FoodwebMotifAnalysis/02_Giling_et_al_motif_analysis.R at master · dgiling/FoodwebMotifAnalysis · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171

#======================================
# 02 Analyse motif counts and z-scores
#    Giling et al.
#======================================

# This code analyses the effect of sown plant
# species richness on the motif results


# 1. Required libraries and definitions
#---------------------------------------

library(tidyverse)
library(lme4)
library(lmerTest)

add.alpha <- function(col, alpha=1)
{
  if(missing(col))
    stop("Please provide a vector of colours.")
  apply(sapply(col, col2rgb)/255, 2,
        function(x)
          rgb(x[1], x[2], x[3], alpha=alpha))
}

col.list <- c(rgb(0,158,115,max=255), rgb(0,114,178,max=255), rgb(230,159,0,max=255),  rgb(204,121,167,max=255), "grey50")


# 2. Import results and perform calculations
#---------------------------------------------

# Import mean results of full network (i.e. grounded and free-floating motifs)
# and consumer subweb (i.e. free-floating motifs only)
full <- read.csv("Giling_et_all_motif_data_full_network.csv") # full network
sub <- read.csv("Giling_et_all_motif_data_sub_network.csv") # consumer sub web

# Select desired scenario
full <- filter(full, alpha==10000 & beta==10000)
sub <- filter(sub, alpha==10000 & beta==10000)

# Calculations and transformations
full$other.mean = apply(full[,c("s3.mean", paste0("d",1:8,".mean"))],1,sum)
sub$other.mean = apply(sub[,c("s3.mean", paste0("d",1:8,".mean"))],1,sum)

full$log.s1 <- log10(full$s1.mean +1)
full$log.s2 <- log10(full$s2.mean +1)
full$log.s4 <- log10(full$s4.mean +1)
full$log.s5 <- log10(full$s5.mean +1)
full$log.other <- log10(full$other.mean +1)

sub$log.s1 <- log10(sub$s1.mean +1)
sub$log.s2 <- log10(sub$s2.mean +1)
sub$log.s4 <- log10(sub$s4.mean +1)
sub$log.s5 <- log10(sub$s5.mean +1)
sub$log.other <- log10(sub$other.mean +1)

full$block <- as.factor(substr(full$plot,1,2))
full$log2.psr <- log2(full$sowndiv)
sub$block <- as.factor(substr(sub$plot,1,2))
sub$log2.psr <- log2(sub$sowndiv)

# Create empty dataframe for predictions
divseq <- seq(min(full$log2.psr), max(full$log2.psr), length.out = 100)
newdata <- data.frame(log2.psr = divseq)


# 3. Analyse motif frequencies (Fig. 2)
#-----------------------------------------

# setup for plotting and modelling
results <- data.frame(name=NA, NumDF=NA, DenDF=NA, F=NA, beta.mean=NA, beta.sd=NA, beta.p=NA, int.mean=NA, int.sd=NA, int.p=NA)

# define function for plotting and analysis of full web and consumer sub web
plot.freq <- function(y, yaxt=TRUE, xaxt=TRUE, col=1, ylow=log10(50), yhigh=log10(10000), mod=TRUE) {

    full$y <- full[,y]
    sub$y <- sub[,y]

    plot(y ~ log2.psr, type='n', data=full, ylim=c(ylow,yhigh), xlim=(c(0,6)),
         xaxt='n', yaxt='n', las=2, ylab = NA, xlab = NA)
    if(xaxt==TRUE)axis(1, at=log2(c(1,2,4,8,16,60)), lab=c(1,2,4,8,16,60))
    if(xaxt==FALSE)axis(1, at=log2(c(1,2,4,8,16,60)), lab=NA)
    if(yaxt==TRUE) axis(2, at=log10(c(1,10,100,1000,10000)+1), lab=c(1,10,100,1000,10000), las=1)
    if(yaxt==FALSE) axis(2, at=log10(c(1,10,100,1000,10000)+1), lab=NA, las=1)

    # add points for full web
    points(y ~ log2.psr, data=full, pch=16, col=add.alpha(col.list[col],0.15))
    lmm <- NULL
    lmm <-lmer(y ~ log2.psr + (1|block/plot) + (1|time.period), data=full)
    #plot(lmm)
    lmma <- anova(lmm); lmmc <- summary(lmm)$coefficients
    results <- rbind(results, c(y, lmma$NumDF, lmma$DenDF, lmma$`F value`, lmmc[2,1], lmmc[2,2], lmmc[2,5], lmmc[1,1], lmmc[1,2], lmmc[1,5]))
    newdata$ypred <- predict(lmm, newdata, re.form=NA) # unconditional (level 0) - for re see https://www.rdocumentation.org/packages/lme4/versions/1.1-15/topics/predict.merMod
    if(mod==TRUE) points(newdata$ypred ~ newdata$log2.psr, typ='l', lwd=2, col=col.list[col])

