occurring, the structure in the data indicates that
the habitat and human influences dominate. These
findings are consistent with the position expressed
by Lack (1976), that most absences of birds from
islands within their geographic range can be
explained by unfulfilled habitat requirements.
Hinsleyet al. (1994) quantified the incidence
functions of 31 bird species in 151 woods of
0.02–30 ha in a lowland arable landscape in eastern
England over three consecutive years (Fig. 10.2).
For many woodland species the probability of
breeding was positively related to woodland area,
although some breeding occurred even in the
smallest of the woods. Only the marsh tit (Parus
palustris), nightingale (Luscinia megarhynchos), and
chiffchaff (Phylloscopus collybita) failed to breed in
woods of area less than 0.5 ha in any year of the
study, and each was uncommon in the area. Small
woods appear to be poor habitat for specialist
woodland species, but are preferred by others.
Their work showed that harsh winters could
temporarily but significantly alter the incidence
functions of particular species. Specialist woodland
species were more likely to disappear from small
woods after severe winter weather than were
generalists, and could take more than a year
to recolonize. Variables describing the landscape
around the woods were important in relation to
woodland use by both woodland and open country
species. For instance, long-tailed tits (Aegithalos
caudatus) were found to favour sites with lots of
hedgerows around them, whereas yellowhammers
(Emberiza citrinella) preferred more open habitats,
small woods, and scrub.
To establish the broader reliability of incidence
functions as indicators of area requirements, they
need to be shown to be reproducible across the
range of a species. The results of Watson et al. (2005)
reported in Box 5.2 for birds in the Canberra area,
Australia, are not encouraging. They calculated
incidence functions for woodland habitat islands in
three different landscapes, with similar ranges in258 ISLAND THEORY AND CONSERVATION
Probability (%)Wren
Dunnock
Great Tit
Treecreeper
G-s Woodpecker
Marsh TitWoodland area (ha)0 2 6 8 10 12 14Probability (%)1990
1991
1992Woodland area (ha)0101008060402001008060402005 15 20 25 30(a)4(b)Figure 10.2Incidence functions for the probability of breeding
as a function of woodland area, based on 151 woods of
0.02–30 ha area, in a lowland arable landscape in eastern
England. (a) Data representative of widespread and common
woodland species (wren, dunnock, great tit) compared with that
for more specialist woodland species (treecreeper, great spotted
(G-s) woodpecker, marsh tit). All relationships are for 1990, except
that for the marsh tit, which is for 1991. (b) Interannual variation in
the incidence functions for the great spotted woodpecker, reflecting
a period of severe weather in February 1991. Small woods were
reoccupied in 1992. Vertical bars represent 1 SE; thosebelow 10%
are not shown. (Redrawn from Hinsley et al. 1994.)