Long-term Population Trajectories and Recent Change of East Maui Landbird Communities - Supplemental Material A: Distance-sampling model selection and Akaike's Information Criterion (AIC) Statistics
Judge S, Warren C, Bak T, Camp RJ, Vallocchia S, Mounce HL. 2026. Long-term Population Trajectories and Recent Change of East Maui Landbird Communities - Supplemental Material A: Distance-sampling model selection and Akaike's Information Criterion (AIC) Statistics. National Park Service
Model parameters and model selection results for the report Long-term Population Trajectories and Recent Change of East Maui Landbird Communities are presented in this appendix. For each species, candidate detection function models were compared using Akaike’s information criterion (AIC). Models are ordered by ΔAIC, the difference between each candidate and the model with the lowest AIC, which is generally considered the best supported model. Candidate models included half normal and hazard-rate key detection functions, optionally combined with cosine (Cos) or Hermite polynomial (Hpoly) series expansions. Covariates were added to the most parsimonious key detection function to improve model fit and precision. Covariates considered as categorical effects included cloud cover (Cloud), amount of rain (Rain), Beaufort wind scale (Wind), Beaufort gust scale (Gust), observer (Observer) and detection type (Detection; 1 = heard, 2 = seen, 4 = heard and seen). Survey year (Year, continuous) was also evaluated. For each model, the tables report the Key model (Model), Adjustment Terms/Covariates used, the number of estimated parameters (Num params), the AIC value, the negative loglikelihood (−LogL), the delta of AIC, and the Akaike model weights. The weights give the relative support for each model and can be interpreted as the probability that a given model is the best among the converged candidate models for that species. Model used for inference is the model in the first row, unless otherwise noted.