Insect collection - a philosophical question

Comments on a recent post in the UK Hoverflies Facebook page raise some important questions about the practice of retaining specimens. I hope the following will help to set some perspective.

The dataset held by the HRS comprises a mixture of field observations and preserved specimens. There would be very few records prior to 1976 without the collections of long-dead entomologists. Indeed, this is true for most datasets involving invertebrates. The one problem with museum collections is that they tend to be dominated (proportionally) by 'scarcer' species and by species that are tricky to identify. This creates an unhelpful imbalance that means that pre-1976 data are not ideal in presenting an image of what the hoverfly fauna looked like in the past. The one thing we do know is that certain species were regarded as 'prizes' (e.g. Volucella zonaria). Some data from this period are definitely helpful in reconstructing our knowledge of particular species. Anybody wanting to see this in action should look at two papers Stuart and I published on Volucella inanis and V. zonaria.

http://www.bacoastal.co.uk/Entomology/2004-Volucella-zonaria.pdf

http://www.bacoastal.co.uk/Entomology/2003-Volucella-inanis.pdf

Many of the species that were most prized were big and bold (e.g. Callicera aurata). It is now possible to read the accounts of field trips by one of the early Dipterists and then actually hold the specimen the describe taking - which is an amazing link back in history.

Evolution of the dataset

Fast speed forward to 1976. Between 1976 and around 2009 the data held by the HRS was primarily supplied by people who at least partially retained specimens. Over 50% of the data were supplied by around 20 people. Those people generally record all taxa - in other words they tackle difficult genera such as Cheilosia and Pipiza. Having voucher specimens is essential if there are questions about a species' identity or if there is a split. Even experienced recorders make mistakes and being able to check is helpful. Only yesterday I spent the afternoon checking the IDs of specimens forwarded for verification. Not all were right, even from very experienced recorders. So, we must assume there are dodgy records in the dataset (some mine I expect).

The degree to which records are dodgy is often difficult to tell, but I have four species in mind where I think the maps definitely include significant misinformation: Lejogaster tarsata, Platycheirus immarginatus, P. nielseni and possibly Eristalinus aeneus. In the case of the Platycheirus I think we will have to restrict maps to cases where males have been recorded - females are too readily confused with others. This means that we will only get a realistic picture from specimens.

That leaves another 50% of data. What I have found is that there is a strong separation between the records of the nucleus who retain specimens and the remainder of the data, with far fewer records of tricky species/genera, and far more of species that can to a greater or lesser degree be identified from photographs or in the wild. Of course, we do get small blocks of data from people who work all taxa but do relatively little recording, so the data are not simply dominated by recorders who don't tackle tricky species. These data exclude lists where I take one look and say 'Chinnery' - there are ways of telling how recorders behave and the sources they use. They also exclude lists where it seems to me that the author has not used the key or bothered to check on a species' distribution - the presence of Scottish or coastal species in a list for a dry southern grassland is a fair indication!

So, moving on to the era where the mantra is: take nothing but photos and leave nothing but footprints. I am just finalising an analysis of species covered by photographs over 2014. That highlights just over 130 species (I have excluded a few photos of rare animals that were posted because I know they will arrive via lists supplied by the author). I also have an analysis of what has not been identified - no great surprises - dominated by Cheilosia, Pipiza, Platycheirus, but also including lots of Syrphus and Eristalis. So, bearing in mind this dataset comprises over 10,000 records (all records extracted from internet sources in 2014), there is a new influence on the dataset that will have to be taken into account when data are analysed for trends.

Based on past analysis of photographic records, I reckon that maybe 150 species will be recorded in this way, but that no more than 10 species will dominate the data, and no more than 35-40 will occur in sufficient numbers to perform any year-on-year analysis. Whether numbers will be sufficiently big to undertake latitudinal analysis remains to be seen, but I suspect the list will be depleted further.

Should specimens be retained?

We therefore hit the point where one must ask the question 'is it necessary to retain specimens?' If one wants to understand what is going on in the natural environment there is a need for quantitative data. In broad terms, that means trapping (and killing). But, trapping is fairly indiscriminate and does not necessarily cover all taxa. Also, trapping can kill large numbers of 'rare' species before this is detected. In the case of poorly known taxa this may never be detected. So, although trapping has its place, it has many drawbacks. If trapping is used to sample a narrow spectrum of animals then what happens to the by-catch? Bearing in mind this can be voluminous, the chances of it being tackled by a specialist are small. BUT, the data are quantitative and can probably be repeated at suitable intervals, thus forming a long time-series. One of the best examples of this is the Rothampstead moth survey. Jenny Owen's malaise trap also produced a really valuable time-series from her Leicester garden. These are rare, but valuable resources, and anything Government funded is under threat, so may not be available in future.

