Figure 1.
Number of studies using the HP package of Walsh and Mac Nally [9] for the R software over time.
The subject categories with most number of studies are shown. Filled circles: number of total studies using HP package (n = 128). Filled squares: “Ecology” subject category. Opened circles: “Biodiversity conservation” subject category. Filled triangles: “Environmental sciences” subject category. Filled diamonds: “Geography physical” subject category. Red filled circles and dashed line: Number of studies using the HP package with more than 9 variables (n = 26). Note that a same study can pertain to more than 1 subject category and thus the sum of the number of studies from all the categories is higher than the number of total of studies.
Table 1.
Percentage of times that a variable changes its position within the ranking.
Figure 2.
Probability of a variable changing ranking position obtained from the averaged general linear mixed model.
Solid lines: Effect of the difference in independent variance explained (IVE) between a particular variable and the previous one in the ranking (established for explaining the response variable) on the probability of changing the position for data sets formed by 10, 11 and 12 explanatory variables in analysis of hierarchical partitioning. Note that no change is expected when the difference in explained variance between a variable and the previous one in the ranking is >17.1. Dotted lines: upper and lower limits according to the standard error of the averaged mixed-model coefficients.
Table 2.
Ranking of all the models explaining probability of a variable changing its position (%).
Table 3.
Percentage of independent, joint and total explained variance for each considered variable.
Table 4.
Variables used for explaining the probability of a variable changing its ranking position.