Preparations

The workshop is part of a three day workshop in R including sessions on data handling and basic and advanced statistical modelling and assumes knowledge of these sections.

If you work through this on your own, you may want to start with this intro on the basics in R, and this [workshop on data handling(http://fdschneider.de/r_datahandling/).

The examples and code snippets require that you install the following packages:

install.packages("ggplot2", "readxl", "data.table", "RColorBrewer", "colorspace", "Shiny")

For the following examples, I will use the datasets from wednesday: LUI_factors, voeller and flowerdata.

voeller <- read.csv("data/voeller.csv", encoding = "UTF-8")
flowerdata <- read.csv("data/flowerdata.csv", encoding = "UTF-8")
LUI_factors <- read.csv("data/LUI_factors.csv", encoding = "UTF-8")

We will refer to this dataset in the exercises:

Data from: Resource-mediated indirect effects of grassland management on arthropod diversity (2014) Simons NK, Gossner MM, Lewinsohn TM, Boch S, Lange M, Müller J, Pašalić E, Socher SA, Türke M, Fischer M, Weisser WW; DOI: https://doi.org/10.5061/dryad.f3b77

library(readxl)
arthropods08 <- read_xlsx("data/Plant_herb_carn/01_Arthropods_2008.xlsx")
arthropods09 <- read_xlsx("data/Plant_herb_carn/01_Arthropods_2009.xlsx")
plantcover08 <- read_xlsx("data/Plant_herb_carn/02_Plant_cover_2008.xlsx")
plantcover09 <- read_xlsx("data/Plant_herb_carn/02_Plant_cover_2009.xlsx")

landuse <- read_xlsx("data/Plant_herb_carn/03_Landuse_2007to2009.xlsx")
semdata08 <- read_xlsx("data/Plant_herb_carn/04_SEMData_2008.xlsx")
semdata09 <- read_xlsx("data/Plant_herb_carn/04_SEMData_2009.xlsx")

library(data.table)
dd_full <- rbindlist(list(semdata08, semdata09), fill = TRUE)

dd_full <- transform(dd_full,
                  PlotID = as.factor(PlotID),
                  Region = as.factor(Region),
                  Lifestock = as.factor(Lifestock)
                  )

Online Reference

The presentation slides can be found here!

R graphics by Florian Schneider is licensed under a Creative Commons Attribution 4.0 International License.

Properties of scientific visualisations

In the first place, we show data. Visuals are adding high-resolution and detail, were text can only transport summary and statistics.

Advantages of graphics:

• high data density

Example: This two-dimensional mapping of multi-dimensional trait-space illustrates the global spectrum of plant form and function (Diaz et al. 2016). The lower panels illustrate subsets of the same data (b: woody plants, c: plant classes) .

• put data in context (e.g. spatial maps!)

Example: besides the spatial context, maps contain information about area, shape, and neighborhood of a data point. It is easy to read where values are high or low because of spatial context and where and at which level to look for explanatory variables.

(Source: E.R. Tufte “The Visual Display of Quantitative Information”)

• increasing objectivity

Example: Anscombes famous quartet of x-y datasets. Each dataset has exactly the same stats regarding means, coefficient of correlation, model fit etc., but visually, they tell a totally different story.

Graphical integrity

Good data analyses are one thing. But the complex results in science can not be transported by statistics alone. In research papers, figures should do a good part of the job. But very often, they do a bad job.

As a fundamental critique to the practice of graphical communication Edward Tufte published this series of books, where he elaborated on why graphics fail at being informative and – more importantly – fail at being truthful about the underlying pattern in the data.

In this workshop, in addition to the technical aspects, I want to discuss three big challenges of visual communication in Data Science:

1. Reduce noise
2. Be truthful
3. Build a narrative

Note that we are not talking about plots for reviewing your data (like Q-plot or analysis of residuals)! those have been covered in the statistics lecture by Caterina.

A word on packages

Many available R packages facilitate plotting into publishable output. Particularly, ggplot2 is a great tool to design graphics since it is build around the logic of plot design, separating content (i.e. the data) from aesthetics (i.e. the design and choice of elements).

However here is a word of caution: Just like other software, like Excel or SASS, plotting packages come with the risk of ridding oneself of the responsability of thinking about plot elements. Indeed, one intention of those software packages is creating an own style that readers would recognise and make them use their software.

I am using base R for my examples because I believe, first comes understanding at the low level of plotting grammar, before we can start using powerful tools.

