Fig. 2: Explained variance R² of the generalised linear models. | Nature Communications

Fig. 2: Explained variance R² of the generalised linear models.

From: Consistent stoichiometric long-term relationships between nutrients and chlorophyll-a across shallow lakes

Fig. 2

The R2 values are separated for long-term variation in the shallow lakes based on 5-year simple moving averages (SMA) (A–C) and short-term variation based on SMA residuals (D–F). Shown are results from generalised linear models with gamma distributions for 5-year SMAs and linear models with normal distributions for SMA residuals for total phosphorus (TP, mg/L, panels A, D), total nitrogen (TN, mg/L, panels B, E), or additive models of TP and TN (panels C, F) and chlorophyll a (Chla, µg/L). These are plotted against the mean molar TN:TP of each randomly sampled dataset (see Fig. 1 for data processing steps). The less transparent the points, the more overlapping solutions for R² were found by the bootstrap procedure (indicating the error of R²). The orange line is the mean response based on a LOESS function. N = 9194 samples for the 5-year SMAs because not all models converged (A–C) and n = 13800 samples for the SMA residuals (D–F).

Back to article page