I’m not going to lie, organising millions of records of species observations, developing and testing confusing models and running the models (often more than once!) to produce the data described in our data paper was not an easy task. So, was it worth it? Absolutely!
Let me give you some background. In the UK, we have very good data on certain groups of species which has enabled scientists to determine trends in these species over time. We know, for instance, that farmland birds have undergone strong declines in numbers since 1970, that specialist butterflies have also declined and that populations of bats have increased. Knowing these trends then enables us to start thinking about what is causing them so that we can prevent declines.
The data we have for these well-studied groups is great because it is collected from across the UK following a strict protocol. This means the data is standardised and it can easily be used to look at how numbers of species have changed over time - trends. Unfortunately, this data is not available for much of the UK’s wildlife, in particular for most insect groups. So, in order to determine how other species are doing, as was our aim, we had to find an alternative source of data.
This data was in the form of biological records. A biological record is an observation of a species at a known time and location. These records are collected and collated by volunteers as part of recording schemes. There are over 90 recording schemes in the UK, each with a focus on a different set of species from ants, bees & wasps, to fungus gnats, soldierbeetles, plants and lichens. Many volunteers collect data and, as a result, there are millions of records available over time. The problem is that these data are not collected using a strict protocol, they are opportunistic, and so have issues with biases such as imperfect detection.
Luckily, there are a class of models out there that enable us to account for the biases associated with this kind of data. They are called occupancy-detection models. They are quite tricky and can take a long time to run when you have a lot of data. While models for some species groups took just a few days to run, others needed the use of a super computer called JASMIN and took months! Needless to say, I was not impressed when I ended up having to rerun some of these models, sometimes more than once (doh!).
But it was worth it! At the end of a lot of work, we have been able to produce annual estimates of species occupancy and associated trends for over 5,000 UK species from 31 taxonomic groups for the time period 1970 to 2015. For many of the species, including many insects, bryophytes and lichens, this is the first time that their trends have been assessed. These estimates have been used to assess change in species distributions to monitor progress towards the UK Biodiversity Indicators and have contributed to the latest State of Nature report. We have also been able to assess the broad patterns of change in species distribution over time, keep an eye out for that paper soon! Without the information found within the biological records, and the hard work of the volunteers in collecting it, the state of many of these species would be unknown. Now, we can work towards understanding these trends and trying to prevent any further declines from occurring.
You can find and use the dataset of species trend estimates here.