Learn design principles! Not usage of tools.

After learning about the basics, we’ll talk about ggplot2 as well.

high-level commands

Examples for high level plotting functions are:

plot(Plant_SpeciesRichness ~ Herbivore_SpeciesRichness, data = dd_full)

plot(co2)

plot(x = 1:12, y = rnorm(12,5,1.2), type = "l", ylim = c(0,10) )

x <- z <- seq(0,1,length = 50)

f <- function(x) 1.2 + 0.4*x - 2.1*x^2 + 1.1*x^3
plot(f, ylim = c(0,2), xlim = c(0,2))

plot(Plant_biomass ~ Lifestock, data = dd_full)

barplot( tapply(dd_full$Plant_biomass, dd_full$Lifestock, mean), 
        xlab = "Lifestock", ylab = "Plant biomass")

hist(voeller$plant_height, breaks = 24)

x <- y <- seq(0,1,length = 25)
dd <- expand.grid(data.frame(x,y))
dd$z <- with(dd, 0.1 + 0.4*x - 1.5*x^2 + 1.15*x^3 + 0.2*y - 0.4*y^2 + 0.3*y^3 + 0.2*x*y - 0.23 *x^2*y^2 + 0.33 *x^3*y^3)
persp(x,y,z = matrix(dd$z, nrow = 25, byrow = FALSE), ticktype = "detailed", zlim = c(0,0.6), shade = 0.4) 

low-level commands

In contrast, low level plotting commands are adding something to an existing plot

• data (points() or symbols())

plot(inflorescences_qcm ~ LUI, data = flowerdata, log = "xy")

## Warning in xy.coords(x, y, xlabel, ylabel, log): 142 y values <= 0 omitted
## from logarithmic plot

points(inflorescences_qcm ~ LUI, data = subset(flowerdata, plant_genus == "Stellaria"),  
      pch = 20, col = "red", cex = 2
      )

plot(n_inflorescences ~ plant_height, data = voeller)
model <- lm(n_inflorescences ~ plant_height, data = voeller )
abline(model) # adds a model line

lines(c(0,100), c(800,800), lwd  = 2)

plot(n_inflorescences ~ plant_height, data = voeller, xlim = c(0,100))
axis(4)

plot(n_inflorescences ~ plant_height, data = voeller, xlim = c(0,100),
     pch = 20, col =c("black", "red", "blue")[as.integer(voeller$eventID)] )
legend(3.5,800, c("HAI", "SCH", "ALB"), pch = c(20), 
       col = c("black", "red", "blue"))

plot(n_inflorescences ~ plant_height, data = voeller, xlim = c(0,100),
     pch = 20, col =c("black", "red", "blue")[as.integer(voeller$eventID)] )
mtext("a)", adj = 0, line = 0.3)

plot(n_inflorescences ~ plant_height, data = voeller, xlim = c(0,100),
     pch = 20, col =c("black", "red", "blue")[as.integer(voeller$eventID)] )
arrows(105,0, 105, 600, xpd = TRUE)

graphical parameters

The graphical parameters are ususally defined before a high-level plot is called using the function par(). They set the scene for the plot(s) that follow and are valid until the graphics device is terminated using dev.off().

Important graphical parameters are:

• margins (inner mar and outer oma, specified in bottom, left, top, right order)
• font (family,font or font.lab specify font, e.g. for axis labels)
• size of elements (cex, cex.lab etc.)
• plot over plot (if new = TRUE the existing frame is not cleaned)

Note that many parameters specific to the plot’s high- and low-level elements can be specfied here generally, to apply to all subsequent plots, e.g. axis type or point character. Calling ?par gives a full list.

The most common settings here are the outer and inner margins.

For reducing space around single plot panels, setting mar with a vector in order bottom, left, top, right will do.

par(mar = c(4,4,4,6))

plot(n_inflorescences ~ plant_height, data = voeller, xlim = c(0,100),
     pch = 20, main = "Inflorescences" )

Detail too high?

It is rarely the case that a graph shows too many data. If a graph appears cluttered, this is mostly because it contains to many elements of data or uses a noisy design. The message delivered from a blurry, dense cloud of data points is also that a large amount of data was available to draw this conclussion. Another message might be the huge complexity of the analysed system (e.g. networks). Thus, reducing data density is not a good choice.

