rules of dynamic biogeography: the number of species, characteristic to the source decreases as the distance to the source increases. This may be explained by historical reasons that Ó Blackwell Science Ltd 2001, Journal of Biogeography, 28, 129±138 Figure 4 Small-to-large and large-to small cumulative area curves for the carabids of the forest patches. the negative correlation is not signi®cant. Nowadays, the distance between the woodlands of the Carpathians and the forest patches of the Bereg Plain are 2.5±33.5 km, and this region is covered by agricultural areas and open (herbaceous) assemblages. The forest specialist species are not able to spread over such a large areas, which are remarkably different in microclimate and vegetation structure from the forest. Nowadays, these forests are entirely isolated from the colonization source. Moreover, each of the forest patches is segregated by unpenetrable barriers (1±10 km wide agricultural area, or grasslands). Therefore, these patches do not serve as stepping stones (Den Boer, 1970) spreading from the Carpathians for the brachypterous species characteristic of the closed canopy forests of the hills and mountains. At the beginning of 1900s, when the forest stands were continuous, or partly continuous, these species were able to spread over the region. According to the prediction of island biogeography, the further the forest stand from the colonization source, then the less species are able to reach it. Summarizing, the distance to the Carpathians had signi®cant in¯uence to the composition of carabid fauna of the forest patches. Nowadays, it has no signi®cant in¯uence because of the isolation of the forest patches, but the negative relationship between the number of ForHim species and the distance to the Carpathians is still observable, although it is not signi®cant statistically. 134 T. Magura, V. KoÈdoÈboÈcz and B. ToÂthmeÂreÂsz Species±area relations Many published papers in zoology report positive correlations between the total number of species of the habitat island and the size of the island (Brown, 1971; Faeth & Kane, 1978; Mader, 1980; Nilsson et al., 1988). A few papers do not corroborate this relationship between the number of species and area (Hopkins & Webb, 1984; 3 Mader, 1984; Bauer, 1989; De Vries, 1994). We found a signi®cant negative relationship between the size of habitat island and the total number of species. This is the opposite of the prediction of the classical theory of island biogeography. Bauer (1989) also ®nds a signi®cant negative relationship between the patch size and the total number of species of carabids of isolated limestone outcrops and adjacent peat. Halme & NiemelaÈ (1993), studying the fauna of conifer fragments report that the number of carabid species decreases as the patch size of the conifer forest increases. Bauer (1989) and De Vries (1994) stress that there should be a difference between species which recognize the habitat as a patch or island, and those which can survive in the neighbouring habitats as well. The ignorance of these two categories may disguise the real patterns. For the ForHim species, the forest patches studied are islands, while the WidGe species can survive in the neighbouring habitats. There is a signi®cant positive correlation between the number of species and the size of the forest patch for ForHim species. The correlation is signi®cantly negative for WidGe species. Bauer (1989) observes a similar pattern. There is a signi®cant negative correlation between the size of limestone outcrops and those widely distributed carabid and staphylinid species that can survive in the neighbouring peat. There is a signi®cant positive relationship between the size of limestone outcrops and the limestone specialist carabids and staphylinids that are restricted to limestone outcrops. Usher et al. (1993) also found that only the species richness of farm woodland carabids correlated positively with the area of woodland, while the total number of arthropod species did not correlate. Similar patterns exist for the carabid data set of conifer forest patches published by Halme & NiemelaÈ (1993). There is a negative correlation between the number of widely distributed species and the size of the conifer forest patches and a positive correlation between the number of species of the forest species and the patch size. The above patterns also occur in other types of habitats, like