Ground Beetle Species (Coleoptera, Carabidae) Associations with Land Cover Variables in Northern England and Southern Scotland
Total Page:16
File Type:pdf, Size:1020Kb
ECOGRAPHY 27: 417Á/426, 2004 Ground beetle species (Coleoptera, Carabidae) associations with land cover variables in northern England and southern Scotland M. D. Eyre and M. L. Luff Eyre, M. D. and Luff, M. L. 2004. Ground beetle species (Coleoptera, Carabidae) associations with land cover variables in northern England and Scotland. Á/ Ecography 27: 417Á/426. Distribution data concerning 172 ground beetle species derived from 1145 pitfall trap sites in northern England and southern Scotland were used to assess the relationship between species distribution and 12 satellite-derived land cover variables at the regional scale. A number of species were strongly associated with one cover type and negatively with others. The major variation was for preferences for covers in upland or lowland parts of the region. Other distinct preferences for some species were covers such as those at the coast whilst a number of common species showed no strong preference for any cover variable. The synthesis of ground beetle species distribution and satellite- derived cover data is discussed in relation to environmental assessment and change. M. D. Eyre ([email protected]) and M. L. Luff, Centre for Life Science Modelling, School of Biology, The University, Newcastle upon Tyne, U.K. NE1 7RU. Ground beetle research within European landscapes scape (Helenius et al. 2001, Kinnunen et al. 2001, tends to be confined to one land cover type or limited Basedow 2002, Pena et al. 2003) whilst Niemela¨ et al. by activities such as farming. The distribution of ground (2002) investigated ground beetle species distribution beetle species and assemblages in wooded habitats have across urban-rural gradients in three countries. received considerable attention with investigations con- Satellite-derived land cover data has been applied cerned with both broadleaved woodland (Assmann 1999, consistently to assess vegetation distribution and change Desender et al. 1999, Judas et al. 2002, Moore et al. and has been used in investigations into changes in such 2002) and conifer forests (Butterfield 1997, Jukes et al. covers as rain forests (Wittmann et al. 2002), boreal 2001, Koivula and Niemela¨ 2002). Work on the ground woodlands (Masek 2001), deserts (Muldavin et al. 2001), beetles of grasslands tends to be linked with other tundra (Walker 2000), wetlands (Townsend and Walsh invertebrate groups (Niemela¨ and Baur 1998, So¨der- 2001) and coasts (Donoghue and Mironnet 2002) and to strom et al. 2001, Vessby et al. 2002), as does work on assess landscape fragmentation (Petit and Lambin 2002). upland areas (Dennis et al. 2002, Eyre et al. 2003a). The use of this cover data has expanded into applica- More limited areas within the landscape such as riverine tions concerned with habitats of mammals (Velazquez et sediments and corridors (Eyre et al. 2001, Bonn et al. al. 2001, Jepsen et al. 2002), birds (Groom and Grubb 2002), saltmarshes (Elkaim and Rybarczyk 2000, Irmler 2002, Daily et al. 2001) and vertebrates (Cassidy et al. et al. 2002) and urban and post-industrial land (Kiel- 2001). Invertebrate habitat assessments using satellite horn et al. 1999, Andersen 2000, Bra¨ndle et al. 2000) imagery has concentrated on disease vectors (Dister et have also been assessed with regard to ground beetle al. 1997, Daniel et al. 1998, Moncayo et al. 2000) and distribution. Research at a larger scale has focused on pests of grasslands, forests and crops (Dreiser 1994, the effects of the composition of the agricultural land- Grilli and Gorla 1997, Brewster et al. 1999, Radeloff et Accepted 12 January 2004 Copyright # ECOGRAPHY 2004 ISSN 0906-7590 ECOGRAPHY 27:4 (2004) 417 al. 2000). There has been some assessment of woodland invertebrate taxa distribution (Va¨sa¨nen and Helio¨vaara 1994, Chust et al. 2000) and some conservation assess- ments with butterflies and dragonflies (Fuller et al. 1998, Moilanen and Hanski 1998, Debinski et al. 1999, Wessels et al. 2000). Eyre et al. (2003b) classified British 10 km grid squares according to pooled ground beetle species lists and related this to satellite-derived land cover data. The incorporation of invertebrate data with this sort of cover data has concentrated on the identification of habitat types for various assemblages, with little work carried out on the assessment of the relationship of individual species with specific land cover types. The present work, therefore, uses ground beetle species data from 1145 sites in northern England and southern Scotland and 12 land cover types derived from Landsat satellite imagery in order to investigate relationship of each species to each cover type. Fig. 1. Map showing the distribution of the 259 1 km squares