In an abandoned pasture in Brittany, we compared artificial small-scal
e disturbances to natural disturbances by wild boar and undisturbed ve
getation. We developed a multivariate statistical approach which analy
ses how species biological attributes explain the response of communit
y composition to disturbances. This technique. which reconciles the in
ductive and deductive approaches for functional classifications, ident
ifies groups of species with similar responses to disturbance and char
acterizes their biological profiles. After 5 months of recolonization,
artificial disturbances had a greater species richness than undisturb
ed vegetation as a result of recruitment of new species without the ex
clusion of pre-existing matrix species. Species morphology, described
by canopy structure, canopy height and lateral spread, explained a lar
ge part (16 %) of community response to disturbance. Regeneration stra
tegies, described by life history, seed mass, dispersal agent, dormanc
y and the existence of vegetative multiplication, explained a smaller
part of community response to disturbance (8 %). Artificial disturbanc
es were characterized by therophyte and compact rosettes with moderate
ly dormant seeds, including a number of Asteraceae and other early suc
cessional species. Natural disturbances were colonized by leafy guerri
lla species without seed dormancy. Few species were tightly related to
undisturbed vegetation and were essentially grasses with a phalanx ro
sette morphology. The functional classification obtained is consistent
with the classification of the community into fugitives, regenerators
and persistors. These groups are structured according to Grubb's mode
l for temperate grasslands, with regenerators and persistors in the ma
trix and fugitives taking advantage of Zaps open by small-scale distur
bances. The conjunction of functional diversity and species diversity
within functional groups is the key to resilience to disturbance, an i
mportant ecosystem function. (C) Elsevier, Paris.