Type I errors linked to faulty statistical analyses of endangered subspecies classifications

• Published in: Journal of agricultural, biological, and environmental statistics Vol. 13; no. 2; pp. 199 - 220
• Main Authors: Skalski, J.R, Townsend, R.L, McDonald, L.L, Kern, J.W, Millspaugh, J.J
• Format: Journal Article
• Language: English
• Published: New York American Statistical Association and the International Biometric Society 01.06.2008; Springer-Verlag; American Statistical Association; International Biometric Society
• Subjects: Agriculture; Agronomy. Soil science and plant productions; Applied statistics; Aves; Biological and medical sciences; Biometrics, statistics, experimental designs, modeling, agricultural computer applications; Biostatistics; California; Clines; Cluster analysis; Discriminant analysis; Ecological genetics; endangered species; Evolution; Evolutionary genetics; fish and wildlife law; Fundamental and applied biological sciences. Psychology; Generalities. Biometrics, experimentation. Remote sensing; Health Sciences; Mathematics and Statistics; Medicine; Monitoring/Environmental Analysis; Polioptila californica; Population genetics; regression analysis; Statistical analysis; Statistics; Statistics for Life Sciences; wild birds; Wildlife conservation; California; United States; North America; America; Cluster analysis; Discriminant analysis; Multivariate statistics; Taxonomy; Spline-regression; ); California gnatcatcher; Stepregression; Subspecies; Spatial statistics; California gnatcatcher (Polioptila californica); Statistical analysis; Endangered species; Regression; Error; Statistics; Step- regression; Classification
• ISSN: 1085-7117; 1537-2693
• DOI: 10.1198/108571108X310771

Legal issues related to subspecies identification frequently occur through the implementation of the 1973 Endangered Species Act (ESA). A listing under the ESA requires management actions to ensure the continued existence of the taxa. However, these actions often have important social, economic, and political implications. We examined the statistical methods of morphological analysis used in subspecies identification. Methods are illustrated using the California gnatcatcher (Polioptila californica), which was incorrectly listed under the ESA due to misinterpretation of morphological data. We found that inferences based on tests of sample means (i.e., t-test, Hotelling's T2-statistic), cluster analysis, and discriminant analysis were subject to high rates of false positives (identification of subspecies when none exist; Type I error). These simple tests ignore the common occurrence of spatial clines in animal tracts. Alternatively, spline-regression and step-regression procedures were found to be quite robust yet had high resolution in finding subspecies break locations.