A bad Example:

With the choice of color and plot style, resulting in heavy overplotting it is difficult to extract any information about the size and price of diamonds, depending on their quality.

A good Example: combine summary and raw-data when showing richess in data (panel E) or illustrate network complexity by density of lines.

(Source: Brose et al. 2016 Biological Reviews)

Chartjunk

Reduce elements of graphs that do not contribute information.

In this graph, the entire legend and filling is redundant with the x-axis labels. The number of observations could be placed on top of the bars.

(Source: E.R. Tufte “The Visual Display of Quantitative Information”)

Improved version:

Edward Tufte suggests that we should maximize the ‘data-ink ratio’:

\[ \textrm{data-ink ratio} = \frac{\textrm{ink that transports information}}{\textrm{total ink of the graph}} \]

R applications to reduce noise

Reduction of noise and chartjunk is key for producing statistical graphics for publication. The following elements are to consider:

plot(Herbivore_SpeciesRichness ~ Predator_SpeciesRichness, data = dd_full, 
     pch = c(18,20,17)[Region],  
     col = c( "black", "red","blue")[Region]
     )

par(las = 1, bty = "l")
plot(reproductive_allocation_ratio ~ plant_height, data = voeller, 
     pch = 19, col = rgb(0,0,0, alpha = 0.1)
     )

ggplot(voeller, aes(x = plant_height, y =  reproductive_allocation_ratio)) +
  stat_bin2d(bins = 50) + 
  scale_fill_gradient(low = "white", high = "black", limits = c(0,25)) + 
  geom_density_2d(colour = "black", n = 200) + 
  xlab("") +
  theme_classic()

## Warning: Removed 622 rows containing non-finite values (stat_bin2d).

## Warning: Removed 622 rows containing non-finite values (stat_density2d).

use of color

Basically, every plot element can be colored. For scientific graphs, color should only be used when it encodes meaningful information! A rule of thum would be: don’t use color, if the difference can be made by shape or by clear labelling.

R’s default color palette is somewhat cruel.

In principle, all plot commands have an argument for color for points, lines or areas. You can specify a color palette by providing a vector with the color names.

As shown above, we apply the squared brackets for assigning each factor level in column eventID an own color value.

For continuous variables this requires binning of the vector into discrete variables, before assigning colors. However, adding a legend for this is not trivial.

plot(Herbivore_SpeciesRichness ~ Predator_SpeciesRichness, data = dd_full, 
     pch = 19,  
     col = gray.colors(25)[cut(Mowing, 25)]
     )

The simplest form of color difference is black and white, or halftone (i.e. grayscale), which is always safe for print and on screen.

There are four basic ways of defining colors.

1. chose from the 657 named colors in R (see this pdf for a previev)

2. use a hexadecimal code for a fine grained 16 million colors: The first two digits give the intensity of red, the second pair gives green, the third pair gives blue; “#000000” is black, “#FFFFFF” is white.

3. use a color mixing function to produce color from a color space: rgb() uses the red-green-blue colorspace, hsv() uses the hue-saturation-value colorspace, hcl() uses hue-chroma-luminance.

4. chose from R’s alternative color palettes: rainbow(), heat.colors(), terrain.colors(), topo.colors()

However the first three options arguably leave you with an overwhelming choice. The latter is designed to produce contingent color tables, rather than divergent palettes that are strong in contrast.

You can create your own color palettes using the function colorRampPalette()

pal <- colorRampPalette(c("springgreen", "darkgoldenrod1", "lightsalmon"))
pal(8)

## [1] "#00FF7F" "#48EB5F" "#91D73F" "#DAC31F" "#FFB51E" "#FFAE3C" "#FFA75B"
## [8] "#FFA07A"

It is not recommended to pick arbitrary or subjectively aesthetic colors for a scientific plot. Rather you should rely on professionally designed color palettes for this purpose.

