Methodology:

Since 1998, station personnel have been recording the wildlife they've observed whilst living and working at Rothera Research Station, as part of the Rothera Time-Series (RaTS) project. Originally, this consisted of a daily count of animals seen recorded in a logbook kept in Bransfield House. More recently, dedicated wildlife survey recording sheets have been made available in both New Bransfield House (NBH) and the Bonner Lab. The handwritten raw data are then collated and input into a spreadsheet, creating a database of wildlife sightings.

The exact methodology has changed a few times over the years. Originally, the largest single count per species per day was recorded in the spreadsheet. However, in January 2015 there was a significant change in how the data were collected. Firstly, the original recording sheets left a (fairly small) space to note down species names. This reduced the number of observations and often led to a highlights-only approach. The new recording sheet lists all the regular species and so is simpler to use, and responders are prompted to record everything they see. Secondly, to smooth out some of the noise, there was a switch from recording daily observations to weekly presence/absence data, together with the maximum number of each species observed at once during that week. This also reduced the impact of the same animals being counted multiple times by different observers.

The different versions of the data recording sheets are available on request from the UK Polar Data Centre

In some years, the Bonner Terrestrial Assistant conducted regular wildlife counts. From 2013, the Air Unit Ground Coordinator requested his own recording sheets and conducted daily surveys throughout the summer months from the Hangar. But for the most part, data have been recorded on a casual basis by anyone on Station.

Data validation

An individual data cell can be of four types

1. A record with numbers

2. A record with words ('x', 'few', 'some', 'many', 'lots', 'loads')

3. Empty, indicating that species was not seen during that recording period (day or week)

4. Empty because no observations were recorded on that day (or during that week)

The first is straightforward, needing no interpretation.

The second required interpretation; these were recoded as follows:

'x' = 1 (simply indicating the presence of at least one individual)

'few' = 2 (based on the frequency of previous observations)

'some' = 5 (based on the frequency of previous observations)

'many', 'lots' and 'loads' = 50 (based on the frequency of previous observations)

For the remaining two types (empty cells):

Where there were data for other species on that day, the cell was coded as 0.

Where there were no data for any other species on that day, or an absence of observations is specifically mentioned, that cell was coded as missing data.

Data quality:

Volunteer data collection

Volunteer-collected observational data has its limitations, with the following impacting data quality and variability:

-The accuracy of species identification depends on the observer's experience. For this dataset, this will have varied hugely.

-The number of individual animals recorded may be rough estimates. The greater the number recorded, the less accurate the estimate is likely to be.

-There will have been variability in whether observers were using a telescope, binoculars, or the naked eye. This will not have been recorded.

-Recording effort varies hugely at every timescale. This relates to how long an observer spent outside observing, but also how many different people completed recording sheets on any particular day, week, or month. This is influenced by the number of people on station on any given day (there are far more people on Station in summer than in winter), the daylight hours available to make observations (the sun doesn't rise for six weeks during midwinter), whether the weather encouraged people to spend time outside, amongst other factors.

Missing data

Certain months have no data at all. This may be due to no-one completing a recording sheet during that month, or because recording sheets were lost before being input to the spreadsheet.