Skip to content

Latest commit

 

History

History
1522 lines (1278 loc) · 76.5 KB

MainAnalyses.md

File metadata and controls

1522 lines (1278 loc) · 76.5 KB

Script information

Script of the article entitled: Drivers of contrasting boreal understory vegetation in coniferous and broadleaf deciduous alternative states

Article submitted to Ecological Monographs

Authors of the article: Juanita C. Rodríguez-Rodríguez, Nicole J. Fenton, Steven W. Kembel, Evick Mestre, Mélanie Jean, and Yves Bergeron

Script developed on 2021 by: Juanita C. Rodríguez-Rodríguez (e-mail: juanitacarolina.rodriguezrodriguez@uqat.ca), for my PhD in Environmental Sciences. A part of the script was developed during the statistical cours of Pierre Legendre.

Load packages

library(dplyr) #For droplevels() library(tidyr) #For spread() library(stats) #For filter(), For PCA: prcomp() library(vegan) #For PCA: decostand(), PCA: rda() library(BiodiversityR) #For ordiequilibriumcircle() in PCA library(ggplot2) #For Figures library(nlme) # (for lme) library(devtools) install_github("sdray/adespatial") library(adespatial) # For beta.div (TBI) library(ade4) # For s.value (Beta div plot) and is.euclid() library(factoextra) # For fviz_pca_biplot library(cowplot) # For ggdraw library(ggpubr) # For ggarrange library(fastDummies) library(ape) # For pcoa() library(emmeans) #For lsmeans() library(ggbeeswarm) #For geom_quasirandom() library(lmerTest) #To include p values in lmer

Load all data and prepare tables

DB_All <- read.csv2("Understory_veg_JR.csv") DB_All2 <- DB_All

Prepare a species table with all variables

DB_All2$Species <- NULL # Leaving just species Abbreviations (Abbr) DB_All2$Func_gr <- NULL # Not for now

Spp_All <- spread(DB_All2, Abbr, Coverage, fill = DB_All2$Coverage) which(is.na(DB_All2$Coverage)) # I checked it because after spread, I get a Warning message: In if (!is.na(fill)) { : the condition has length > 1 and only the first element will be used head(Spp_All) # Species table with all variables

Prepare tables from Spp_All

Env <- select(Spp_All, 1:5) Spp <- select(Spp_All, 6:70)

Env$Year <- as.factor(Env$Year) Env$Block <- as.factor(Env$Block)

Organize levels of the factor Treatment

Env$Treatment <- factor(Env$Treatment, levels = c("Li","Nu","1F","2F","To","Ti","C"))

Separate matrices by forest type (Canopy) to do alpha analysis (with all years included)

Canopy <- split(Spp_All, Spp_All$Canopy) BS <- Canopy$'BS' TA <- Canopy$'TA'

Alpha diversity of understory communities

Figure 4

Alpha diversity (Shannon-Weiner index) of understory vegetation in 2018 among treatments in each forest type (Black spruce - BS and Trembling aspen - TA), over time (2013-2018) (n=9). Treatments correspond to light (Li, in yellow), nutrients (Nu, in purple), single amount of leaves (1F, in orange) and double amount of leaves (2F, in red), transplants-out (To, in blue), Transplants-in (Ti, in green) and control conditions (C, in grey).

Diversity plot with Shannon index - facet_grid with both Year and Canopy

####Ecosystem approach: "gold","mediumpurple1","tan1","lightcoral","paleturquoise2","chartreuse","azure3" ####Community approach: "goldenrod","darkorchid3","orangered","red3","lightseagreen","forestgreen","azure4"

div.plot2 <- ggplot(Spp, aes(x = Env$Treatment, y = diversity(Spp, index = "shannon"), fill=Env$Treatment)) + geom_boxplot()+ stat_summary(fun = mean, geom="point", pch=21, col = "black", bg = "white", size = 2.8)+ labs(x="Treatments", y = "Alpha diversity (Shannon index)")+ facet_grid(cols=vars(Env$Year), vars(Env$Canopy)) + scale_fill_manual(values=alpha(c("gold","mediumpurple1","tan1","lightcoral","paleturquoise2","chartreuse","azure3")), name = "Treatments") + theme(panel.background = element_rect(fill = "white", colour = "grey50")) + theme(strip.background = element_rect(colour = "black", fill = "white")) div.plot2

ANOVA of alpha diversity

Table S2.

Differences in α-diversity (Shannon-Weiner index) based on the ANOVA of the linear mixed effect model of understory species abundance to evaluate the factors of Year, Forest type, Treatment and all interactions, using Sites and Blocks as nested random factors.

set.seed(1989) div.aov5 <- lmer(Div.spp ~ YearCanopyTreatment + (1|Site/Block), data = Env) summary(div.aov5) anova(div.aov5) plot(div.aov5)

Comparisons per Year and Canopy, between C and each Treatment

div.emmeans <- emmeans(div.aov5, specs = trt.vs.ctrl ~ Treatment|Year*Canopy, type = "response") div.emmeans

Obtain the letters for those comparisons

div.mod_means <- multcomp::cld(object = div.emmeans$emmeans,

                           Letters = letters,no.readonly=TRUE)

div.mod_means

Post-hoc test

Table S3.

Post-hoc emmeans contrasts of the lineral mixed effect model (Table S2) of the α-diversity (Shannon index) of plant understory species in each forest type (Black spruce – BS, Trembling aspen - TA), comparing each treatment to the control. Significant differences correspond to P < 0.0001 ***, <0.001 **, < 0.05 *, denoted in letters for Figure 4.

lsmeans(div.aov2, pairwise~YearCanopyTreatment, adjust="tukey") t.test(diversity(Spp) ~ Env$Canopy)

Re-Order levels of the factor so I do comparisons of C versus all

Env$Treatment <- factor(Env$Treatment, levels=c("C","Li","Nu","1F","2F","To","Ti"))

Beta diversity

Make matrix of 2018 data

Spp_Year <- split(Spp_All, Spp_All$Year) Spp_2018 <- Spp_Year$2018 Spp_18 <- select(Spp_2018, 6:70) Env_18 <- select(Spp_2018, 1:7)

Non-directional approach: Partitioning beta diversity

Compute beta diversity with function beta.div() of adespatial

B.div for each objective separately and each canopy

Separate by Canopy

Spp_18Can <- split(Spp_2018, Spp_2018$Canopy) Spp_18BS <- Spp_18Can$BS Spp_18TA <- Spp_18Can$TA

####Separate by Treatment - Objective approach ####BS Spp_18BS_Tbio <- filter(Spp_18BS, Treatment %in% c("C","1F","2F","Li","Nu")) Spp_18BS_Tss <- filter(Spp_18BS, Treatment %in% c("C","Ti","To")) ####TA Spp_18TA_Tbio <- filter(Spp_18TA, Treatment %in% c("C","1F","2F","Li","Nu")) Spp_18TA_Tss <- filter(Spp_18TA, Treatment %in% c("C","Ti","To"))

####Divide into Spp and Env tables ####BS, Tbio BS_Tbio_spp <- select(Spp_18BS_Tbio, 6:70) BS_Tbio_env <- select(Spp_18BS_Tbio, 1:5) ####BS, Tss BS_Tss_spp <- select(Spp_18BS_Tss, 6:70) BS_Tss_env <- select(Spp_18BS_Tss, 1:5) ####TA, Tbio TA_Tbio_spp <- select(Spp_18TA_Tbio, 6:70) TA_Tbio_env <- select(Spp_18TA_Tbio, 1:5) ####BS, Tss TA_Tss_spp <- select(Spp_18TA_Tss, 6:70) TA_Tss_env <- select(Spp_18TA_Tss, 1:5)

Tbio_cols <- c("salmon1","firebrick2","antiquewhite4","gold","mediumorchid") Tss_cols <- c("antiquewhite4","chartreuse3","lightseagreen")

Beta-diversity for each approach

Figure 5

Partitioning of β-diversity into local contributions of single plots (LCBD = Local Contributions to Beta Diversity) for treatments in each forest type (Black spruce and Trembling aspen) of (a, b) the ecosystem approach corresponding to Light (Li, in yellow), Nutrients (Nu, in purple), Single-litter (1F, in orange) and Double-litter (2F, in red), and for (c, d) the community approach corresponding to Transplants-out (To, in blue), Transplants-in (Ti, in green), as well as Control conditions (C, in grey) for both approaches. Treatments in black spruce forests in light colors and in trembling aspen forests in dark colors. Tables below each figure present the total sum-of-squares (SStotal), total β-diversity (BDtotal) and the contributions (decreasing order) of individual species (SCBD = Species Contributions to Beta Diversity), for each forest type in each approach. Data correspond to 2018, n = 9, from the combination of Site (A, B, C) and Blocks (1, 2, 3) in each forest type.

