Does growth in the volume of records equate to growth in the recorder skill-base?

I did a very simple analysis this afternoon, looking at the numbers of records generated from social media in the period 2015 to 2018. Most of those records come from the UK Hoverflies Facebook page which is the main mentoring tool for the Hoverfly Recording Scheme. The results are worth looking at because they tell an important story.

It might be assumed that a large increase in the numbers of records reaching recording schemes is a direct reflection of the growth in skills amongst recorders. At least, we would like to think so.  The headline message is that 2,228 people participated between 2015 and 2018 (Figure 1) via social media in that period and contributed 83,309 records. It is very encouraging to say the least.

Figure 1. Recruitment of contributors to HRS data derived from social media between 2015 and 2018.

The message is a little more sobering, however, when one looks more rigorously at the data. Of those 2,229 recorders, 619 have contributed 10 or more records and 227 have contributed 50 or more records (Figure 2). There is no point in trying to take this analysis further because at least 35 of the 133 really active recorders (who have contributed in excess of 100 records) have moved over to running their own spreadsheets. That is a result in its own right because, between them, this nucleus contributed 26,616 records, or approximately 30% of the data.

Figure 2. Recruitment of active recorders between 2015 and 2018.

There are two elements to recruitment:

The first is, of course, assembly of a bigger and more riobust dataset, which is really needed if we are to understand what is happening to hoverfly numbers. Although this sort of recording tends to concentrate on the commoner and more readily recorded species, big blocks of data will help to idenbtify the bellwethers of change.

I think the more important issue is probably the generation of a new cohort of specialists who can take on the mentoring role that the current scheme organisers fulfil. In this respect, we are doing very well with five or six very active members now regularly assisting newcomers with identications. Getting members to this point is really important and it is encouraging to see how they adopt suitably cautious approaches that demonstrate that they have learned and understand the limitations to what can and cannot be done from photographs. Thus, I think the results to date are extremely positive, especially as we will doubtless get a few new recriuits who will join that experienced pool each year.

Effects of the heatwave

Sometimes one gets a gut feeling that something is happening but it is difficult to work out why. That is the case for Volucella pellucens, which has been remarkably scarce this summer. I've noticed this in my own field work and in the posts on the Facebook group. It is a great subject because it is an animal that is noticed by novices and experienced recorders alike, and is probably over-represented in the data because it is big and relatively easy to photograph. So, when there seems to be a drop in numbers it is readily noticeable. The overall phenology (Figure 1) for 2018 compared against the past 3 years seems to tell a story but is it the heatwave or something else that is causing a problem?

Figure 1. All social media records for Volucella pellucens in 2018 compared against the three-year average of records extracted from social media.

What we see is the first peak being strongly attenuated before the second peak commences. We can be reasonably sure that the population was behaving quite normally until early June when the heatwave started to take a hold. BUT, thereafter numbers have been very low.. The second peak seems to be coincident with the normal peak, but again at much lower numbers. What is going on?

I think the answer lies in the emergence times of males and females. In most Volucella, males start to emerge a week to ten days before females and there is a very definite male peak followed by a female peak that is delayed by a couple of weeks. This is likely to be linked to delayed female emergence which is fairly common in insects. So, how does the phenology of males and females in 2018 differ from the three-year average? this we can see in Figures 2 and 3.

Figure 2. Phenology of male Volucella pellucens in 2018 compared to the three-year average 2015 to 2017.

Figure 3. Phenology of female Voucella pellucens in 2018 compared with the 3-year average from 2015 to 2017

These results are quite startling. We can see that emergence in 2018 was delayed by about a week and is likely to be a response to the longer winter. Once emergence had commenced it followed a normal trajectory for the first month in the males but females seem to have started to respond after just a couple of weeks.

What is interesting is that there appear to be double peaks in the phenology of both sexes. Prior to this year I would probably have thought that the second peak was a second brood and that larval development was quite rapid. I rather doubt that is the case now and suspect that there is always a partial delay in emergence from north to south but I don't think we really have enough records to test this theory.

What does seem to be clear is that the male population has been substantially attenuated by the heatwave, whilst there may have been some bounce-back amongst females. We know from our own mark-release-recapture work that some individuals survive for as much as six weeks, but that the vast bulk of the population only survives for about 3 weeks. So, there could be doubt about the numbers of females fertilised by the much-diminished male population.

My instincts are that these animals probably don't stay in diapause in response to external temperatures and that the drop in numbers in 2018 is related to survival rates and adult longevity. Perhaps this could be tested if we can secure sufficient larvae and breed them up in controlled conditions?

