Communities of Harvestmen 5: European communities.

The communities considered in earlier sections are from a wide span of sources as listed in Table 1, but do not refer equitably to different habitats in different continents (which is evident when examining Table 6). To examine how opilionid communities vary with environmental conditions, avoiding the large-scale geographical variations, consider the sub-set of 211 continental European and British samples, involving just 68 species. These species are distributed heterogeneously between countries and habitats, as shown in Table 7, with only two species approaching a 50% constancy, i.e. present in nearly half of the samples: Mitopus morio (48%) and Oligolophus tridens (46%). Even for these widely-distributed species, their proportional abundance in communities varies markedly, as can be seen in their average abundances between countries and also habitats). These data are derived solely from studies investigating communities and thus exclude single-species accounts or faunistic lists; accordingly the species’ patterns of co-occurrence in communities can be described by both ordination and classification.

A DCA ordination generated the plot for axes I and II shown in Figure 6a. Axis I (λ = 0.897) is dominated by the separation of three communities, all from high altitudes in the Austrian Alps, so that the data points are mostly packed in a triangular cluster along axis II (λ = 0.839) [what's λ?]. More details can be seen when the Austrian montane outliers are omitted from the plot (but not the analysis); then the scattering of data points for each of the habitats becomes more obvious. Additionally, the relative positioning of the habitats is made slightly clearer by locating their centroids, i.e. mean values on each axis, for the habitat types, as shown in Figure 6b. A plot of axis II against axis III (λ = 0.733), as shown in Figure 6c, reveals more separation between communities from different environments. Axis I is quite short in terms of species turnover, with a gradient of only 1.29sd so there is only about 50% change in species composition. Both axes II (7.59sd) and III (6.58sd) are longer gradients, with marked species differences between samples along them. Even the fourth ordination axis, with a length of 5.85sd (λ = 0.632), shows complete species turnover.

Although there is overlap between the samples from different environmental categories, analysis of variance supports significant differences (p < 0.001) in their scores on all three axes. The confusion generated by the scatter of so many points is reduced by considering just the mean values on each axis for each of the habitat types; these are presented in the three-dimensional plots of Figure 7, though obviously this is a gross simplification. Visual inspection of Figure 7 shows a clump, towards the centre of the plot, comprising samples from open, wetland and woodland habitats, from which axis I separates out the montane samples and axis III those from cultivated land, while the spread along axis II places the built environment at its extremity. Grassland and rocky samples cluster together midway along axis III.

Using the Euclidean distance measure [what's that?] to evaluate the separations between all pairs of points, cluster analysis shows how the habitats relate to each other in terms of their scores on the three ordination axes. The resulting dendrogram, based on average distances between the grouped centroids as the clustering proceeds, is shown in Figure 8. Built and cultivated habitats remain separate, but other types of samples form clusters, one comprising cave and montane samples, the other divisible into one cluster of rocky and grassland samples and another cluster of open and wetland samples plus woodland ones.

The cluster analysis of the centroids is very much a simplification of the relationship between the opilionid communities of these habitats. A much more thorough classificatory analysis is provided by TWINSPAN, incorporating some quantitative information by differentiating between species at abundance levels <25%, <50%, <75% and ≥75% (pseudospecies cut levels 0, 25, 50 and 75). More detail can also be extracted by using finer habitat categories, especially different types of woodland (see Table 7b). The resulting classification is described by the dendrogram in Figure 9, in which all divisions have fairly high eigenvalues suggesting clear differences in species composition between samples on positive and negative sides of each division. The first division, not surprisingly, separates off the three distinct samples from the Austrian Alps, using Mitopus glacialis with at least 25% proportional abundance in the samples on the positive side. Subsequent divisions tend to be more subtle, involving multiple indicator species. A consideration of group sizes, relationship to countries and habitats, and levels of divisions suggested the end-groups designated in Figure 9 so that we have twelve classes, A – L, of opilionid communities based on our 211 European samples. These classes are profiled in Table 8, which lists the species recorded in the samples in each class.

Table 9 shows how these twelve classes are distributed between countries and across habitat types. The spread of these classes between countries is shown in Figure 10.

Class A only contains three samples, all in Austrian woodland, which are characterised by Astrobunus helleri, although there are eight species altogether found in these samples with Lacinius dentiger, Carinostoma carinatum and Trogulus nepaeiformis occurring in two samples.

In class B only three species are involved, again in just three samples, which all contain Leiobunum rupestre, while Amilenus auranticus is present in two of them; these are cave communities in Slovenia.

Class C, although only comprising five samples again in Slovenia but in a variety of habitats (Table 9b), is more species-rich with 12 species of which A. auranticus is in all five samples and another species with high constancy is L. rupestre (60%, i.e. in three samples).

