Skip to contents

Diversity analyses

Source: vignettes/diversity.Rmd
diversity.Rmd

This vignette shows how to compute diversity indices from camera trap data.

Import data 

data(mica, package = "camtraptor")
records <- mica$data$observations

Summarize data

To compute diversity indices, we need the number and proportion of sightings for each species per camera. For that, we use the function summarize_species with by_cam = TRUE:

spp_cam <- summarize_species(records, 
                             spp_col = "vernacularNames.en", 
                             cam_col = "deploymentID",
                             obstype_col = "observationType",
                             count_col = "count",
                             by_cam = TRUE,
                             NA_count_placeholder = 1)

knitr::kable(spp_cam)
vernacularNames.en observationType deploymentID sightings individuals sightings_prop individuals_prop
Eurasian beaver animal 577b543a-2cf1-4b23-b6d2-cda7e2eac372 1 1 0.1111111 0.1111111
European polecat animal 577b543a-2cf1-4b23-b6d2-cda7e2eac372 3 3 0.3333333 0.3333333
beech marten animal 577b543a-2cf1-4b23-b6d2-cda7e2eac372 1 1 0.1111111 0.1111111
gadwall animal 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 3 6 0.3000000 0.2727273
great herons animal 62c200a9-0e03-4495-bcd8-032944f6f5a1 2 2 0.4000000 0.4000000
grey heron animal 62c200a9-0e03-4495-bcd8-032944f6f5a1 1 1 0.2000000 0.2000000
human human 7ca633fa-64f8-4cfc-a628-6b0c419056d7 1 2 0.3333333 0.5000000
mallard animal 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 4 13 0.4000000 0.5909091
red fox animal 577b543a-2cf1-4b23-b6d2-cda7e2eac372 1 1 0.1111111 0.1111111
NA blank 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 1 1 0.1000000 0.0454545
NA blank 577b543a-2cf1-4b23-b6d2-cda7e2eac372 3 3 0.3333333 0.3333333
NA unknown 29b7d356-4bb4-4ec4-b792-2af5cc32efa8 2 2 0.2000000 0.0909091
NA unclassified 62c200a9-0e03-4495-bcd8-032944f6f5a1 2 2 0.4000000 0.4000000
NA unclassified 7ca633fa-64f8-4cfc-a628-6b0c419056d7 2 2 0.6666667 0.5000000

Compute diversity indices

Then, we can compute diversity indices by feeding this summary dataframe to the get_diversity_indices function.

div <- get_diversity_indices(spp_cam,
                             spp_col = "vernacularNames.en",
                             cam_col = "deploymentID")

knitr::kable(div)
deploymentID richness shannon simpson
29b7d356-4bb4-4ec4-b792-2af5cc32efa8 4 1.0237155 0.4069264
577b543a-2cf1-4b23-b6d2-cda7e2eac372 5 1.4648164 0.1666667
62c200a9-0e03-4495-bcd8-032944f6f5a1 3 1.0549202 0.2000000
7ca633fa-64f8-4cfc-a628-6b0c419056d7 2 0.6931472 0.3333333

Let \(i = 1 ... S\) denote the species and \(j = 1 ... C\) denote the cameras. \(S_j\) is the number of species seen at camera \(j\). \(N_j\) is the total number of individuals of all species seen at a camera \(j\). \(n_{ij}\) and \(p_{ij}\) represent respectively the count and proportion if individuals of species \(i\) seen at a camera \(j\). This function computes:

  • richness \(R_j = S_j\) (the number of species seen at camera \(j\)).
  • Shannon index \(H_j = -\sum_{i = 1}^{S_j} p_{ij} \ln(p_{ij})\). It ranges between 0 and \(+\infty\), zero indicating the lowest diversity.
  • Simpson index \(D_j = \sum_{ = 1}^{S_j} \frac{n_{ij}(n_{ij} - 1) }{N_j(N_j - 1)}\). It ranges between 0 and 1, one indicating the lowest diversity.

Optionally, you can also provide the name of the columns containing counts and proportions (but they default to the name given by the summarize_species function).

div <- get_diversity_indices(spp_cam,
                             spp_col = "vernacularNames.en",
                             cam_col = "deploymentID",
                             count_col = "individuals",
                             prop_col = "individuals_prop")

Plot diversity indices

Although it is then simple to use the diversity dataframe to make a custom plot, the helper function plot_diversity is provided.

Here, we plot the richness.

plot_diversity(div, 
               div_col = "richness", 
               cam_col = "deploymentID") +
  ggtitle("Species richness")

The plot can also be made interactive.

pr <- plot_diversity(div, 
                     div_col = "richness", 
                     cam_col = "deploymentID", 
                     interactive = TRUE) +
  ggtitle("Species richness")

girafe(ggobj = pr)

In case you want to show more cameras than the ones in the div dataframe, you can use the cam_vec argument.

div_less <- div[1:3, ]
cameras <- unique(div$deploymentID)

# We deleted one camera
unique(div_less$deploymentID)
#> [1] "29b7d356-4bb4-4ec4-b792-2af5cc32efa8"
#> [2] "577b543a-2cf1-4b23-b6d2-cda7e2eac372"
#> [3] "62c200a9-0e03-4495-bcd8-032944f6f5a1"

# But it is still in cameras
cameras
#> [1] "29b7d356-4bb4-4ec4-b792-2af5cc32efa8"
#> [2] "577b543a-2cf1-4b23-b6d2-cda7e2eac372"
#> [3] "62c200a9-0e03-4495-bcd8-032944f6f5a1"
#> [4] "7ca633fa-64f8-4cfc-a628-6b0c419056d7"
pr <- plot_diversity(div_less, 
                     div_col = "richness", 
                     cam_col = "deploymentID", 
                     cam_vec = cameras,
                     interactive = TRUE) +
  ggtitle("Species richness")

girafe(ggobj = pr)

Finally, by changing the div_col argument you can plot multiple diversity indices.

ph <- plot_diversity(div, 
                     div_col = "shannon", 
                     cam_col = "deploymentID", 
                     interactive = TRUE) +
  ggtitle("Shannon diversity index")

pd <- plot_diversity(div, 
                     div_col = "simpson", 
                     cam_col = "deploymentID", 
                     interactive = TRUE) +
   ggtitle("Simpson diversity index")

girafe(ggobj = ph) 
girafe(ggobj = pd)