Many recent quantitative studies of avian population ecology have used strip transect methods to estimate bird densities (see Emlen 1971, 1977). These methods, and indeed all census methods, suffer from complications and limitations, some of which pertain to the observation conditions (e.g., weather and time of day). Traditionally the bias introduced by diurnal variations in bird detectability has been met by limiting censusing to early morning “when birds are most active” (Pettingill 1970). However, recent findings that detectability may vary inversely with time of day in winter, but directly with time of day in summer (Anderson and Ohmart 1977, Shields 1977), or in some species (Robbins 1981), emphasize the need for further studies.
Since 1978 we have been intensively censusing bird communities along a 2600-m altitudinal gradient in California’s Santa Rosa Mountains, using the narrow strip transect method (Merikallio 1946, 1958). In order to sample all of our transects monthly, we have had to census at various times of day. Our early results suggested that, in the open desert habitats of our study region, time of day had little influence on census results, provided air temperature was below about 35 C. Indeed, midday censuses seemed to yield density estimates comparable to those obtained at sunrise. To examine this further, we censused two desert habitats ten times each, twice daily — once around sunrise and once at midday — between 25 March and 9 April 1980.