Class D involves even more species, 19 in all, and contains 17 samples of which 15 are in deciduous woodland in Austria; the most constant species in this class are Egaenus convexus and Trogulus tricarinatus, both at 88%, other frequent species being Rilaena triangularis (76%), T. nepaeiformis (53%) and Nelima semproni (59%).

Class E is the largest group (possibly could be further divided) with 53 samples, and the most species-rich with 49 species, of which only two have constancy near or above 50%: Oligolophus tridens (79%), R. triangularis (60%) and Lophopilio palpinalis (49%). With constancy values >30%, other common species in this class are Paranemastoma quadripunctatum (38%), Leiobunum blackwalli (34%), Leiobunum rotundum (36%), Phalangium opilio (38%) and Mitostoma chrysomelas (32%). Most (37) of the class E communities were in Austria and the remainder mostly in eastern Europe, covering a wide range of habitats, although 19 samples were in Austrian coniferous woodland.

Class F was also located mainly in the eastern countries in a variety of habitats, but also includes eight samples in England - two cultivated (Owen’s (1991) garden 1980-84 and 1985-89) and six rocky (quarries). The 22 samples in class F include 26 species of which only one is ubiquitous: Phalangium opilio; the only other species with constancy >30% were Opilio saxatilis, L. blackwalli and Paroligolophus agrestis (all at 32%), although Lacinius horridus comes close at 27%.

As can be seen in Figure 10, classes G, H, I and J are more characteristic of western Europe.

Class G, apart from a single rocky sample in England, is restricted to Spain where 12 samples are distributed between grassland (1 sample), scrub (2) and deciduous (4) and conifer (5) woodland. Class G involves 13 species, none of which are ubiquitous although seven have constancy >50%: Odiellus troguloides (92%), L. rotundum and P. opilio (both at 85%), Oligolophus hansenii (77%), Anelasmocephalus cambridgei, L. blackwalli and Homalenotus quadridentatus (these three at 62%).

Class H comprises 48 samples, all but three (1 in Germany, 2 in Wales) being located in Scotland; half of these samples are in deciduous woodland (see Table 9b). Class H involves 15 species, with just six of them exceeding 50% constancy: P. agrestis (96%), Nemastoma bimaculatum (92%), O. tridens (83), Mitopus morio (81%), O. (Odiellus) palpinalis (56%) and Lacinius ephippiatus (54%), while R. triangularis comes close at 48%.

Class I is a much smaller group with 16 samples, of which seven are from Scotland and four from England, from moorland, wetland and woodland habitats; remaining samples are two in German conifer woodland, two Swedish caves and one in built environment in Slovenia. There are 15 species included in the class I communities, with the dominant species being L. ephippiatus (94% constancy), M. morio (88%) and N. bimaculatum (56%).

The 21 samples of class J encompass 13 species, of which M. morio is present in all samples and only two other species occur in about half or more of the samples: N. bimaculatum (62%) and P. agrestis (48%). Class J communities are located in open moorland, wetland and woodland habitats in Scotland, montane moorland in Wales and rocky (quarry) habitat in England

We return to Austrian montane habitats for all but one of the samples in the remaining two classes.

Class K has just a single wetland sample along with six montane communities, and the most characteristic species are M. morio and M. chrysomelas, both with a constancy of 71%, other frequent species being Ischyropsalis kollari (57%), Nemastoma triste, Platybunus bucephalus and Dicranopalpus gasteinensis (all three at 43%).

Finally, in class L we have the three Austrian high altitude montane samples, characterised by just three species: M. glacialis abundant in all three samples, D. gasteinensis and M. morio each in just one sample.

Thus we can see distinctive classes of opilionid communities, characterised by different species. Furthermore, these community types show significant differences (p<0.001, Kruskal-Wallis test) in their species diversity, in terms of species richness (S) and Simpson’s index (B) [ more details are available in §2 ] . The sequence of classes in order of increasing diversity, median values of S and B, is thus: L(1, 1), K (6, 1.4), F (4, 1.8), J (4, 1.8), B (2, 2), D (8, 2), G (6, 2.4), I (4, 2.5), H (6, 2.95), E (7, 3), A (5, 3.8), C (4, 4). Notwithstanding the small sample size for some countries and habitats, χ² values are large enough to suggest statistically significant variation in class frequencies over countries (χ² = 657.5 with 110 degrees of freedom, p < 0.001) as well as habitats (χ² = 461.2 with 143 degrees of freedom, p < 0.001), confirming the patterns evident in Table 9.

As shown in Table 10, these community types are distinctively located in Europe (χ² = 172.87, with 11 degrees of freedom, p < 0.001), with a contrast between western samples influenced by the Atlantic Ocean and the more continental locations towards the east. This presents an impression of contrasting species assemblages across the continent and it would be interesting to have sufficient samples so that we could see if similar patterns pertain for other continental regions.

Back to Opilionid Communities

or

Back to Arachnologia

Back to Home page