####BS, Tbio = BS_Tbio_spp, BS_Tbio_env res_BS_Tbio <- beta.div(BS_Tbio_spp, method="hellinger", nperm = 999) # Hell transformation of data summary(res_BS_Tbio) res_BS_Tbio$beta res_BS_Tbio$LCBD which(res_BS_Tbio$p.LCBD <= 0.05) # Which are the significant LCBD indices? res_BS_Tbio$SCBD res_BS_Tbio$SCBD[res_BS_Tbio$SCBD >= mean(res_BS_Tbio$SCBD)] res_BS_Tbio$p.adj

Figure 5a

LCBD.plot_BST_bio <- ggplot(BS_Tbio_spp, aes(x = BS_Tbio_env$Treatment, y = res_BS_Tbio$LCBD), fill=res_BS_Tbio$LCBD) + geom_boxplot(fill=Tbio_cols) + labs(x="Treatments", y = "LCBD") + facet_grid(cols=vars(BS_Tbio_env$Canopy)) + ylim(0, 0.08) + theme(panel.background = element_rect(fill = "white", colour = "grey50")) + theme(strip.background = element_rect(colour = "black", fill = "white"))

LCBD.plot_BST_bio_aov <- aov(res_BS_Tbio$LCBD ~ BS_Tbio_env$Treatment / BS_Tbio_env$Site*BS_Tbio_env$Block) summary(LCBD.plot_BST_bio_aov)

####data.nest.agg = data.nest %>% group_by(A, Sites) %>% summarize(y = mean(y)) ejemp <- how(nperm = 9999, blocks = BS_Tbio_env$Site,
plots = Plots(BS_Tbio_env$Block), Within(type = "free"))

ejemp_ran <- adonis2(res_BS_Tbio$LCBD ~ BS_Tbio_env$Treatment, permutations = ejemp, method = "euc", by = "term") ejemp_ran

####TA, Tbio = TA_Tbio_spp, TA_Tbio_env res_TA_Tbio <- beta.div(TA_Tbio_spp, method="hellinger", nperm = 999) # Hell transformation of data summary(res_TA_Tbio) res_TA_Tbio$beta res_TA_Tbio$LCBD which(res_TA_Tbio$p.LCBD <= 0.05) res_TA_Tbio$SCBD res_TA_Tbio$SCBD[res_TA_Tbio$SCBD >= mean(res_TA_Tbio$SCBD)] res_TA_Tbio$p.adj

Figure 5b

LCBD.plot_TAT_bio <- ggplot(TA_Tbio_spp, aes(x = TA_Tbio_env$Treatment, y = res_TA_Tbio$LCBD), fill=res_TA_Tbio$LCBD) + geom_boxplot(fill=Tbio_cols) + labs(x="Treatments", y = "LCBD") + facet_grid(cols=vars(TA_Tbio_env$Canopy)) + ylim(0, 0.08) + theme(panel.background = element_rect(fill = "white", colour = "grey50")) + theme(strip.background = element_rect(colour = "black", fill = "white"))

LCBD.plot_BST_bio_aov <- aov(res_BS_Tbio$LCBD ~ BS_Tbio_env$Treatment / BS_Tbio_env$Site*BS_Tbio_env$Block) summary(LCBD.plot_BST_bio_aov)

ejemp_TA <- how(nperm = 9999, blocks = TA_Tbio_env$Site,
plots = Plots(TA_Tbio_env$Block), Within(type = "free"))

ejemp_ran_TA <- adonis2(res_TA_Tbio$LCBD ~ TA_Tbio_env$Treatment, permutations = ejemp, method = "euc", by = "term") ejemp_ran_TA

####BS, Tss = BS_Tss_spp, BS_Tss_env res_BS_Tss <- beta.div(BS_Tss_spp, method="hellinger", nperm = 999) # Hell transformation of data summary(res_BS_Tss) res_BS_Tss$beta res_BS_Tss$LCBD which(res_BS_Tss$p.LCBD <= 0.05) res_BS_Tss$SCBD res_BS_Tss$SCBD[res_BS_Tss$SCBD >= mean(res_BS_Tss$SCBD)] res_BS_Tss$p.adj

Figure 5c

LCBD.plot_BST_ss <- ggplot(BS_Tss_spp, aes(x = BS_Tss_env$Treatment, y = res_BS_Tss$LCBD), fill=res_BS_Tss$LCBD) + geom_boxplot(fill=Tss_cols)+ labs(x="Treatments", y = "LCBD") + facet_grid(cols=vars(BS_Tss_env$Canopy)) + ylim(0, 0.08) + theme(panel.background = element_rect(fill = "white", colour = "grey50")) + theme(strip.background = element_rect(colour = "black", fill = "white"))

#TA, Tss = TA_Tss_spp, TA_Tss_env res_TA_Tss <- beta.div(TA_Tss_spp, method="hellinger", nperm = 999) # Hell transformation of data summary(res_TA_Tss) res_TA_Tss$beta res_TA_Tss$LCBD which(res_TA_Tss$p.LCBD <= 0.05) res_TA_Tss$SCBD res_TA_Tss$SCBD[res_TA_Tss$SCBD >= mean(res_TA_Tss$SCBD)] res_TA_Tss$p.adj

Figure 5d

LCBD.plot_TAT_ss <- ggplot(TA_Tss_spp, aes(x = TA_Tss_env$Treatment, y = res_TA_Tss$LCBD), fill=res_TA_Tss$LCBD) + geom_boxplot(fill=Tss_cols)+ labs(x="Treatments", y = "LCBD") + facet_grid(cols=vars(TA_Tss_env$Canopy)) + ylim(0, 0.08) + theme(panel.background = element_rect(fill = "white", colour = "grey50")) + theme(strip.background = element_rect(colour = "black", fill = "white"))

Plot all 3 Figures together. Save PDF as 5x7 landscape

Figure 5a-d

Plot all four Figures together. Save PDF as 5x7 landscape

(LCBD_Allplots <- ggarrange(LCBD.plot_BST_bio, LCBD.plot_TAT_bio, LCBD.plot_BST_ss, LCBD.plot_TAT_ss + rremove("x.text"), labels = c("a)", "b)", "c)", "d)"), ncol = 2, nrow = 2))

(LCBD_Allplots_Tbio <- ggarrange(LCBD.plot_BST_bio, LCBD.plot_TAT_bio + rremove("x.text"), labels = c("a)", "b)"), ncol = 2, nrow = 2))

(LCBD_Allplots_Tss <- ggarrange(LCBD.plot_BST_ss, LCBD.plot_TAT_ss + rremove("x.text"), labels = c("c)", "d)"), ncol = 2, nrow = 2))

Species turnover: Temporal analysis of multivariate data (TBI)

####Note: As adespatial package did not worked, I had to upload the whole script of TBI from Legendre script.

TBI analysis with all results together for each forest type, between years 2013 and 2018

Preparing data tables

DBTbio <- DB_All

Filter by canopy dominance (BS or TA)

BS Matrix

BS.mat <- DBTbio %>% filter(Canopy %in% c("BS")) %>% droplevels()

TA Matrix

TA.mat <- DBTbio %>% filter(Canopy %in% c("TA")) %>% droplevels()

Make two matrices, one with 2013 data and the second with 2018 data. I'll get two matrices for each canopy composition.

Matrices of BS, filtering by year.

2013 Matrix of BS

BS.2013 <- BS.mat %>% filter(Year %in% c("2013")) %>% droplevels()

2018 Matrix of BS

BS.2018 <- BS.mat %>% filter(Year %in% c("2018")) %>% droplevels()

Matrices of TA, filtering by year.

2013 Matrix of TA

TA.2013 <- TA.mat %>% filter(Year %in% c("2013")) %>% droplevels()

2018 Matrix of TA

TA.2018 <- TA.mat %>% filter(Year %in% c("2018")) %>% droplevels()

Now, I already have my four matrices of 2013 and 2018 for BS and 2013 and 2018 for TA.

Join columns of Site, Block and Treatment

For BS, 2013:

Mat1.BS <- unite(BS.2013, TP,c("Site","Block","Treatment"), sep = ".", remove = TRUE) Mat1.BS$Year <- NULL Mat1.BS$Canopy <- NULL Mat1.BS$Species <- NULL Mat1.BS$Func_gr <- NULL

For BS, 2018:

Mat2.BS <- unite(BS.2018, TP,c("Site", "Block","Treatment"), sep = ".", remove = TRUE) Mat2.BS$Year <- NULL Mat2.BS$Canopy <- NULL Mat2.BS$Species <- NULL Mat2.BS$Func_gr <- NULL

For TA, 2013:

Mat1.TA <- unite(TA.2013, TP,c("Site", "Block","Treatment"), sep = ".", remove = TRUE) Mat1.TA$Year <- NULL Mat1.TA$Canopy <- NULL Mat1.TA$Species <- NULL Mat1.TA$Func_gr <- NULL

For TA, 2018:

Mat2.TA <- unite(TA.2018, TP,c("Site", "Block","Treatment"), sep = ".", remove = TRUE) Mat2.TA$Year <- NULL Mat2.TA$Canopy <- NULL Mat2.TA$Species <- NULL Mat2.TA$Func_gr <- NULL

Now, I have all matrices (Mat1.BS, Mat2.BS, Mat1.TA, Mat2.TA) that I need for the analysis (filtered by Year, Canopy, with Abbr and with one single column with TP). However, I need a "species table", where the first column is my "sites" <- and the other columns are each Func_gr, filled by the Coverage in each cell.

####BS Mat1.BS.spp <- Mat1.BS %>% pivot_wider(names_from = Abbr, values_from = Coverage) Mat2.BS.spp <- Mat2.BS %>% pivot_wider(names_from = Abbr, values_from = Coverage) ####TA Mat1.TA.spp <- Mat1.TA %>% pivot_wider(names_from = Abbr, values_from = Coverage) Mat2.TA.spp <- Mat2.TA %>% pivot_wider(names_from = Abbr, values_from = Coverage)

Now I have spp tables ((Mat1.BS, Mat2.BS, Mat1.TA, Mat2.TA).spp) as I need.

I take a new table with the Treatment-plot names, because I need to eliminate it to do the analysis (must be numeric). I do the same for all the tables but they are the same, so I take just one in TP at the end.

TP.BS.Mat1 <- as.matrix(Mat1.BS.spp$TP) TP.BS.Mat2 <- as.matrix(Mat2.BS.spp$TP) TP.TA.Mat1 <- as.matrix(Mat1.TA.spp$TP) TP.TA.Mat2 <- as.matrix(Mat2.TA.spp$TP) (TP <- as.matrix(Mat1.BS.spp$TP))

(TP2 <- Mat1.BS.spp$TP)

Then, I need to add an ID to each row and then eliminate the TP columns of each DB that I'll use in the analysis.

ID <- data.frame("ID" = c(1:63)) #I create an ID with 9 numbers

I add the ID to all tables.

BS1 <- Mat1.BS.spp BS1$TP <- NULL

BS2 <- Mat2.BS.spp BS2$TP <- NULL

TA1 <- Mat1.TA.spp TA1$TP <- NULL

TA2 <- Mat2.TA.spp TA2$TP <- NULL

Now, I got finally all the tables for the analysis: BS1.spe, BS2.spe, TA1, TA2

TBI Analysis per forest type

BS1 <- as.data.frame(BS1) BS2 <- as.data.frame(BS2) TA1 <- as.data.frame(TA1) TA2 <- as.data.frame(TA2)

TBI - BS

TBI.BS <- TBI(BS1,BS2,method="%diff", pa.tr=FALSE, nperm=9999, BCD=TRUE, replace=FALSE, test.BC=TRUE, test.t.perm=TRUE, save.BC=TRUE, seed.=NULL, clock=FALSE) TBI.BS # Non Sig - change: 0.3238 (P>0.05) summary(TBI.BS) BS.BCD.mat <- TBI.BS$BCD.mat

TBI - TA

TBI.TA <- TBI(TA1,TA2,method="%diff", pa.tr=FALSE, nperm=9999, BCD=TRUE, replace=FALSE, test.BC=TRUE, test.t.perm=TRUE, save.BC=TRUE, seed.=NULL, clock=FALSE) TBI.TA # Sig + change: 1e-04 * (P<0.05) summary(TBI.TA) TA.BCD.mat <- TBI.TA$BCD.mat

Figure 6.

