Extended Data Fig. 4: Temporal variability of the Skewness-Kurtosis Relationships (SKRs) over the thirteen years of observation.
From: Thresholds of functional trait diversity driven by land use intensification
(a) We show the changes in the SKR-parameters expressed in a 2D space (Y-intercept [α] against slope [β]; Ntot = 1950 observations), and their comparison to null models (small dots). We pinpoint the centroids observed at low (blue dots), intermediate (white dots) and high LUI (red dots), calculated as the averaged parameters observed across the thirteen years. Our analysis showed that the distributions at high LUI behaved as an outlier in 2018, a year known for an extreme summer drought in Central Europe40,41 (Schuldt et al. 2020, Moravec et al. 2021). (b) We show the variability (coefficient of variation, CV) of the SKR-parameters expressed in a 2D space (Y-intercept [α] vs. slope [β]) at low (LUI < 1.25; N = 665 observations; blue dots), intermediate (1.25 < LUI < 1.85; N = 695 observations; white dots), and high LUI (LUI > 1.85; N = 590 observations; red dots), and their comparison against the CVs of 1000 random SKRs (grey violin). The CVs of random SKRs are represented as grey violins where the middle line is the median, the lower and upper hinges correspond to the first and third quartiles, the upper and lower lines show the 0.95 confidence intervals. The observed CVs significantly differed from null expectations when they were either below the 0.05 or above the 0.095 percentile of the distribution of random CVs. *** indicate p values < 0.001 when comparing with random distributions. Ns indicate non-significant differences from null expectations. The variability was calculated as the CV of Euclidean distances among SKR-parameters in the standardized 2-dimensional space (Y-intercept α / slope β) to their mean value observed over the years such as: \(SKRs\,temporal\,{variability}=CV\sqrt{{(\,\underline{\beta }-{\beta }_{i})}^{2}+{(\underline{\alpha }-{\alpha }_{i})}^{2}}\) (Eq. 1).
