Thursday, 2 January 2020

Changing recorder behaviours

When I started extracting data from the internet, almost all records were of single species from a locality on the same day. The value of the internet as a medium for recording was only starting to become apparent and it did not really gain much traction until the advent of iSpot in 2009. That was quite a 'game changer'. The next big leap came in 2013 when the UK Hoverflies Facebook group was established. The effects of these two events can be seen in figure 1.
Figure 1. Numbers of full and partial records of hoverflies from 2002 to 2019 extracted from internet sources. Numbers started to rise markedly in 2010 with a much bigger jump in 2014 which was the first full year after the establishment of the UK Hoverflies Facebook group. I only really started recording partial identifications in 2015; hence the low data for preceding years.

The numbers of active recorders is probably a better test of the popularity of social media amongst the photographic recording community. This is illustrated in Figure 2.
Figure 2. Numbers of recorders contributing to the data presented in Figure 1. The decline since 2013 reflects several factors; principally a reduction in the time I spend extracting data from Flickr and other sites, together with a gradual shift of recorders using spreadsheets or iRecord to submit data. The big drop in 2018 is almost certainly a combination of people shifting to the use of spreadsheets plus a big hit from the heatwave of June and July in which hoverfly numbers crashed.

One of the big problems for recording schemes is that recorders can be extremely selective over what they submit. Many would only take an interest in the rare or unusual; a situation that probably continues in many other schemes. The HRS has actively encouraged the submission of full lists on all occasions, and I think this process has been very effective as can be seen in Figure 3. This shows the benefits of schemes directly interacting with recorders in order to develop new ways of working. In many ways, it mirrors BirdTrack which only uses full lists in its interactive feedback.
Figure 3. The average numbers of records per recorder since 2009. This graph shows how there was a gradual rise in numbers until 2013 but in the following two years there was a paradigm shift in recorder behaviour. Part of this shift would have come from the reduction in my scanning of Flickr for occasional records, but even so, it does not wholly explain the results. It is noticeable how the average plummeted in 2018 but has rebounded very strongly in 2019.
I think this is a good indication of the way forward for other schemes and for other recording media. The important issue is to shift thinking away from the idea that recording is about dot maps and towards the idea that total lists may help to provide the detail deened to track at least some of the critical environmental influences on Britain's wildlife.

Saturday, 13 July 2019

Picking up signals from the data. Some examples

I have spent a sizeable part of the day trying to make something of the data that are emerging this year. My feeling is that there are probably quite a few species that have suffered but there will be a few that have benefitted. The same species may have lost out in one part of the country and gained elsewhere. I have started by looking at five examples:
  • Episyrphus balteatus, which seems to be having a good year and was far more abundant than expected in the winter.
  • Eristalis pertinax, which has appeared to be poorly represented in recent data.
  • Eristalis tenax, which appears to have been more dominant in recent data.
  • Helophilus pendulus, which has generated regular comments about its apparent scarcity.
  • Rhingia campestris, which has been notable by its absence even though it is a highly distinctive species.
To do this, I have compared the proportions of records rather than the absolute numbers of records. This approach is necessary because the numbers of records and recorders vary from year to year; as does the geographic spread of recording. So, context remains an important part of the analysis. Each species is compared against the data for the previous six years (2013 to 2018) extracted from Facebook and other social media.

Episyrphus balteatus

This is an aphidophagous species that occurs throughout the year but in particular dominates the records at certain times. In mid-summer, it vastly outnumbers all other hoverflies but peaks at differing times.
Figure 1. Episyrphus balteatus
As the data has grown, it has been clear that E. balteatus has been far more dominant than in other years. The precise reasons are unclear, but it seems likely that the warm winter allowed dults to be far more active and for the larvae to get started earlier in the year.

Eristalis pertinax

The larvae of Eristalis (and Helophilus) are aquatic. E. pertinax usually has two peaks, one in the spring and the other in late-summer. Numbers should start to rise in July but as yet that has not been detected.Late autumn abundance is very variable, depending upon both temperature and sunshine. It usually starts to disappear in late October but can be abundant into December during warm autumns.

Figure 2. Eristalis pertinax
Current evidence suggests that tthe spring generationwas smaller than the long-term average. It is probably too early to be sure that the autumn generation will be smaller than normal, but a recent dip in numbers may be indicative of this species having experienced a poor year.