Analysis of Temporal Beta-diversity Index (TBI) of understory communities in black spruce (left panel) and trembling aspen (right panel) stands. Figures correspond to B-C plots of understory vegetation comparing all treatments from 2013 to 2018, where 63 plots per forest type were plotted using losses (B/den statistics) and gains (C/den statistics) computed from the abundance data of understory species. These sites correspond to the combination of sampling design of Sites (A,B,C), Blocks (1, 2, 3) and seven Treatments: Control conditions (C, in grey), to treatments of the ecosystem approach: Light (Li, in yellow), Nutrients (Nu, in purple), Single-litter (1F, in orange) and Double-litter (2F, in red), and of the community approach: Transplants-out (To, in blue), Transplants-in (Ti, in green). Treatments in black spruce forests in light colors and in trembling aspen forests in dark colors. Notice also that axes are different to allow clearer data visualisation. The green line (with a slope of 1) is drawn to through the origin and it represent the theoretical positions of a net-zero species gain/loss scenario. The position of green line respective to red line indicates an average of species losses (green line above red line, as in left panel) or species gains (green line below red line, as in right panel) across the sites. Distinctive symbols are used for the sites dominated by gains (squares) and by losses (circles). Symbols sizes represent the values of the D = (B+C) statistics: larger points found in the upper-right corner of each plot represent a higher temporal β-diversity than the smaller points in the lower-left corner.

B-C Plots

#####plot.TBI # Appeler la fonction avant! From: https://github.com/sdray/adespatial/blob/master/R/plot.TBI.R plot.TBI(TBI.BS) plot.TBI(TBI.TA)

Save both Figs together as 12x6 portrait PDF

par(mfrow = c(1, 2))

BS

plot.TBI(TBI.BS, type = "BC", s.names = TP2, pch.loss = 21, pch.gain = 22, cex.names = 0.5, col.rim = "black", col.bg = "dodgerblue3", cex.symb = 3, diam = TRUE, main = "Black spruce forests", cex.main = 1, cex.lab = 1, xlim = NULL, ylim = NULL, silent = TRUE)

TA

plot.TBI(TBI.TA, type = "BC", s.names = TP2, pch.loss = 21, pch.gain = 22, cex.names = 0.5, col.rim = "black", col.bg = "hotpink", cex.symb = 3, diam = TRUE, main = "Trembling aspen forests", cex.main = 1, cex.lab = 1, xlim = NULL, ylim = NULL, silent = TRUE)

par(mfrow = c(1, 1))

Ordinations for each objective (ecosystem and community approaches)

Figure 7.

Principal Component Analysis (PCA) for the Objective 1 (a - Ecosystem approach) and the Objective 2 (b - Community approach) illustrating community change between 2013 and 2018. Points correspond to the centroids per treatment (means of 9 replicate plots from nested design of 3 Sites x 3 Blocks) of the Hellinger-transformed abundance of understory species from the beginning (2013, circles) and the end of the experiment (2018, triangles), united by lines indicating change over time. Solid lines correspond to black spruce stands (BS) and dotted lines correspond to trembling aspen stands (TA). Colors correspond to treatments of the ecosystem approach: Single-litter (1F, in orange), Double-litter (2F, in red), Control (C, in grey), Light (Li, in yellow) and Nutrients (Nu, in purple) and of treatments of the community approach: Control (C, in grey), Transplants-in (Ti, in green) and Transplants-out (To, in blue). Treatments in black spruce forests in light colors and in trembling aspen forests in dark colors. Species ordination is in the middle of the panel, with only the main species to allow a clear visualization. Notice that the scales are different.

Table S4.

###PERMANOVA based on the Hellinger-transformed data (Bray-Curtis distance) of understory vegetation abundance for the ecosystem and community approaches to test the variables of Year, Forest type, Treatment and all possible interactions, using Site and Blocks as nested random factors.

Prepare a species table with all variables

DB_All$Species <- NULL DB_All$Func_gr <- NULL

Spp_All <- spread(DB_All, Abbr, Coverage, fill = DB_All$Coverage) Spp_All$Y.C.T <- paste0(Spp_All$Year,"", Spp_All$Canopy, "", Spp_All$Treatment) Spp_All <- Spp_All[, c(1:5, 71, 6:70)] # To order columns

Ecosytem approach (Tbio) - Analysis with both Canopies

Select treatments of Tbio (Exclude treatments Ti and To)

Spp.Tbio <- Spp_All %>% filter(Treatment %in% c("C", "Li", "Nu", "1F", "2F")) %>% droplevels()

Separate years

Year <- split(Spp.Tbio, Spp.Tbio$Year) Y2013 <- Year$2013 Y2018 <- Year$2018

BSTA.Year <- rbind(Y2013, Y2018) BSTA.sxs <- BSTA.Year[,6:71] #From column Y.C.T to the end of spp columns BSTA.env <- BSTA.Year[,1:5] #All sampling info

Combining the data of 2013 and 2018 for each stand

BSTA <- rbind(Y2013, Y2018) BSTA.PCA <- BSTA[,6:71] #From column Y.C.T to the end of spp columns

#Make a matrix with Site and Block columns to add random factors in the following PERMANOVAs BSTA_SB <- rbind(Y2013, Y2018) BSTA_SB <- BSTA_SB[,2:3]

Centroid for Cntrl

Data.BSTA<-droplevels(subset(BSTA.PCA)) Data0.BSTA<- Data.BSTA[, colSums(Data.BSTA != 0) > 0] # To eliminate empty (0) columns

Data transformation and PCA

PCA.crd.dec <- decostand(Data0.BSTA[,-1], "hellinger") PCA.BSTA <- prcomp(PCA.crd.dec, center = TRUE, scale. = FALSE) PCA.crd.dec <- cbind(Data0.BSTA$Y.C.T, PCA.crd.dec) #Adding column with names colnames(PCA.crd.dec) names(PCA.crd.dec)[names(PCA.crd.dec) == "Data0.BSTA$Y.C.T"] <- "Y.C.T" PCAD.BSTA <- as.data.frame(PCA.BSTA$x)

PCAD.BSTA$Year <- with(PCA.crd.dec, substr(Y.C.T, 1,4)) PCAD.BSTA$Canopy <- with(PCA.crd.dec, substr(Y.C.T, 6,7)) PCAD.BSTA$Treatment <- with(PCA.crd.dec, substr(Y.C.T, 9,nchar(as.character(Y.C.T))))

Mean for each treatment and year

pca.centroids.BSTA <- aggregate(PCAD.BSTA[,1:2], list(Year = PCAD.BSTA$Year, Canopy=PCAD.BSTA$Canopy, Treatment=PCAD.BSTA$Treatment), mean) pca.centroids.BSTA$LAB<-with(pca.centroids.BSTA, ifelse(Year==2018, "18", "13"))

Figure 7 (part), Save as 5x5 Landscape

Main.BSTA<-ggplot(data=pca.centroids.BSTA, aes(PC1, PC2, col=Treatment))+ theme_bw()+ geom_hline(yintercept = 0, lty=2, col="gray")+ geom_vline(xintercept = 0, lty=2, col="gray")+ geom_point(aes(shape = Year, color=Treatment),size = 2)+ scale_color_manual(values=c('darkorange1', 'red3', 'gray53', 'gold', 'darkmagenta'))+ geom_line(aes(PC1, PC2, group=interaction(Canopy, Treatment), linetype=Canopy)) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),legend.position="bottom", plot.title = element_text(size=12,hjust = 0.5,face="bold"), axis.title=element_text(size=11),text = element_text(size=12), axis.text.x = element_text(angle=0, hjust=0.5, size=8), axis.text.y = element_text(angle=0,hjust=0.5,size=8))+ labs(x= "PC1 (37.8%)", y ="PC2 (9.7%)", title="Ecosystem approach")

summary(PCA.BSTA)$importance[2,1] # PCA1: 0.37872 summary(PCA.BSTA)$importance[2,2] # PCA2: 0.09751 summary(PCA.BSTA)$importance[2,3] # PCA2: 0.07542 summary(PCA.BSTA)$importance[2,4] # PCA2: 0.05592

Screen plot and broken stick model to test the significance of PCA axes

screeplot(PCA.BSTA, bstick = TRUE, npcs = length(PCA.BSTA$center)) bstick(PCA.BSTA) # I get four axes

Subplot of species

scores(PCA.BSTA$center) #To get the scores of species and select the most important ones

Figure 7 (part-Species)

#Ancient spp selected: "AUR", "ARN", "CLB", "CON", "EQP", "GAH", "LIB", "LYA", "MAC", "PLS", "POA", "PTC", "LEG", "RIG", "RUP", "SPS", "VAM", "VIE", "VIS" #Selecteed species with the highest scores (0.39-0.04): PLS,CON,GAH,PTC,RUP,LYA,MAC,VIS,LIB,CLB,VAM,VIE,ARN Sub.BSTA<-fviz_pca_biplot(PCA.BSTA, label = "var",labelsize = 3, habillage="none",title="", xlab="", ylab="", addEllipses=FALSE, geom.ind = "",col.var = "black", select.var=list(name=c('PLS','CON','GAH','PTC','RUP','LYA','MAC','VIS','LIB','CLB','VAM','VIE','ARN')), repel =TRUE, lable="ind", fill.ind = "white", legend.title = "SP")+ theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), legend.text=element_text(size=13), legend.title=element_text(size=15, face="bold"))+ theme_void()

Figure 7 (both joined), Save as 5x5 Landscape

plot.with.inset.BSTA <-ggdraw() + draw_plot(Main.BSTA) + draw_plot(Sub.BSTA, x = 0.191, y = 0.22, width = .60, height = .60)