conifer forest patches, limestone outcrops and the neighbouring peat, and farm woodland patches. This pattern is related to the following factors, which are discussed in details below: (1) edge effect (2) habitat heterogeneity; and (3) asyncronity of subpopulations and (4) isolation. Habitat islands consist of edge zones and core zones. Several studies demonstrate that edges of the forest stands have greater species richness than the forest interior (Helle & 4 Muona, 1985; Bedford & Usher, 1994; Magura & ToÂthmeÂreÂsz, 1997, 1998; Magura et al., 2001). This feature can be explained by the fact that the forest edge has species characteristic of the adjoining habitat and edge-associated species. These edge-associated species may require the particular abiotic and biotic environmental conditions resulting from the proximity of two structurally different habitats (Murcia, 1995). However, forest edges have an inverse effect on the forest biota. Previous studies analysing carabid beetles prove that with decreasing area of forest patches the number of generalists species increases, while the species richness of the forest specialists decreases (Mader, 1980; Halme & NiemelaÈ, 1993; Usher et al., 1993; Spence et al., 1996). Our results also demonstrate that the number of WidGe species signi®cantly increases, while the number of the ForHim species signi®cantly decreases as the area of the forest patches decreases (Table 2 & Fig. 2). These facts can be explained by the increase of the edge-area ratio and the decrease of the area of the forest patches. Increased edgearea ratio and decreased forest area produce microclimatic conditions of the forest stands that make possible the colonization by generalist species (Halme & NiemelaÈ, 1993; Spence et al., 1996). The decrease of the core zone (forest interior) and the decrease of the habitat heterogeneity also have harmful impacts on the forest specialist species (Murcia, 1995; Spence et al., 1996; Davies & Margules, 1998). According to Laurence & Yensen (1991) the shape index to some extent expresses the edge-area ratio of a forest patch. We have not found signi®cant relationships between the shape index and the number of the total species, the WidGe or the ForHim species. Blouin & Connor (1985) examine the relationship between species richness and island shape expressed by the shape index and conclude that the island shape does not explain a signi®cant amount of the variation in species richness. Perhaps the application of a suitable measure of the edge-area ratio, such as the total length of edges wider than 2 m, as proposed by Honnay et al. (1999), can reveal the relationship between the edge to area ratio of forest patch and the species richness of carabids in the studied region. It is a general rule that with the increase of the area the habitat heterogeneity also increases (Forman & Godron, 1986). In this paper, signi®cant positive relationship between the area of the forest patch and the number of ForHim species can also be explained by the habitat heterogeneity. De Vries & Den Boer (1990) prove that Agonum ericeti (Panzer 1809) needs a minimum area of habitat because this species requires special environmental conditions (microclimate, dead and decaying trees, cover of leaf litter and herbs, etc.). NiemelaÈ et al. (1987) published similar results for Agonum mannerheimii (Dejean 1828). In the present study Carabus intricatus Linnaeus 1761 (a ForHim species) requires dead and decaying trees for its existence because this species overwinters in these trees. A larger forest patch is more likely to have greater habitat heterogeneity permitting the forest specialists to ®nd their special microhabitat (Magura et al., 2000). This fact can explain the signi®cant increase in the number of ForHim species with the increase of the forest patch area. 5 A theory proposed by Den Boer (1981, 1985) may also explain the signi®cant positive relationship between the area of the forest patch and the number of ForHim species. According to this theory, the number of interaction groups Ó Blackwell Science Ltd 2001, Journal of Biogeography, 28, 129±138 1 Effects of habitat fragmentation on carabids 135 (subpopulations) of a large habitat patch may ¯uctuate in an asynchronous way. Thus, when the size of a subpopulation decreases, that of an another subpopulation increases. If a population consists of several interaction groups (so-called