What is the difference in percent coverage of intertidal seaweed Ulva lactuca between wave-exposed versus wave-sheltered shores?

Hypothesis:

The rocky intertidal zone experiences a vast number of abiotic and biotic stresses as it is a unique system that functions as the transition zone between terrestrial and marine environments. Wave action is one of the stresses this zone undergoes that is operative in determining intertidal species abundance and diversity. Ulva lactuca is a sessile organism that can act as an indicator for environmental stress. We predict that U. lactuca will be more abundant at sheltered sites that receive less wave action and therefore less stress.

Sampling scheme:

In order to avoid low and high intertidal bias between the wave-exposed and wave-sheltered shore and create a random stratified sample, 15 sample sites were randomly selected using a random number generator from the 30 photo quadrats captured at the low intertidal and 15 from the 30 photo quadrats captured at the high intertidal for both sites. By dividing the photos into smaller groups (low and high intertidal), we accounted for shared characteristics or attributes the different intertidal zones might have (1). By taking an even amount of random samples from both, we were able to get a subset that represents the larger population (1).

We chose 30 samples for both of our variables (wave-exposed versus wave-sheltered) to make our total sample size 60 samples. Having 30 samples per variable ensures that the sample we take will be a relatively accurate representation of the population. Having a sample size too small often does not yield valid results (2). Larger sample size also helps to account for variability and extremes among samples. The larger the sample size is, the smaller the margin of error (2).

Metadata:

This study was conducted at Scott’s Bay, West Bamfield (48.8340803, -125.1468488) on 23/06/2020. At each of our two sites (wave-exposed shore and wave-sheltered shore) two 30 m transects were set up parallel to the shoreline. At both sites, a transect was placed at one meter above chart datum and the other at two meters above chart datum. Quadrats were 0.5 m by 0.5 m, divided into 100 sections, and were arrayed along each of the four 30 m transects. Thirty photo quadrats were captured per transect at one-meter intervals. 15 sample sites were randomly selected using a random number generator from the 30 photo quadrats captured at the low intertidal and 15 from the 30 photo quadrats captured at the high intertidal for both sites. The photographs were later analyzed and the percent coverage of U. lactuca was calculated to the nearest 0.25% for each of the samples. Seaweed abundance was quantified with percent coverage because alternative measures cannot be reliably determined for clonal organisms such as seaweed.

Excel column headers:

Wave Exposure: If sample was taken from wave-exposed or wave-sheltered shoreline

Percent Cover (%): Percent coverage of Ulva lactuca per sample calculated to the nearest 0.25% from 30 samples taken from the wave-exposed shoreline (15 random samples from the low transect, 15 random samples from the high transect) and 30 samples taken from the wave-sheltered shoreline (15 random samples from the low transect, 15 random samples from the high transect)

knitr::opts_chunk$set(echo = TRUE) #activates the knitting function

#load packages for data analysis
library(ggplot2)
library(tidyr)
library(dplyr)
## 
## Attaching package: 'dplyr'
## The following objects are masked from 'package:stats':
## 
##     filter, lag
## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union
library(ggtext)

theme_set(theme_classic()) #setting theme to classic for the entire document
#Cleaning data, looking at data to see what needs to be done
seaweed = read.csv("Seaweed.csv")
head(seaweed) #looking at the column names
##   Wave.Exposure Percent.Cover....
## 1       Exposed              4.25
## 2       Exposed              2.75
## 3       Exposed             11.00
## 4       Exposed              1.00
## 5       Exposed              1.00
## 6       Exposed              1.25
str(seaweed) #looking at the structure of the data frame
## 'data.frame':    60 obs. of  2 variables:
##  $ Wave.Exposure    : Factor w/ 2 levels "Exposed","Sheltered": 1 1 1 1 1 1 1 1 1 1 ...
##  $ Percent.Cover....: num  4.25 2.75 11 1 1 1.25 2.25 1.75 4.75 2 ...
#Column names too complicated, changing column names

seaweed=rename(seaweed,exposure=Wave.Exposure,cover=Percent.Cover....)

#Checking to see if it worked
head(seaweed)
##   exposure cover
## 1  Exposed  4.25
## 2  Exposed  2.75
## 3  Exposed 11.00
## 4  Exposed  1.00
## 5  Exposed  1.00
## 6  Exposed  1.25
#It worked so now looking for outliers, weird or incorrect data

seaweed_boxplot = ggplot(data=seaweed, aes(x=exposure, y=cover))+
  geom_boxplot()+
  labs(x="Wave Exposure", y=expression('Percent Cover of *Ulva lactuca* (%)'))+ #labeling axes
  scale_y_continuous(breaks=seq(0,60,10))+ #changing y axis ticks
  theme(axis.title.y=ggtext::element_markdown())#italics for Ulva lactuca name

seaweed_boxplot

#Looks like what was expected, the outliers of exposed checked against the quadrat photos and are accurate.