One of the best collection of colors is provided in the RColorBrewer package.

library(RColorBrewer)
plot_palette(brewer.pal(8, "Dark2"))

You should constrain your use to the many colorblind friendly options offered by this package:

display.brewer.all(colorblindFriendly =  TRUE)

There is also the very convenient package ‘colorspace’ which lets you chose your color palette in a graphical interface:

library(colorspace)
pal <- choose_palette()

will return you a function that can be called to produce n steps of the choosen color palette:

The palettes produced by this tool maximize contrast even for colorblind viewers.

par(las = 1, bty = "l")
plot(n_inflorescences ~ plant_height, data = voeller,
     pch = 19, xlim = c(0,100),
     col = c( "#E88797", "#BA92E6", "#00B8C6")[eventID]
     )

Finally, ggplot2 specifies color within the aes() argument of the ggplot() call, i.e. a third level of information mapped to color. It comes with tools for color ramps defined as scale functions.

ggplot(voeller, aes(x = plant_height, y = n_inflorescences, colour = eventID)) + 
  geom_point() + 
  scale_color_brewer(palette = "Dark2") +
  xlim(0,100) + theme_classic()

## Warning: Removed 3461 rows containing missing values (geom_point).

• scale_color_brewer() applies the palettes of the RColorBrewer package
• scale_color_manual() allows you to specify own colors
• scale_color_hue() alters the shade/hue of your selected color palette

For continuous variables, we use scale_color_gradient():

ggplot(voeller, aes(x = plant_height, y = n_inflorescences, colour = latitude)) + 
  geom_point() + 
  scale_colour_gradient(low = "salmon", high = "red3") +
  xlim(0,100) + theme_classic()

## Warning: Removed 3461 rows containing missing values (geom_point).

Note that ggplot gracefully produces an exact legend!

# offset grid on grey (ggplot base style) 
ggplot(d1, aes(x = x,y = y) ) + geom_point()

boxplot(Plant_SpeciesRichness ~ Region, data = dd_full)

boxplot(Plant_SpeciesRichness ~ Region, data = dd_full,
        ylim = c(0,75), las = 1,
        pars = list(boxwex = 0.3, staplewex = 0, 
                    whiskwex = 0, whisklty = 1, outpch = 20, outcex = 0.8)
        )

ggplot(dd_full, aes(x = Region, y = Plant_SpeciesRichness)) + 
  geom_boxplot() + ylim(0,75)

R recipes

In the following sections, we will discuss properties in R plotting that are relevant for matching your graphical effects with the effect size in the data.

model <- lm(Plant_SpeciesRichness ~ Herbivore_SpeciesRichness, data = dd_full)
S_new <- seq(0, 60, length = 51)  # generates new x-values

plot(Plant_SpeciesRichness ~ Herbivore_SpeciesRichness, data = dd_full, pch = 20)
abline(model)

model <- lm(Plant_SpeciesRichness ~ Herbivore_SpeciesRichness, data = dd_full)
S_new <- seq(0, 60, length = 51)  # generates new x-values

plant_pred <- predict(model,     # simulate y-values
                    newdata = list(
                      Herbivore_SpeciesRichness = S_new
                      ), interval =  "confidence"
                    )

plot(Plant_SpeciesRichness ~ Herbivore_SpeciesRichness, data = dd_full, pch = 20)
lines(S_new, plant_pred[,"fit"])
lines(S_new, plant_pred[,"lwr"], lty = 3)
lines(S_new, plant_pred[,"upr"], lty = 3)

model <- glm(total_biomass ~ plant_height, data = voeller, family = quasipoisson)

summary(model)

## 
## Call:
## glm(formula = total_biomass ~ plant_height, family = quasipoisson, 
##     data = voeller)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -7.2572  -2.5005  -0.8708   1.3318  11.0920  
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  -0.476239   0.303382   -1.57    0.117    
## plant_height  0.024266   0.001957   12.40   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for quasipoisson family taken to be 10.75206)
## 
##     Null deviance: 6222.0  on 439  degrees of freedom
## Residual deviance: 4197.5  on 438  degrees of freedom
##   (3306 observations deleted due to missingness)
## AIC: NA
## 
## Number of Fisher Scoring iterations: 5

x_new <- seq(0, 200, length = 51)  # generates new x-values

plant_pred <- predict(model,  newdata = list(plant_height = x_new), type="response", se.fit = TRUE)

plot( total_biomass ~ plant_height, data = voeller, pch = 20)
lines(x_new, plant_pred$fit)
lines(x_new, plant_pred$fit + 1.96*plant_pred$se.fit, lty = 3)
lines(x_new, plant_pred$fit - 1.96*plant_pred$se.fit, lty = 3)

plot( total_biomass ~ plant_height, data = voeller, pch = 20, log = "y")
lines(x_new, plant_pred$fit)
lines(x_new, plant_pred$fit + 1.96*plant_pred$se.fit, lty = 3)
lines(x_new, plant_pred$fit - 1.96*plant_pred$se.fit, lty = 3)

library(ggplot2)

ggplot(dd_full, aes(x = Herbivore_SpeciesRichness, y = Plant_SpeciesRichness)) +
  geom_point() + 
  geom_smooth(method='lm',formula=y~x) +
  theme_classic()

add error bars

The base barplot does not have a native option to add error bars. The arrows() function can be used for that:

show correlation

Co-variation of multivariate data can be shown in multi-panel plots. More useful for data exploration.