multipartite population) and the numbers of subpopulations ¯uctuate differently, then the survival time of the composite population may signi®cantly increase. Den Boer (1985) by simulation of the ¯uctuation of the observed population size, shows that survival time of a population consisting of only one interactive group is merely a few decades or less. According to Den Boer (1981, 1985), out of the factors that can determine the number of subpopulations in a composite population, the one of most importance is the area of the habitat. Therefore, it is expected that the survival time and the extinction rate of a population are greatly in¯uenced by the area of the habitat (MacArthur & Wilson, 1967; Den Boer, 1981; De Vries & Den Boer, 1990). In the present situation, the ForHim species in larger forest patches probably consist of more subpopulations, therefore their survival time is longer. The signi®cant positive relationship between the inverse isolation index and the number of ForHim species emphasizes that increased isolation of a forest patch signi®cantly decreases the number of ForHim species present there. Thus, the isolation also has a remarkable effect on the carabid fauna of a forest patch. If a ForHim species becomes extinct in a given small forest patch because of one of the above mentioned factors (edge-effect, decrease of the habitat heterogeneity and synchronicity of the subpopulations) or combination of these factors, then this extinct ForHim species cannot recolonize in the forest patch because the forest patch is isolated. ForHim species are large and wingless, so they are poor colonists. Therefore, the continuous and expanded agricultural area and grasslands between the Carpathians, and the forest patches, are impenetrable barriers for these species. The same is true for the area between the forest patches, namely these areas are also barriers for the ForHim species. Therefore, the recolonization of the ForHim species in a given forest patch through the `from patch to patch' mechanism [cf. stepping stones Den Boer (1970)] is also prevented. By contrast, the good colonist WidGe species may more easily colonize in the `empty' forest patches (Spence et al., 1996). Implications for nature conservation Our study shows that carabid species characteristic of closed canopy deciduous forest of hills and mountains still persists in the isolated forest patches of the Bereg Plain. These species are not typical in other lowland areas of Hungary. Moreover, according to the prediction of Higgs & Usher (1980), many small forest patches contain more species than one large patch of the same total area. This is explained by the heterogeneity of the region studied. It is also con®rmed by the high value of b-diversity of the fauna of the forest patches. Usually, just a quarter of the total species pool was present in a forest patch in the region studied. The relatively low ratio of the species present in a patch compared with the species pool is not related to the small size of a patch, Ó Blackwell Science Ltd 2001, Journal of Biogeography, 28, 129±138 because the number of species of the patches is negatively related to the size of the forests (Fig. 2). Therefore, the large heterogeneity of the fauna is explained by isolation and other historical reasons. For the conservation of the carabid species characteristic of the deciduous forests of hills and mountains in this region the following measures are recommended from the point of view of an active nature conservation management programme: (1) All of the forest patches should be preserved to serve as a source habitat; (2) Our results show a signi®cant positive relationship between the number of carabid species characteristic of the deciduous forest of hills and mountains and the area of forest patches. Therefore, the additional fragmentation of the forest patches has to stop for the conservation of these species to be successful; (3) In the forest patches, mainly in the patches with a small area, a large area-to-edge ratio is recommended contributing to the recolonization of the species characteristic of the deciduous forest of hills and mountains; (4) Corridors of forest should be established to connect forest patches with one another and with the Carpathians. Creation of windbreaks between the forest patches might seem a reasonable solution because several studies have proved that forest specialist carabids can