Figure 4.

Differences in prediction error between USA‐NPN and LTER‐derived models. Density plots for comparisons of predictions on LTER data (top row) and USA‐NPN data (bottom row). Each plot represents the difference between the RMSE for LTER‐derived model and the USA‐NPN derived model, meaning that values less than zero indicate more accurate prediction by LTER‐derived models and values greater than zero indicate more accurate prediction by NPN‐derived models. P < 0.001 for all t tests. Differences are calculated pairwise for the 38 species/phenophase comparisons.