Stacks

Positive or negative correlation between response values along the same predictor value can be shown in stacks:

(Source: Schneider et al. 2012 J.Anim.Ecol.)

(Source: Jochum et al. 2012 Phil. Trans. R.Soc. B)

Multiples

Ideal for showing correlative effects in parameter space are multiples. Once the x-y-coordinate system is clearly laid out, the other data can be shown in smaller layout or without the noise of labels and legends.

(Source: E.R. Tufte “The Visual Display of Quantitative Information”)

The transfer achieved when reading multiples is extraordinary, as the example of sea ice extend in the arctic shows. It shows seasonal variation along y and yearly decline on x.

(Source: NASA)

This combination of Micro pattern and Macro pattern allows for the reader to zoom in and out and acheive a contextual view of things, a understanding of events across time.

Arrangement

Other arrangements are possible to illustrate relation between different plots. E.g. a central schematic figure which is explained by statistical plots.

(Source: Schneider et al. 2012 Ecol Lett)

par(mfrow = c(2,3))
layout.show(n = 6)

par(mfrow = c(3,4), las = 1)
for(i in levels(flowerdata$plant_genus)[c(2,4,7,10,11,12,14,17:20,22)]) {
    plot(inflorescences_qcm ~ LUI, 
         data = subset(flowerdata, plant_genus == i), 
         pch = 19,
         xlim = c(0.001,1), ylim = c(1,10000),
         main = i,
         log = "xy"
         ) 
} 

ggplot(subset(flowerdata, as.integer(plant_genus) %in% c(2,4,7,10,11,12,14,17:20,22)), aes(y = inflorescences_qcm, x =  LUI)) + 
  geom_point() +
  scale_x_log10() + scale_y_log10() +
  facet_wrap( ~ plant_genus, nrow = 3) 

(m <- matrix(c(1,1,1,2,3,6,4,5,6), byrow = TRUE, ncol = 3))

##      [,1] [,2] [,3]
## [1,]    1    1    1
## [2,]    2    3    6
## [3,]    4    5    6

layout(m)
layout.show(n = 6)

layout(m , height = c(1,2,2), width = c(2,2,1))
layout.show(n = 6)

m <- matrix(rep(0, 36), byrow = TRUE, ncol = 6)
m[1:5,1:5] <- 1
m[6,1:2] <- 2
m[6,3:4] <- 3
m[6,5:6] <- 4
m[1:2,4:5] <- 5
m[1:3,6] <- 6
m

##      [,1] [,2] [,3] [,4] [,5] [,6]
## [1,]    1    1    1    5    5    6
## [2,]    1    1    1    5    5    6
## [3,]    1    1    1    1    1    6
## [4,]    1    1    1    1    1    0
## [5,]    1    1    1    1    1    0
## [6,]    2    2    3    3    4    4

layout(m , height = c(1,1,1,1,1,3), width = c(1,2,1,2,1,2))
layout.show(n = 6)

par(mar = c(4,4,1,3))
plot(Plant_SpeciesRichness ~ Herbivore_SpeciesRichness, data = dd_full, col = pal(3)[Region], pch = 20, cex = 2)
mtext("a)", adj = 0, line = -1.5)
par(mar = c(5,4,1,1))
boxplot(Plant_SpeciesRichness ~ I(Grazing > 0), data = dd_full)
mtext("b)", adj = 0, line = -1.5)
boxplot(Plant_SpeciesRichness ~ I(Fertilization > 30), data = dd_full)
mtext("c)", adj = 0, line = -1.5)
boxplot(Plant_SpeciesRichness ~ I(Mowing > 0), data = dd_full)
mtext("d)", adj = 0, line = -1.5)

par(mar = c(4,4,2,4))
hist(dd_full$Plant_SpeciesRichness, main = NA, col = "black", breaks = 24, xlab = NA)
par(mar = c(1,1,1,1))
contour(unique(dd$x),unique(dd$y), matrix(dd$z, nrow = 25, byrow = FALSE))
mtext("e)", adj = 0, line = -1.5)

devices

By default, R plots figures into the screen graphics device (windows in Windows, X11 in Ubuntu and Mac), which pops up in a new window. Via buttons, this picture can be exported to files. Alternatively, you can print plots directly into a file, or a virtual ‘device’. When producing figures for production, i.e. files for submission or to embed in presentations, it is important to chose the right file format.

