R script to analyze 3D habitat complexity and coral point-based annotations
includes data wrangling, visualizations, and statistical tests
# load packages
library(readr)
library(tidyverse)
library(ggplot2)
library(vegan)
# load data
KAHO_3D <- read_csv("Raw_Data/KAHO_3Dmetrics_df.csv") # 3D data (obtained from KAHO_Habitat.R)
KAHO_Coral <- read_csv("Raw_Data/CoralNetAnnotations.csv") # Coral data (obtained from CoralNet)Data wrangling – Calculate % Coral Cover from raw CoralNet data
KAHO_Cover <- KAHO_Coral %>% # from our raw CoralNet data
# create "Site" column where values are from "Name" characters before the underscore
mutate(Site = str_extract(Name, "^[^_]+")) %>%
# group data based on the Site and Labels
group_by(Site, Label) %>%
# for each site, sum up all values for each label
count() %>%
# widens data frame (increasing n of columns & decreasing n of rows)
# column names will come from "Label" and values from "n"
# if count is NA, fill with 0
pivot_wider(names_from = Label, values_from = n, values_fill = 0) %>%
# remove non-coral labels (for this analyses only focusing on coral cover)
select(-c("*TURFH","*TURFR", "Unk","Sand", "CCA", "TAPE", "*RDMA")) %>%
group_by(Site) %>%
# for each site, perform the following calculations on a row-based manner
rowwise() %>%
# for each numeric column (coral species), calculate % cover
mutate(across(where(is.numeric), ~ (.x / 1000) * 100)) %>%
# calculate overall % coral cover by summing up % cover of each species
mutate(coral_cover = sum(c_across(where(is.numeric)))) %>%
ungroup()Data wrangling – 3D
# keep only 1cm resolution
KAHO_3D <- KAHO_3D[KAHO_3D$fac == 1,]
# clean up the site names by keeping only characters before the 1st underscore
KAHO_3D <- KAHO_3D %>% mutate(site = sub("_.*", "", site))Visualize complexity metrics by site
ggplot(KAHO_3D, aes(x = reorder(site, -mean_vrm), y=mean_vrm)) +
geom_bar(aes(fill=site), stat = "identity", position = "dodge") +
theme_classic()
ggplot(KAHO_3D, aes(x = reorder(site, -mean_slope), y=mean_slope)) +
geom_bar(aes(fill=site), stat = "identity", position = "dodge") +
theme_classic()
ggplot(KAHO_3D, aes(x = reorder(site, -mean_profile_curvature), y=mean_profile_curvature)) +
geom_bar(aes(fill=site), stat = "identity", position = "dodge") +
theme_classic()
Visualize complexity metrics distributions
# Convert data to long format with variables of interest
KAHO_3D_long <- pivot_longer(KAHO_3D, cols = c(mean_slope, mean_vrm,
mean_profile_curvature, mean_plan_curvature, surf_complex), names_to = "variable")
# Create a facetted density plot
ggplot(KAHO_3D_long, aes(x = value, fill = variable)) +
geom_density(alpha = 0.5) + facet_wrap(~ variable, scales = "free") +
theme_minimal() + theme(legend.position = "none")
Visualize % coral cover by site
ggplot(KAHO_Cover, aes(x = reorder(Site, -coral_cover), y=coral_cover)) +
geom_bar(aes(fill=Site), stat = "identity", position = "dodge") +
theme_classic()
ggplot(KAHO_Cover, aes(x = reorder(Site, -PLOB), y=PLOB)) +
geom_bar(aes(fill=Site), stat = "identity", position = "dodge") +
theme_classic()
ggplot(KAHO_Cover, aes(x = reorder(Site, -PCOM), y=PCOM)) +
geom_bar(aes(fill=Site), stat = "identity", position = "dodge") +
theme_classic()
Visualize coral cover distributions
# Convert data to long format
KAHO_Cover_long <- pivot_longer(KAHO_Cover, cols = -Site,
names_to = "variable", values_to = "value")
# Create a facetted density plot
ggplot(KAHO_Cover_long, aes(x = value, fill = variable)) +
geom_density(alpha = 0.5) + facet_wrap(~ variable, scales = "free") +
theme_minimal() + theme(legend.position = "none")
Visualize associations between complexity & coral
## combine 3D & Coral data into one dataframe ##
KAHO <- KAHO_Cover %>% full_join(KAHO_3D, by = c("Site" = "site"))
KAHO %>% ggplot(aes(x = coral_cover, y = mean_vrm)) +
geom_point(size = 2) +
geom_smooth(method = "lm", se = FALSE) + theme_classic() ## `geom_smooth()` using formula = 'y ~ x'
KAHO %>% ggplot(aes(x = coral_cover, y = mean_profile_curvature)) +
geom_point(size = 2) +
geom_smooth(method = "lm", se = FALSE) + theme_classic() ## `geom_smooth()` using formula = 'y ~ x'
Statistical analysis = simple linear regression - complexity ~ coral
summary(lm(mean_vrm ~ coral_cover, data = KAHO))##
## Call:
## lm(formula = mean_vrm ~ coral_cover, data = KAHO)
##
## Residuals:
## Min 1Q Median 3Q Max
## -0.018742 -0.012320 -0.008822 0.018622 0.023308
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.0585855 0.0125168 4.681 0.00339 **
## coral_cover 0.0014526 0.0004232 3.433 0.01392 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.01866 on 6 degrees of freedom
## Multiple R-squared: 0.6626, Adjusted R-squared: 0.6064
## F-statistic: 11.78 on 1 and 6 DF, p-value: 0.01392summary(lm(mean_profile_curvature ~ coral_cover, data = KAHO))##
## Call:
## lm(formula = mean_profile_curvature ~ coral_cover, data = KAHO)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.6688 -1.8224 -0.6926 1.2759 5.2245
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -2.71177 1.90685 -1.422 0.2048
## coral_cover 0.14145 0.06446 2.194 0.0707 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.843 on 6 degrees of freedom
## Multiple R-squared: 0.4452, Adjusted R-squared: 0.3527
## F-statistic: 4.815 on 1 and 6 DF, p-value: 0.07065Multiple Linear Regressions
vrm <- lm(mean_vrm ~ PLOB + PCOM + PMEAN, data = KAHO)
sjPlot::plot_model(vrm, p.shape = TRUE, show.legend = TRUE, show.p = TRUE)
curvature <- lm(mean_profile_curvature ~ PLOB + PCOM + PMEAN, data = KAHO)
sjPlot::plot_model(curvature, p.shape = TRUE, show.legend = TRUE, show.p = TRUE)
for questions, sofiaf@hawaii.edu and johnhr@hawaii.edu