migrate great distance in windbreaks (Burel, 1989; Gruttke, 1994; SÏustek, 1994). However, the distance between the source of the colonization (the Carpathians) and the forest patches, and between each of the forest patches is considerable (1±30 km). The creation of windbreaks is thus not sensible. The planting of forest stands composed of native deciduous trees between the source of colonization and the forest patches, or between each of the forest patches, may be a useful, alternative solution. The distance between these plantings, as stepping stones (Den Boer, 1970), is recommended to be 500±600 m, because the poor colonist forest specialists can cover this distance through the inhospitable habitats (Thiele, 1977). The realization of the above mentioned recommendation may contribute to the connection of the carabid populations living in the isolated forest patches. This connection can support the survival of the carabid species characteristic of the deciduous forest of hills and mountains and the maintenance and conservation of biodiversity. 6REFERENCES Bauer, L.J. (1989) Moorland beetle communities on limestone `habitat islands'. I. isolation, invasion and local species diversity in carabids and staphylinids. Journal of Animal Ecology, 58, 1077±1098. BaÂldi, A. & Kisbenedek, T. (1999) Orthopterans in small steppe patches: an investigation for the best±®t model of the species± 136 T. Magura, V. KoÈdoÈboÈcz and B. ToÂthmeÂreÂsz area curve and evidences for their non-random distributions in the patches. Acta Oecologica, 20, 125±132. Bedford, S.E. & Usher, M.B. (1994) Distribution of arthropod species across the margins of farm woodlands. Agriculture Ecosystem, and Environment, 48, 295±305. Blouin, M.S. & Connor, E.F. (1985) Is there a best shape for nature reserves? Biological Conservation, 32, 277±288. Brown, J.H. (1971) Mammals on mountain tops: nonequilibrium insular biogeography. American Naturalist, 105, 467±478. Burel, F. (1989) Landscape structure effects on carabid beetles spatial patterns in western France. Landscape Ecology, 2, 215±226. Davies, K.F. & Margules, C.R. (1998) Effects of habitat fragmentation on carabid beetles: experimental evidence. Journal of Animal Ecology, 67, 460±471. Den Boer, P.J. (1970) On the signi®cance of dispersal power for populations of carabid-beetles (Coleoptera, Carabidae). Oecologia, 4, 1±28. Den Boer, P.J. (1981) On the survival of populations in a heterogeneous and variable environment. Oecologia, 50, 39±53. Den Boer, P.J. (1985) Fluctuations of density and survival of carabid populations. Oecologia, 67, 322±330. De Vries, H.H. (1994) Size of habitat and presence of ground beetle species. Carabid beetles: ecology and evolution (eds K. Desender, M. DufreÃne, M. Loreau. M.L. Luff and J.-P. Maelfait), pp. 253±259. Kluwer Academic Publishers, Dordrecht. De Vries, H.H. & Den Boer, P.J. (1990) Survival of populations of Agonum ericeti Panz. (Col., Carabidae) in relation to fragmentation of habitats. Netherlands Journal of Zoology, 40, 484±498. Douglas, M. & Lake, P.S. (1994) Species richness of stream stones: an investigation of the mechanisms generating the species±area relationship. Oikos, 69, 387±396. Faeth, S.H. & Kane, T.H. (1978) Urban biogeography ± city parks as islands for Diptera and Coleoptera. Oecologia, 32, 127±133. Forman, R.T.T. & Godron, M. (1986) Landscape ecology. Wiley & Sons, New York. Freude, H., Harde, K.W. & Lohse, G.A. (1976) Die KaÈfer Mitteleuropas. Goecke and Evers-Verlag, Krefeld. Gruttke, H. (1994) Dispersal of carabid species along a linear sequence of young hedge plantations. Carabid beetles: ecology and evolution (eds K. Desender, M. DufreÃne, M. Loreau. M.L. Luff and J.-P. Maelfait), pp. 299±303. Kluwer Academic Publishers, Dordrecht. Halme, E. & NiemelaÈ, J. (1993) Carabid beetles in fragments of coniferous forest. Annales Zoologici Fennici, 30, 17±30. Helle, P. & Muona, J. (1985) Invertebrate numbers in edges between clear-fellings and mature forests in northern Finland. Silva Fennica, 19, 281±294. Higgs, A.J. & Usher, M.B. (1980) Should nature reserves be larger or small? Nature, 285, 568±569. Honnay, O., Hermy, M. & Coppin, P. (1999) Effects of area, age and diversity of forest patches in Belgium on plant species richness, and implications for conservation and reforestation. Biological Conservation, 87, 73±84. Hopkins, P.J. & Webb, N.R. (1984) The composition of the beetle and spider faunas on fragmented heathlands. Journal of Applied Ecology, 21, 935±946. KoÈdoÈboÈcz, V. & Magura, T. (1999) Biogeographical connections of the carabid fauna (Coleoptera: Carabidae) of the Beregi-sõÂksaÂg to the Carpathians. Folia Entomologica Hungarica, 60, 195±203. Laurence, W.F. & Yensen, E. (1991) Predicting the impacts of edge effects in fragmented habitats. Biological Conservation, 55, 45±67. MacArthur, R.H. & Wilson, E.O. (1963) An equilibrium theory of insular zoogeography. Evolution, 17, 373±387. MacArthur, R.H. & Wilson, E.O. (1967) The theory of island biogeography. Princeton, NJ. Mader, H.