Ecotourism Tutorial

Joining Occurrence, Tourism and Weather Data- Are Glowworm sightings driven by tourism?

Author

Shreya Gupta

Objectives

In this tutorial you will investigate whether glowworm sightings in Tasmania are driven by tourist activity or local hobbyists.

By the end you should be able to:

Filter occurrence records by state and month
Join occurrence data to weather using ws_id and data
Join occurrence data to tourism data via tourism_region
Interpret whether tourism and sightings move together over time

Preparation

Install the package if you haven’t already:

Code

install.packages("pak")
pak::pak("vahdatjavad/ecotourism")

We will use four datasets from the ecotourism package:

glowworms: glowworm occurrence records (2014–2024)
weather: daily weather by station
tourism_quarterly: quarterly domestic visitor counts by region
tourism_region: tourism region names linked to weather stations

Exercises

We focus on glowworm sightings in Tasmania across all years, investigating whether sightings and tourism activity move together.

Explore the data first

Explore the table above to get familiar with the data. Notice the hour, date, and ws_id columns which we will use for joining later.

The map below shows all Tasmanian glowworm sightings from 2014–2024. Each gold dot represents a recorded sighting. Click any marker for date and time details.

Question 1 . When are glowworms spotted?

Filter the glowworms dataset to Tasmania in December. Join with the weather dataset using ws_id and date.

What hours of the day do most sightings occur?
What was the average temperature (temp) on sighting days?
Were most sightings on rainy or clear days?

What does this tell you about glowworm spotting conditions?

Solution.

Code

# filter Tasmania December sightings
tas_dec <- glowworms |>
  filter(obs_state == "Tasmania", month == 12)

# join with weather on both ws_id and date for exact area and day match
tas_dec_weather <- tas_dec |>
  left_join(weather, by = c("ws_id", "date"))

# summarise key weather stats on sighting days
tas_dec_weather |>
  summarise(
    n_sightings = n(),                          # total sightings
    avg_temp = mean(temp, na.rm = TRUE),        # mean temperature
    prop_rainy = mean(rainy, na.rm = TRUE)      # proportion rainy days
  )

# A tibble: 1 × 3
  n_sightings avg_temp prop_rainy
        <int>    <dbl>      <dbl>
1          30     9.19          1

What do you notice?

Most sightings occur in the afternoon (3pm) rather than at night which is rather surprising for a bioluminescent organism! This likely reflects observer activity patterns rather than glowworm behaviour. People visit caves and trails during daylight hours and record what they see.

Question 2 — Do sightings and tourism peak together?

Using all years of Tasmania glowworm data, join glowworms to tourism_region via ws_id, then to tourism_quarterly via region_id.

Count glowworm sightings per year and quarter
Calculate average tourism trips per year and quarter
Do sightings and tourism trips rise and fall together?

Solution.

Code

# 1st joining glowworms to region info
tas_region <- glowworms |>
  filter(obs_state == "Tasmania") |>
  left_join(tourism_region, by = "ws_id")         

# 2nd join to tourism quarterly data
tas_tourism <- tas_region |>
  left_join(tourism_quarterly,                      # bring in trip counts
            by = c("region_id", "ws_id")) |>
  filter(!is.na(region))                            # drop unmatched rows

# summarise sightings and trips per year and quarter
yearly <- tas_tourism |>
  group_by(year.x, quarter) |>                     # year.x = glowworm year
  summarise(
    n_sightings = n(),                              # count sightings
    avg_trips = mean(trips, na.rm = TRUE),          # average tourism trips
    .groups = "drop"
  )

yearly

# A tibble: 32 × 4
   year.x quarter n_sightings avg_trips
    <dbl>   <int>       <int>     <dbl>
 1   2015       1          20      8.20
 2   2015       2          21      5.85
 3   2015       3          13      4.26
 4   2015       4          18      5.67
 5   2017       1          20      8.20
 6   2017       2          21      5.85
 7   2017       3          13      4.26
 8   2017       4          18      5.67
 9   2018       1          80     13.9 
10   2018       2          63     11.9 
# ℹ 22 more rows

What do you notice?

