Data in action: Time travelling with butterfly specimens | Digital Collections Programme

A guest blog by Galina Jönsson

From Left: Galina Jönsson in the Museum collection, The Museum’s Data Portal and graph showing scietific paper rates and numbers since 2015.

Digital Collections support over 1000 scientific papers

The Museum’s Data Portal was launched in December 2014 to provide access to Museum collections and research, enabling to explore, download and re-use these data for their own purposes. Museum collections include specimens collected over the last 200 years, a critical time period, during which humans have had a major impact on the distribution of biodiversity.

Since 2015, more than 1000 research papers have cited data from the Data portal and partner platforms like the Global Biodiversity Information Facility (GBIF), covering topics including agriculture, biodiversity, evolution, ecology, species distributions and human health. This blog looks at just one of the studies using Museum data, PhD candidate Galina Jönsson’s research using data to examine how human activity has impacted butterfly populations over the 20th Century.

Insects are declining at alarming rates, but we do not precisely know why. From wasps to butterflies, Galina is looking for answers in the Museum’s pinned insect collection and extending time series to span the period of accelerating human pressures like agricultural intensification and deforestation. ‘At first glance, my results suggested that British insects fared pretty well, but I quickly realised there is much more to this than meets the eye.’

Blame eccentric Victorians or lazy statisticians?

A graph showing rates of butterfly distribution change over the last century

Natural history collections’ pinned insect specimens have revealed fascinating changes over the last centuries but have rarely been used to map how, and why, some species increase while other decrease. Nearly everything we know about insect responses to human activities comes from survey data collected by national schemes like the UK Butterfly Monitoring Scheme (UKBMS), which was launched in 1976. One of the benefits of using survey data is that it is standardised, meaning that all species at a particular location are recorded in the same way, at the same time of the year, for multiple years in a row. This makes it easy to compare how different species change in population or geographical location over the years. In the UK, which is unusually well-documented, our knowledge from such survey data is limited to the period since 1970. This period falls after most large-scale transformations of the British landscape such as the agricultural intensification of the 1950s with its deforestation and increased pesticide-use. As a result, we find ourselves without baselines reflecting the state of biodiversity prior to major human pressures.

In contrast to survey data, museum specimens do go back much further in time to give us these baselines – but they were not systematically sampled. This challenges conventional statistics. Labels inform us where and when specimens were collected, but not how. Just like millennial houseplant enthusiasts, Victorian bug collectors had individual preferences. Some travelled far to collect one specimen of every species, others collected every tiny variation within their favourite species. Some collectors were working scientists, but a lot of the collection comes from amateurs and those that collected as a hobby, so the type of specimens and data that was recorded also varies due to the collector.

Amateur collectors can match or exceed the standards of scientists who were paid for their work. Robert ‘Porker’ Watson (1916-84) was a tax accountant and ‘amateur’ butterfly collector and breeder whose setting of specimens was described as ‘a miracle of perfection’ Aurelian Legacy, British Butterflies and their collectors (Salmon.M.A et al, 2000 p230)

Ambitious digitisation projects are making collections available with the click of a button; and in addition, now, citizen science projects generate enormous amounts of contemporary data in addition to data from collections and systematic surveys. Smartphone applications let anyone submit wildlife sightings in seconds but, just as collections reflect eclectic Victorians, citizen scientists’ preferences introduce their own set of biases to the data. We need new statistical models to extract the valuable yet varied information museum specimens, survey data and citizen science sightings hold, but the models also need to handle their respective biases.

The European hornet (Vespa crabro) and its distribution in the UK over the last 120 years.

Solving the statistical riddle


As a masters student, I naturally felt drawn to solving the statistical riddle and embarked on modelling social wasp trends using the Museum’s collection alongside survey data from the Bees Wasps and Ants Recording Society. Our study extended existing trends by 70 years, and indicated that agricultural intensification drove a 70% decline in English hornets (Vespa crabro) between 1950 and 1970. This was followed by a northward range expansion facilitated by climate change-induced warming. Today, hornets have bounced back to 1950-levels in terms of numbers but are more sparsely distributed over a larger area. Through this study, we demonstrated that specimen data from collections can produce long-term population trends, but thoroughly addressing questions of human influence requires more museum data, both species and specimens per species.