in northern England and southern Scotland containing the 1145 Methods pitfall sites, with national grid 100 km numbers. Data Species combined to produce an urban cover type; the shrub The occurrence of 172 species from 1145 pitfall sites heath and shrub/grass heath were combined to produce in 259 1 km national grid squares in northern England one shrub heath cover. and southern Scotland were used for analysis. The data were generated between 1985 and 2002. Some of the data has been previously used in a number of publica- tions concerned with the distribution of species assemblages (Gardner 1991, Luff et al. 1992, Butterfield Table 1. The Countryside Information System (CIS) land cover classes and the derived land classes used in the analyses, with et al. 1995) but they have not been used in work indications of altitudinal distribution. relating individual species to specific land cover vari- ables. The 1145 sites were in the 259 1 km national CIS cover Derived covers grid squares shown in Fig. 1. Ground beetle presence/ Bog (herbaceous) Bog (mainly upland) absence species lists, for each site, were generated Bracken Bracken (mainly upland) using the pitfall trapping method outlined by Sea/estuary Coastal bare ground (beach/ Luff (1996), with 5Á/10 traps (8.5 cm diameter, mudflats/cliffs) 10 cm deep, part filled with preservative) used bet- Saltmarsh Coastal (lowland) Coniferous/evergreen woodland Coniferous woodland ween May and September. Nomenclature follows Luff (mainly upland) (1998). Deciduous/mixed wood Deciduous wood (mainly lowland) Grass moor/dune grass/rough Heath grassland (upland) Land cover pasture Inland water Inland water (upland and The Landsat Thematic Mapper-based land cover data lowland) were released in the Countryside Information System Pasture/meadow/amenity grass Managed grassland (mainly lowland) (CIS) (Anon. 1995) as a list of the number of hectares of Rough grass/marsh Rough grass (mainly low- cover per 1 km national grid square (%) of 17 cover types land) and unclassified pixels. The data were derived from Shrub heath images generated between 1987 and 1990 (Fuller et al. Shrub/grass heath Shrub heath (upland) 1994). The 17 land classes in the CIS were reduced to 12 Tilled land (arable crops) Tilled land (lowland) Suburban/rural development classes for these analyses and Table 1 shows the classes Urban development used and an indication of altitudinal distribution. Three Inland bare ground Urban (lowland) sea and coast covers were merged to form a coastal Unclassified Á/ cover; urban, suburban and inland bare ground were 418 ECOGRAPHY 27:4 (2004) Analysis tus, Pterostichus versicolor (Sturm)) and other with shrub heath (e.g. Bradycellus ruficollis (Stephens), Mis- CANOCO Ver. 4 (Ter Braak and Smilauer 1998) was codera arctica, Olisthopus rotundatus (Paykull), Tricho- used to analyse the 1145 ground beetle species lists with cellus cognatus (Gyllenhal)). However some species were the proportional land cover data from appropriate 1 km found to be strongly associated with both these covers squares. The 1 km land class data was used since it was (Agonum ericeti (Panzer), Carabus nitens Linnaeus, this was the smallest scale satellite-derived cover data Dyschirius globosus (Herbst)). Some species were posi- available. This version of CANOCO produces a species- tively related to managed grassland (Carabus granulatus environment table showing the weighted averages with Linnaeus, Pterostichus minor (Gyllenhal), P. vernalis respect to each standardised environmental variable, in (Panzer)) and some to the tilled land (Bembidion this case the twelve cover types. This table indicates the quadrimaculatum (Linnaeus), Stomis pumicatus (Pan- relationship, or lack thereof, between cover types and zer), Trechus micros (Herbst)) but there are a number species. An indication of which were the most important (e.g. Agonum dorsale (Pontoppidan), Bembidion aeneum land covers was provided using automatic forward Germar, B. guttula (Fabricius), B. tetracolum Say, selection of the 12 cover variables within CANOCO Clivina fossor (Linnaeus), Pterostichus melanarius (Illi- with their significance calculated using Monte Carlo ger)) associated with both of these covers. permutation tests. Species related to the urban variable tended to be also related to coastal (e.g. Broscus cephalotes, Calathus cinctus Motschulsky, Dromius melanocephalus Dejean), to tilled land (e.g. Bembidion obtusum Serville, Harpalus Results rubripes (Duftschmid), H. rufipes (De Geer)) or to both (e.g. Dromius notatus Stephens, Harpalus affinis The relationships between individual species and the 12 (Schrank)) and reflected the lowland land cover prefer- land cover variables, shown as weighted averages, are ences of these species. Of the species positively associated given in Table 2. Seven species were related to the bog with