Eristalis tenax

Adults usually hibernate, but in recent years this has been considerably less clear from the data. Winter 2018-19 was relatively warm but the numbers of E. tenax recorded seem to have been lower than the longer-term trend.
Figure 3. Eristalis tenax
Despite comparatively low numbers of winter records, it seems that E. tenax is experiencing a good year, with numbers noticeably higher than the longer-term trend.

Helophilus pendulus

Monitoring the Facebook group has pointed to this species having not been as numerous in previous years. The phenology for 2019 is somewhat odd, with a much shorter spring generation.

Figure 4. Helophilus pendulus
I wondered whether there might be a regional reason for this difference so split the data into five regions as shown.
Figure 5. Helophilus pendulus phenology in England & Wales
Figure 6. Helophilus pendulus in Scotland
The data present a very confusing picture that is difficult to interpret. If anything, numbers in Scotland appear to be marginally up on the longer-term trend, and South-East England seems t have somewhat elevated numbers but a shorter emergence season. Elsewhere, numbers seem to be lower than recent trends.

Rhingia campestris

This is a highly distinctive species whose larvae develop mainly in cow dung, although its distribution suggests that a wider range of dung is probably utilised. It is a species that has, historically, been considered to respond to hot summers and for some generations to almost completely fail. This was detected in 2018 and the response in 2019 is therefore of interest.

What emerges is a response that seems to be completely different in Scotland. Whilst there seems to have been a uniform crash in England and Wales, numbers in Scotland appear to be robust and possibly higher than in recent years. I suspect that the data for northern England may ultimately show a similar pattern to Scotland once records from various very active recorders are absorbed into the dataset (at the end of the year). Those from more southerly regions seem to suggest that numbers are exceptionally low; which is consistent with the failure of the summer generation in 2018 and thus a failure of breeding success to create the spring generation.

Figure 7. Rhingia campestris in England and Wales
Figure 8. Rhingia campestris in Scotland
I suspect that the elevated numbers in Scotland can be explained by the higher temperatures in 2018, which allowed greater breeding success in a region where higher temperatures would have allowed more rapid population build-up in late summer and consequently more larvae available to create the spring generation.


Wednesday, 10 July 2019

Getting started as a serious Dipterist

There was a recent post on the UK Hoverflies Facebook Group concerning 'starter kits' for new recruits. The query did not specifically seek funding but a response showed how some potential young Dipterists might be constrained by funds.

My days might be numbered, so I need to think about how best to do something positive, and one way of doing so is to help the next generation. I am therefore thinking about how best I might set up a fund to help young Dipterists. What do they need?

I have always believed that people should help themselves but I know that the costs of getting started are steep. So, if I was to set up a fund through one of the major societies what should I be thinking about? No promises!

What should a starter pack comprise? Who should qualify? How should such a fund be administered? What should its objectives be? There are numerous questions to answer before I do anything formal but in the meantime I need feedback to understand what would work?

Thoughts please? (sensible ones)

Thursday, 20 June 2019

Common species matter

I return to that perennial chestnut – whether there is any value in records of common species? New recruits to recording schemes often wonder whether their very limited repertoire has any value, whilst more experienced recorders are perhaps less excited by common species and therefore ignore them. I think it is essential that scheme organisers continue to emphasise the benefits of recording commoner and more easily identified species.

I am reminded of the fate of the House Sparrow. Ringers tell me that when they started they were actively discouraged from ringing house sparrows. So, when these cheeky little chaps started to decline nobody had any data! Even now, I’ll bet the data are incomplete but at least these days BirdTrack asks for full lists and the way it is designed makes you feel obliged to try to compile a full list. So, an opportunity was missed but future issues are now addressed by a hugely successful data-capture system.

With invertebrates we have an even bigger problem. There have been several stages in the decline of our wonderful insect populations. Firstly we had the development of more intensive agriculture post WW2. By the 1950s farmers were being actively encouraged to drain and plough and ‘improve’ previously fallow land. We lost vast areas of grassland and wet pasture. No more orchid fields, just a green monoculture of grass and cereals. That must have had a devastating effect on invertebrates but we don’t have any data to consider it.

