1. A 3-year study using arrays of artificial refuges to survey for rep
tiles was carried out on a 20-ha area of mature Calluna heathland in W
areham Forest, south Dorset, UK. between 1993 and 1995. 2. During each
of the three years different arrays were used, each duplicated, to in
vestigate the relationships between reptile numbers and array area, re
fuge number and refuge density. A total of 18 arrays was used with a r
ange in area of 0.335-0.597 ha, a range in refuge number per array of
7-127 and a range of refuge densities of 14-378 ha(-1). 3. Arrays were
checked for the presence of reptiles either 25 (1994 and 1995) or 28
(1993) times during the period March/April-October by walking a pre-de
termined transect route (range: 150-731 m) visiting each refuge in tur
n. 4. Individuals of all three snake species were uniquely marked usin
g PIT (Passive Integrated Transponder) tags. 5. During the study 321 s
mooth snake, 44 grass snake, six adder, 96 sand lizard, 286 slow worm
and one common lizard observations were recorded. The majority of smoo
th snakes (96%), grass snakes (86%) and slow worms (99%) were found be
neath refuges, whilst the majority of sand lizards (75%) and the singl
e common lizard were found in the open between refuges (transect walk)
. Adders were found equally under refuges or in the open. The total nu
mber of snake observations represented 77 individual smooth snakes (71
PIT-tagged, six juveniles), 25 individual grass snakes and four indiv
idual adders. 6. The total number of smooth snake and sand lizard capt
ures/sightings showed logarithmic relationships with refuge number and
refuge density such that at refuge numbers of 37 array(-1) and refuge
densities of 80-120 ha(-1) they levelled off, indicating that these t
wo species were probably resident within the study area. 7. Similar lo
garithmic relationships for grass snake and slow worm captures showed
no levelling off. This indicated that the grass snake, a free-ranging
species, could not be considered a resident of the study area, whilst
the slow worm, a sedentary species, although a resident, was probably
not adequately 'sampled' by the range of refuge densities used. 8. The
findings of this study have resulted in a proposed standard method fo
r the surveying of reptiles on dry lowland heath that would enable com
parisons to be made both between sites and within sites over time. Thu
s, with the exception of the sand lizard, for which transect walks res
ulted in the most sightings, lowland heath should be surveyed for smoo
th snakes, grass snakes and slow worms using hexagonal arrays of 37 ar
tificial refuges spaced 10 m apart, and visited throughout the year be
tween April and October, but with most effort being concentrated in Ma
y, June and September. The ability to manage successfully sites known
to contain reptiles will be enhanced by the application of the results
of this study to the determination of reptile presence and abundance.