Table S4.(Tbio) - PERMANOVA

PERMANOVA Including both blocks and Sites in permutations

perm_Tbio <- how(nperm = 9999, blocks = BSTA_SB$Site,
plots = Plots(BSTA_SB$Block), Within(type = "free"))

set.seed(1989) ado.BSTA_random2 <- adonis2(PCA.crd.dec[, -1] ~ PCAD.BSTA$Year * PCAD.BSTA$Canopy * PCAD.BSTA$Treatment, permutations = perm_Tbio, method = "euc", by = "term" ) ado.BSTA_random2

Community approach (Tss) - Analysis with both Canopies

Select treatments of Tss (Exclude treatments "C", "Li", "Nu", "1F", "2F")

Spp.Tss <- Spp_All %>% filter(Treatment %in% c("C", "Ti", "To")) %>% droplevels()

Separate years

Year.Tss <- split(Spp.Tss, Spp.Tss$Year) Y2013.Tss <- Year.Tss$2013 Y2018.Tss <- Year.Tss$2018

Combining the data of 2013 and 2018 for each stand

BSTA.Tss <- rbind(Y2013.Tss, Y2018.Tss) BSTA.PCA.Tss <- BSTA.Tss[,6:71] #From column Y.C.T to the end of spp columns

####Make a matrix with Site and Block columns to add random factors in the following PERMANOVAs BSTA.Tss_SB <- rbind(Y2013.Tss, Y2018.Tss) BSTA.Tss_SB <- BSTA.Tss_SB[,2:3]

Centroid for Cntrl

Data.BSTA.Tss<-droplevels(subset(BSTA.PCA.Tss)) Data0.BSTA.Tss<- Data.BSTA.Tss[, colSums(Data.BSTA.Tss != 0) > 0] # To eliminate empty (0) columns

Data transformation and PCA

PCA.crd.dec.Tss <- decostand(Data0.BSTA.Tss[,-1], "hellinger") PCA.BSTA.Tss <- prcomp(PCA.crd.dec.Tss, center = TRUE, scale. = FALSE) PCA.crd.dec.Tss <- cbind(Data0.BSTA.Tss$Y.C.T, PCA.crd.dec.Tss) #Adding colmn with names colnames(PCA.crd.dec.Tss) names(PCA.crd.dec.Tss)[names(PCA.crd.dec.Tss) == "Data0.BSTA.Tss$Y.C.T"] <- "Y.C.T" str(PCA.crd.dec.Tss) PCAD.BSTA.Tss <- as.data.frame(PCA.BSTA.Tss$x)

PCAD.BSTA.Tss$Year <- with(PCA.crd.dec.Tss, substr(Y.C.T, 1,4)) PCAD.BSTA.Tss$Canopy <- with(PCA.crd.dec.Tss, substr(Y.C.T, 6,7)) PCAD.BSTA.Tss$Treatment <- with(PCA.crd.dec.Tss, substr(Y.C.T, 9,nchar(as.character(Y.C.T))))

Mean for each treatment and year

pca.centroids.BSTA.Tss <- aggregate(PCAD.BSTA.Tss[,1:2], list(Year = PCAD.BSTA.Tss$Year, Canopy=PCAD.BSTA.Tss$Canopy, Treatment=PCAD.BSTA.Tss$Treatment), mean) pca.centroids.BSTA.Tss$LAB<-with(pca.centroids.BSTA.Tss, ifelse(Year==2018, "18", "13"))

Figure 7 (part), Save as 5x5 Landscape

Main.BSTA.Tss<-ggplot(data=pca.centroids.BSTA.Tss, aes(PC1, PC2, col=Treatment))+ theme_bw()+ geom_hline(yintercept = 0, lty=2, col="gray")+ geom_vline(xintercept = 0, lty=2, col="gray")+ geom_point(aes(shape = Year, color=Treatment),size = 2)+ scale_color_manual(values=c('gray53', 'green', 'blue'))+ geom_line(aes(PC1, PC2, group=interaction(Canopy, Treatment), linetype=Canopy)) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(),legend.position="bottom", plot.title = element_text(size=12,hjust = 0.5,face="bold"), axis.title=element_text(size=11),text = element_text(size=12), axis.text.x = element_text(angle=0, hjust=0.5, size=8), axis.text.y = element_text(angle=0,hjust=0.5,size=8))+ labs(x= "PC1 (38%)", y ="PC2 (9%)", title="Species abundance under BS and TA stands")

summary(PCA.BSTA.Tss)$importance[2,1] # PCA1: 0.38404 summary(PCA.BSTA.Tss)$importance[2,2] # PCA2: 0.09073

Subplot of species

scores(PCA.BSTA.Tss$center) #To get the scores of species and select the most important ones

Figure 7 (part-Species)

#Selected species with the highest scores (0.36-0.04): PLS,CON,GAH,PTC,RUP,LIB,MAC,VIS,LYA,CLB,VAM,DIP,PES,POA,RIG,POC,GAA Sub.BSTA.Tss<-fviz_pca_biplot(PCA.BSTA.Tss, label = "var",labelsize = 3, col.var = "black",habillage="none",title="", xlab="", ylab="", addEllipses=FALSE, geom.ind = "", select.var=list(name=c('PLS','CON','GAH','PTC','RUP','LIB','MAC','VIS','LYA','CLB','VAM','DIP','PES','POA','RIG','POC','GAA')), repel =TRUE, lable="ind", fill.ind = "white", legend.title = "SP")+ theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), legend.text=element_text(size=13), legend.title=element_text(size=15, face="bold"))+ theme_void()

Figure 7 (both joined), Save as 5x5 Landscape

plot.with.inset.BSTA.Tss <-ggdraw() + draw_plot(Main.BSTA.Tss) + draw_plot(Sub.BSTA.Tss, x = 0.212, y = 0.326, width = .60, height = .60)

Table S4.(Tss) - PERMANOVA

PERMANOVA Including both blocks and Sites in permutations! (08/08/2022). If Year is included in the equation, I don't need to include it as random factor (As discussed with Steve and Valentina)

perm <- how(nperm = 9999, blocks = BSTA.Tss_SB$Site,
plots = Plots(BSTA.Tss_SB$Block), Within(type = "free"))

set.seed(1989) ado.BSTA.Tss_random2 <- adonis2(PCA.crd.dec.Tss[, -1] ~ PCAD.BSTA.Tss$Year * PCAD.BSTA.Tss$Canopy * PCAD.BSTA.Tss$Treatment, permutations = perm, method = "euc", by = "term" ) ado.BSTA.Tss_random2

Ordinations for each objective and each forest type

Figure 8.

Principal Coordinate Analysis (PCoA) based on the Bray-Curtis distance with Cailliez correction of the understory species abundance in 2013 (crossed squares) and 2018 (filled squares), with their corresponding percentage of variances in each axis. Treatments of the ecosystem approach in black spruce (a) and trembling aspen forests (b) correspond to Light (Li, in yellow), Nutrients (Nu, in purple), Single-litter (1F, in orange), Double-litter (2F, in red) and Control (C, in grey), with the corresponding centroids (standard deviation). Treatments of the community approach in black spruce (c) and trembling aspen forests (d) correspond to Control (C, in grey), Transplants-in (Ti, in green) and Transplants-out (To, in blue). Treatments in black spruce forests in light colors and in trembling aspen forests in dark colors. Arrows correspond to the most important understory species defining the community composition (Envfit, P < 0.05). Notice that scales are different.

Table 3.

Differences in composition of understory plant communities between treatments and forest types.

Table 4.

###Homogeneity of multivariance dispersion of understory plant abundance among treatments and years for each forest type.

Tbio (Ecosystem approach)

PCoA separated by Canopy with 2013 and 2018 data, Treatments Tbio (Ecosystem approach)####

View(BSTA.Year) BSTA.Year0<-droplevels(subset(BSTA.Year)) BSTA_All<- BSTA.Year0[, colSums(BSTA.Year0 != 0) > 0] # To eliminate empty (0) columns (1 was eliminated) BSTA_All$Y.C.T <- NULL View(BSTA_All) #Table with Year 2013 and 2018, with all other info, to separate between canopies

###Separate by Canopy BSTA_Canopy<- split(BSTA_All, BSTA_All$Canopy) BS_BegEnd <- BSTA_Canopy$'BS' TA_BegEnd <- BSTA_Canopy$'TA'

Separate by Table of Vars and Spp

####BS BS_BegEnd_Var <- select(BS_BegEnd, 1:5) BS_BegEnd_Spp <- select(BS_BegEnd, 6:68)

####TA TA_BegEnd_Var <- select(TA_BegEnd, 1:5) TA_BegEnd_Spp <- select(TA_BegEnd, 6:68)

####Preparing data for PCoA Farben <- c("gold","mediumorchid","salmon1","firebrick2","antiquewhite4","lightseagreen","chartreuse3")

Tbio-TA

BS.bray <- vegdist(BS_BegEnd_Spp, method = "bray") # Distance matrix with method = bray (it's good for spp abundance) is.euclid(BS.bray) #FALSE (Bad because then it has negative eigenvalues)

PCoA with cmdscale (by adding 'add = TRUE', I get the caillez correction)

BS.bray.pcoa <- cmdscale(BS.bray, k=2, eig=TRUE, add = TRUE) #Multidimensional scaling (also known as PCoA)

