Animals: Data Summaries

Data generated from animal surveys can be summarised at multiple levels (e.g. area, period, animal groups), in order to make direct comparisons between these groups. For example, one may be interested in differences in the number of species between different areas, or across two different time periods. This article demonstrates how such summaries may be performed and visualised.

First, load the necessary packages to run the analysis:

library("biodivercity")
library("dplyr") # data processing
library("ggplot2") # data visualisation

Summarise data

Load the example data. Refer to help(animal_observations) and help(animal_surveys) for more details on these datasets, as well as vignette("process-animal") on how the animal surveys may be processed.

data("animal_observations")
data("animal_surveys")

Next, define the unique combinations of the grouping variables of interest. These will be used to summarise the data. For example, animal surveys can be summarised for each of the six towns (column area), two survey periods (column period), as well as four animal groups (column taxon):

unique <- animal_surveys %>%
  distinct(area, period, taxon) %>% 
  arrange(area)

unique

## # A tibble: 40 × 3
##    area  period taxon      
##    <chr>  <dbl> <chr>      
##  1 BS         1 Odonata    
##  2 BS         1 Lepidoptera
##  3 BS         1 Aves       
##  4 BS         1 Amphibia   
##  5 JW         1 Aves       
##  6 JW         1 Lepidoptera
##  7 JW         1 Odonata    
##  8 JW         1 Amphibia   
##  9 JW         2 Aves       
## 10 JW         2 Lepidoptera
## # … with 30 more rows

For each of these (rows of) unique combinations, calculate the species accumulation curve using the function calculate_sac(). Species accumulation curves provide useful information on how much sampling effort is required to be reasonably certain of the total number of species (‘gamma’ diversity) within a given sampling unit. In this example, they allow the number of species between different animal groups, areas and survey periods to be compared at the same amount of sampling effort, i.e., same point along the curves.

plot_sac <- data.frame()

for(i in 1:nrow(unique)){ # calculate for each row of 'unique'
  
  sac_data <- calculate_sac(observations = animal_observations, 
                            survey_ref = animal_surveys,
                            specify_area = unique$area[i], 
                            specify_period = unique$period[i], 
                            specify_taxon = unique$taxon[i])
  
  # append results to plot_sac (overwrite), and order each grouping factor
  plot_sac <- plot_sac %>%
    bind_rows(sac_data) %>% 
    mutate(area = factor(area, levels = c("PG", "QT", "TP", "JW", "BS", "WL"))) %>%
    mutate(taxon = factor(taxon, levels = c("Aves", "Lepidoptera", "Odonata", "Amphibia"))) 
}

The output plot_sac contains data points along the species accumulation curves, i.e., the estimated number of species (column richness) with increasing survey effort (column sites), as well as the standard deviation (column sd) for the estimates:

head(plot_sac)

##   area period   taxon sites richness        sd
## 1   BS      1 Odonata     1 2.533333 0.6900281
## 2   BS      1 Odonata     2 4.430508 1.1270489
## 3   BS      1 Odonata     3 5.902016 1.4111596
## 4   BS      1 Odonata     4 7.083565 1.6020460
## 5   BS      1 Odonata     5 8.063188 1.7352009
## 6   BS      1 Odonata     6 8.898539 1.8317823

Make comparisons

For each animal group, we can only compare between the survey areas/periods based on similar levels of sampling effort. Thus, we need to extract the estimates (column richness) which have a similar value for the column sites. To improve the certainty of comparisons, the sampling effort used for comparisons should be as high as possible (where the curves plateau); accordingly, choose the highest available value for sites that are present among all grouping variables (area, period, taxon):

n_refer <- plot_sac %>%
  group_by(area, period, taxon) %>%
  summarise(n_max = max(sites)) %>% # highest value for 'sites' across grouping vars
  group_by(area, period, taxon) %>%
  summarise(sites = min(n_max)) %>% # lowest common denominator among the grouping vars (value for 'sites' present among all)
  group_by(taxon) %>%
  summarise(sites = min(sites)) %>% # analyse each taxon separately
  inner_join(plot_sac) # join data of 'richness' for all grouping variables at corresponding value for 'sites'

The resulting dataframe n_refer allows direct comparisons to be made between the different areas/periods, based on the estimated number of species (column richness):

n_refer

## # A tibble: 40 × 6
##    taxon sites area  period richness    sd
##    <fct> <int> <fct>  <dbl>    <dbl> <dbl>
##  1 Aves     84 BS         1     66    3.18
##  2 Aves     84 JW         1     61.5  4.55
##  3 Aves     84 JW         2     64.0  4.16
##  4 Aves     84 PG         1     73.8  5.27
##  5 Aves     84 PG         2     73.6  4.40
##  6 Aves     84 QT         1     82.3  3.68
##  7 Aves     84 QT         2     68.8  4.23
##  8 Aves     84 TP         1     69.9  4.19
##  9 Aves     84 TP         2     70.7  4.91
## 10 Aves     84 WL         1     64.1  4.86
## # … with 30 more rows

Finally, the species accumulation curves (plot_sac) and comparisons based on similar sampling efforts (n_refer) can be visualised:

plot_sac %>%
  
  ggplot() +
    facet_grid(taxon ~ area, scales = "free_y") +
    geom_line(aes(x = sites, y = richness, color = factor(period))) +
    geom_ribbon(aes(x = sites,
                    ymin = (richness - 2 * sd),
                    ymax = (richness + 2 * sd),
                    fill = factor(period)),
                alpha = 0.2) +
    geom_point(data = n_refer,
               aes(x = sites, y = richness), shape = 4) +
    geom_segment(data = n_refer,
                 aes(x = sites, y = 0, xend = sites, yend = richness),
                 linetype = "dashed") +
    geom_segment(data = n_refer,
                 aes(x = 0, y = richness, xend = sites, yend = richness), 
                 linetype = "dashed") +
    ylab("Cumulative no. of species") + 
    xlab("No. of surveys") +
    labs(color = "Survey\nPeriod", fill = "Survey\nPeriod") +
    theme_bw() + 
    theme(panel.grid.major.x = element_blank(),
          panel.grid.minor.x = element_blank())

Figure: Species accumulation curves showing the number of species for each area and taxon, across two survey periods. The average number of species (solid lines) and two standard deviations of the mean (shaded region) are shown. Note that the scale of the y-axis varies between the animal groups.

Note that some of the curves have yet to reach a clear plateau, despite the high sampling effort. At each of the six towns (area), six surveys (cycles) were conducted per sampling point (point_id), per survey period (duration of one year). This was conducted for each of the four animal groups (column taxon): Birds (Aves), Butterflies (Lepidoptera), Odonates (Odonata) and Amphibians (Amphibia).

Compare local diversity between sampled areas

2022-12-08

Summarise data

Make comparisons