In allanecology/BetaDivMultifun: Testing effects of multitrophic beta-diversity and land-use intensity (LUI) on ecosystem multifunctionality

maybe not up to date.

not used for analysis.

note: This file requires the functions dataset raw_functions which is constructed in the previous scripts 1read_raw_dataset.Rmd and 2calc_raw_dataset.Rmd. Here, it is named raw_grlfuns.

library(naniar)

A selection of the functions which are included in the BetaDivMultifun analysis is stored in grlfuns.

Missing data

vis_miss(grlfuns[, !colnames(grlfuns) %in% c("Explo","Plotn", "Plot"), with=F]) # plot and explo don't contain any missing data

Values with over 21% missing data are excluded.

#USER : set treshold for missing values
treshold <- 0.21
t <- apply(grlfuns, 2, function(x) sum(is.na(x)))
exclude <- names(which(t > 150 * treshold))

small_grlfuns <- grlfuns[, !colnames(grlfuns) %in% exclude, with=F]

Visualisation of new situation

vis_miss(small_grlfuns[, !colnames(grlfuns) %in% c("Explo","Plotn", "Plot"), with=F])

Are there plots which are missing in all variables?

gg_miss_upset(small_grlfuns)

No plots are missing in all variables.

Correlations

M <- cor(small_grlfuns[, !colnames(grlfuns) %in% c("Explo","Plotn", "Plot"), with=F], use="pairwise.complete.obs")
corrplot::corrplot(M,type="lower",addCoef.col = "black",method="color",diag=F, tl.srt=1, tl.col="black", mar=c(1,0,1,0), number.cex=0.6)

Correlations for analysis should not be over 0.7.

treshold <- 0.7
M[M < 0.7 & M > -0.7] <- 0
corrplot::corrplot(M,type="lower", tl.srt=1, tl.col="black", diag = F, title = "Correlations over 0.7",  mar=c(0,0,2,0))

High correlations between following variables :

• Soil.C.stock vs. SoilOrganicC : chose only one and leave out the other
• Groundwater.recharche vs. soil.depth : ?
• PRI and NRI (because they both contain Bulk density 2014 AND Biomass 2013)
• PRI is contained in P loss comb as a mini-multifunctionality

Variable exclusion

Either Soil.C.stock or SoilOrganicC.

vis_miss(small_grlfuns[, colnames(small_grlfuns) %in% c("Soil.C.stock","SoilOrganicC"), with=F])

Soil Organic C is probably contained in Soil C stock, as the stock consists of both organic and anorganic C. Are both available to the ecosystem? Who consumes it? Microbes? Plants? animals?

allanecology/BetaDivMultifun documentation built on Nov. 9, 2023, 8:47 p.m.