Comparative evaluation of new approaches to survey birds.

Two recently devised approaches for sampling bird communities ' time-balanced area-proportionate transects and standardised searches ' aim to yield bird surveys of improved resolution specifically for studying distribution of terrestrial birds in patchy landscapes. Here, I compare these two approaches with the two most common methods presently used in Australian ornithological research: the fixed-effort (or 20-min) search and the repeat transect. The woodland-dependent avifauna of four reserves was sampled using all four methods. Total woodland richness of the four reserves was estimated by combining exhaustive surveys and incidental records to provide a benchmark for subsequent comparisons. The standardised search yielded the most complete richness estimates (i.e. closest to total woodland richness), averaging 78.4% under the strict stopping rule after an overall sampling effort of 740 min. Using a more lenient rule, mean completeness was slightly lower (72.5%) but overall sampling effort fell to 320 min. Coincidentally, this was the same total effort as required for the three fixed-effort methods, but resultant surveys were all less consistent and less complete. Fixed-effort searches (four 20-min searches per patch) yielded richness estimates with a mean completeness of 68.3%, compared with 37.1% for area-proportionate transects and 33.8% for repeat transects. Thus, for the same sampling effort, the lenient standardised search yielded data approximately twice as complete as transect-based methods. Moreover, resultant data are of uniform completeness and are expressed at the patch scale, and thus have greater biological value and relevance for management. In addition to being the most efficient and yielding the most complete data, the standardised search also yielded incidence estimates for all species recorded. While subject to the same limitations as abundance estimates, this information can complement richness data and allows more subtlecomparisons of habitat preference and site quality. The use of transect counts and other fixed-effort sampling methods is discouraged for studies comparing variable sites, and standardised searches and other approaches relying on results-based stopping rules are advocated.