Community Ecology Processes, Models, and Applications

variability are observed to those in litter that de-
graded over time.
The relatively high short-term variability of
soil fauna communities implies that soil food
webs, at least in seemingly stable and homoge-
neous temperate forests, do not have a fixed
structure but are instead dynamic. Therefore,
conclusions from community and food web ana-
lyses based on yearly averages should be treated
with caution, especially in ecosystems with a fast
turnover of populations and basic resources. For
example, connectedness food webs, as shown in

Fig. 6.1, are a template for simulation models

that calculate the amount of C and N processed

by a functional group or the whole food web.

Using one of these models we derived the min-

eralization rates for C and N from the trophic

interactions among groups of organisms, based

on the annual biomass of the functional groups

depicted in Fig. 6.1. The values of C and N

mineralization rates indicated by the model

were compared with observed C and N losses

from organic matter in litterbags. The food web

model underestimated C mineralization in litter

0.8

0.6

BC similarity

4 FG

48 sp

0.4

0.2

Temp Hor Vert

L-F

Vert

L-H

Vert

F-H

0

Figure 6.3Temporal variability, and spatial variability in the micro-arthropod species composition (circles, 48 species
combined) and the micro-arthropod functional group composition (squares, four functional groups combined
(springtails, oribatid browser mites, oribatid grazer mites and predaceous mites)) in which the species were lumped, for a
period of 2 years. The spiders and omnivorous beetles (no species determined) and prostigmatid mites (dominated for
95% by one species) were excluded from these analyses. The observed values for functional groups (squares) are similar
to the values observed for the whole community. Compositional similarity clearly declines when the taxonomical level of
resolution (from functional group to species) increases. The Bray–Curtis (BC) similarity index (Legendre and Legendre
1998) is used as a measure of community variability. Mean values of BC similarity are shown for the litter horizon over
time (Temp), the litter horizon in horizontal space (Hor) and between the adjacent litter and fragmented litter horizons
(Vert) and fragmented litter and humus horizon (Vert) and between the non-adjacent litter and humus horizons (Vert) in
vertical space. Bars indicate the 95% confidence intervals. Lower BC-similarity values indicate higher variability. The
mean biomass values from the six litterbags in each horizon collected at the same time were taken to calculate temporal
similarity. Hence, the measures of temporal similarity are made on the spatial scale of the whole study plot. For analyses
of horizontal spatial similarity, pair-wise similarities were calculated between the micro-arthropod functional group or
species composition in each of the six litterbag sets sampled at each time. Each horizon was examined separately. For
vertical similarity on the plot scale, the average abundances in each horizon of the six litterbag sets was used to calculate
similarity between composition in adjacent (L–F and F–H) and non-adjacent (L–H) horizons. For more detail about
analyses and for additional analyses in community composition at other levels of taxonomic resolution, see Berg and
Bengtsson (2007).

74 SPACE AND TIME