And in flew century-old butterfly specimens, forming the basis of my PhD research. In the interest of honesty, perhaps I should say ‘in flew iCollections’, NHM’s pilot mass digitisation project that digitised over half a million British butterflies and moths.
My current research explores temporal patterns of British butterfly trends across centuries, looking at how the timings of major changes to butterflies coincide with habitat changes, and how species-specific characteristics affect population-level change. There are 59 British butterfly species; another five species have become extinct in the last 150 years. Butterflies are sensitive to temperature and weather conditions, and caterpillars are picky eaters, some accept nothing but one specific host plant. These factors render them particularly vulnerable to, and simultaneously good indicators of, greater habitat and climate changes.

Generalisations hide uncomfortable truths

After a couple of years formulating the perfect model (hint: there is no such thing as a ‘perfect model’), I summarised the trends across all British butterfly species. The preliminary results were surprising. Averaging across species, there has been a 15% decrease since 1900. But we know that humans have extensively altered 75% of Earth’s surface, so this had me wondering – is a 15% decrease over 120 years really that bad?

              Next, I grouped species according to whether they are specialists requiring specific habitats (the picky eaters) or generalist wider countryside species that can use a range of habitats. The generalist species nearly doubled since 1900, whilst the specialists had halved. Separating specialists from generalists also showed that the most dramatic changes occurred before the 1970s baseline that many recording schemes give us. Just like the hornets, specialist butterflies started to plummet around 1950, but in contrast to hornets, they did not recover after 1970. It appears that agricultural intensification in the 1950s triggered the troubling subsequent declines (or at least was the straw that broke the specialist’s thorax). Wider countryside species also began expanding in the 1950s, and this expansion continued into the 2000s.

What is wrong with generalists?


Overall, preliminary results show that we’ve lost around 15% of British butterflies since 1900. Specialised species have plummeted, but generalist wider countryside species are making up for the losses. Sometimes people ask ‘what is wrong with generalists?’ – does it really matter which butterfly species are in the ascendant? It all comes down to biodiversity. The diversity of life on Earth, which we need for human well-being, prosperity and ultimately, survival.


Species richness is the number of different species in an area, a way of measuring biodiversity. When the number of species thriving in an area declines or becomes unbalanced, certain species that are doing well can come to increasingly dominate the area. The species that can’t adapt are put under further pressure from the increasing generalist species eating their food or nesting in their areas. A change to the delicate balance of the species in an area can reduce biodiversity and species richness, cause extinctions and dramatically change ecosystems.

The wall butterfly (Lasiommata megera) distribution change over 20th century


However, is it fair to divide all butterflies into either habitat generalist or specialists? And assume that, within each group, every species shows the same long-term trends? Although a habitat-use separation can give useful indications, the reality is much more complex. For instance, the wall butterfly (Lasiommata megera) has suffered worrying declines despite enjoying a variety of habitats. With rising temperatures, the cold-loving wall butterfly has been forced northwards and risks joining the list of butterflies that are extinct in Britain, when it reaches John O’Groats. Biologists often divide species by habitat-use, but the dramatic decline of the wall butterfly shows us that every species has its own particular quirks, extending beyond habitat-use. In addition to temperature-tolerance, species differ in a number of characteristics like their reproduction strategies (for example, many tiny eggs but few survive or a few huge eggs with high survival), the ease with which they find a mate, and how strong flyers they are (which determines if they can colonise new habitats). I am currently using several such species-specific characteristics to identify combinations of characteristics that predispose species to being particularly vulnerable and give others the ability to rapidly expand.

Natural history collections’ specimens are vital to gather the data needed to extend time series of species’ trends to periods prior to extensive anthropogenic pressures and provide important novel insights into our effects on biodiversity. However, most specimens world-wide are relatively inaccessible to research, hidden away in undigitised collections. Mobilising digitisation projects that provide open access to this important biodiversity data will allow us to refine models, produce more accurate future projections, and make effective conservation decisions to bend the curve of global biodiversity loss.

We would love to hear from you if you are using data from data.nhm.ac.uk please get in touch or stay up to date with Digital Collections news by following us on Twitter and Instagram. Keep up to date with our blog posts for more examples of our data in action.

 If you are spotting butterflies this summer please log your findings on a recording scheme so that researchers like Galina can make use of your work. You can also follow Galina on Twitter to keep up with her research.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.