The Rothampstead programme has already been cut. Its loss would be catastrophic because it has highlighted a terrible decline in moth populations. Those changes are not down to retaining specimens - they are a mixture of climatic factors and anthropogenic changes to the countryside. Similar declines can be detect for many other taxa, including hoverflies. About 40% of our hoverfly fauna is in decline. We know this from the HRS dataset, which is supplied by a mixture of lethal and non-lethal sources, collected largely in an ad-hoc manner. These data are not ideal, but they are the best we have.

So, we now return to the question of retaining specimens in this modern World. A relatively small number of people retain specimens and many of them contribute extensive lists to the HRS. My guess is that around 150 people across Britain retain any level of specimens, and most retain very little. The contribution they make to the dataset is considerable, however. Perhaps not every specimen needs to be retained, but overall the impact of specimen retention is negligible. Conversely, the impact of the contribution to the dataset is considerable. Like it or not, we actually need people who are prepared to retain specimens and are driven by this interest. If we rely on non-lethal techniques then we will see a diminution in the value of the dataset.

It is therefore important to put specimen retention into context. Anybody who has watched social wasps hunting will know that they are voracious killers. A single nest may kill many tens of thousands of insects. Stuart and I did a bit of work in a small wood in Northamptonshire a few years ago that showed there was a strong probability of about 1 wasp's nest for every 1.5 ha of woodland. (i.e. around 40 nests in 60ha). That mounts up to a huge volume of invertebrate biomass. So social wasps are clearly important regulators of invertebrates. They happily munch hoverflies and if they have the same abilities of Anthrenus (museum beetle) my bet is that they take rare species disproportionately. So, our active Dipterist, working over the year, probably equates in impact to little more than a few workers within a social wasp nest, spread over a much larger area.

Ah, I hear you say, but if the 'collector' did not collect, then there would be more insects. TRUE! FRACTIONALLY! But also there would be CONSIDERABLY less data.

A further moral dilemma

So let us move on to the take nothing but footprints argument. No hoverflies were killed in the making of this day out? Well, maybe. But if you drive to your venue that may not be so. We know that huge volumes of insects are killed in collision with cars. So, those who choose not to take specimens but enjoy watching wildlife in special places are also taking an informed decision to kill. It is just that they do so indiscriminately and with little knowledge of what they are doing.

I have had this debate with several proponents of this philosophy and am amazed by the answers they give - very largely about the fact that this improves their quality of life and the death of the insects they kill on the road is not intentional. How can this be? After all, they know that cars kill insects and will be irritated by the splat marks on the windscreen when they impair visibility. So, I fear, anybody who uses powered transport to conduct their hobby is having a negative impact on insects. Who knows if their victims are common or rare? The result is the same, but there is a difference between this approach and that of the entomologist who retains specimens. One generates partial data and the other more comprehensive data. Maybe this is a broad generalisation, but I think the key philosophical question relates to the contribution the deaths of insects make to our knowledge of insect ecology.

I therefore hold the view that the immoral issue is not about whether or not specimens are retained; it is about the degree to which we pursue our interests without making a contribution back to the wildlife we love. Detailed recording at whatever level can make a difference in highlighting the plight of wildlife. At least with robust data there is something to influence politicians and decision-makers. The biggest threat to wildlife is arguably the loss of taxonomically competent recorders whose contributions are essential if we are to track the trajectory of wildlife abundance.

Phenology of Episyrphus balteatus in 2014 and 2013

In a few moments of inactivity I took a look at the phenology of a few species seen this year. One that I thought worth looking at was Episyrphus balteatus as it is reported in very large numbers. The attached histograms are strictly for photographic records - I will look in more detail once the garden records have arrived.

Data were plotted in four blocks:

•  All records for 2014 and 2013 plotted separately
• South of a line created by the northern limits of ST/SU/TQ plotted as separate years 2014 and 2013.
• Midlands - up from the northern limits of ST/SU/TQ to the southern limits of SE/SD plotted as separate years 2014 and 2013.
• North - north of the limits of the southern limits of SE-SD plotted as separate years 2014 and 2013.

The results were a bit of a surprise:

1. It looks as though there might be as many as four generations in southern England - I expected 3 but there is a bit of a March generation too.
2. In 2014 I found no migration peak at the end of July but there is quite a spike by late June - and a pronounced spike in the Midlands.
3. The number of generations in the midlands is a bit messy - probably 3, but I cannot be sure. It begs the question - can one come up with a series of thermoclines to come up with differences in phenology?
   
4. The northern population seems to have a distinct single generation.

What I think it starts to tell us is that every year is very different, and that this species probably has slightly different emergence strategies even between southern England and the Midlands. In 2014, in which we had an exceptionally mild spring and autumn it looks as though there were probably at least 4 generations in southern England, whereas the further north one gets it seems to be broadly univoltine.What I find particularly puzzling is that the late July migration spike was not picked up this year. I have a feeling that this may only be detected by absolute counts.