Then came the pesticide revolution. DDT and a wider range of Organo-phosphates laid waste to our insect populations in the 1960s and 1970s. I can still remember a field in France in 1970 that had been treated – it was awash with the corpses of so many lovely animals. The shock has stayed with me for the following 50 years (I was 11 or 12 at the time). So, by the mid-1970s a lot of damage had been done, but still we had no decent recording system.

Then came the onset of recording schemes. The HRS started in 1976 – just in time for the first devastating heatwave and drought. Again, we have precious little data! There is a theme developing here. So, today, we have had a sequence of extreme heatwaves and droughts but can we analyse their effects? Probably not. The data are just too patchy. Yet, if every field naturalist had simply noted all the ‘common’ and readily identifiable species for the past 50 years we would have a fighting chance of picking up some signals.

So, the issue of insect decline is not going to go away! Indeed this spring we have witnessed some pretty worrying shortages of all manner of insects. It is time we addressed the data shortfall and valued records of ‘common’ species. After all, they could become the modern equivalent of the passenger pigeon!

So, if you have just joined the recording process, don't worry if your records are of a narrow suite of species. The data are useful. Similarly, if you are experienced please submit full lists. They really do count.

Friday, 24 May 2019

If only .......

This spring is starting to show signs of stress in the insect world. The numbers of hoverflies I am recording have dropped alarmingly and there are some genera that are seemingly absent on my local patch (Mitcham Common). Where are the Cheilosia and Pipiza? Platycheirus and Melanostoma are extremely thin on the ground, as are Eristalines (although I did get a very odd one yesterday that I cannot identify). My instincts are that these are the impact of last year's heatwave, but it may not be possible to make the causal link.

What is needed is a decent block of data on a range of relatively widespread and abundant species across a range of taxa. I started casting my mind around and hit upon the spectacular red and black hopper Cercopis vulnerata. Surely I had recorded this in sufficient detail? It seems not! I know it was recorded from Mitcham Common in 1984 at least in two locations (by John Hollier). I don't seem to have logged it. Has it disappeared? If so, it joins Leucozona lucorum as an unexplained loss.

I don't see how these two widespread and often abundant species can be the victims of anything other than climate change. The Common has not changed so much as a result of adjacent land management but might of course have been affected by atmospheric pollution.

Unfortunately, I have rarely been around sufficiently consistently to record every year in May, but have had the opprtunity these last two years. It is now a sigh of 'if only'. So, the mantra for all field naturalists needs to be log everything all the time. That is not to say I am proposing 'pan-listing'. I am not. But, most of us can recognise some taxa beyond the narrow confines of our specialism; some of which may just be useful in developing our understanding of the impacts of climate change!

Friday, 26 October 2018

Who wants records and how do we submit them?

A thread on the UK Hoverflies Facebook group yesterday raised a host of questions:


Whether there is any need to submit records of the same species on different days

It is worth remembering that we cannot go back and create data once the day has past, so making as full a record of what we see is a way of preserving a point in time that may be useful in the future. If you see something and can be sure of its identity, then log it and submit it! Modern databasing systems cope with vast quantities of data so there is no problem about volume.

The bigger problem comes when people selectively submit data. Are all species treated equally, or is there bias towards the more ‘interesting’ species? If there is bias, then the commoner species appear to be rarer and the rarer species commoner. The most telling lesson comes from the BTO’s ringing programme, which actively discouraged ringing of house sparrow until it was too late and the decline (crash) had happened. Commoner species are often the bellwether of the natural world because they tell us a lot about the wider landscape. Rare species are rare because their ecological requirements are more precise, and these are much more scattered (but may have been widespread in the past).

The HRS encourages daily recording as this helps to paint a picture of the seasons and years. Sadly, we have lots of evidence of sensitivities to changing weather patterns, but older data are so limited that we cannot go back and test ideas against previous major events such as the 1976 drought. BUT, if we make sure we have the data now, our successors will be able to do so much more.

The problem comes because there is a commonly held misconception that recording schemes are simply about mapping distribution. That was the case when schemes started in the 1960s and 1970s but once modern computing because available and powerful a wide range of new opportunities emerged. I’ve written previously on the use of opportunistic data and its use in occupancy modelling. Most, if not all, recent reports on the status of British wildlife are based on this modelling (and HRS data are part of this).

