Modelling the distributions of Falkland Island plants, part one - disentangling datasets
Rebecca Upson from Kew's UK Overseas Territories team is investigating the impacts of climate change on plant distributions in the Falkland Islands. Here she provides an insight into the vital first stage of the project.
If there was just one message to take home from this post, it would be that it takes a lot of work to prepare botanical records prior to their use in spatial modelling. This is certainly not something that is always appreciated and yet certainly is something for which adequate time must be allowed in the planning of similar projects. I for one know that many were perplexed by my repetition of the answer 'I’m organising the data' several months into the current project.
Surely once you have a file of GPS coordinates and species names you are ready to go? Not so... there is invariably a huge amount of data processing that needs to take place before you can start to unlock the ecological trends held in your records.
Get to know your data
In our case we are using a dataset of botanical records (largely gathered by staff and volunteers at Falklands Conservation) for the Falkland Islands to develop species distribution models. These models are being developed in order to investigate the possible impacts, through temperature increases, that climate change may have on plant diversity across the archipelago.
Species distribution modelling depends on access to reliable and accurate records so it is vital to take the time to go through and cross-check the data you have available. No small task when you have over 50,000 records! We are fortunate in that we can still contact data collectors from the last major surveying effort in the Falkland Islands to the present day (I am one of them). This and our own familiarity with the flora of the Islands have helped to solve a range of data issues, from plant and place name typos to erroneous GPS coordinates.
Know your plant names
Up-to-date knowledge of the taxonomy of the flora is also vital to ensure that all records of the same taxon are referred to by the same name. The name used for a plant can sometimes change, often as a result of advances in our knowledge of plant diversity, such as when a very variable species becomes recognised as two or more separate species, or is found to belong to a different genus.
For example, the scarce coral fern has historically been referred to by the Latin name Gleichenia cryptocarpus but is now classified within the closely related genus Sticherus - so I need to make sure that all records for this can be located under a single search term. In other cases a single species can mistakenly be referred to by more than one name, when only one correct name is allowed.
The scarce native coral-fern (Sticherus cryptocarpus) growing on a sheltered slope on West Falkland.
Know your study area
The British Isles are extremely intensively surveyed, to the point that it is possible for separate time snaps to be taken of the distributions of different plant species across particular years. In the Falklands we do not have this luxury and must treat our dataset as one single ‘snapshot’. It is therefore important to use our current knowledge of the Islands and their ecology to inform which historical records we accept and which we exclude from our analyses.
For example, recent survey work on an island out in the far west of the archipelago indicates that a historical record of the endemic snake plant Nassauvia serpens from the American botanist G.H.Snyder in 1853 should no longer be included in studies. This species is both vulnerable to grazing and drought and it is likely that a combination of these factors is responsible for its absence at the site today.
The endemic snakeplant (Nassauvia serpens) growing out from a stone run in the Hill Cove Mountain.
Know your place names
Many historical records are not associated with an accurate latitude and longitude but may rather be associated solely with a place name. Another major task was therefore to go through the dataset and assign a ‘resolution’ to each record. By this I mean it was necessary to indicate what the maximum diameter is of the area within which the record could have been made. The place name may be a hill which may place the record within several square kilometres or the summit of that hill may have been explicitly named, therefore narrowing down the possible location to perhaps several 100 square meters.
Many records are associated with the tallest peak in the Falkland Islands, Mount Usborne (705 m a.s.l.). However, this covers an extremely large area with plants possibly recorded anywhere within an area with a maximum diameter of 4 km. In contrast other records have more specific information, pinpointing the location of plants to an outcrop on the south side of Mount Usborne known as ‘Ceritos Rocks’. When I don’t know a specific area myself and can’t use the Falkland Islands official gazetteer to find it, then I have often contacted local Falkland Islanders – I am very grateful for all their help on this!
Make sure your records are at the same resolution or higher than your environmental variables
Our climate data is on a scale of 1 km grid squares, which means that each 1 km grid square covering the Falkland Islands has a particular set of climate data associated with it. So to study relationships between the distribution of a species and these climatic gradients we need, as a minimum, to know for sure that any given record could only have come from one 1 km grid cell rather than, say, any of the surrounding grid squares.
We also have data that relates to local environmental conditions such as water availability and surface temperature – these data are mapped at a finer scale so that any given 100 m square has a particular value associated with it. These data allow us to investigate the relationship between the distribution of a species and a range of habitat features that influence the resources available at any given site. This work is being carried out in collaboration with the GIS team at Kew. We have now classified our records according to their resolution, leaving us with three sets of data, two of which are at a high enough resolution for use in distribution modelling:
|Resolution||Number of vascular plant records|
|≤ 100 m||32,111|
|≤ 1 km||57,628|
|> 1 km||4,108|
Some of the Falkland’s most widespread and common plant species have over 1,000 records – for example, the dominant heathland shrub, Empetrum rubrum, has 2,045 records – whereas 27 of the 180 native vascular plants have fewer than 20 records. This is largely because they are genuinely rare, but is also because there are still areas of the Falkland Islands that have not been surveyed.
In my next post I'll be presenting a case study of two Falkland species with very different distributions across the Islands.
- Rebecca -