If the two lines move together, i.e, sightings spike when tourism spikes, this suggests tourists are the ones recording glowworm sightings rather than local hobbyists. Look for years where both dip (hint: 2020!) as extra evidence.

Question 3 — Holiday tourists or business travellers?

From your joined dataset in Question 2, split the tourism data by purpose (Holiday vs Business).

Calculate the correlation between sightings and trips separately for Holiday and Business purpose
Which type of tourism correlates more strongly with glowworm sightings?
What does this suggest about who is recording glowworms?

Solution.

Code

# summarise by year, quarter AND purpose this time
purpose_yearly <- tas_tourism |>
  group_by(year.x, quarter, purpose) |>        # split by purpose
  summarise(
    n_sightings = n(),                          # sightings per group
    avg_trips = mean(trips, na.rm = TRUE),      # trips per group
    .groups = "drop"
  ) |>
  filter(!is.na(purpose))                       # drop missing purpose rows

# calculate correlation for each purpose separately
purpose_yearly |>
  group_by(purpose) |>
  summarise(
    correlation = cor(                          # cor() gives correlation
      n_sightings, 
      avg_trips, 
      use = "complete.obs"                      # ignore NA pairs
    )
  )

# A tibble: 2 × 2
  purpose  correlation
  <chr>          <dbl>
1 Business       0.317
2 Holiday        0.205

Key Finding

If Holiday trips correlate more strongly with sightings than Business trips, this is strong evidence that tourists on holiday are driving glowworm recordings and not so much locals or business travellers. This has implications for the ecotourism package itself that sighting data may be biased toward tourist-heavy seasons and regions.

Finishing Up

In this tutorial you investigated whether glowworm sightings in Tasmania are driven by tourist activity. You practiced:

Joining glowworms -> tourism_region ->tourism_quarterly
Joining occurrence data with weather using ws_id and date
Interpreting correlations between sightings and tourism purpose

Want to explore further?

Try repeating Question 2 with a different organism. Do gouldian_finch or manta_rays show the same tourism pattern?
Filter Question 2 to exclude 2020. Does the correlation between tourism and sightings get stronger or weaker without the pandemic year?
Filter Question 3 to only Q4 (October–December). Does the Holiday correlation get stronger in peak tourist season?