Figure 8a

PCoA plot (with bray-curtis distance) - Save PDF as portrait 6x5 (name InR_PCoA_BS_Tbio)

ordi.pcoa_BS <- ordiplot(BS.bray.pcoa, display="sites", type="points", cex = 0.2)

points(ordi.pcoa_BS$sites[BS_BegEnd_Var$Year == '2013' & BS_BegEnd_Var$Treatment == '1F',], pch=7, col="salmon1", bg="salmon1", cex = 1.3) points(ordi.pcoa_BS$sites[BS_BegEnd_Var$Year == '2018' & BS_BegEnd_Var$Treatment == '1F',], pch=15, col="salmon1", bg="salmon1", cex = 1.3)

points(ordi.pcoa_BS$sites[BS_BegEnd_Var$Year == '2013' & BS_BegEnd_Var$Treatment == '2F',], pch=7, col="firebrick2", bg="firebrick2", cex = 1.3) points(ordi.pcoa_BS$sites[BS_BegEnd_Var$Year == '2018' & BS_BegEnd_Var$Treatment == '2F',], pch=15, col="firebrick2", bg="firebrick2", cex = 1.3)

points(ordi.pcoa_BS$sites[BS_BegEnd_Var$Year == '2013' & BS_BegEnd_Var$Treatment == 'C',], pch=7, col="antiquewhite4", bg="antiquewhite4", cex = 1.3) points(ordi.pcoa_BS$sites[BS_BegEnd_Var$Year == '2018' & BS_BegEnd_Var$Treatment == 'C',], pch=15, col="antiquewhite4", bg="antiquewhite4", cex = 1.3)

points(ordi.pcoa_BS$sites[BS_BegEnd_Var$Year == '2013' & BS_BegEnd_Var$Treatment == 'Li',], pch=7, col="gold", bg="gold", cex = 1.3) points(ordi.pcoa_BS$sites[BS_BegEnd_Var$Year == '2018' & BS_BegEnd_Var$Treatment == 'Li',], pch=15, col="gold", bg="gold", cex = 1.3)

points(ordi.pcoa_BS$sites[BS_BegEnd_Var$Year == '2013' & BS_BegEnd_Var$Treatment == 'Nu',], pch=7, col="mediumorchid", bg="mediumorchid", cex = 1.3) points(ordi.pcoa_BS$sites[BS_BegEnd_Var$Year == '2018' & BS_BegEnd_Var$Treatment == 'Nu',], pch=15, col="mediumorchid", bg="mediumorchid", cex = 1.3)

ordiellipse(ordi.pcoa_BS, BS_BegEnd_Var$Treatment, label=TRUE, col = Farben, cex=1.2, font=2, lty=2, lwd=1)

plot(envfit(ordi.pcoa_BS, BS_BegEnd_Spp), p.max=0.05, col="gray25", cex=1, font=2)

#Legend 1 legend(0.58, 0.4, legend = c("2013","2018"), #names to display col = c("gray25", "gray25"), pch = c(7,15), #symbol type bty = "n", #type of box around the legend pt.cex = 1.3, #symbol size cex = 1, #text size text.col = "gray25", horiz = F , inset = c(1.5, 1.5)) #% (from 0 to 1) to draw the legend away from x and y axis. You can also give the X and Y coordinate of the legend: legend(3, 5, ...) #Legend 2 legend(0.58, 0.4, legend = c("Li","Nu","1F","2F","C","Ti","To"), #names to display col = Farben, pch = c(0,0,0,0,0,0,0), #symbol type bty = "n", #type of box around the legend pt.cex = 1.3, #symbol size cex = 1, #text size text.col = Farben, horiz = F)

Eigenvalues and their proportion to the sum

BS.bray.pcoa$eig[1:3] # 14.650921 10.682960 7.390236

Calculate the percent of variance explained by first two axes

100*BS.bray.pcoa$eig[1:3]/sum(BS.bray.pcoa$eig) # 15.020097 10.952151 7.576457

Table 3 - PERMANOVA Tbio-BS

Including both blocks and Sites in permutations

perm_BS_Tbio <- how(nperm = 9999, blocks = BS_BegEnd_Var$Site,
plots = Plots(BS_BegEnd_Var$Block), Within(type = "free"))

set.seed(1989) BS.bray.pcoa_perm_bray2 <- adonis2(BS.bray ~ Treatment*Year, data=BS_BegEnd_Var, permutations = perm_BS_Tbio, method = "bray") BS.bray.pcoa_perm_bray2

Table 4. Tbio-BS

Posthoc tests - PERMUTEST

BS_BegEnd_Var2 <- BS_BegEnd_Var BS_BegEnd_Var2$YT <- paste(BS_BegEnd_Var2$Year,BS_BegEnd_Var2$Treatment)

dispersion_BS_Tbio <- betadisper(BS.bray, group=BS_BegEnd_Var2$YT,type = c("median","centroid"), bias.adjust = TRUE) dispersion_BS_Tbio # !!! To get the contrast (average distance to median) perm_BSTbio <- permutest(dispersion_BS_Tbio, pairwise = TRUE, permutations = 9999) # !!! TO get the P value in contrasts

anova(dispersion_BS_Tbio)

mod.HSD_BS_Tbio <- TukeyHSD(dispersion_BS_Tbio) mod.HSD_BS_Tbio plot(mod.HSD_BS_Tbio) pairwise.adonis2(BS.bray ~ YT, data = BS_BegEnd_Var2)

Tbio-TA

TA.bray <- vegdist(TA_BegEnd_Spp, method = "bray") # Distance matrix with method = bray (it's good for spp abundance) is.euclid(TA.bray) #FALSE

PCoA with cmdscale (by adding 'add = TRUE', I get the caillez correction)

TA.bray.pcoa <- cmdscale(TA.bray, k=2, eig=TRUE, add = TRUE) #Multidimensional scaling (also known as PCoA)

Figure 8b

PCoA plot (with bray-curtis distance) - Save PDF as portrait 6x5 (name InR_PCoA_TA_Tbio)

ordi.pcoa_TA <- ordiplot(TA.bray.pcoa, display="sites", type="points", cex = 0.2)

points(ordi.pcoa_TA$sites[TA_BegEnd_Var$Year == '2013' & TA_BegEnd_Var$Treatment == '1F',], pch=7, col="salmon1", bg="salmon1", cex = 1.3) points(ordi.pcoa_TA$sites[TA_BegEnd_Var$Year == '2018' & TA_BegEnd_Var$Treatment == '1F',], pch=15, col="salmon1", bg="salmon1", cex = 1.3)

points(ordi.pcoa_TA$sites[TA_BegEnd_Var$Year == '2013' & TA_BegEnd_Var$Treatment == '2F',], pch=7, col="firebrick2", bg="firebrick2", cex = 1.3) points(ordi.pcoa_TA$sites[TA_BegEnd_Var$Year == '2018' & TA_BegEnd_Var$Treatment == '2F',], pch=15, col="firebrick2", bg="firebrick2", cex = 1.3)

points(ordi.pcoa_TA$sites[TA_BegEnd_Var$Year == '2013' & TA_BegEnd_Var$Treatment == 'C',], pch=7, col="antiquewhite4", bg="antiquewhite4", cex = 1.3) points(ordi.pcoa_TA$sites[TA_BegEnd_Var$Year == '2018' & TA_BegEnd_Var$Treatment == 'C',], pch=15, col="antiquewhite4", bg="antiquewhite4", cex = 1.3)

points(ordi.pcoa_TA$sites[TA_BegEnd_Var$Year == '2013' & TA_BegEnd_Var$Treatment == 'Li',], pch=7, col="gold", bg="gold", cex = 1.3) points(ordi.pcoa_TA$sites[TA_BegEnd_Var$Year == '2018' & TA_BegEnd_Var$Treatment == 'Li',], pch=15, col="gold", bg="gold", cex = 1.3)

points(ordi.pcoa_TA$sites[TA_BegEnd_Var$Year == '2013' & TA_BegEnd_Var$Treatment == 'Nu',], pch=7, col="mediumorchid", bg="mediumorchid", cex = 1.3) points(ordi.pcoa_TA$sites[TA_BegEnd_Var$Year == '2018' & TA_BegEnd_Var$Treatment == 'Nu',], pch=15, col="mediumorchid", bg="mediumorchid", cex = 1.3)

ordiellipse(ordi.pcoa_TA, TA_BegEnd_Var$Year, label=TRUE, col = "gray25", cex=1.2, font=2, lty=1, lwd=1) #the default for the ellipse is the standard deviation (kind="sd") ordiellipse(ordi.pcoa_TA, TA_BegEnd_Var$Treatment, label=TRUE, col = Farben, cex=1.2, font=2, lty=2, lwd=1)

plot(envfit(ordi.pcoa_TA, TA_BegEnd_Spp), p.max=0.05, col="gray25", cex=1, font=2)

#Legend 1 legend(0.58, 0.4, legend = c("2013","2018"), #names to display col = c("gray25", "gray25"), pch = c(7,15), #symbol type bty = "n", #type of box around the legend pt.cex = 1.3, #symbol size cex = 1, #text size text.col = "gray25", horiz = F , inset = c(1.5, 1.5)) #% (from 0 to 1) to draw the legend away from x and y axis. You can also give the X and Y coordinate of the legend: legend(3, 5, ...) #Legend 2 legend(1.5, 0.4, legend = c("Li","Nu","1F","2F","C","Ti","To"), #names to display col = Farben, pch = c(0,0,0,0,0,0,0), #symbol type bty = "n", #type of box around the legend pt.cex = 1.3, #symbol size cex = 1, #text size text.col = Farben, horiz = F)

Eigenvalues and their proportion to the sum

TA.bray.pcoa$eig[1:3] # 10.419424 7.126796 4.062613

Calculate the percent of variance explained by first two axes

100*TA.bray.pcoa$eig[1:3]/sum(TA.bray.pcoa$eig) # 14.647095 10.018487 5.711014

Table 3 - PERMANOVA Tbio-TA

Including both blocks and Sites in permutations

perm_TA_Tbio <- how(nperm = 9999, blocks = TA_BegEnd_Var$Site,
plots = Plots(TA_BegEnd_Var$Block), Within(type = "free"))

set.seed(1989) TA.bray.pcoa_perm_bray2 <- adonis2(TA.bray ~ Treatment*Year, data=TA_BegEnd_Var, permutations = perm_TA_Tbio, method = "bray") TA.bray.pcoa_perm_bray2

Table 4. Tbio-TA

Posthoc tests - PERMUTEST

TA_BegEnd_Var2 <- TA_BegEnd_Var TA_BegEnd_Var2$YT <- paste(TA_BegEnd_Var2$Year,TA_BegEnd_Var2$Treatment)

dispersion_TA_Tbio <- betadisper(TA.bray, group=TA_BegEnd_Var2$YT,type = c("median","centroid"), bias.adjust = TRUE) dispersion_TA_Tbio # !!! To get the contrast (average distance to median) perm_TATbio <- permutest(dispersion_TA_Tbio, pairwise = TRUE, permutations = 9999) # !!! TO get the P value in contrasts

anova(dispersion_TA_Tbio)

mod.HSD_TA_Tbio <- TukeyHSD(dispersion_TA_Tbio) mod.HSD_TA_Tbio plot(mod.HSD_TA_Tbio) pairwise.adonis2(TA.bray ~ YT, data = TA_BegEnd_Var2)

