A Multivariate Regression Tree Approach to Defining Population Spatial Units Using Frequencies of Individual Characteristics and Time Series of Abundance

• Activity Number: 356
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Cleridy E. Lennert-Cody*+ and Mark N. Maunder and Alexandre Aires-da-Silva and Mihoko Minami
• Companies: Inter-American Tropical Tuna Commission and Inter-American Tropical Tuna Commission and Inter-American Tropical Tuna Commission and Keio University
• Keywords: multivariate regression tree ; Kullback-Leibler divergence ; population management unit ; stock structure

Abstract

A critical aspect of the assessment of population status is identification of areas containing relatively independent populations or individuals with similar characteristics. We present an approach for defining spatial units that combines results of multivariate regression tree analyses of spatial structure in frequency distributions of individual characteristics and in temporal trends in abundance. We illustrate this approach with length-frequency and catch-per-unit-effort data for bigeye tuna (Thunnus obesus) from the Japanese longline fisheries in the eastern Pacific Ocean. The approach proves useful as a means of quantitatively defining dominant spatial structure, and provides options for incorporating non-quantitative subject-matter knowledge into the choice of spatial assessment units. Although our application is focused on population management, this approach could be applied to any system for which data describing the system dynamics were available in the form of time series and frequencies of characteristics.