-J. (1980) Die Verinselung der Landschaft aus tieroÈkologischer Sicht. Natur und Landschaft, 55, 91±96. Mader, H.-J. (1984) Animal habitat isolation by roads and agricultural ®elds. Biological Conservation, 29, 81±96. Magura, T. & ToÂthmeÂreÂsz, B. (1997) Testing edge effect on carabid assemblages in an oak±hornbeam forest. Acta Zoologica Academiae Scientiarum Hungaricae, 43, 303±312. Magura, T. & ToÂthmeÂreÂsz, B. (1998) Edge effect on carabids in an oak±hornbeam forest at the Aggtelek National Park (Hungary). Acta Phytopathologica et Entomologica Hungarica, 33, 379±387. Magura, T., KoÈdoÈboÈcz, V., ToÂthmeÂreÂsz, B., MolnaÂr, T., Elek, Z., SzilaÂgyi, G. & Hegyessy, G. (1997) Carabid fauna of the Beregi-sõÂksaÂg and its biogeographical relations (Coleoptera: Carabidae). Folia Entomologica Hungarica, 58, 73±82. Magura, T., KoÈdoÈboÈcz, V. & ToÂthmeÂreÂsz, B. (1999) Carabid species in the forests of the Bereg Plain in the Ukraine and in Hungary (Coleoptera: Carabidae). Journal of Ukrainian 7 Entomological Society, in press. Magura, T., ToÂthmeÂreÂsz, B. & BordaÂn, Zs. (2000) Effects of nature management practice on carabid assemblages (Coleoptera: Carabidae) in a non-native plantation. Biological Conservation, 93, 95±102. Magura, T., ToÂthmeÂreÂsz, B. & MolnaÂr, T. (2001) Forest edge and diversity: carabids along forest±grassland transects. 8 Biodiversity and Conservation, 10, (in press). Murcia, C. (1995) Edge effects in fragmented forests: implications for conservation. Trends in Ecology and Evolution, 10, 58±62. NiemelaÈ, J., Haila, Y., Halme, E., Pajunen, T., Punttila, P. & Tukia, T. (1987) Habitat preferences and conservation status of Agonum mannerheimii Dej. in HaÈme, southern Finland. Notulae Entomologicae, 67, 175±179. NiemelaÈ, J., Kotze, J., Ashworth, A., Brandmayr, P., Desender, K., New, T., Penev, L., Samways, M. & Spence, J. (2000) The search for common anthropogenic impacts on biodiversity: a global network. Journal of Insect Conservation, 4, 3±9. Nilsson, S.G., Bengtsson, J. & AÊs, S. (1988) Habitat diversity or area per se? Species richness of woody plants, carabid beetles and land snails on islands. Journal of Animal Ecology, 57, 685±704. Patton, D.R. (1975) A diversity index for quantifying habitat edge. Wildlife Society of Bulletin, 3, 171±173. Quinn, J.F. & Harrison, S.F. (1988) Effects of habitat fragmentation and isolation on species richness: evidence from biogeographic patterns. Oecologia, 75, 132±140. Ó Blackwell Science Ltd 2001, Journal of Biogeography, 28, 129±138 1 Effects of habitat fragmentation on carabids 137 Saunders, D.A., Hobbs, R.J. & Margules, C.R. (1991) Biological consequences of ecosystem fragmentation: a review. Conservation Biology, 5, 18±32. Spence, J.R., Langor, D.W., NiemelaÈ, J., CaÂrcamo, H.A. & Currie, C.R. (1996) Northern forestry and carabids: the case for concern about old-growth species. Annales Zoologici Fennici, 33, 173±184. SÏustek, Z. (1994) Windbreaks as migration corridors for carabids in an agricultural landscape. Carabid beetles: ecology and evolution (eds K. Desender, M. DufreÃne, M. Loreau. M.L. Luff and J.-P. Maelfait), pp. 377±382. Kluwer Academic Publishers, Dordrecht. SzeÂl, Gy. (1996) Rhysodidae, Cicindelidae and Carabidae (Coleoptera) from the BuÈkk National Park. The fauna of the BuÈkk national park, II (eds S. Mahunka, L. Zombori and L. AÂdaÂm), pp. 159±222. Hungarian Natural History Museum, Budapest. Thiele, H.U. (1977) Carabid beetles in their environments. Springer-Verlag, Berlin. Usher, M.B., Field, J.P. & Bedford, S.E. (1993) Biogeography and diversity of ground-dwelling arthropods in farm woodlands. Biodiversity Letters, 1, 54±62. Varga, Z. (1995) Geographical patterns of biological diversity in the Palearctic region and the Carpathian Basin. Acta Zoologica Academiae Scientiarum Hungariese, 