Recently, there has been a lot of debate about an ongoing, worrisome decline of insect diversity and biomass, both in the scientific community, but also among policy makers and the general public. Terms like “insect apocalypse”1 or “insectageddon”2 have been doing the rounds. Although previous studies had provided evidence for rapid insect declines, the topic only really made it to the global agenda after the "Krefeld study" that showed declines of over 75% of insect biomass in less than 30 years in German nature protection areas3. Subsequent studies on insect declines have led to a more ambiguous picture, some showing stable or even increasing numbers in the last decades. The reasons for the different findings could be many. For example, different insect groups could have changed in different directions, or different drivers could have been acting in different regions. To better understand the reasons for these inconsistencies, we need more research. And particularly, we need to understand the drivers of insect trends better. Climate change and land-use change have already been recognised as main drivers4–7, but their combined and particularly their interactive effects are not well understood, which is however important to sustain effective conservation measures. Also, many studies that investigate these drivers work with a space-for-time substitution approach. While this approach is very valuable to find important drivers of insect diversity or community composition8, time-series approaches may be much better suited to understand what has been driving changes in insect populations in the past.
These knowledge gaps were also recognised by the group of researchers of different Swiss research institutes (Agroscope, WSL, FiBL, Swiss Ornithological Institute) and the Swiss fauna information centre info fauna when they came together to start a new project on the state of insect diversity and biomass in Switzerland. Given the unique topography of this small country situated along the Alps, Switzerland provides an interesting case to study the different drivers of insect trends, particularly of climate and land use which vary greatly between the different regions. As the goal was to reconstruct time series of the state of insect communities, the researchers started off by screening available long-term data on insect communities. At the one hand, they managed to put together a huge dataset of pitfall-trapped insect samples collected by the different institutes. But this is another story, which will be told another time. At the other hand, they had a main expert of long-term insect data in Switzerland on board: info fauna, which has the purpose of gathering records data for all animal species of Switzerland. As such, it curates a huge database of species records, also including many insect species. And although accepting data from various sources, experts make sure that only valid records are included in the database. A real treasure for long-term data, which should proof very useful to study insect trends.
At this stage, I joined the project as a postdoc. To make use of the huge, but highly heterogeneous dataset of species records, we decided to use occupancy-detection models9,10. This statistical tool has been successfully used for similar data repeatedly in recent years11–13. To have the necessary flexibility to adapt the model to our needs, we decided to use a Bayesian implementation in Stan. This meant a lot of specific coding. And a lot of dealing with (unforeseen) technical issues. Like for example, how to get results within a reasonable time frame if fitting the model for one single species already needs more than 100 hours on average? (A cheer for HPC clusters!) Or how to handle the huge amount of data produced per model run? (More than 1 TB of data were produced by our models.) After overcoming these difficulties, we finally had 40 years of trend estimates ready for 215 butterfly species. And we were very pleased to see that the first resulting species occupancy trends aligned broadly with distribution trends estimated from standardised monitoring data for most of the study species.
I will spare the details of how we managed to get a good dataset of spatially and temporally resolved climate and regional land-use change variables. It was a journey as well. Luckily, many things have been recorded by someone somewhere, such as the number of cows registered in a year and municipality. Which allowed us to infer a set of variables for large-scale climate and land-use change across the study period, which we could then relate to the observed species trends. Which we did, and we found particularly strong relations of species trends to climate warming. This may be surprising at first, but climate warming in Switzerland was not only apparent in the recent years, but the country experienced an above-average increase of mean annual temperatures by almost 2 degrees between 1980 and 2020. And we found the signal of this warming trend in the 40-year butterfly trends. Land-use change in the same period was much more mixed, some sites experienced intensification, whereas other experiences extensification, for example due to increasing regards for ecological measures in the Swiss subsidies system. Still, short-term insect trends were related to regional land-use change. And not only to land-use change, but also to the interaction between climate change and land-use change. Which means that if both pressures occur simultaneously, insect populations may suffer more than we would expect from their single main effects. A finding that should be considered in future studies and when extrapolating climate or land-use effects to other regions or to the future.
It was somewhere here when we realised that while we had these sound analyses of butterfly trends, including other groups might make our results much more interesting and generally applicable. Thus, we went back to the info fauna database to find similarly resolved data for two other major insect groups: Grasshoppers and dragonflies. Although this meant running an extra set of models for 175 species and reworking quite a bit of the analyses, we took the extra step to include these other groups in our analyses as well. We were pleased to find that the overall results did not change when we included the two groups. However, the findings became much more nuanced. For example, while we had found that a majority of butterfly species has been declining, there were more increases than declines in the distribution of species of these other groups. For the investigated period, we thus did not find a general insect decline in terms of distribution, but a significant turnover of insect communities, strongly driven by climate warming. Declines were particularly affecting cold-adapted, specialised, and rare species. Another indication that climate warming played a crucial role in the observed changes to insect distribution. Thus, to conserve insect diversity in the future, both climate and land-use change need to be addressed in concert.
- Goulson, D. The insect apocalypse, and why it matters. Curr. Biol. 29, R967–R971 (2019).
- Monbiot, G. Insectageddon: farming is more catastrophic than climate breakdown. The Guardian (2017).
- Hallmann, C. A. et al. More than 75 percent decline over 27 years in total flying insect biomass in protected areas. PLoS ONE 12, e0185809 (2017).
- Forister, M. L. et al. Fewer butterflies seen by community scientists across the warming and drying landscapes of the American West. Science 371, 1042–1045 (2021).
- Hemberger, J., Crossley, M. S. & Gratton, C. Historical decrease in agricultural landscape diversity is associated with shifts in bumble bee species occurrence. Ecol. Lett. 24, 1800–1813 (2021).
- Seibold, S. et al. Arthropod decline in grasslands and forests is associated with landscape-level drivers. Nature 574, 671–674 (2019).
- Soroye, P., Newbold, T. & Kerr, J. Climate change contributes to widespread declines among bumble bees across continents. Science 367, 685–688 (2020).
- Blüthgen, N., Staab, M., Achury, R. & Weisser, W. W. Unravelling insect declines: can space replace time? Biol. Lett. 18, 20210666 (2022).
- MacKenzie, D. I. et al. Estimating site occupancy rates when detection probabilities are less than one. Ecology 83, 2248–2255 (2002).
- Isaac, N. J. B., Strien, A. J. van, August, T. A., Zeeuw, M. P. de & Roy, D. B. Statistics for citizen science: extracting signals of change from noisy ecological data. Methods Ecol. Evol. 5, 1052–1060 (2014).
- Outhwaite, C. L., Gregory, R. D., Chandler, R. E., Collen, B. & Isaac, N. J. B. Complex long-term biodiversity change among invertebrates, bryophytes and lichens. Nat. Ecol. Evol. 4, 384–392 (2020).
- Bowler, D. E. et al. Winners and losers over 35 years of dragonfly and damselfly distributional change in Germany. Divers. Distrib. 27, 1353–1366 (2021).
- Engelhardt, E. K. et al. Consistent signals of a warming climate in occupancy changes of three insect taxa over 40 years in central Europe. Glob. Change Biol. 28, 3998–4012 (2022).
Please sign in or register for FREE
If you are a registered user on Nature Portfolio Ecology & Evolution Community , please sign in