The effects of habitat fragmentation may include the loss of species from i
solated Fragments or changes in species abundances among habitats that diff
er in area, structure, or edge characteristics. We measured the species ric
hness and abundance of ground-dwelling insects in a 1.14-ha old field that
was mowed to produce patches of unmowed vegetation which differed in size,
degree of isolation, and the amount of habitat edge. Four treatments - rang
ing from unfragmented (169-m(2)) to highly fragmented (1-m(2)) patches - we
re replicated four times in a Latin square design, and insects were sampled
twice during 1995 using 177 pitfall traps. Species richness showed a non-m
onotonic response to fragmentation, with the fewest species occurring in th
e slightly fragmented treatment. Responses of rove beetles and ants, the mo
st species-rich and abundant taxa, respectively, were similar to the overal
l insect community but ants had a stronger and more consistent treatment ef
fect in both sample months. Ordinations of ant and rove-beetle assemblages
using nonmetric multidimensional scaling showed that the slightly fragmente
d treatment differed from other treatments in species occurrence and abunda
nce. The lower species richness in the slightly fragmented treatment was pr
imarily due to a subset of ant and rove beetle species that showed a lower
abundance than in other treatments, possibly because this treatment had the
greatest amount of habitat edge. We hypothesize that the non-monotonic spe
cies response to fragmentation was due to the differential effects of habit
at edge on species movements across the habitat boundary between unmowed pa
tches and mowed areas. A greater effect due to the amount of habitat edge r
ather than total patch area, at least among the range of patch sizes studie
d, suggests that the length of habitat edge may be quite important to the d
istribution and abundance of ground-dwelling animals in fragmented habitats
.