The more complete the data the better. It is worth recording the gender of the animal, what it was doing and whether it was associated with a particular plant. BUT, there are complications – some insects will sit on flowers but not be taking pollen or nectar – so be sure that it is actively foraging and say, ‘at flower of …’. A note saying ‘on lupin’ is meaningless – it might be sunning on lupin leaves or feeding at the flowers; we get lots of data of this sort and, sadly, it is impossible to use.
As a simple illustration of what we can do with data on gender, Figure 1 illustrates the differences in phenology of male and female Eristalis tenax – without full data extracted from records from social media we would have a far less complete picture because most people submit records just naming the animal with no associated information.

So, the best mantra is ‘see it, log it, submit it
Figure 1. Phenology of male and female Eristalis tenax in 2018. Without regular and detailed recording of a seemingly common species we would not have such a complete picture of its over-wintering ecology.


What is the preferred route for records to reach schemes?

There are numerous ways of submitting data to schemes: spreadsheet, Mapmate synch, downloads from other databases (e.g. Recorder), iRecord and posting on Facebook (where schemes extract directly).

Scheme organisers are not a uniform bunch, and each has their own preferences. Some schemes definitely prefer iRecord; others do not! Again, I have written about the issues we have with iRecord, so take a look! Speaking strictly for the HRS, we prefer spreadsheets or databases downloads/synchs for big blocks of data. I do all of the verification for hoverflies on iRecord and it is probably the most frustrating job of the year! So often people submit a record based on a photo where you can tell what the gender is, but it is not logged. If I go in and adjust every record then there will be months of work and I simply don’t have the time, so the information is lost. I also get immensely frustrated when I see the same photograph used to support a series of records from different dates – can I really believe the records? Equally, I get frustrated when several posts of the same species for the same date and place – why not one post and save both me and you time?

I am far more sanguine about extracting data direct from the Facebook group. Until very recently I have done this but thanks to a wonderful team of volunteers we now have a group of data extractors (a HUGE thanks to David Rayner and the team – Chickena Lurve, Sue Kitt, Katie Stanney and Adam Kelsey). This data has been an important stage in developing the data because there is much more consistency in the validation of records; I now use this independent of the main dataset to look at some of the ongoing responses to the environment – it provides real-time data to work with.


What happens to the data?

We must always remember that scheme organisers are simply the custodians of the scheme and the data that have been assembled. Schemes should (hopefully) pass between generations and provide the foundations for our successors. Each scheme operates independently, and some are far more active than others. There are a few schemes that are essentially moribund because the organiser has ceased to be actively engaged but cannot be persuaded to pass on the data and the responsibility. Most are active to some degree but may not engage with media such as iRecord. Some schemes (e.g. moths, dragonflies, plants) have a network of county records, but many are effectively ‘one-man-bands’. The numbers of active recorders for most taxonomic groups are very limited and there are fewer still people who are prepared to take on administrative roles.

Where schemes are active and engaged, there should be data transfer to the NBN on a regular basis. Here, I must put my hands up and say that the most recent HRS data are not on the NBN because we are trying to clean up the dataset and mark-up the dodgy records (it is a monumental task). We do, however, pass on data to research groups as and when we are asked. So, when you see reports based on wildlife statistics you can be sure that HRS data are part of the mix. I cannot remember how many requests we have had this year – maybe half a dozen.

In the case of the HRS, incoming data are loaded into RECORDER as the storage platform – Stuart Ball is the data manager, whilst I am simply the front end who deals with the wider administration. The data are used to inform a range of products, most notably the maps in the WILDGuide and the analysis behind the 2014 Species Status Review. We have a problem with the original HRS website (Stuart managed to break the mapping system and has never found a way of repairing it). Stuart is developing a new site, but it is not on active release (links are often provided to the FB group) – it is one of the multitude of jobs that are reliant upon a small nucleus of people who run the schemes.

In theory, Stuart and I should be writing a series of research papers based on Stuart’s occupancy modelling. We have a lot to say but seem to be stuck in the drafting stage! Meanwhile, I do try (intermittently) to provide a bit of feedback on this blog – scroll through the posts and maybe something will be of interest.