Ecological Armageddon is a bit dramatic. But the message from this paper published in PLOS One is important. The study shows an 82% decline in mid-summer flying insect biomass since 1989 over multiple sites in Germany. Mid-summer is usually peak insect activity, so this is weird.
But every ecological study has a context. This context is described in the Methods section – the most important but least-read section of a scientific paper. For this study, most of the media stories glossed over or overextended the context.
This study doesn’t mean the ‘global ecosystem’ is rapidly collapsing. The data were collected at 63 sites clustered in two regions in the east and west of Germany (see this map (S1 Fig) – the yellow dots are site locations). To quote the authors, these sites are “representative of Western European low-altitude nature protection areas embedded in a human-dominated landscape”. Not the whole world. Not even other kinds of agricultural landscapes in other parts of the world. Most of the media stories missed the opportunity to highlight two key messages from this: (i) we need more long-term monitoring like this from other locations; (ii) to understand these impacts, we need more information on the hundreds of thousands of unknown invertebrates in the world that we know absolutely nothing about where they live, how they benefit humans, and what they need to persist.
Biomass is not the same as abundance. Nearly all media stories mentioned abundance, either in the headline or article text. Biomass is not abundance. Biomass is the collective mass of all insect bodies caught in a trap. Biomass doesn’t tell us whether these insects are pests, pollinators or natural enemies. This study does not provide any data on abundance (the number of insect individuals) or richness (the number of different types of insects). The authors are honest about this, and also that it’s pretty hard to identify millions of insect individuals from 1503 trap samples without lots of funding to pay experienced people to do this. But the media stories didn’t really clarify that without information on abundance and richness it’s hard to make assumptions about environmental effects. It’s also misleading to link this study to the collapse of pollinators, or declines in butterflies etc., as many media stories and tweets have done.
27 years vs. 27 years. To identify a true decline in a taxonomic group at a single location, you need to collect data at that same location every year for multiple (preferably consecutive) years. In this study, over half (59%) of the trap locations were only surveyed for one year during the 27 year period. And only 26 sites were surveyed in multiple (2, 3 or 4) years – these were not all consecutive years. This is an important point to remember, because phrases in media stories like “waded through 27 years of data” or “insect abundance has fallen by 75% in the last 27 years” are slightly misleading (hang on, not to mention the study didn’t even measure abundance). Animal and plant populations naturally vary from year to year, for a variety of ecological reasons. If you only collect data at a site once, it is impossible to know whether that sampling event was a ‘normal’ year or an ‘odd one out’. It’s okay to do this, and many ecological studies do – you can still build knowledge of a site from one year, and you can build knowledge of a region over multiple sites. But without any baseline knowledge of insect populations in the study region, it’s hard to know whether insect populations at each individual site are ‘normal’ or responding to environmental factors like drought, extreme storms, the conditions of the previous winter, or resource pulses. The authors haven’t looked at these factors in this study. I think the most interesting figure is buried in the supplementary material (S4 Fig) – it shows the trends just at the 26 sites that were sampled more than once. You can clearly see that biomass measured in 2013-2016 is a lot lower than biomass measured in earlier years (1989-2006).
‘It must be bad, because this is happening in protected areas’. I couldn’t find any information in the paper about the size of the reserves, although the authors mention toward the end of the paper that reserves were of ‘limited size’. Insect populations can increase rapidly, as most species produce lots of offspring in every reproduction event and some can produce multiple generations in a season. But if essential resources for survival (food and nest sites) are limited, then more individuals will need to disperse to other habitats to survive as the population increases. Small reserves (e.g. a couple of hundred metres square) that are isolated from other reserves will be less likely to have enough habitat resources to maintain insect populations compared to larger, or more connected, reserves. We still know very little about how fragmented habitats influence most insects.
Yes, insects are in trouble. Despite the inevitable media beat-up, this study is an important scientific story. We know that overuse of agricultural chemicals, particularly pesticides, and intensification of agricultural landscapes have negative effects on non-target insects. There are so many more questions that can be asked of this dataset, I hope the researchers have the time and funds to find all these answers. I also hope the story gives much-needed attention to the fact that entomologists and insect ecologists all over the world need more support and funding to answer similar questions in other environments. The only way to identify ‘ecological Armageddon’ is to show that these patterns of decline exist across multiple scales.
© Manu Saunders 2017
Thoughtful and perceptive, as always – well done
LikeLiked by 1 person
An important corrective. Also important to note that huge changes in biomass in a malaise sample can arise from the loss of a single big bug.
LikeLiked by 2 people
Thanks, excellent point.
I am not an insect ecologist, so maybe someone can lighten me whether it is suitable to make monitoring with the same method in exactly the same over the years (as you propose the long term study should be), especially when populations are really small. I remember the following pattern from our samplings that were carried out during insect ecology fieldcourses – going to new place to for fieldcourse and carrying out sampling with similar methods in the same place over next years resulted in “catch-out” of some species (most memorable example was the massive decline of Serica brunnea over 3 years of catching). Are there any methodological constraints in this?
To identify changes in a particular population at a particular location, monitoring should occur on that same population at that location repeatedly. If you sample in a different location each time, you may be able to identify broader regional patterns, but you won’t be sure that differences in populations are actual declines or just habitat differences.
And yet the EU is about to ban seed treatment with neonics and force farmers to return to aerial spraying. Makes absolutely no sense.
I think several important points need to be made. In no way is any evidence presented to contradict the media messages. Absence of evidence, is not evidence of absence. There is a reason much of this is only anecdotal and conjecture, and this is because total insect numbers in terms of the size of populations and biomass, was generally not being measured, and this was information is largely lost to science because of this.