--- title: "Ecotourism Tutorial" subtitle: "Joining Occurrence, Tourism and Weather Data- Are Glowworm sightings driven by tourism?" author: "Shreya Gupta" format: html: embed-resources: true code-fold: true code-tools: true toc: true theme: cosmo editor: visual --- ```{r} #| label: setup #| message: false #| warning: false #| echo: false library(ecotourism) library(tidyverse) library(plotly) library(DT) library(leaflet) ``` ## Objectives In this tutorial you will investigate whether glowworm sightings in Tasmania are driven by tourist activity or local hobbyists. By the end you should be able to: - Filter occurrence records by state and month - Join occurrence data to weather using `ws_id` and `data` - Join occurrence data to tourism data via `tourism_region` - Interpret whether tourism and sightings move together over time ## Preparation Install the package if you haven't already: ```{r} #| eval: false install.packages("pak") pak::pak("vahdatjavad/ecotourism") ``` We will use four datasets from the `ecotourism` package: - `glowworms`: glowworm occurrence records (2014–2024) - `weather`: daily weather by station - `tourism_quarterly`: quarterly domestic visitor counts by region - `tourism_region`: tourism region names linked to weather stations ## Exercises We focus on **glowworm** sightings in **Tasmania** across all years, investigating whether sightings and tourism activity move together. ### Explore the data first Explore the table above to get familiar with the data. Notice the `hour`, `date`, and `ws_id` columns which we will use for joining later. ```{r} #| label: preview-table #| echo: false # filters to Tasmania glowworms only tas_glow <- glowworms |> filter(obs_state == "Tasmania") # shows interactive scrollable table (DT package) datatable( tas_glow |> select(date, month, hour, obs_lat, obs_lon, sci_name, ws_id), options = list(pageLength = 8), caption = "Tasmania Glowworm Sightings (2014-2024)" ) ``` The map below shows all Tasmanian glowworm sightings from 2014–2024. Each gold dot represents a recorded sighting. Click any marker for date and time details. ```{r} #| label: preview-map #| echo: false # removes any rows where coordinates are missing tas_map <- tas_glow |> filter(!is.na(obs_lon), !is.na(obs_lat)) # builds leaflet map and addTiles() adds the base map # addCircleMarkers plots each sighting as a dot leaflet(tas_map) |> addTiles() |> addCircleMarkers( lng = ~obs_lon, lat = ~obs_lat, radius = 6.5, color = "coral", fillColor = "orange", fillOpacity=0.7, weight=1, popup = ~paste0( # click a dot to see details "Date: ", date, " Time: ", time, " Species: ", sci_name ) ) ``` ### Question 1 . When are glowworms spotted? Filter the `glowworms` dataset to Tasmania in December. Join with the `weather` dataset using `ws_id` and `date`. 1. What hours of the day do most sightings occur? 2. What was the average temperature (`temp`) on sighting days? 3. Were most sightings on rainy or clear days? What does this tell you about glowworm spotting conditions? #### Solution. ```{r} #| label: q1-solution #| code-fold: true #| warning: false # filter Tasmania December sightings tas_dec <- glowworms |> filter(obs_state == "Tasmania", month == 12) # join with weather on both ws_id and date for exact area and day match tas_dec_weather <- tas_dec |> left_join(weather, by = c("ws_id", "date")) # summarise key weather stats on sighting days tas_dec_weather |> summarise( n_sightings = n(), # total sightings avg_temp = mean(temp, na.rm = TRUE), # mean temperature prop_rainy = mean(rainy, na.rm = TRUE) # proportion rainy days ) ``` ```{r} #| label: q1-plot #| echo: false #| warning: false # count sightings per hour hour_counts <- tas_dec |> count(hour) # plotly bar chart for sightings by hour plot_ly( hour_counts, x = ~hour, # x axis-hour y = ~n, # y axis-count type = "bar", marker = list(color = "indianred") ) |> layout( title = "Glowworm Sightings by Hour of Day (Tasmania, December)", xaxis = list(title = "Hour of Day (24hr)"), yaxis = list(title = "Number of Sightings") ) ``` ::: {.callout-note collapse="true"} ## What do you notice? Most sightings occur in the **afternoon (3pm)** rather than at night which is rather surprising for a bioluminescent organism! This likely reflects **observer activity patterns** rather than glowworm behaviour. People visit caves and trails during daylight hours and record what they see. ::: ### Question 2 — Do sightings and tourism peak together? Using all years of Tasmania glowworm data, join `glowworms` to `tourism_region` via `ws_id`, then to `tourism_quarterly` via `region_id`. 