The journal requirements often ask for these things

• figure size
• serif or sans font
• raster or vector format

there are different graphics devices and they can be sorted into raster formats or vector formats:

Raster formats:

• bmp() for bitmap files (no compression)
• png() for gif and png files (selective color)
• jpeg() for jpg (pixel compression)
• tiff() for tiff files (best quality/compression trade-off)

Vector formats:

• pdf() for pdf plotting
  
• postscript() for .eps files

Mostly, vector format is what journals want, because its scalable without loss and the file size is small, unless your data points are several thousands. Use raster formats for plots with many (>1000) datapoints or for presentations on screen, because they are more reliable. Use png() for black-and-white plots in high-resolution!

the raster output

Raster plots scale with the resolution that you specify. You can either give a precise pixel dimensions by providing width and height, or define a target print dimension in centimeters, by setting units = "cm" and define width, height as well as resolutinon res. A minimal resolution for printing is 300 dots per inch. For screen output, e.g. on websites, 72 dpi is sufficient.

jpeg("filename.jpg", 
    width = 460, height = 720, #in pixels
    antialias = "none" # no smoothing of pixelation 
    family = "Times" # for journals that want serif
    )

plot(...)

dev.off()

the pdf output

pdf is the standard format for paper submission because it is lossless. The file will grow very big if your plot contains many data points. You can chose whether you print your plot on a regular A4 paper format, or cropped to the given plot dimensions. The useDingbats argument is important for making your plot independent of Windows-specific fonts.

pdf("filename.pdf", 
    width = 7, height = 6,  #in inches: 1in = 2.54cm
    paper = "special",  # or "a4"
    useDingbats = FALSE, # avoids weird symbols
    family = "Times" # for journals that want serif
    )

plot(...)

dev.off()

Note that multiple high-level plot calls in a pdf device create multi-page pdfs!

ggplot saving

Saving ggplot plots is slightly different. A simple way is to wrap your plot in the print() function for writing it to a pdf or jpeg device.

pdf("filename.pdf", 
    width = 7, height = 6,  #in inches: 1in = 2.54cm
    paper = "special",  # or "a4"
    useDingbats = FALSE, # avoids weird symbols
    family = "Times" # for journals that want serif
    )

print(ggplot(...) + geom_point() + theme_classic())

dev.off()

The other option is the function ggsave() which is called just following a plot call.

ggplot(...) + geom_point() + theme_classic()

ggsave("filename.pdf", width = 20, height = 20, units = "cm")

shiny apps (web- output)

RStudio offers a feature for interactive plotting which is a great way to explore complex datasets or to visualize model behaviour depending on parameters.

#
# This is a Shiny web application. You can run the application by clicking
# the 'Run App' button above.
#
# Find out more about building applications with Shiny here:
#
#    http://shiny.rstudio.com/
#

library(shiny)

# Define UI for application that draws a histogram
ui <- fluidPage(
   
   # Application title
   titlePanel("Biodiversity Exploratories : Land Use Intensity on Plots"),
   
   # Sidebar with a slider input for number of bins 
   sidebarLayout(
      sidebarPanel(
        selectInput(inputId = "exploratory",
                    label = "Exploratory region:",
                    choices = c("ALB", "HAI", "SCH")
                    ),
      sliderInput("year",
                    "Year of observation:",
                    min = 2006,
                    max = 2016,
                    value = 2014)
      ),
      
      # Show a plot of the generated distribution
      mainPanel(
         plotOutput("distPlot")
      )
   )
)

# Define server logic required to draw a histogram
server <- function(input, output) {
   
  output$distPlot <-  renderPlot({
    
    dd <- subset(LUI_factors, Year == input$year & Exploratory == input$exploratory )
    
   
    plot(LUI ~ EP_PlotID, 
         data = droplevels(dd),
         ylim = c(0,1)
         )
   })
}

# Run the application 
shinyApp(ui = ui, server = server)

animation package

The package animation offers the possibility to create animated gifs or videos from your plot, e.g. to create a visual of changin parameters on a model.