PARENT SESSION
Poster Session # 2: Statistical Ecology and Ecological Modeling.
Monday, August 4 Presentation from 5:00 PM to 6:30 PM. SITCC Exhibit Hall B.
Evaluating the use of statistical decision trees for modeling avian habitats and regional range distributions in the Great Plains.
Vaitkus, Milda*,1, Henebry, Geoffrey1, Putz, Brian1, Merchant, James1, 1 Center for Advanced Land Management Information Technologies (CALMIT), Lincoln, NE, USA
ABSTRACT- Attempts to regionalize species models by mosaicking range distributions produced by neighboring state Gap Analysis Projects have proved problematic. Variations in habitat modeling have resulted in significant differences in predicted species distributions within and across state lines. The use of national geospatial data to map surrogates of habitat enables a regional scope. Yet, there is a decided knowledge gap between the scales of biogeography and those of wildlife management. Can the flexibility of statistical decision trees help fill this gap? We generated regional distributions of 20 selected breeding birds in a six-state region (IA, KS, MN, ND, NE, SD) using four recursive partitioning algorithms (CART, QUEST, CRUISE, GUIDE). BBS route level summaries over two time periods (last 10 and 30 years) were used for the occurrence data (presence/absence and abundance). Environmental variables included land cover, daily climatic means and variances, soil texture, and terrain. Multiple statistical decision trees were generated for each target species to evaluate the relative strengths and weaknesses of the different algorithms. Principal considerations were speed of tree identification, interpretability of the cross-validated tree, and plausibility of the predicted range distribution resulting from tree inversion. As expected, classification trees (from CART, QUEST, CRUISE) yielded predicted range distributions different from regression trees (from CART and GUIDE). Unbiased variable selection in QUEST, CRUISE, and GUIDE appeared to facilitate the identification of parsimonious, robust models and plausible range distributions.
Key words: statistical trees, GAP, breeding bird survey, habitat modeling