Arthropod decline in grasslands and forests is associated with landscape-level drivers

Results:

In total, we sampled more than 1 million individuals and 2,700 arthropod species. Those include beetles, true bugs, cicada, grasshoppers and planthoppers as well as spiders.

Grassland:

• Biomass, abundance and species numbers per site as well as gamma diversity (i.e. the total number of species across all grasslands per year) decreased significantly over time.
• All trophic groups declined; only for predators this trend was not significant.
• Although biomass and species number decreased with increasing local land-use intensity at the sites (i.e. mowing, grazing and fertilization intensity), the direction and strength of the temporal trends were not affected by local land-use intensity.
• The decline in species numbers increased significantly with cover of arable fields in the surrounding landscape. In other words, species numbers declined more strongly at sites embedded in a landscape with many arable fields than at sites with few arable fields in the surrounding.
• Both species with low and high dispersal ability decreased in biomass, abundance and species number. However, weak dispersers declined more strongly in biomass at sites with high cover of arable fields in the surrounding landscape.

Forest:

• Biomass, species number per site and gamma diversity (i.e. the total number of species across all forests per year) decreased significantly over time. The trend for abundance was also negative but not statistically significant.
• Myceto-/detritivores, omnivores and carnivores declined, but herbivores increased over time.
• The temporal trends were not affected by local land-use intensity. However, declines in species numbers were weaker at sites with high natural or anthropogenic tree mortality, possibly due to increased local habitat heterogeneity (e.g. more deadwood, higher herb cover, higher sun exposure).
• Cover of arable fields and grasslands surrounding the forest sites had no effect on temporal trends. However, the range in land-cover variables around forest sites was low (0-30%) compared to grassland sites (0-90%) and thus the statistical power was limited.
• Only strong dispersers decreased over time, while weak disperser increased over time.

Conclusions:

• The decline in arthropods concerns not only open habitats but also forests.
• Arthropods decline not only in biomass but also in abundance, species number and gamma diversity, suggesting that populations are shrinking and species may disappear from entire regions.
• Declines concern arthropods of all trophic levels, except for herbivores in forests.
• Local land-use intensity did not explain the magnitude of the declines. There are, however, some hints that increasing habitat heterogeneity in forests is lowering the decline.
• While our results indicate that drivers of arthropod decline in grasslands are associated with arable farming, they remain unclear for forests. The result that only strong dispersers declined in forests suggests that drivers of the decline in forests also act at the landscape level.

3. Further questions

Will there be no insects left in some years?

• What we currently observe is a decline in local populations, which can finally lead to extinction of some species.

• There will always be some insect species that are less sensitive to the drivers of insect decline and which might even benefit from reduced competition.

Does it matter if insects decline? What will be the impact of insect decline?

• Insects are the most numerous group of animals: they make up 2/3 of all animal species.
• Insects are integral components of food webs and ecosystems and play important roles for ecosystem processes, such as pollination and decomposition, and ecosystem services, such as food production, pest control and soil fertility.
• The importance of each individual species is hard to assess and some species may be functionally redundant. Redundancy of species can be important for the stability of ecosystem processes because species with the same function might respond differently to environmental changes (e.g. climate change). Nevertheless, the observed decline affects the overall number and biomass of insects and a large number of species. It is very likely that this decline has immediate effects on ecosystems.

Are the observed trends and patterns reliable?

We fitted linear models to test whether there was a change in biomass, abundance and species number per site. The estimated trends and the derived percentage decline depend on the years considered, particularly the high insect numbers in 2008. There are several reasons why we are confident that our patterns represent true trends:

• For gamma diversity we did not fit linear models but we can compare years directly. This shows that there was a statistically significant decline over several years. In grasslands, most of the decline happened during the first three study years, but for forests, the declined continued over the complete time series.
• Weather conditions in 2008 were similar to weather conditions of some later years, which showed, however, lower insect numbers. So there is no reason to think 2008 was exceptional due to weather.
• There were no systematic changes in management due to the onset of the project and due to limited activities on the research plots, we do not think that scientific activities had strong effects on insect populations.
• The strongest argument why our results are valid, is that the temporal patterns in our time series and in the one of Hallmann et al. 2017 match quite well: insect biomass declined from 2008 to 2010 in both datasets; then comes an increase in 2011 which is more pronounced in Hallmann et al. ´s data but also observable in ours, particularly in forests; after 2011, insect numbers again declined in both data sets; This suggests that our time series represents part of the true long-term decline.

As the observed patterns indicate dramatic declines, we preferred not to wait for a few more years of monitoring but report the findings now to initiate action as soon as possible. But we will continue our monitoring and see whether the decline pattern is valid also for a longer period.
However, ten years is rather short for time series analyses and the overall decline in percent will change with every year of data that is added. These values should therefore not be overrated. The emphasize of the paper is therefore

• that grasslands and forests (so far neglected) are considered,
• that not only biomass, but biomass, abundance, species number and gamma diversity are evaluated,
• that dispersal and trophic guilds are compared and
• that effects of local and landscape land-use on insect trends are evaluated.

Could the decline be caused by climate change?

• Indeed, we observed that winter temperatures increased and precipitation decreased during our study.
• We also found effects of weather on insect numbers. However, these effects were the opposite for forests and grasslands: both winter temperature and precipitation during the growing period had positive effects in grasslands but negative effects in forests.
• Based on these results, both positive and negative effects on arthropod numbers may be expected from these climatic changes: In grasslands, increasing winter temperatures may be beneficial for arthropods, but lower precipitation during the growing period may be detrimental. In forests, decreasing precipitation may be beneficial, but higher winter temperature may be detrimental. We are currently not able to quantify the net effect of these climatic changes on arthropod numbers and thus, if and how much the observed trends in arthropod numbers were affected by them.
• Considering these results it is, however, unlikely that the observed declines are only due to climate change