Biogéographie et Écologie des Communautés de Carabidae en Wallonie | Biogeography and Ecology of Carabid assemblages in Southern Belgium

1. Regional distribution of Carabid beetles in Belgium (1) The study of regional distribution of Carabid beetles is preceded by a detailed analysis of potential ecological factors such as, altitude, climate, geology, pedology and lithology. The analysis sets up a mapping of geographically homogeneous areas. Ordination methods reveal a strong gradient in the set of potential ecological factors taken to account. This gradient is strongly correlated with altitude. Although maximum altitude in Belgium is 700 m, this factor explains most of the geographical structure, because all the other variables are correlated with it. Yet, edaphic conditions isolate the Belgian Lorraine from the other Belgian regions. In spite of the existence of the altitudinal gradient, several homogeneous regions separated from one another by stable frontiers are evident. These frontiers correspond to the maximal discontinuity zones of the local distribution of all potential ecological variables examined. The established frontiers are not artefacts produced on a continuous gradient but are real natural geographic boundaries. (2) Although the biogeographic structure is not as clear as the geographic structure, analysis of the distribution of Carabid beetles in Belgium allows us to define five biogeographic regions: the Coast, the sandy-loamy Flanders, the chalky regions (Condroz-Fagne-FamenneCalestienne and Belgian Lorraine) and the Ardennes plateau. A classification of the species in eight clusters, corresponding to different types of distribution, is established. Beside species groupes with geographical affinities explained by their ecological requirements, a cluster of rare species and one of species found on a large scale were evident. The role of the different potential factors of distribution has been evaluated. 15 to 20% of the total variance of the distribution data is directly explained by the factors of distribution. Among these, the principle factor is the topographic gradient. The climatic and edaphic factors seem to occur with the same intensity in the distribution. The role of these factors is strongly related to the spatial structure. More than half of the information they bring is explained by the geographic coordinates of U.T.M. squares. The ecological factors tend to reveal homogeneous biogeographic sets. HENGEVELD's hypothesis (1985) conceming the artificial character of the biogeographic regions is not confirmed in this case. Yet, it is clear that some biogeographic frontiers are better defined than others. The important overlapping of the different types of distributions makes it difficult to precisely define the biogeographic boundaries. (3) Study of the distribution dynamics reveals that about a third of the Carabid beetle species are in regression. This proportion is thought to be under-evaluated. Strong relationships are observed between species natural history traits, their type of distribution and the dynamics of their distribution. Species that regress are mostly xérophitic, and live in open habitats. Belgium is at the margin of their European distributional range. These species make up 70.6 and 50.0% of the total number of species and predominate respectively in chalky regions and Campine. Spreading species are generally dipolymorphic and eurytope. A detailed analysis of the methodology generally used for defining the types of distribution dynamics has some important limitations due to the type of methods used. We suggest the use of another estimator of sampling intensity rather than the number of species recorded by U.T.M. squares. 2. Species assemblage structures (1) The structure of Carabid beetle communities in open environments is studied by the sampling of 80, well distributed plots, in Wallonie. 65 of them have been followed for two years. The plots represent 10 types of habitat defined a priori. These habitats are: fringes of ponds, swamps, floating mires, raised mires, peat bogs, sandy heathlands, mineral heathlands, alluvial meadows, xeric chalky grasslands and serpentine soils. The analysis of botanic relevés shows a good concordance with the physionomic classification established a priori. A complex gradient, taking to account humidity and acidity, orders the different habitats. It contrasts first, dry alkaline plots to dry acid plots, then dry acid plots to wet acid plots. Data describing the stations edaphic characteristics show that the major chemical factor is the richness in mineral elements. The poorest plots are sandy heathlands and certain mineral heathlands or bogs. The richest are xeric chalky grasslands, swamps and raised mires. The second chemical factor is the acidity. Richness in mineral elements and acidity need to be compared by texture and classes of draining to better discriminate the different types of habitats. (2) Pit fall trap samples of the 80 plots studied, collected 43.096 specimens and 190 species, some of them very rare. Three species reported missing in Wallonie since 1950 were found. New distribution data were obtained for many species, especially those for which the habitat was poorly prospected by entomologistes. ~ diversity, ordinations and grouping indicate that the data are very well structured. About 60% of the data structure is conserved if the annual cycles of trapping are grouped in 8 classes corresponding to a classification of the plots based on Carabid beetles. This typology includes: xeric chalky grasslands, mineral heathlands, serpentine soils, very dry heathlands, wet heathlands, grassy heathlands, eutrophic wet habitats (swamps and alluvial meadows) and oligotrophic wet habitats (floating mires and raised mires). Analysis of distribution factors of Carabid beetles show, in general, that abiotic factors better explain the species distributions than biotic factors. Among biotic factors, parasitism isn't a major regulator of population densities. The number of Carabid beetles infected by laboulbeniales and acari is neglegable. The main effects of competition are not proved by the exclusion of congeneric species, taxonomically and ecologically related. However, strong indications show that certain pairs of taxonomically related species are characterised by disjoint distributions and therefore, have different ecological requirements. Ecological variables retained for explaining the Carabid beetle distribution expresse only a part of the information found by the grouping of annual pit fall trapping cycles. The grouping of annual pit fall trapping cycles explains about 60% of the original structure. The only major gradient is determined by the humidity of the substrata. Spatial autocorrelation is week. Data structures are better represented by partitions than ordinations which indicates that species assemblages is very compartmentalized. (3) Preliminary analysis of the fit of HANSKI's (1982) and BROWN's (1984) regional distribution models to Carabid beetle distribution indicates that Carabid beetle species assemblages are dominated by rare species which occupy only certain assemblages. There are so few species in most of the sites from each habitat types that HANSI's model and the associated dassification of core and satellite species are not supported. The high number of rare species in Carabid beetle assemblages has two origins. About half of them are effectively rare at other spatial levels. The others belong to well represented eurytopic species and must be considered as erratic elements. 3. Conclusions Species assemblages differentiate from one another at a regional scale or for different types of habitats. At a local scale, Carabid beetle assemblages are characterised by a big number of rare species and few species common to all the sampling units, however. High heterogeneity is observed in a habitat type or even in a station. These observations seem to indicate that Carabid species assemblages are open systems, where local equilibrium is rarely reached. Subsequent researches, comparing natural history traits of these assemblages to those of forest habitats, should conflrm this observation. Multivariate analysis are powerful tools to extract the most objectivety as possible from the structure of a data set. Nevertheless, the choice of the methods remains fundamental. Only a cautious process, not limited to graphs and results obtained by software, but further research of the strong and recurrent forms, is the only way to guarantee the validity of the revealed structure. This work has allowed to establish that Carabid beetles assemblages are better explained by the stochastic community model rather than the determinist model. Physical components of the environment are more important that interspecific interactions. Distributions areas and species groups are revealed for the first time in Belgium, with multivariate analysis. Some enhancements are brought to these multivariate methods. We emphasize also the importance of the relationships between the different study scales of biological phenomenon.