Tables:
Table 1. Model fit (BIC scores, adjusted R 2), and prediction error (RMSE, RMSE obs > 0) for each PLN model using n =36 species for Fall data, and n =35 for Spring data. Models were constructed using 78 (Fall) and 86 (Spring) observations where a single observation is a single sampling abundance measure for multiple species made at a specific point in a stream at one of the nine NEON sampling locations at a specific day. “RMSE” (root mean square error) indicates the predictive ability of each model and predicts for 36 (Fall) and 35 (Spring) species based on the two most recent observations for each of the nine NEON sites (n =648 [Fall] and n =630 [Spring]). “RMSE obs > 0” indicates the predictive ability (root mean squared error) of each model for presence-only abundances (i.e. observations >0) (n =87 [Fall] and n=79 [Spring]).