see Morin 1999). This aspect of spatial scale has so
far been largely ignored in empirical and theoreti-
cal food web studies. In most studies that focus on
horizontal space trophic interactions are aggre-
gated over broader spatial scales, the scale of a
forest or even a landscape, and not the scale of
the organisms (Broseet al.2005).Thisresultsin
potential food web architectures that depict all
physiological possible feeding links. In reality,
when sampling on a small spatial scale, not all
potential feeding links are realized, because not
every species occurs everywhere (Fig. 6.4). The
larger the species, the larger its home range, but
the lower its numerical abundance and frequency
in a local community (Cohenet al. 2003; Jennings
and Mackinson 2003). Similarly, not all small-sized
organisms occur locally, owing to environmental
heterogeneity. In Fig. 6.4, three samples at differ-
ent locations yield three different local connected-
ness food webs. The top predator is not always
present, i.e. the right-hand example in Fig. 6.4c,
owing to its low abundance and large home
range. The spatial dimensions of resource use of
the top predator links the other two local webs
(Holt 1996; Pokarzhevskiiet al. 2003; Hedlund
et al. 2004). The more local webs are sampled and
combined, the more the potential food web is ap-
proached (Broseet al. 2005). Many local versus one
potential food web and coupling of local webs by
predators are a challenge for future food web re-
search, especially from a modelling point of view.
However, the first attempts to deal with these as-
pects of spatial variability have already been made
(Kondoh 2003, 2005; Teng and McCann 2004).
6.6 Variability across vertical space in soil communities is high
The distribution of groups of soil organisms often
shows a distinct vertical spatial pattern (Faber 1991;
Ponge 1991; Berget al. 1998a). Likewise, in our
forest example community composition has a dis-
tinct vertical pattern. The community is significant-
ly more variable between non-adjacent organic
horizons, the litter–humus comparison (BCL–H¼
0.470.088), than between adjacent horizons, the
litter–fragmented litter comparison (BCL–F¼0.62
0.084) and fragmented litter–humus comparison
(BCF–H¼0.570.085). Moreover, vertical spatial
variability in the community composition of non-
adjacent horizons is greater than both temporal and
horizontal variability (Fig. 6.3). Thus, a high pro-
portion of the variation in community structure in
soils is likely to be attributed to different soil hor-
izons, even if these horizons are only a few centi-
metres apart.
Soil temperature and soil moisture content,
organic matter texture, quantity and quality are
factors that differ between organic horizons.
Differences in soil temperature and moisture con-
tent between successive horizons are generally
small (Berg and Verhoef 1998), whereas significant
modifications in organic matter morphology (from
needle to amorphous colloids) are accompanied by
a decrease in particle size, organic matter quality
and, subsequently, degradation rates. Decay rates
of organic matter decline from litter to humus
(mass loss/year: L¼17.6% versus 1.7% and 1.1%
for F and H, respectively). These morphological
and chemical transitions, which are more pro-
nounced the further the horizons are apart, may
cause considerable variation in the composition of
soil communities. When ecosystem engineers are
present, the scale of vertical gradient in organic
matter texture, quantity and quality can change
from a few centimetres, in coniferous forests, to
decimetres or even metres in other ecosystems,
owing to mixing of detritus through the soil by
earthworms.
The observation of vertical structure in commu-
nity composition implies that, even more so than
for horizontal space, food webs can be structured
into compartments in which interactions are loca-
lized (Fig. 6.5). Most soil food web studies, howev-
er, do not consider vertical space, and feeding links
in connectedness food webs are aggregated over
the whole organic layer, or over the depth of soil
core samples. However, on the scale of the organ-
isms, not all physiologically possible trophic inter-
actions are realized, as not all species are
everywhere. For instance, the small particle size of
humus prevents large-bodied animals, such as spi-
ders, occurring in the humus horizon (Berget al.
1998a), whereas the decline in resource quality with
depth strongly affects succession, hence species
composition, of microbes (Kendrick and BurgesSPATIO-TEMPORAL STRUCTURE 77