Figure 1. Total 2014 records of Episyrphus balteatus against week number.

Figure 2. 2014 records of Episyrphus balteatus from southern England.

Figure 3. 2014 records of Episyrphus balteatus from the English midlands and Wales.

Figure 4. 2014 records of Episyphus balteatus from northern England and Scotland.

Figure 5. Total 2013 records of Episyrphus balteatus against week number

Figure 6. 2013 Episyrphus balteatus records from southern England

Figure 7. 2013 Episyrphus balteatus records from the English Midlands and Wales.

Figure 8. 2013 Episyrphus balteatus records from northern England and Scotland

Reflections on 2014 - a good year for hoverfly recording?

As autumn draws in and the numbers of hoverflies seen rapidly declines, it is perhaps worth reflecting on what has happened in 2014. Obviously only a proportion of data are available for analysis, but I think the photographic dataset is pretty robust, so I have taken a quick look to see what story it conveys.

If one looks at absolute numbers of records, it looks as though hoverfly recording has gained speed and the numbers of records have grown exponentially (Figure 1). Part of this is undoubtedly the establishment of the UK Hoverflies Facebook group, which has provided a very active forum to promote recording and where occasional casual records can be submitted. There will be big blocks of data from others coming in during the Autumn and Winter, so we will not really know for some time how well recorded 2014 will be. Even so, the data are encouraging. There are lots more casual recorders and a fair number of people who now regularly record hovers. Two photographers have each contributed in excess of 450 records each and there are seven others who have contributed between 100 and 250 records.  Big blocks of data are a great boost to the scheme, especially when they paint a picture for a single site.

Figure 1. Growth in the photographic dataset between 2011 and 2014.

 


The second aspect of photographic recording in 2014 relates to the abundance of hoverflies represented. We saw lots of activity throughout the winter and a really productive April, followed by a cold wet May. The graph of photographic records for 2014 (as a percentage of all records) shows how hoverfly abundance came to a shuddering halt! (Figure 2) If compared against a similar graph for the period 2002 to 2014 covering all of the photographic records I hold (Figure 3), it is clear just how different May was to the general pattern What is less clear is the degree to which this had a knock-on effect into June and July. Membership of the Facebook group grew very rapidly at this time and a growth in recorder numbers may have masked some of these effects.

Figure 2. Monthly totals of photographic records extracted for 2014.

Figure 3. Monthly totals for photographic records covering the years 2002 to 2014.

The graph for 2014 illustrates one of the developing problems for many invertebrates - they are definitely emerging earlier in the year (see earlier posts), but just when there is a need for warm sunny weather there is a cold wet spell that affects both adults and developing larvae. This may be one reason why some species appear not to be doing well. Obviously, this is just a single year's data and there are limitations because the membership of the UK Hoverflies Facebook Group group has been growing throughout the summer. Inevitably, the relative numbers of records may be different, but the overall picture would probably be more stark if this was to be taken into account.

I think it is a great endorsement of what is possible when an interactive medium such as Facebook is used to encourage biological recording, but there may be issues to deal with in due course. Readers will know from previous posts that relatively few representatives of some genera are photographed, and that there are some genera that really cannot be identified from photos. This means that there will be an inevitable need to look at analytical techniques to make sure that interpretations of trends are not influenced by the recording technique. Stuart has started to look at this and has concluded that there is a skew developing (as I have previously predicted).

Data skews are not necessarily a problem provided they are recogised and taken into account when undertaking analysis. The big question now is how to make sure that some forms of recording are promoted to generate comprehensive data sets. Stuart Ball has looked at the data that he, me and a couple of other very active recorders were generating and clearly there is a difference in what occurs. My own analysis (unpublished) showed that there was just over a 50% overlap between the 30 commonest represented species recorded using photography and those recorded by my own sampling - which comprehensively covers all species encountered. Data held by the HRS lies somewhere midway, which indicates that there is undoubtedly a skewing effect. BUT, does this matter? - Maybe not.

My instincts are that we will have to split data for certain analytical processes and that the next target for me to pay attention to is developing a set of replacements for the people such as Stuart, me, Alan Stubbs etc. I guess we need at least 20 such recorders to form the nucleus for the next 30 years (and must hopefully encourage a few to take on the role of growing a further generation).

What I think is greatly encouraging is that there is so much more interest in hoverflies and that so many new people are making active contributions that really are useful. The Hoverfly Recording Scheme is about so much more than mapping. Provided we recruit a broad constituency of new recorders we will hopefully be able to continue to keep hoverflies high on the agenda of research into the effects of landscape and climate change on Britain's wonderful wildlife.