    # add points for consumer sub web
    points(y ~ log2.psr, data=sub, pch=21, col=add.alpha(col.list[col],0.15))
    lmm <- NULL
    lmm <-lmer(y ~ log2.psr + (1|block/plot) + (1|time.period), data=sub)
    #plot(lmm)
    lmma <- anova(lmm); lmmc <- summary(lmm)$coefficients
    results <- rbind(results, c(y, lmma$NumDF, lmma$DenDF, lmma$`F value`, lmmc[2,1], lmmc[2,2], lmmc[2,5], lmmc[1,1], lmmc[1,2], lmmc[1,5]))
    newdata$ypred <- predict(lmm, newdata, re.form=NA) # unconditional (level 0) - for re see https://www.rdocumentation.org/packages/lme4/versions/1.1-15/topics/predict.merMod
    if(mod==TRUE) points(newdata$ypred ~ newdata$log2.psr, typ='l', lwd=2, col=col.list[col], lty=2)

    return(results)
}

# run analysis and print figure
par(mfrow=c(3,2), oma=c(3,3,1,0), mar=c(1,1,2,1))
results <- plot.freq(y="log.s1", TRUE, FALSE, 1)
results <- plot.freq("log.s2", FALSE, FALSE, 2)
results <- plot.freq("log.s4", TRUE, FALSE, 3)
results <- plot.freq("log.s5", FALSE, TRUE, 4)
results <- plot.freq("log.other", TRUE, TRUE, 5, ylow=0,yhigh=log10(500), mod=FALSE)


# 4. Analyse normalisedz-scores (Fig. 3)
#-----------------------------------------

# function for plotting and analysis
plot.zscores <- function(y, yaxt=TRUE, xaxt=TRUE, col=1) {

  full$y <- full[,y]
  sub$y <- sub[,y]

  plot(y ~ log2.psr, type='n', data=full, ylim=c(-0.8,0.8), xlim=(c(0,6)),
       xaxt='n', yaxt='n', las=2, ylab = NA, xlab = NA)
  abline(h=0, lty=2)
  if(xaxt==TRUE)axis(1, at=log2(c(1,2,4,8,16,60)), lab=c(1,2,4,8,16,60), cex.axis=0.8)
  if(xaxt==FALSE)axis(1, at=log2(c(1,2,4,8,16,60)), lab=NA)
  if(yaxt==TRUE) axis(2, at=seq(-0.8,0.8,0.4), las=1, cex.axis=0.8)
  if(yaxt==FALSE) axis(2, at=seq(-0.8,0.8,0.4), lab=NA, las=1)

  # add points for full web
  points(y ~ log2.psr, data=full, pch=16, col=add.alpha(col.list[col],0.15), cex=0.9)
  lmm <- NULL
  lmm <-lmer(y ~ log2.psr + (1|block/plot) + (1|time.period), data=full)
  #plot(lmm)
  lmma <- anova(lmm); lmmc <- summary(lmm)$coefficients
  results <- rbind(results, c(y, lmma$NumDF, lmma$DenDF, lmma$`F value`, lmmc[2,1], lmmc[2,2], lmmc[2,5], lmmc[1,1], lmmc[1,2], lmmc[1,5]))
  newdata$ypred <- predict(lmm, newdata, re.form=NA) # unconditional (level 0) - for re see https://www.rdocumentation.org/packages/lme4/versions/1.1-15/topics/predict.merMod
  points(newdata$ypred ~ newdata$log2.psr, typ='l', lwd=2, col=col.list[col])

  # add points for consumer sub web
  points(y ~ log2.psr, data=sub, pch=21, col=add.alpha(col.list[col],0.15), cex=0.9)
  lmm <- NULL
  lmm <-lmer(y ~ log2.psr + (1|block/plot) + (1|time.period), data=sub)
  #plot(lmm)
  lmma <- anova(lmm); lmmc <- summary(lmm)$coefficients
  results <- rbind(results, c(y, lmma$NumDF, lmma$DenDF, lmma$`F value`, lmmc[2,1], lmmc[2,2], lmmc[2,5], lmmc[1,1], lmmc[1,2], lmmc[1,5]))
  newdata$ypred <- predict(lmm, newdata, re.form=NA) # unconditional (level 0) - for re see https://www.rdocumentation.org/packages/lme4/versions/1.1-15/topics/predict.merMod
  points(newdata$ypred ~ newdata$log2.psr, typ='l', lwd=2, col=col.list[col], lty=2)

  return(results)
}

# run analysis and print figure
par(mfrow=c(2,2), oma=c(2,2,1,0), mar=c(1,1,2,1))
results <- plot.zscores("s1.z.mean", TRUE, FALSE, 1)
results <- plot.zscores("s2.z.mean", FALSE, FALSE, 2)
results <- plot.zscores("s4.z.mean", TRUE, TRUE, 3)
results <- plot.zscores("s5.z.mean", FALSE, TRUE, 4)