Tss (Community approach)

PCoA separated by Canopy with 2013 and 2018 data, Treatments Tss (Community approach)####

View(BSTA.Tss) BSTA.Tss0<-droplevels(subset(BSTA.Tss)) BSTA_All_Tss<- BSTA.Tss0[, colSums(BSTA.Tss0 != 0) > 0] # To eliminate empty (0) columns (12 were eliminated) BSTA_All_Tss$Y.C.T <- NULL View(BSTA_All_Tss) #Table with Year 2013 and 2018, with all other info, to separate between canopies

####Separate by Canopy BSTA_Tss_Canopy<- split(BSTA_All_Tss, BSTA_All_Tss$Canopy) BS_BegEnd_Tss <- BSTA_Tss_Canopy$'BS' TA_BegEnd_Tss <- BSTA_Tss_Canopy$'TA'

Separate by Table of Vars and Spp

####BS BS_BegEnd_Tss_Var <- select(BS_BegEnd_Tss, 1:5) BS_BegEnd_Tss_Spp <- select(BS_BegEnd_Tss, 6:59)

####TA TA_BegEnd_Tss_Var <- select(TA_BegEnd_Tss, 1:5) TA_BegEnd_Tss_Spp <- select(TA_BegEnd_Tss, 6:59)

####Preparing data for PCoA Farben_Tss <- c("lightseagreen","chartreuse3","antiquewhite4")

Tss-BS

BS.bray_Tss <- vegdist(BS_BegEnd_Tss_Spp, method = "bray") # Distance matrix with method = bray (it's good for spp abundance) is.euclid(BS.bray_Tss) #FALSE

PCoA with cmdscale (by adding 'add = TRUE', I get the caillez correction)

BS.bray.pcoa_Tss <- cmdscale(BS.bray_Tss, k=2, eig=TRUE, add = TRUE) #Multidimansional scaling (also known as PCoA)

Figure 8c

PCoA plot (with bray-curtis distance) - Save PDF as portrait 6x5 (name InR_PCoA_BS_Tss)

ordi.pcoa_BS_Tss <- ordiplot(BS.bray.pcoa_Tss, display="sites", type="points", cex = 0.2)

points(ordi.pcoa_BS_Tss$sites[BS_BegEnd_Tss_Var$Year == '2013' & BS_BegEnd_Tss_Var$Treatment == 'C',], pch=7, col="antiquewhite4", bg="antiquewhite4", cex = 1.3) points(ordi.pcoa_BS_Tss$sites[BS_BegEnd_Tss_Var$Year == '2018' & BS_BegEnd_Tss_Var$Treatment == 'C',], pch=15, col="antiquewhite4", bg="antiquewhite4", cex = 1.3)

points(ordi.pcoa_BS_Tss$sites[BS_BegEnd_Tss_Var$Year == '2013' & BS_BegEnd_Tss_Var$Treatment == 'Ti',], pch=7, col="chartreuse3", bg="chartreuse3", cex = 1.3) points(ordi.pcoa_BS_Tss$sites[BS_BegEnd_Tss_Var$Year == '2018' & BS_BegEnd_Tss_Var$Treatment == 'Ti',], pch=15, col="chartreuse3", bg="chartreuse3", cex = 1.3)

points(ordi.pcoa_BS_Tss$sites[BS_BegEnd_Tss_Var$Year == '2013' & BS_BegEnd_Tss_Var$Treatment == 'To',], pch=7, col="lightseagreen", bg="lightseagreen", cex = 1.3) points(ordi.pcoa_BS_Tss$sites[BS_BegEnd_Tss_Var$Year == '2018' & BS_BegEnd_Tss_Var$Treatment == 'To',], pch=15, col="lightseagreen", bg="lightseagreen", cex = 1.3)

ordiellipse(ordi.pcoa_BS_Tss, BS_BegEnd_Tss_Var$Treatment, label=TRUE, col = Farben_Tss, cex=1.2, font=2, lty=2, lwd=1)

plot(envfit(ordi.pcoa_BS_Tss, BS_BegEnd_Tss_Spp), p.max=0.05, col="gray25", cex=1, font=2)

#Legend 1 legend(0.58, 0.4, legend = c("2013","2018"), #names to display col = c("gray25", "gray25"), pch = c(7,15), #symbol type bty = "n", #type of box around the legend pt.cex = 1.3, #symbol size cex = 1, #text size text.col = "gray25", horiz = F , inset = c(1.5, 1.5)) #% (from 0 to 1) to draw the legend away from x and y axis. You can also give the X and Y coordinate of the legend: legend(3, 5, ...) #Legend 2 legend(0.58, 0.4, legend = c("C","Ti","To"), #names to display col = Farben_Tss, pch = c(0,0,0), #symbol type bty = "n", #type of box around the legend pt.cex = 1.3, #symbol size cex = 1, #text size text.col = Farben_Tss, horiz = F)

Eigenvalues and their proportion to the sum

BS.bray.pcoa_Tss$eig[1:3] # 12.605008 4.847055 2.556525

Calculate the percent of variance explained by first two axes

100*BS.bray.pcoa_Tss$eig[1:3]/sum(BS.bray.pcoa_Tss$eig) # 27.376608 10.527239 5.552475

Table 3 - PERMANOVA Tss-BS

Including both blocks and Sites in permutations

perm_BS_Tss <- how(nperm = 9999, blocks = BS_BegEnd_Tss_Var$Site,
plots = Plots(BS_BegEnd_Tss_Var$Block), Within(type = "free"))

set.seed(1989) BS.bray_Tss_perm_bray2 <- adonis2(BS.bray_Tss ~ Treatment*Year, data=BS_BegEnd_Tss_Var, permutations = perm_BS_Tss, method = "bray") BS.bray_Tss_perm_bray2

Table 4. Tbio-TA

Posthoc tests - PERMUTEST

BS_BegEnd_Tss_Var2 <- BS_BegEnd_Tss_Var BS_BegEnd_Tss_Var2$YT <- paste(BS_BegEnd_Tss_Var2$Year,BS_BegEnd_Tss_Var2$Treatment)

dispersion_BS_Tss <- betadisper(BS.bray_Tss, group=BS_BegEnd_Tss_Var2$YT,type = c("median","centroid"), bias.adjust = TRUE) dispersion_BS_Tss # !!! To get the contrast (average distance to median) perm_BSTss <- permutest(dispersion_BS_Tss, pairwise = TRUE, permutations = 9999) # !!! TO get the P value in contrasts

anova(dispersion_BS_Tss)

mod.HSD_BS_Tss <- TukeyHSD(dispersion_BS_Tss) mod.HSD_BS_Tss plot(mod.HSD_BS_Tss) pairwise.adonis2(BS.bray_Tss ~ YT, data = BS_BegEnd_Var2)

Tss-TA

TA.bray_Tss <- vegdist(TA_BegEnd_Tss_Spp, method = "bray") # Distance matrix with method = bray (it's good for spp abundance) is.euclid(TA.bray_Tss) #FALSE

PCoA with cmdscale (by adding 'add = TRUE', I get the caillez correction)

TA.bray.pcoa_Tss <- cmdscale(TA.bray_Tss, k=2, eig=TRUE, add = TRUE) #Multidimensional scaling (also known as PCoA)

Figure 8d

PCoA plot (with bray-curtis distance) - Save PDF as portrait 6x5 (name InR_PCoA_TA_Tss)

ordi.pcoa_TA_Tss <- ordiplot(TA.bray.pcoa_Tss, display="sites", type="points", cex = 0.2, col="cornsilk2")

points(ordi.pcoa_TA_Tss$sites[TA_BegEnd_Tss_Var$Year == '2013' & TA_BegEnd_Tss_Var$Treatment == 'C',], pch=7, col="antiquewhite4", bg="antiquewhite4", cex = 1.3) points(ordi.pcoa_TA_Tss$sites[TA_BegEnd_Tss_Var$Year == '2018' & TA_BegEnd_Tss_Var$Treatment == 'C',], pch=15, col="antiquewhite4", bg="antiquewhite4", cex = 1.3)

points(ordi.pcoa_TA_Tss$sites[TA_BegEnd_Tss_Var$Year == '2013' & TA_BegEnd_Tss_Var$Treatment == 'Ti',], pch=7, col="chartreuse3", bg="chartreuse3", cex = 1.3) points(ordi.pcoa_TA_Tss$sites[TA_BegEnd_Tss_Var$Year == '2018' & TA_BegEnd_Tss_Var$Treatment == 'Ti',], pch=15, col="chartreuse3", bg="chartreuse3", cex = 1.3)

points(ordi.pcoa_TA_Tss$sites[TA_BegEnd_Tss_Var$Year == '2013' & TA_BegEnd_Tss_Var$Treatment == 'To',], pch=7, col="lightseagreen", bg="lightseagreen", cex = 1.3) points(ordi.pcoa_TA_Tss$sites[TA_BegEnd_Tss_Var$Year == '2018' & TA_BegEnd_Tss_Var$Treatment == 'To',], pch=15, col="lightseagreen", bg="lightseagreen", cex = 1.3)

ordiellipse(ordi.pcoa_TA_Tss, TA_BegEnd_Tss_Var$Year, label=TRUE, col = "gray25", cex=1.2, font=2, lty=1, lwd=1) #the default for the ellipse is the standard deviation (kind="sd") ordiellipse(ordi.pcoa_TA_Tss, TA_BegEnd_Tss_Var$Treatment, label=TRUE, col = Farben_Tss, cex=1.2, font=2, lty=2, lwd=1)

plot(envfit(ordi.pcoa_TA_Tss, TA_BegEnd_Tss_Spp), p.max=0.05, col="gray25", cex=1, font=2)

#Legend 1 legend(0.58, 0.4, legend = c("2013","2018"), #names to display col = c("gray25", "gray25"), pch = c(7,15), #symbol type bty = "n", #type of box around the legend pt.cex = 1.3, #symbol size cex = 1, #text size text.col = "gray25", horiz = F , inset = c(1.5, 1.5)) #% (from 0 to 1) to draw the legend away from x and y axis. You can also give the X and Y coordinate of the legend: legend(3, 5, ...) #Legend 2 legend(-0.5, 0.4, legend = c("C","Ti","To"), #names to display col = Farben_Tss, pch = c(0,0,0), #symbol type bty = "n", #type of box around the legend pt.cex = 1.3, #symbol size cex = 1, #text size text.col = Farben_Tss, horiz = F)