Unfortunately the argument put forward here in the blog also fails to address the context. It’s all very well calling for more longitudinal studies. However, this fails to address the elephant in the room not being acknowledged. Expert observers, entomologists, naturalists and other observers have noted a huge decline in insect numbers sometime between the mid-1970s and early 2000s. Yet no one knows exactly when this occurred, and at what rate. This is unknown simply because no one was measuring insect populations during this time.
Setting up longitudinal studies now will not establish when this huge and observable declines happened. Science does not have time machines available to it.
This study used a fortuitous pre-existing study by amateur entomologists, which allowed this data to be interrogated to determine a decline in insect biomass. Most data from the past does not allow this, because generally insects and other arthropod surveys were presence and absence data i.e. it says nothing about the population which existed at the time.
Science, and in particularly scientific ecology is all about accurately defining, measuring and describing the actuality. The undoubted actuality is that there have been massive declines in overall insect biomass i.e. populations of many species. This has for instance led to huge declines in birds of agricultural environments, because there is simply no longer enough biomass to feed their young.
To understand the actuality, we have to use what limited methods there are to assay what the biomass was in the past. This was a fortuitous example. The argument put forward here misleadingly gives the impression this can be tested with further studies. No it cannot. Further longitudinal studies will only establish the declines from what is almost certainly a very low baseline. It will not reveal the level of these massive declines in the preceding 4 decades.
Science can only deal with what there is hard evidence for. When this evidence was not gathered as reliable data, to grasp the actuality we have to refer to anecdote, memory etc. Whilst no ideal in scientific terms, it is the only way of accessing the actuality. Those who wrote the primary media articles such as Damian Carrington and George Monbiot are seasoned writers on the natural environment, with scientific qualifications in a relevant field. They were not just after the headlines. Concerned and expert observers have long believed that this was what was happening from observation. This was confirmation of it, and it is not confirmation bias.
Well, we can not collect new data from the past, but we can certainly make new studies in other places. We need studies from landscapes less formed by modern farming or in parts of Europe where modernisation of agriculture and changes in land use has happened recently or are in full progress. We can also study how much size of a habitat matter, as well as the density of natural habitats in a landscape dominated by farmland. What’s the difference between having 2% isolated small spots of natural land compared to (for instance) 20%? Are there a breaking line for many species, under which the loss of metapopulations make them doomed?
Yes I fully agree we need more studies. This was in my original comment before I edited it, and forgot to add it. I more than support more studies. We should be putting all the resources in place for these studies.
However, this is the big question. Why weren’t these studies being set up at least a decade ago, or in fact before then. It has been obvious to many expert observers for quite some time that there was a very serious decline in insect numbers taking place. Essentially we have several layers of declines on top of each other. First there is a general trend of decline which goes back to the 19th Century, and probably earlier. Another accelerating decline happened in the immediate post-war era, then another layer of accelerating decline trend appears to have set in with the so called green revolution. Rachel Carson’s Silent Spring woke everyone up 55 years ago.
But then another accelerating decline trend has become apparent within the last 15 years or so. This was noticeable on top of general long term trends. I noticed it as a macro, insect photographers. I noticed about 12 years ago, that it was now very difficult to find the abundance of insect pollinators willdflowers attractive to them, to make it easy to photograph them. I started commenting online about this, and other macro photographers and naturalists said they had noted it. I was at a bioblitz attended by quite a lot of the county recorders of many invertebrate groups. It was a bit disappointing in records, from what is only a lightly grazed area of an ancient hillfort. It got round to discussing this. Most of the county recorders all said they had seen these big declines themselves since the mid-1990s.
Given all of this, why weren’t these studies being set-up years ago? After all the authors of this study were trying to test out whether these reported declines were real, and to quantify them. They only had to use the limited data collected by amateur entomologists, because there were so few longitudinal studies in place, which allowed them to interrogate this data. Meta-studies are very difficult simply because of the paucity of baseline data. But why? It is 55 years since the publication of Silent Spring. When I studied ecology as a mature undergrad 25 years ago, this was the big theme in population ecology modules at the university I studied at.
Overall my main point was nitpicking this study as if it was not set up well is missing the elephant standing in the room. Often science has to work with very little. Science and scientific ecology is about establishing the actuality. Given that no one was collecting hard data about declines in insect population numbers and the decrease in insect biomass as a whole, scientists need to use less than ideal information to establish what is happening.
I will be very overt about this. The reason these studies and the impacts of industrial farming, and especially agro-chemicals such as insecticides was not studied was because of vested interest pressure from the farming industry and the agro-chemical companies. They didn’t want there to be a record of what was happening.
LikeLiked by 2 people
Possibly we can use old data sets from other locations, and update them by re-surveying as well. Anecdote and memory and so on are valid as a last resort I think, but we should also look for opportunities to use more reliable data if and where it exists!
All important points, and in addition they did not measure biomass; they measured the total of mass of insects wetted with an unknown mass of alcohol. They did not even take the trouble to show how wetted mass correlated with dry mass, and at least some of the weighing was done on balances with too low a resolution for the masses they were weighing.
LikeLiked by 1 person
There are good long-term datasets that show evidence of decline, but as far as I am aware none are really suited to analysing biomass. One might do an extrapolation from the Rothampstead moth trap data but it would be a guestimate, albeit using constant metrics. Analysis of opportunistic data certainly also points to good evidence of declines, but again at a broad scale and not based on biomass.
The German study has its weaknesses, but its real strength lies in the scale of change, which seem to me to be a strong indicator of overall change, even though one can pick holes in aspects of the analysis and the headline message, which is potentially more sensational than might be found in the underlying data.