Current agricultural practices are believed to have contributed to the decl
ines of many farmland bird species. especially seed-eaters, throughout Euro
pe. We investigated associations between the spatial distribution of fourte
en granivorous farmland bird species and agricultural land-use in Britain,
using breeding bird atlas data and national agricultural statistics. Analys
es were spatially-referenced by 10 x 10 km square and variation due to broa
d-scale geography and spatial auto-correlation was controlled for. Generali
zed linear modelling analyses were used to select models describing variati
on in distribution explained by the available land-use variables. The resul
ts show that relationships between distribution and agriculture tend to be
species-specific. but that some general effects can be identified. Features
of intensive arable farming including large areas of sugar beet, wheat and
oilseed rape tended to be associated with low frequencies of occurrence fo
r 9-11 species, while large areas of younger (re-seeded) grassland and high
sheep stocking densities were associated with low frequencies in pastoral
farmland for up to 12 species. One key feature of lower intensity farming,
the presence of larger areas of fallow land, was positively related to freq
uency index for up to 11 species. The proportion of barley sown in spring a
nd agricultural diversity were each associated with a range of complex rela
tionships with frequency index across species, probably reflecting combinat
ions of positive influences and artefacts of scale and geography. A variabl
e describing the heterogeneity of farmland (the extent to which it is a mix
of arable and pastoral land-use) was negatively related to frequency index
for eight species, but other results suggested that farming which is mixed
at a smaller spatial scale is widely beneficial. The results reveal relati
onships between agriculture and the occurrence of granivorous farmland bird
species which suggest both hypotheses for the causes of population change
and directions for management action. However, data on several key features
of agricultural practice (such as pesticide use) were unavailable, so thei
r effects could not be tested, and the effects of the variables included co
uld not be separated from those of other factors which are subject to compl
ex geographical variation. Experimental comparisons of the effects of land-
use at the farm scale are needed to investigate such confounded influences
on farmland bird occurrence.