The use of portable thermal imaging for estimating deer population density in forest habitats.

Published online
16 Jan 1998
Content type
Journal article
Journal title
Journal of Applied Ecology

Gill, R. M. A. & Thomas, M. L. & Stocker, D.

Publication language


The reliability of deer population management could be improved with good density estimates, but current methods are either labour-intensive or suffer from uncertainties regarding accuracy. Visibility varies substantially in forests depending on stand type, age and understorey vegetation. In such conditions distance sampling would be an efficient estimation method, but observer disturbance often results in bias when the method is applied to deer. The performance of thermal infrared imaging for estimating deer density by distance sampling was assessed in deer populations in 7 forest areas in mainland UK. Thermal imaging equipment can detect the long-wave energy radiated by natural objects, clearly revealing warm-bodied animals even if partly obscured by vegetation. Many more deer were detected at night using a thermal imager than along the same transect routes in daytime. Detection distances were correlated with visibility but were substantially longer than the average distances at which most animals were disturbed. Most deer were detected without causing prior disturbance. Densities were estimated with a coefficient of variation ranging from 10.2 to 28.4%. Precision depended on sampling effort and sample sizes obtained. A Monte Carlo simulation revealed a quadratic relationship between accuracy and visibility, with accuracy increasing with average visibility and a tendency for deer to select more open habitats within a forest. Under conditions that are likely to be typical of temperate forests (<40% thicket and neutral selection, or <70% thicket if thicket is avoided), accuracy was generally good and changed relatively little in relation to visibility and habitat selection. Likely sources of bias as well as alternatives to thermal imaging are discussed. It is concluded that the method would be suitable for estimating ungulate densities in forests with an adequate network of tracks.

Key words