The study was carried out in October 1999, at Mortagne-du-Nord (Nord-Pas-de-Calais, France), where a zinc smelter complex operated from 1901 to 1962 [9,10]. Sampling was carried out at five sites located within the metallurgical zone and nine sites at different distances away from this source of pollution, in the direction of the prevailing winds. Among the five sites within the metallurgical complex, two were located on a metallophyte grassland, and three on a poplar plantation. Among the nine sites of the agricultural perimeter potentially concerned by metal pollution, three grasslands, two fields, two poplar plantations and two forests were surveyed (Table 1).

At each site, eight soil cores (25×25×15 cm) distributed every 2 m on two transects, were sampled, air-dried, sieved and mixed thoroughly to form a composite sample. The resulting soil samples were analysed to quantify soil pollution by Zn, Cd, Pb (NF X31-151, [11]). In addition, the following supplementary variables were measured: C and N contents, CEC, pH and granulometry (Table 1). Soil macro-organisms were sampled on the same transects. Soil macrofauna was extracted using a modified Tropical Soil Biology and Fertility protocol, which consists of an application of a 0.2% formaldehyde solution to the upper 10 cm of soil, followed by soil hand sorting. Individuals collected were immediately fixed in 4% formaldehyde for further identification. The macro-organisms were identified at the level of morphospecies. The identification of most adult specimens was done at the level of species, while the remaining adults and larvae were identified to genus, family or class on the basis of the morphological differences between individuals.

2.3.1 Species diversity and species richness

We have chosen to gather the 14 investigated sites into six groups, in function of our knowledge of pollution status and land use [12] (Table 1, Fig. 1). This a priori classification was very close to the site clustering obtained using the K-means clustering method applied to the three first axes of the PCoA of species abundance [13]. In each case, the observed species richness was recorded and the Shannon diversity index computed. The evenness was also determined as the ratio of the Shannon index to its maximum value, i.e. ln2(species richness) [14]. The estimator of species richness and its associated standard deviation proposed by Chao [15] were computed to compare with observed values. Computations were realised using the software EstimateS [16].

2.3.2 The indicator value

Sites presented in Table 1 were empirically grouped into six units, according to their pollution status and the type of vegetation cover (Fig. 1). The first level of the classification grouped all sites. The second level separated sites according to the litter layer depth and distinguished herbaceous unpolluted sites (low litter quantity) from unpolluted woody and polluted sites (rich in litter). Among sites rich in litter, level 3 distinguished polluted sites from unpolluted wood. Level 4 separated metallophyte grassland from the polluted poplar plantation. Depending on the management practice, herbaceous unpolluted sites were divided at level 5 into grasslands and crops. The last level separated unpolluted natural forest from unpolluted poplar plantation.

A total of 339 morphospecies was found among the 112 sampling units, with a cumulative number of 3465 individuals. Oligochaeta, Hymenoptera, Coleoptera, Diptera larvae, Arachnida and Isopoda were the most abundant groups (Table 2). Oligochaeta [Lumbricidae and Enchytraeidae] had higher density than the other taxa in the majority of sites. Their density decreased principally according to the land use type from 428 ind m⁻² to 57 ind m⁻² in unpolluted poplar plantation and unpolluted crops respectively, and according to the level of soil pollution, from 342 ind m⁻² in unpolluted grassland to 4 ind m⁻² in polluted metallophyte grassland. Coleoptera, Arachnida and Hymenoptera were the most abundant taxa in this metallophyte grassland. Rare species were generally found in unpolluted sites.

Species diversity analyses were performed using all the morphospecies (339). Results showed differences in species richness and diversity between the six groups (Table 3). Observed species richness was mainly affected by the type of land use and, to a lesser extent, by soil pollution. Cultivated plots had much lower species richness than grasslands and woody sites (21, 76–80 and 105–126, respectively). Polluted grassland had higher species richness than unpolluted grassland, even if the differences were not important. The same trend appeared for polluted and unpolluted poplar plantations. The Shannon diversity index was higher in the polluted poplar plantation than in the unpolluted forest or poplar plantation. In grasslands, a similar trend was observed, with 50% higher values in the polluted site than in the unpolluted grasslands. Patterns of evenness and the Shannon index were similar across sites, except for the unpolluted crop that had a high evenness index, but a low Shannon index. This means that this site had a low diversity, because of the very low species richness and rare or very abundant species were less frequent than in the other sites. The Chao index was linearly correlated to the observed richness and thus accounting for it did not change the between-site comparison. This species richness estimation accounts for occurrences of both singletons and doubletons, i.e. those species present in only one and two samples respectively [15]. Seemingly, the proportion of these species was of no effect upon the pattern of species richness across our sites.

To determine if a species was an indicator, we first examined the significance of the index and arbitrarily retained a threshold of 25%. This means that a characteristic species is present in at least 50% of one site group and that its relative abundance in that group reaches 50% ([13] (p. 356)). Two hundred fifty-seven species were encountered less than five times and were removed from indicator values analysis (i.e., 75% of the total). Among the 82 species examined, only 21 were indicator species at least at one level of the hierarchy, i.e. five Coleoptera, two Diplopoda, three Chilopoda, three Gastropoda, six Oligochaeta, one Diptera, and one Crustacea (Table 4). Among the 21 indicator species, three were generalists, i.e. the IndVal index was maximum for the lower hierarchy level. An example of such a species (Lithobius crassipes) is provided in Fig. 2A. Nine species were found to be specialised species (satellite) like the centipede (scolopendridae) Cryptops savignyi (Fig. 2B) or the diplopod Polydesmus complanatus (Fig. 2C). Nine species like the geophilid Haplophilus subterraneus (Fig. 2D), indicated an intermediate level of the hierarchical typology.

Fig. 3 presents all the species that have a significant indicator value index larger than 25. Two immature pigmented and apigmented lumbricids had their indicator values as maximum for the first level of the hierarchy, i.e. all sites (level 1). At the second level, pigmented and apigmented immature worms and Aporrectodea caliginosa, appeared as indicators of the herbaceous unpolluted zone. At the next level, the unpolluted grassland group was characterised by the same three earthworm taxa plus two endogeic Coleoptera larvae (level 5). No indicator species was found for crops (level 5). Four epigeic taxa were indicators of sites with a thick litter system, with a maximum value for the centipede Lithobius crassipes (level 2). At level 3, seven taxa were indicators of unpolluted woods: Enchytraeidae (87.25), Philocia muscorum (49.29), Haplophilus subterranus (34.88), Lumbricus castaneus (30.85), Elateridae larvae (30.30), Diptera larvae (28.13) and Cryptops savignyi (28.13). Unpolluted woods were further divided into forest and poplar plantations (level 6). Forests were characterised by five specialist indicator species (one Diptera larva, 2 Staphilinidae, an earthworm and a millipede) and by two intermediate indicator species (Elateridae larvae and Enchytraeidae). Two gastropods, an earthworm and an Oniscida, were indicator species of poplar plantations (level 6). At level 4, polluted sites are characterised by three indicator species of the polluted poplar plantation: 2 Polydesmus sp. and a snail. No indicator species was found in metallophyte grassland.