Historically, ecologists have been more interested in organisms feedin
g at the tops of food chains than in organisms feeding at or near the
bottom. The problem of taxonomic and trophic inconsistency within and
among described food webs is central to criticisms of contemporary foo
d web research. To study the effects of taxonomic and trophic aggregat
ion on food web properties, 38 published food webs, each containing a
large fraction of investigator-defined biological species, were system
atically aggregated by taxonomy and trophic (functional) group similar
ity. During each step of taxonomic and trophic aggregation, eight food
web properties (MIN, MAX. mean chain lengths; the fractions of basal,
intermediate and top species; the ratio of all links by the total num
ber of species, L/S; and rigid circuits) were calculated and their dep
artures from the original, unaggregated version were recorded. We foun
d only two properties showing wide systematic departure from initial v
alues after both taxonomic and trophic group aggregation: the fraction
of basal species and L/S. One reason for the relative 'constancy' of
the six other properties was due in part to large numbers of trophical
ly equivalent species (species with identical sets of prey and predato
rs) found in these and other published webs. In the 38 webs. the avera
ge number of trophically equivalent species was 45% and ranged from a
low of 13% in aquatic webs to a high of 71% in certain terrestrial sys
tems (i.e., carrion webs). Six of the eight properties (MIN, MAX and m
ean chain lengths, the fractions of top and basal species, and the L/S
ratio) were found to be more sensitive to taxonomic than to trophic a
ggregation. The relatively smaller variations observed in trophically
lumped versions suggest that food web properties more aptly reflect fu
nctional, rather than taxonomic, attributes of real food webs. These f
indings parallel earlier trophic-based results, and bolster the conclu
sion that uneven lumping of taxonomic and trophic groups in published
food web reports do not modify markedly the scaling behaviour of most
of their descriptive properties.