Eigenvalues and their proportion to the sum

TA.bray.pcoa_Tss$eig[1:3] # 10.131209 4.704789 3.861210

Calculate the percent of variance explained by first two axes

100*TA.bray.pcoa_Tss$eig[1:3]/sum(TA.bray.pcoa_Tss$eig) # 22.128320 10.276077 8.433552

Table 3 - PERMANOVA Tss-TA

Including both blocks and Sites in permutations

perm_TA_Tss <- how(nperm = 9999, blocks = TA_BegEnd_Tss_Var$Site,
plots = Plots(TA_BegEnd_Tss_Var$Block), Within(type = "free"))

set.seed(1989) TA.bray_Tss_perm_bray2 <- adonis2(TA.bray_Tss ~ Treatment*Year, data=TA_BegEnd_Tss_Var, permutations = perm_TA_Tss, method = "bray") TA.bray_Tss_perm_bray2

Table 4. Tbio-TA

Posthoc tests - PERMUTEST

TA_BegEnd_Tss_Var2 <- TA_BegEnd_Tss_Var TA_BegEnd_Tss_Var2$YT <- paste(TA_BegEnd_Tss_Var2$Year,TA_BegEnd_Tss_Var2$Treatment)

dispersion_TA_Tss <- betadisper(TA.bray_Tss, group=TA_BegEnd_Tss_Var2$YT,type = c("median","centroid"), bias.adjust = TRUE) dispersion_TA_Tss # !!! To get the contrast (average distance to median) perm_TATss <- permutest(dispersion_TA_Tss, pairwise = TRUE, permutations = 9999) # !!! TO get the P value in contrasts

anova(dispersion_TA_Tss)

mod.HSD_TA_Tss <- TukeyHSD(dispersion_TA_Tss) mod.HSD_TA_Tss plot(mod.HSD_TA_Tss) pairwise.adonis2(TA.bray_Tss ~ YT, data = TA_BegEnd_Var2)

Abundance of species/Functional groups in 2018

Supp.Table 1 and Boxplots plot (Figure 2) for functional groups in 2018

DB_All3 <- DB_All DB_All3_Year <- split(DB_All3, DB_All3$Year) DB_All3_2018 <- DB_All3_Year$2018

Table S1.

Average abundance of understory plants of the cumulative effect of treatments in 2018 in black spruce and trembling aspen forests, for the Control (C, in grey) and the different treatments of the ecosystem approach (Objective 1): Light (Li, in yellow), Nutrients (Nu, in purple), Single-litter (1F, in orange) and Double-litter (2F, in red), and of the community approach (Objective 2): Transplants-out (To, in blue), Transplants-in (Ti, in green). Species, with their corresponding abbreviations, were classified in functional groups: Bryophytes, Sphagnaceae, Ericaceae, Pteridophyta, Graminoids, Herbs, and Shrubs, and Trees. Data correspond to the average and standard deviation (in grey numbers) of blocks and sites for each species among treatments and forest types. Values for each functional group (in colors) correspond to the average abundance per functional group for each treatment.

####Do a sum of Coverage per Func_gr Func18 <- DB_All3_2018 Func18$Year <- NULL

####make a joint column Func18_united <- unite(Func18, Site, Canopy, Treatment, Species, Abbr, Func_gr, col = "SCTSAF", sep = "_")

####Do a mean of Coverage per Block Func18_mean <- aggregate(Func18_united$Coverage, list(Func18_united$SCTSAF), FUN=mean)

####Arrange columns again Func18_mean2 <- separate(Func18_mean, Group.1, into = c("Site","Canopy","Treatment","Species","Abbr","Func_gr"), sep = "") Func18_mean3 <- unite(Func18_mean2, Canopy, Treatment, Species, Abbr, Func_gr, col = "CTSAF", sep = "")

####Do a mean of Coverage per Site (from mean of blocks) Func18_mean4 <- aggregate(Func18_mean3$x, list(Func18_mean3$CTSAF), FUN=mean) Func18_SD <- aggregate(Func18_mean3$x, list(Func18_mean3$CTSAF), FUN=sd)

####Arrange columns again Func18_mean5 <- separate(Func18_mean4, Group.1, into = c("Canopy","Treatment","Species","Abbr","Func_gr"), sep = "") Func18_mean6 <- unite(Func18_mean5, Canopy, Species, Abbr, Func_gr, col = "CSAF", sep = "")

Func18_SD2 <- separate(Func18_SD, Group.1, into = c("Canopy","Treatment","Species","Abbr","Func_gr"), sep = "") Func18_SD3 <- unite(Func18_SD2, Canopy, Species, Abbr, Func_gr, col = "CSAF", sep = "")

####Spread Treatmetns Func18_mean7 <- spread(Func18_mean6, Treatment, x, fill = Func18_mean6$x) Func18_SD4 <- spread(Func18_SD3, Treatment, x, fill = Func18_SD3$x)

####Merge to get a tbale with mean and SD for each treatment Func18_merged <- merge(Func18_mean7,Func18_SD4, by="CSAF") Func18_merged2 <- separate(Func18_merged, CSAF, into = c("Canopy","Species","Abbr","Func_gr"), sep = "_")

####Merge again from tables before spreading to get a table with mean and SD for each treatment (but witouth spread to arrange in Excel) Func18_merged3 <- merge(Func18_mean4,Func18_SD, by="Group.1") Func18_merged4 <- separate(Func18_merged3, Group.1, into = c("Canopy","Treatment","Species","Abbr","Func_gr"), sep = "_") ####write.csv(Func18_merged4, "Func18_merged4.csv")

Prepare data for Boxplots (Figure 2) for functional groups in 2018 only C treatment

DB3_2018_Treat <- split(DB_All3_2018, DB_All3_2018$Treatment) DB3_2018_C <- DB3_2018_Treat$'C'

#####Calculate cover by C plot for each of each Func-gr DB3_2018_C2 <- DB3_2018_C DB3_2018_C2$Species <- NULL DB3_2018_C2$Abbr <- NULL DB3_2018_C2$SBC <- paste0(DB3_2018_C2$Site,"", DB3_2018_C2$Block, "", DB3_2018_C2$Canopy) #Create a single column for variables DB3_2018_C2$Year <- NULL DB3_2018_C2$Treatment <- NULL

####Do a sum of Coverage per Func_gr Func18_C <- DB3_2018_C2 %>% group_by(Site, Block, Canopy, Func_gr) %>% summarise(Coverage = sum(Coverage))

make a column joining Site and Block

Func18_C2 <- unite(Func18_C, Site, Canopy, Func_gr, col = "Site_Can_Func", sep = " ") ####Do a mean of Coverage per Block Func18_C_mean <- aggregate(Func18_C2$Coverage, list(Func18_C2$Site_Can_Func), FUN=mean) ####Arrange columns again Func18_C_mean2 <- separate(Func18_C_mean, Group.1, into = c("Site", "Canopy", "Func_gr"), sep = " ") Func18_C_mean3 <- unite(Func18_C_mean2, Canopy, Func_gr, col = "Can_Func", sep = " ") ####Do a mean and SD of Coverage (x) per Site Func18_C_mean4 <- aggregate(Func18_C_mean3$x, list(Func18_C_mean3$Can_Func), FUN=mean) Func18_C_SD <- aggregate(Func18_C_mean3$x, list(Func18_C_mean3$Can_Func), FUN=sd) Func18_C_mean5 <- merge(Func18_C_mean4,Func18_C_SD, by="Group.1") ####Arrange columns again colnames(Func18_C_mean5)[2] <- "Mean" colnames(Func18_C_mean5)[3] <- "SD" Func18_C_mean6 <- separate(Func18_C_mean5, Group.1, into = c("Canopy", "Func_gr"), sep = " ") ####With Func18_C_mean6 I got the table with the values of Mean and SD per Func_gr per Canopy, but for the figure, I must use data before calculating the mean, so (Func18_C_mean3)

####Arrange Func18_C_mean3 (with Mean per Blocks) for the boxplot Func18_C_mean7 <- separate(Func18_C_mean3, Can_Func, into = c("Canopy", "Func_gr"), sep = " ") colnames(Func18_C_mean7)[4] <- "Mean" Func18_C_mean8 <- Func18_C_mean7 Func18_C_mean8$Site <- NULL

Figure 2.

Experimental design for overstory – understory relationships. a) The study area is located in the boreal forest of eastern Canada; b) above the parallel 49⁰N in western Quebec. c) The experiment was conducted in three sites A, B and C (Sites B and C are 0.5 km apart and Site A is 5.3 km away). d,e,f) Each site has adjacent stands (separated from 34 to 115 m) dominated by black spruce (BS, as triangles) and trembling aspen (TA, as circles), in which we placed 3 blocks (B1, B2, B3) separated from 16 to 49 m. g). In each block, one plot per treatment was placed to obtain a nested block experimental design with a total of 126 plots (3 sites x 2 forest types x 3 blocks x 7 treatments = 126 plots).. Vegetation abundance in all treatments was measured every year from 2013 to 2018 by using a 50 x 50 cm grid placed in the middle of each plot. Treatments in different colors correspond to Control conditions (C, in grey) and to treatments of the ecosystem approach: Light (Li, in yellow), Nutrients (Nu, in purple), Single-litter (1F, in orange) and Double-litter (2F, in red); and of the community approach: Transplants-out (To, in blue), Transplants-in (Ti, in green). Treatments in black spruce forests in light colors and in trembling aspen forests in dark colors.

Boxplot Coverage of Control plots in 2018 (average per Site (from averages per Blocks)) Save as PDF 8x6 portrait

cov.plot <- ggplot(Func18_C_mean8, aes(x = Func_gr, y =Mean, fill=Func_gr)) + geom_boxplot()+ facet_grid(cols=vars(Canopy)) + scale_fill_brewer(palette="Dark2") + theme_light(base_size = 12)

cov.plot.lm <- lm(Mean ~ Func_gr + Canopy + Func_gr:Canopy, data = Func18_C_mean8) summary(cov.plot.lm)

cov.emmeans <- emmeans(cov.plot.lm, specs = pairwise ~ Func_gr|Canopy, type = "response") cov.emmeans

plot(cov.emmeans, comparisons = TRUE)

cov.mod_means <- multcomp::cld(object = cov.emmeans$emmeans,

                           Letters = letters,no.readonly=TRUE)

cov.mod_means

Figure S3 and Figure S4 - Smoothlines

Variation in abundance of functional groups (Bryophytes, Ericaceae, Pteridophyta, Herbs and Shrubs) over time (from 2013 to 2018) for black spruce (left panels) and trembling aspen stands (right panels).

