Table 3. Conditional probability tables of node A, i.e., species presence in the Bayesian belief network (BBN) model before (beta-level model, Fig. 3a) and after (gamma-level model, Fig. 3b) incorporating the case field data (Table 1) using the expectation maximization (EM) algorithm (Appendix 1). Outcome state probabilities sum to 100% for each model across each row and represent frequencies or probabilities of each outcome state given combinations of input parent node states. BRNO = Bridgeoporus nobilissimus. Parent node states Beta-level model Gamma-level model Abies procera substrate Vegetative cover (%) Local mesic habitat BRNO present BRNO  absent BRNO present BRNO absent adequate 0–50 adequate 60.000 40.000 53.700 46.300 adequate 0–50 inadequate 15.000 85.000 14.067 85.933 adequate 50–80 adequate 90.000 10.000 98.458 1.542 adequate 50–80 inadequate 70.000 30.000 68.067 31.933 adequate 80–100 adequate 100.00 0.000 100.00 0.000 adequate 80–100 inadequate 80.000 20.000 42.928 57.072 inadequate 0–50 adequate 10.000 90.000 10.125 89.875 inadequate 0–50 inadequate 5.000 95.000 5.014 94.986 inadequate 50–80 adequate 15.000 85.000 7.534 92.466 inadequate 50–80 inadequate 10.000 90.000 6.450 93.550 inadequate 80–100 adequate 20.000 80.000 3.598 96.402 inadequate 80–100 inadequate 15.000 85.000 6.569 93.431