This article reflects upon a recent paper1 co-authored by BC’s Senior Ecological Statistician, Emily Dennis, in collaboration with Alison Johnston (University of St Andrews) and Eleni Matechou (University of Kent), and published in Methods in Ecology and Evolution, which reviews the outstanding challenges for biodiversity monitoring using citizen science data.
Robust measures of change are vital for providing reliable evidence of changes in the populations of butterflies and macro-moths, both in the UK and beyond. Data gathered through Butterfly Conservation’s extensive and long-running monitoring and recording schemes provide a valuable resource for producing such measures.
Butterfly Conservation, working closely with partners, produces trends and indicators for butterflies and moths for a variety of outputs and purposes. For example, the recent State of UK Butterflies2 report involved analysis of both monitoring data (UK Butterfly Monitoring Scheme, UKBMS) and recording data (Butterflies for the New Millennium) to produce species’ trends and multi-species indicators at both UK and country-level. Similar outputs were also produced in the recent State of Britain’s Larger Moths report3, and analyses of large datasets are also involved in Red Listing4, UKBMS reporting5, UK government indicators6, as well as multi-taxa outputs, such as the State of Nature7, with an updated report due later in 2023, and for analyses beyond the UK e.g. in Europe8.
However, the production of functional outputs, such as individual species trends and multi-species indicators, from raw data, is not necessarily simple, with rapidly evolving statistical approaches and challenges presented, for example by non-standardised sampling.
There is an increasing availability and quantity of citizen science data for biodiversity research and conservation, in particularly from less structured data and casual recording. The recent paper by Johnston et al. outlines ten challenges for biodiversity monitoring using citizen science data, categorized generally as those that arise due to (a) observer behaviour, (b) data structures, (c) statistical models, and (d) communication, and each is relevant to the collection and analysis of data for butterflies and moths. Some challenges have to a certain extent already been addressed, for example, certain biases in recording scheme data are addressed by using occupancy modelling9 to describe the detection process, but other challenges are ongoing, partly due to new opportunities arising.
The increasing quantity of data contributing to existing sources (e.g. the ongoing growth of the UKBMS and BNM for UK butterflies) increases the potential to develop and utilize new analyses which can gain further insights from the data, but the increasing quantity (and variety) of new data sources provides further opportunities. For example, relatively new schemes, typically based on reduced effort or less structured methodologies, such as the Big Butterfly Count and Garden Butterfly Survey, are providing valuable new data from potentially under-sampled urban and garden habitats, in addition to citizen engagement benefits.
New data sources have the potential to be analysed in combination with other data sets, for example, transect data. The development of integrated modelling for citizen science datasets is recent but growing and typically seeks to combine the depth of smaller, structured datasets with the breadth of larger, unstructured datasets. Data integration offers the potential to maximise the use of available data (and the effort undertaken to collect it), increase spatial coverage (and reduce bias), and improve the accuracy and precision of model parameters of interest, such as trends in abundance or distribution (occupancy). However, there are also challenges and outstanding questions, for example when integrated modelling is most suitable and beneficial, and whether larger and potentially biased data sources may dominate a smaller, less biased dataset. Furthermore, computational challenges can be greater for integrated models, reducing their feasibility for use in common practice e.g. in producing species’ trends.
The ongoing growth of data presents challenges for storing and modelling large data sets and this increase in both the volume of data and often model complexity requires more and more computing power. Hence the development of efficient methods and tools for analysing our butterfly and moth data sets is vital, particularly when regular analysis is crucial for monitoring species’ change. Occupancy models are typically used to produce measures of species’ distribution change, e.g. in 'State of ' reports, but Bayesian analysis approaches can be very slow. A new “fast occupancy” approach10, led by the University of Kent, provides an efficient framework that also accounts for both spatial and temporal autocorrelation. The need for efficient methods is also relevant for modelling species’ abundance, with R packages available to account for the seasonal nature of butterfly and moth populations11,12, but producing measures of uncertainty often requires bootstrapping which can be computationally demanding and is an ongoing challenge.
Johnston et al. state that communication is critical in many elements of citizen science, which is of course true for Lepidoptera data. Butterfly Conservation, and many partners, are producing an increasing number and variety of trend and indicator outputs for butterflies and moths, with a need for consistency and clear communication among them, and a duty to provide clear presentation and interpretation to a variety of audiences.
These trend and indicator outputs will continue to evolve and hopefully grow in response to challenges and opportunities presented by the available data, as BC ultimately aims to optimize the use of available data, to get the best assessment and understanding of species’ status to assist in policy development, management and conservation for butterflies and moths.
1. Johnston, A., Matechou, E. & Dennis, E.B. (2023). Outstanding challenges and future directions for biodiversity monitoring using citizen science data. Methods in Ecology and Evolution, 14, 103-116.
2. Fox, R., Dennis, E.B., Purdy, K.M., Middlebrook, I., Roy, D.B., Noble, D.G., Botham, M.S. & Bourn, N.A.D. (2023). The State of the UK’s Butterflies 2022. Butterfly Conservation, Wareham, UK.
3. Fox, R., Dennis E.B., Harrower, C.A., Blumgart, D., Bell, J.R. et al. (2021) The State of Britain’s Larger Moths 2021. Butterfly Conservation, Rothamsted Research and UK Centre for Ecology & Hydrology, Wareham, Dorset, UK.
4. Fox, R., Dennis, E.B., Brown, A. & Curson, J. (2022). A revised Red List of British butterflies. Insect Conservation and Diversity, 15, 485-495.
7. Hayhow, D.B., Burns, F., Eaton, M.A., Al Fulaij, N., August, T.A., et al. (2016) State of Nature 2016. The State of Nature partnership.
9. Dennis, E.B., Morgan, B.J.T., Freeman, S.N., Ridout, M.S., Brereton, T.M., Fox, R., et al. (2017) Efficient occupancy model-fitting for extensive citizen-science data. PLoS ONE 12(3): e0174433. https://doi.org/10.1371/journal.pone.0174433
10. Diana, A., Dennis, E.B., Matechou, E. & Morgan, B.J.T. (2023). Fast Bayesian inference for large occupancy datasets. Biometrics DOI:10.1111/biom.13816
11. Schmucki, R., Harrower, C.A., Dennis, E.B. (2022) rbms: Computing generalised abundance indices for butterfly monitoring count data. R package version 1.1.3. https://github.com/RetoSchmucki/rbms
12. Dennis, E.B., Fagard-Jenkin, C. & Morgan, B.J.T. (2022). rGAI: An R package for fitting the generalized abundance index to seasonal count data. Ecology and Evolution 12, e9200.