Load main Table with all information and with the five Func-gr

DB_All_FG <- DB_All

Reduce table to leave just Func_gr and sum coverage (eliminating Species and Abbr)

DB_All_FG2 <- DB_All_FG %>% group_by(Year, Site, Block, Canopy, Treatment, Func_gr) %>% summarise(Coverage = sum(Coverage)) head(DB_All_FG2) ####Rename levels of Canopy DB_All_FG2$Canopy <- recode_factor(DB_All_FG2$Canopy, BS = "Black spruce", TA = "Trembling aspen")

Tbio

Figure S2.

Variation in abundance of functional groups over time (from 2013 to 2018) for black spruce (left panels) and trembling aspen stands (right panels). Functional groups: (a) Bryophytes, (b) Sphagnaceae, (c) Ericaceae, (d) Peridophyta, (e) Herbs, (f) Shrubs, and (g) Trees. Treatments correspond to ecosystem approach: Light (Li, in yellow), Nutrients (Nu, in purple), Single-litter (1F, in orange) and Double-litter (2F, in red) and Control conditions (C, in grey). Smooth lines are based on the linear model of understory species abundances, each point in different colors (treatments) from data each year of the study.

Filter by treatments Tbio to leave only (C, Li, Nu, 1F and 2F)

DB_Tbio_FG <- DB_All_FG2 %>% filter(Treatment %in% c("C", "Li", "Nu", "1F", "2F")) %>% droplevels() unique(DB_Tbio_FG$Treatment)

Tbio, Func_gr: Herbs

DB_Tbio_Her <- DB_Tbio_FG %>% filter(Func_gr %in% c("Herbs")) %>% droplevels() unique(DB_Tbio_Her$Func_gr)

####Herbs (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tbio_Her <- ggplot(DB_Tbio_Her, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Herbs")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("salmon1","firebrick2","antiquewhite4","gold","mediumorchid"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tbio_Her

Tbio, Func_gr: Bryophytes

DB_Tbio_Bry <- DB_Tbio_FG %>% filter(Func_gr %in% c("Bryophytes")) %>% droplevels() unique(DB_Tbio_Bry$Func_gr)

####Bryophytes (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tbio_Bry <- ggplot(DB_Tbio_Bry, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Bryophytes")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("salmon1","firebrick2","antiquewhite4","gold","mediumorchid"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tbio_Bry

Tbio, Func_gr: Ericaceae

DB_Tbio_Eri <- DB_Tbio_FG %>% filter(Func_gr %in% c("Ericaceae")) %>% droplevels() unique(DB_Tbio_Eri$Func_gr)

####Ericaceae (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tbio_Eri <- ggplot(DB_Tbio_Eri, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Ericaceae")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("salmon1","firebrick2","antiquewhite4","gold","mediumorchid"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tbio_Eri

Tbio, Func_gr: Pteridophyta

DB_Tbio_Pte <- DB_Tbio_FG %>% filter(Func_gr %in% c("Pteridophyta")) %>% droplevels() unique(DB_Tbio_Pte$Func_gr)

####Pteridophyta (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tbio_Pte <- ggplot(DB_Tbio_Pte, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Pteridophyta")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("salmon1","firebrick2","antiquewhite4","gold","mediumorchid"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tbio_Pte

Tbio, Func_gr: Shrubs

DB_Tbio_Shr <- DB_Tbio_FG %>% filter(Func_gr %in% c("Shrubs")) %>% droplevels() unique(DB_Tbio_Shr$Func_gr)

####Shrubs (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tbio_Sch <- ggplot(DB_Tbio_Shr, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Shrubs")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("salmon1","firebrick2","antiquewhite4","gold","mediumorchid"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tbio_Sch

Tbio, Func_gr: Sphagnaceae

DB_Tbio_Sph <- DB_Tbio_FG %>% filter(Func_gr %in% c("Sphagnaceae")) %>% droplevels() unique(DB_Tbio_Sph$Func_gr)

####Sphagnaceae (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tbio_Sph <- ggplot(DB_Tbio_Sph, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Sphagnaceae")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("salmon1","firebrick2","antiquewhite4","gold","mediumorchid"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tbio_Sph

Tbio, Func_gr: Trees

DB_Tbio_Tre <- DB_Tbio_FG %>% filter(Func_gr %in% c("Trees")) %>% droplevels() unique(DB_Tbio_Tre$Func_gr)

####Trees (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tbio_Tre <- ggplot(DB_Tbio_Tre, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Trees")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("salmon1","firebrick2","antiquewhite4","gold","mediumorchid"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tbio_Tre

Tbio, Func_gr: Graminoids

DB_Tbio_Gra <- DB_Tbio_FG %>% filter(Func_gr %in% c("Graminoids")) %>% droplevels() unique(DB_Tbio_Gra$Func_gr)

####Graminoids (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tbio_Gra <- ggplot(DB_Tbio_Gra, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Graminoids")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("salmon1","firebrick2","antiquewhite4","gold","mediumorchid"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tbio_Gra

Tss

Figure S3.

Variation in abundance of functional groups over time (from 2013 to 2018) for black spruce (left panels) and trembling aspen stands (right panels). Functional groups: (a) Bryophytes, (b) Sphagnaceae, (c) Ericaceae, (d) Peridophyta, (e) Herbs, (f) Shrubs, and (g) Trees. Treatments correspond to ecosystem approach: Light (Li, in yellow), Nutrients (Nu, in purple), Single-litter (1F, in orange) and Double-litter (2F, in red) and Control conditions (C, in grey). Smooth lines are based on the linear model of understory species abundances, each point in different colors (treatments) from data each year of the study.

Filter by treatments Tss to leave only (C, Ti, To)

DB_Tss_FG <- DB_All_FG2 %>% filter(Treatment %in% c("C", "Ti", "To")) %>% droplevels() unique(DB_Tss_FG$Treatment)

Tss, Func_gr: Herbs

DB_Tss_Her <- DB_Tss_FG %>% filter(Func_gr %in% c("Herbs")) %>% droplevels() unique(DB_Tss_Her$Func_gr)

####Herbs (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tss_Her <- ggplot(DB_Tss_Her, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Herbs")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("antiquewhite4","lightseagreen","chartreuse3"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tss_Her

Tss, Func_gr: Bryophytes

DB_Tss_Bry <- DB_Tss_FG %>% filter(Func_gr %in% c("Bryophytes")) %>% droplevels() unique(DB_Tss_Bry$Func_gr)

####Bryophytes (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tss_Bry <- ggplot(DB_Tss_Bry, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Bryophytes")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("antiquewhite4","lightseagreen","chartreuse3"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tss_Bry

Tss, Func_gr: Ericaceae

DB_Tss_Eri <- DB_Tss_FG %>% filter(Func_gr %in% c("Ericaceae")) %>% droplevels() unique(DB_Tss_Eri$Func_gr)

####Ericaceae (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tss_Eri <- ggplot(DB_Tss_Eri, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Ericaceae")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("antiquewhite4","lightseagreen","chartreuse3"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tss_Eri

Tss, Func_gr: Pteridophyta

DB_Tss_Pte <- DB_Tss_FG %>% filter(Func_gr %in% c("Pteridophyta")) %>% droplevels() unique(DB_Tss_Pte$Func_gr)

####Pteridophyta (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tss_Pte <- ggplot(DB_Tss_Pte, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Pteridophyta")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("antiquewhite4","lightseagreen","chartreuse3"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tss_Pte

Tss, Func_gr: Shrubs

DB_Tss_Shr <- DB_Tss_FG %>% filter(Func_gr %in% c("Shrubs")) %>% droplevels() unique(DB_Tss_Shr$Func_gr)

####Shrubs (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tss_Sch <- ggplot(DB_Tss_Shr, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Shrubs")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("antiquewhite4","lightseagreen","chartreuse3"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tss_Sch

Tss, Func_gr: Sphagnaceae

DB_Tss_Sph <- DB_Tss_FG %>% filter(Func_gr %in% c("Sphagnaceae")) %>% droplevels() unique(DB_Tss_Sph$Func_gr)

####Sphagnaceae (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tss_Sph <- ggplot(DB_Tss_Sph, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Sphagnaceae")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("antiquewhite4","lightseagreen","chartreuse3"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tss_Sph

Tss, Func_gr: Trees

DB_Tss_Tre <- DB_Tss_FG %>% filter(Func_gr %in% c("Trees")) %>% droplevels() unique(DB_Tss_Tre$Func_gr)

####Trees (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tss_Tre <- ggplot(DB_Tss_Tre, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Trees")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("antiquewhite4","lightseagreen","chartreuse3"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tss_Tre

Tss, Func_gr: Graminoids

DB_Tss_Gra <- DB_Tss_FG %>% filter(Func_gr %in% c("Graminoids")) %>% droplevels() unique(DB_Tss_Gra$Func_gr)

####Graminoids (ggplot with smoothed lines - Abundance per Func_gr over time) (PDF 4x10 landscape) Plot_Tss_Gra <- ggplot(DB_Tss_Gra, aes(x=Year, y=Coverage, group=Treatment, color=Treatment, pch=Treatment)) + facet_grid(~Canopy) + geom_point (size = 1.5, alpha = 0.5)+ geom_smooth(method = lm, formula = y ~ splines::bs(x, 3), se = FALSE, size=1.5)+ theme_bw()+ labs(x="Year", y = "Abundance of Graminoids")+ scale_shape_manual(values=c(17, 16, 17, 16, 17))+ scale_color_manual(values=c("antiquewhite4","lightseagreen","chartreuse3"))+ theme(axis.text=element_text(size=11, face="bold"), axis.title=element_text(size=14,face="bold"), plot.title = element_text(size=14,face="bold"), legend.text=element_text(size=14, face="bold"), legend.title = element_text(size=14, face="bold")) + theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank(), plot.title = element_text(hjust = 0.5)) Plot_Tss_Gra

-- END