Fig. 2: Covariation among the community temporal trends and characterization of community trajectories.
From: Past and recent anthropogenic pressures drive rapid changes in riverine fish communities
a,b, PCA biplot of the community temporal trends and their cluster assignment where the sites are coloured according to their cluster assignment (a, first and second PCA axes; b, third and fourth PCA axes). c, Boxplots displaying the distribution of the temporal trends by cluster. The centre of the box depicts the median while the bounds depict the 25% and 75% percentiles. The whiskers depict the extreme values within 1.5× interquartile range beyond the bounds of the box. d, Cluster frequencies across the three main biogeographic realms. The ellipses in a and b display the 95% intervals around the clusters assuming a Student’s t distribution. The clusters were named according to the most noticeable characteristic of changes across all the biodiversity metrics (c). In particular, ‘medium change’ cluster was associated with sites presenting moderate changes along all the biodiversity metrics considered. Sites not assigned to a cluster because of affiliation uncertainty (N = 641, 14%) are displayed in Extended Data Fig. 4. N = 46,932 sampling events across 4,476 sites.