41, 71±92. Vos, C.C. & Stumpel, H.P. (1995) Comparison of habitat isolation parameters in relation to fragmented distribution patterns in the tree frog (Hylea arborea). Landscape Ecology, 11, 203±214. Ó Blackwell Science Ltd 2001, Journal of Biogeography, 28, 129±138 Whittaker, R.H. (1960) Vegetation of the Siskiyou Mountains, Oregon and California. Ecological Monography, 30, 279±338. BIOSKETCHES Tibor Magura (PhD) is currently the regional programme manager of the Hungarian National Biomonitoring System at the HortobaÂgy National Park Directorate. He is the head of the Carabidological Research Group at the University of Debrecen. His main research interests are the faunistics, the biogeography and the ecology of carabid beetles. Viktor KoÈdoÈboÈcz is a PhD student at the Department of Zoology at the University of Debrecen. He is writing his thesis on the biogeography of carabids. His main ®elds of interest are faunistics, the biogeography and the ecology of carabid beetles. BeÂla ToÂthmeÂreÂsz is professor of quantitative ecology at the Ecological Institute of the Debrecen University. He has published papers on diversity measurement, with special emphasis on scale-dependent characterization, as well as scalable proximity measures and multivariate analyses of ecological communities. 138 T. Magura, V. KoÈdoÈboÈcz and B. ToÂthmeÂreÂsz Appendix Number of individuals and the distribution category of the collected carabid species in the Bereg Plain during the period of 1995±99. Legends: ForHim species ± species of the closed canopy deciduous forest of the hills and mountains; WidGe species ± widely distributed generalist species. Species Number of individuals Distribution category in the studied area Abax carinatus (Duftschmid 1812) Abax parallelepipedus (Piller et Mitterpacher 1783) Abax parallelus (Duftschmid 1812) Agonum micans (Nicolai 1822) Agonum moestum (Duftschmid 1812) Platynus obscurus (Herbst 1784) Amara saphyrea Dejean (1828) Amara similata (Gyllenhal 1810) Anisodactylus binotatus (Fabricius 1787) Badister bullatus (Schrank 1798) Bembidion biguttatum (Fabricius 1779) Blethisa multipunctata (Linnaeus 1758) Brachinus crepitans (Linnaeus 1758) Calosoma inquisitor (Linnaeus 1758) Carabus arcensis carpathus (Born 1902) Carabus cancellatus (Reitter 1896) Carabus clathratus (Linnaeus 1761) Carabus convexus (Fabricius 1775) Carabus coriaceus (Linnaeus 1758) Carabus granulatus (Linnaeus 1758) Carabus hampei ormayi (Reitter 1896) Carabus intricatus (Linnaeus 1761) Carabus ullrichi (Germar 1824) Carabus violaceus (Linnaeus 1758) Chlaenius nitidulus (Schrank 1781) Clivina fossor (Linnaeus 1758) Cychrus caraboides (Linnaeus 1758) Cymindis cingulata (Dejean 1825) Elaphrus cupreus (Duftschmid 1812) Harpalus latus (Linnaeus 1758) Harpalus dimidiatus (Rossi 1791) Harpalus ru®pes (De Geer 1774) Licinus depressus (Paykull 1790) Leistus piceus (FroÈlich 1799) Molops piceus (Panzer 1793) Notiophilus palustris (Duftschmid 1812) Oodes helopioides (Fabricius 1792) Ophonus nitidulus Stephens 1828 Patrobus atrorufus (Stroem 1768) Platynus assimilis (Paykull 1790) Platynus krynickii (Sperk 1835) Platynus livens (Gyllenhal 1810) Poecilus cupreus (Linnaeus 1758) Pterostichus anthracinus (Illiger 1798) Pterostichus latoricaensis (Pulpan 1965) Pterostichus macer (Marsham 1802) Pterostichus melas (Creutzer 1799) Pterostichus melanarius (Illiger 1798) Pterostichus minor (Gyllenhal 1827) Pterostichus niger (Schaller 1783) Pterostichus oblongopunctatus (Fabricius 1787) Pterostichus ovoideus (Sturm 1824) Pterostichus strenuus (Panzer 1797) Stomis pumicatus (Panzer 1796) Synuchus vivalis (Illiger 1798) Trechus quadristriatus (Schrank, 1781) 1260 34 643 30 683 7 21 2 1 1 14 1 10 32 1369 722 10 79 226 141 21 32 199 791 51 4 66 1 1 16 1 267 5 12 74 2 5 5 8 18 43 20 148 193 8 4 2148 59 2 196 903 151 27 18 53 145 WidGe species WidGe species ForHim species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species ForHim species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species ForHim species WidGe species WidGe species WidGe species WidGe species ForHim species ForHim species WidGe species WidGe species WidGe species WidGe species WidGe species ForHim species ForHim species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species ForHim species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species WidGe species Ó Blackwell Science Ltd 2001, Journal of Biogeography, 28, 129±138