1. Count glowworm sightings per year and quarter 2. Calculate average tourism trips per year and quarter 3. Do sightings and tourism trips rise and fall together? #### Solution. ```{r} #| label: q2-solution #| code-fold: true #| warning: false # 1st joining glowworms to region info tas_region <- glowworms |> filter(obs_state == "Tasmania") |> left_join(tourism_region, by = "ws_id") # 2nd join to tourism quarterly data tas_tourism <- tas_region |> left_join(tourism_quarterly, # bring in trip counts by = c("region_id", "ws_id")) |> filter(!is.na(region)) # drop unmatched rows # summarise sightings and trips per year and quarter yearly <- tas_tourism |> group_by(year.x, quarter) |> # year.x = glowworm year summarise( n_sightings = n(), # count sightings avg_trips = mean(trips, na.rm = TRUE), # average tourism trips .groups = "drop" ) yearly ``` ```{r} #| label: q2-plot #| echo: false #| warning: false # plotly dual line chart -> sightings and trips over time plot_ly(yearly, x = ~year.x) |> add_lines(y = ~n_sightings, # sightings line name = "Sightings", line = list(color = "forestgreen")) |> add_lines(y = ~avg_trips, # tourism line name = "Avg Trips", line = list(color = "skyblue")) |> layout( title = "Glowworm Sightings vs Tourism Trips in Tasmania", xaxis = list(title = "Year"), yaxis = list(title = "Count"), legend = list(x = 0.1, y = 0.9) ) ``` ::: {.callout-note collapse="true"} ## What do you notice? If the two lines move together, i.e, sightings spike when tourism spikes, this suggests **tourists are the ones recording glowworm sightings** rather than local hobbyists. Look for years where both dip (hint: 2020!) as extra evidence. ::: ### Question 3 — Holiday tourists or business travellers? From your joined dataset in Question 2, split the tourism data by `purpose` (Holiday vs Business). 1. Calculate the correlation between sightings and trips separately for Holiday and Business purpose 2. Which type of tourism correlates more strongly with glowworm sightings? 3. What does this suggest about **who** is recording glowworms? #### Solution. ```{r} #| label: q3-solution #| code-fold: true #| warning: false # summarise by year, quarter AND purpose this time purpose_yearly <- tas_tourism |> group_by(year.x, quarter, purpose) |> # split by purpose summarise( n_sightings = n(), # sightings per group avg_trips = mean(trips, na.rm = TRUE), # trips per group .groups = "drop" ) |> filter(!is.na(purpose)) # drop missing purpose rows # calculate correlation for each purpose separately purpose_yearly |> group_by(purpose) |> summarise( correlation = cor( # cor() gives correlation n_sightings, avg_trips, use = "complete.obs" # ignore NA pairs ) ) ``` ```{r} #| label: q3-plot #| echo: false #| warning: false # scatter plot sightings vs trips, coloured by purpose plot_ly( purpose_yearly, x = ~avg_trips, # x-tourism trips y = ~n_sightings, # y-sightings color = ~purpose, # colour by Holiday vs Business colors = c("limegreen", "purple"), type = "scatter", mode = "markers", marker = list(size = 8), text = ~paste0( # hover text "Year: ", year.x, " Quarter: ", quarter, " Purpose: ", purpose, " Trips: ", round(avg_trips), " Sightings: ", n_sightings ), hoverinfo = "text" ) |> layout( title = "Do Holiday or Business Tourists Record More Glowworms?", xaxis = list(title = "Average Tourism Trips"), yaxis = list(title = "Glowworm Sightings"), legend = list(title = list(text = "Purpose")) ) ``` ::: {.callout-important collapse="true"} ## Key Finding If Holiday trips correlate more strongly with sightings than Business trips, this is strong evidence that **tourists on holiday are driving glowworm recordings** and not so much locals or business travellers. This has implications for the ecotourism package itself that sighting data may be biased toward tourist-heavy seasons and regions. ::: ## Finishing Up In this tutorial you investigated whether glowworm sightings in Tasmania are driven by tourist activity. You practiced: - Joining `glowworms` -\> `tourism_region` -\>`tourism_quarterly` - Joining occurrence data with `weather` using `ws_id` and `date` - Interpreting correlations between sightings and tourism purpose ### Want to explore further? - Try repeating Question 2 with a different organism. Do `gouldian_finch` or `manta_rays` show the same tourism pattern? - Filter Question 2 to exclude 2020. Does the correlation between tourism and sightings get stronger or weaker without the pandemic year? - Filter Question 3 to only Q4 (October–December). Does the Holiday correlation get stronger in peak tourist season?