#Now looking at a scatter plot.

seaweed_scatterplot = ggplot(data=seaweed, aes(x=exposure, y=cover))+
  geom_point()+
  labs(x= "Wave Exposure", y=expression('Percent Cover of *Ulva lactuca* (%)'))+ #labeling axes
  scale_y_continuous(breaks=seq(0,60,10))+ #changing y axis ticks
  theme(axis.title.y=ggtext::element_markdown())#italics for Ulva lactuca name

seaweed_scatterplot

#Everything looks good.
#Looking at the descriptive statistics of our data.

#Going to look at calculation of mean, standard deviation, and standard error for percent cover of both exposed and sheltered and put it into a data.frame

statistics=seaweed %>%
  group_by(exposure) %>%
  summarize(count=n(),mean=mean(cover,na.rm=TRUE),sd=sd(cover,na.rm=TRUE),se=sd(cover)/sqrt(count))
## `summarise()` ungrouping output (override with `.groups` argument)
#Looking to see if it was successful which it was
str(statistics)
## tibble [2 × 5] (S3: tbl_df/tbl/data.frame)
##  $ exposure: Factor w/ 2 levels "Exposed","Sheltered": 1 2
##  $ count   : int [1:2] 30 30
##  $ mean    : num [1:2] 3.88 15.53
##  $ sd      : num [1:2] 4.36 19.55
##  $ se      : num [1:2] 0.795 3.57
#Now that we know the raw numbers for mean and standard deviation and standard error, we are going to plot them by creating a bar graph with standard error bars

seaweed_bargraph = ggplot(statistics,aes(x=exposure,y=mean))+
  geom_bar(stat="identity",width=0.7, color="black",position=position_dodge())+
  geom_errorbar(aes(ymin=mean-se,ymax=mean),width=0.2)+ #adding error bars
  geom_errorbar(aes(ymin=mean,ymax=mean+se),width=0.2)+
  labs(x="Wave Exposure",y=expression('Mean Percent Cover of *Ulva lactuca* (%)'),caption="Figure 1. Comparison of the mean total *Ulva lactuca* coverage for a wave-exposed and a wave-sheltered site. The percent \n coverage at 30 randomly selected sites for both exposed shore and sheltered shore at Scott's Bay, West Bamfield were \n calculated using a divided quadrat and photo analysis. Data was analyzed to determine the difference in mean percent \n coverage amongst different wave exposures. Error bars represent standard error.")+
  theme(axis.title.x=element_text(size=10),axis.title.y=element_text(size=10))+#changing axes title font sizes
  theme(plot.caption = element_text(hjust = 0))+#moving caption to left
  theme(axis.title.y=ggtext::element_markdown())+ #italics for Ulva lactuca name in y-axis title
  theme(plot.caption=ggtext::element_textbox_simple()) #italics for Ulva lactuca name in caption
  

seaweed_bargraph

Results:

The data in this study consist of percent coverage for U. lactuca at a wave-exposed shore and a wave-sheltered shore. As seen in Figure. 1, the highest mean coverage for U. lactuca was at the sheltered site (15.53%±3.57). The mean coverage for the exposed shore was much lower than the sheltered at 3.88%±0.795. This is a fairly large difference. An explanation for the higher observed coverage found at the sheltered site could be due to the reduced wave-exposure when compared to the exposed site. As a sessile organism, seaweed species rely on their holdfast to root them in place (De Olivera et al., 2006). Higher wave exposure and turbulence may make growth conditions less favorable. This would account for the difference in percent coverage means observed.The sheltered shore also had a lot more variability in percent coverage of U. lactuca. This could be the result of the interaction of multiple variables not measured in this study. Potential reasons include other species outcompeting U. lactuca in certain areas or the recruitment patterns of U. lactuca. To determine the cause of variability further study should be conducted.

References:

  1. Hayes, A. (2020) Reading Into Stratified Random Sampling. Investopedia. URL https://www.investopedia.com/terms/stratified_random_sampling.asp [accessed 1 July 2020]

  2. Zamboni, J. (2020) The Advantages Of A Large Sample Size. Sciencing. URL https://sciencing.com/advantages-large-sample-size-7210190.html [accessed 1 July 2020]

  3. De Oliveira, E., J. Populus and B. Guillaumont. 2006. Predictive modelling of coastal habitats using remote sensing data and fuzzy logic: A case for seaweed in Brittany (France). EARSeL eProceedings 5: 208–223.