ESTIMATES OF MACROPOD DENSITY FROM LINE TRANSECT SURVEYS RELATIVE TO ANALYST EXPERTISE

Analysis of line transect data to estimate population density often requires judgments by the analyst. These judgments often must be made without a firm theoretical basis and include issues such as truncation and grouping of the distance data prior to analysis, the choice of model used to estimate the detection function, and methods used to estimate variance components. Our objective was to examine bias and precision of density estimates obtained by experts in distance sampling theory. We asked 7 experts in the field of distance sampling to cooperate in a blind analysis of 6 sets of line transect data on 3 macropod species in Australia where true density was known. Expert analysts chose several different approaches to analyze these 6 datasets. Experts underestimated density by an average of 10-12%, but precision was good; coefficients of variation ranged from 3.4 to 9.3%. Three people, termed students, volunteered to perform a similar blind analysis of these datasets after they completed a 1-week course in distance sampling and analysis theory. While these students had good quantitative skills, we found little difference in their performance compared with that of the experts; both groups underestimated density by 10-12%. The subjective aspects of the analysis of distance sampling may be overcome with some education, reading and experience with computer software. At that point, a person's performance can be near that of experts in the field of distance sampling, at least in analysis of fairly simple datasets, collected under an adequate design, and when assumptions were relatively valid.