Unlocking the potential of historical abundance datasets to study biomass change in flying insects

• Published in: Ecology and evolution 2020-08, Vol.10 (15), p.8394-8404
• Main Authors: Kinsella, Rebecca S., Thomas, Chris D., Crawford, Terry J., Hill, Jane K., Mayhew, Peter J., Macgregor, Callum J.
• Format: Article
• Language: English
• Subjects: Abundance; biodiversity decline; Biomass; Body mass; Butterflies & moths; Datasets; Empirical analysis; forewing length; Identification; Insects; Lepidoptera; Mathematical models; Model testing; moths; Original Research; predictive model; Species; Statistical methods; Statistical models; Taxonomy; Trends
• ISSN: 2045-7758; 2045-7758
• DOI: 10.1002/ece3.6546

Trends in insect abundance are well established in some datasets, but far less is known about how abundance measures translate into biomass trends. Moths (Lepidoptera) provide particularly good opportunities to study trends and drivers of biomass change at large spatial and temporal scales, given the existence of long‐term abundance datasets. However, data on the body masses of moths are required for these analyses, but such data do not currently exist. To address this data gap, we collected empirical data in 2018 on the forewing length and dry mass of field‐sampled moths, and used these to train and test a statistical model that predicts the body mass of moth species from their forewing lengths (with refined parameters for Crambidae, Erebidae, Geometridae and Noctuidae). Modeled biomass was positively correlated, with high explanatory power, with measured biomass of moth species (R2 = 0.886 ± 0.0006, across 10,000 bootstrapped replicates) and of mixed‐species samples of moths (R2 = 0.873 ± 0.0003), showing that it is possible to predict biomass to an informative level of accuracy, and prediction error was smaller with larger sample sizes. Our model allows biomass to be estimated for historical moth abundance datasets, and so our approach will create opportunities to investigate trends and drivers of insect biomass change over long timescales and broad geographic regions. Assessment of trends in insect biomass is hampered by a lack of data. To address this, we developed a method to estimate the biomass of historical samples of moths, where abundance and species identity have been recorded. Estimates made using this method explain >85% of total variation in sample biomass, allowing biomass to be predicted to an informative level of accuracy.