Summarizing remarks: environmental influences

I will make a few comments on the eight individual papers of this session, followed with comments on the role of factors influencing bird counts and the use of counts to estimate bird abundance. Anderson and Ohmart (198 1) report on an extensive study which had a good study design. Data presentation was, however, inadequate and the estimation of density (J. T. Emlen’s [1971] method was used) from counts and distances would be improved by using recently developed, comprehensive analysis methods. It would be very informative to present the data as graphs of estimated densities over time with 95% confidence intervals indicated. The data presentation in terms of kurtosis and skewness was uninformative. It was also an incorrect analysis to examine for a normal distribution because the data were first combined over factors such as season or habitat type and for these combined data to follow a normal distribution there would have to be no variation in bird density by season or habitat type. However, the authors’ analysis showed that bird densities did vary by season and habitat type. The caveats in their discussion section should be memorized by ornithologists. The paper by Best (1981) lacks a conceptual basis for relating the data (counts of birds) to the parameter of interest, bird density. The detectability profiles are based only on observed counts, with apparently no attempt made to estimate true density. These seasonal profiles, therefore, reflect a confounding of three factors: bird density, the rate of cue production, and the detectability of the cues. I believe Best is saying that these seasonal profiles are only useful as a qualitative basis for improved study design, in which case the confounding of these factors is not of concern. I agree with this idea, but wonder if such intensive studies are really needed to document what ornithologists probably already know about the optimal timing of bird studies. Sampling in rugged terrain raises some theoretical problems about what to record for a distance in both line transect or circular plot sampling. Ms. Dawson’s paper (1981) appropriately raises this question. I believe the guiding principle should be that we are sampling area (to the bird) in these methods; perhaps, therefore, the distance recorded should follow the contour of the landscape. This matter needs more thought.