Time will tell, but there is clearly some interesting research potential in this species.

Impact of drought upon numbers of hoverfly species recorded

Ad-hoc, or opportunistic, data are always very difficult to interpret. There is no consistency in the method of recording, with the numbers of people involved varying from year to year. The relative skills of recorders will also change as people change interests, lose interest or, sadly, depart this world! So, all analysis must be accompanied with caveats.

I have been trying to make sense of this year's drought in the UK. Can we use the numbers of records? For this exercise I looked at the numbers of hoverfly records extracted from social media per week in the preceding 5 years and produced two graphs. One compares the years (Figure 1) and the other compares 2018 against the average for the preceding 5 years (Figure 2). It is clear from Figure 1 that each year is so different that it is almost impossible to make any comparison. It should be noted that the numbers of records in 2013 and 2014 were small in comparison with 2016, and that the numbers of records in late summer and Autumn 2016 were much greater than in any of the other 4 years. 2016 was the peak year for data extraction direct from the UK Hoverflies Facebook page, since when many of the most active recorders of the time have switched to maintaining their own spreadsheets. In addition, the autumn of 2016 was unusually warm and saw recording extend far longer than normal (into early November).

Figure 1. Numbers of hoverfly records extracted from social media between 2013 and 2018

There is more to be made of Figure 2 in so far as it is clear that the general trend is similar with peak numbers occurring in August. A clear drop in the numbers of records during the drought is also apparent, as is the approximate 2-week difference in the start of the season as a result of the length of the last winter. Nevertheless, the range in numbers of records between 2013 and 2017 is substantially skewed by the first year when the UK Hoverflies Facebook group was launched (mid summer 2013). In that year, there was relatively little activity if one bears in mind the levels of activity in 2016.

Figure 2. Numbers of hoverfly records extracted from social media in 2018 against an average for the previous 5 years (2013 to 2017)

So, we can detect a general narrative, but it would be unwise to rely simply on the numbers of records. Is there an alternative metric that might tell us more? I have previously discussed the effects of the drought on recorder activity (28 July 2018: Recorder activity - a possible proxy for looking at the impact of weather on datasets?). In that analysis it seemed that recorder activity had diminished at a time when it might be expected to see growth in activity. So, with diminished recorder activity it may be that the numbers of records is directly related?

This time, my attention turned to the numbers of species recorded. Again, the year by year totals vary hugely, making it difficult to place 2018 into context (Figure 3). When placed into the context of the 5-year average, 2018 does stand out quite markedly (Figure 4). I think the crucial point is that the overall trend for 2018 was similar to both 2016 and 2017 so I have also plotted 2018 against the average for 2015 to 2017, which are the three years in which recorder effort was similar to, or exceeded 2018 (Figure 5). The result work very nicely, with 2018 clearly fitting the 3-year average until the third week of June, when the numbers of species recorded crash. Numbers of species seem to be on the rise now. The rise is partly explained by the arrival of second generations of some species and possibly also the effects of the big mass occurrence event ten days ago.

The big question is whether the numbers of species recorded will recover by September? If not, we must also ask 'what will be the knock-on effects into 2019 and how can we establish whether any reduction in numbers arises because of the 2018 extreme weather?

Figure 3. Numbers of species recorded from social media in the years 2013 to 2018. Note that whilst the totals for 2013 and 2014 are lower than those for later years, the disparity is not as great as in the numbers of records (Figure 1).

Figure 4. Numbers of species recorded in 2018 compared with the 5-year average from 2013 to 2017. Using this metric we might assume that 2018 was actually species-rich until the crash in June; however, the effects of lower levels of recording in 2013 and 2014 are clear when the graph excludes these years (Figure 5).

Figure 5. Numbers of species in 2018 compared with the 3-year average for 2015 to 2017. Using this metric it seems that 2018 was broadly comparable with previous years until the third week of June when numbers crashed. There seems to be recovery in the last week, perhaps as a result of cooler wetter conditions.

Can we interpret current mass occurrence events?

In the past few days there have been a number of posts of spectacular numbers of hoverflies covering every nectar source or source of fluids and sugar (e.g. a banana skin). These reports seem to be confined to coastal zones in northern England, eastern and northern Scotland. Superficially at least, we might assume that these are mass migrations. An equally valid interpretation, that has been advanced previously, is that these are mass emergence of local provenance. Can we resolve this conundrum from the data?

The dominant species in these events is Eupeodes corollae, with smaller numbers of Episyrphus balteatus and Scaeva pyrastri. All of these species are aphidophagous (larvae are predaceous upon aphids) and have a relatively short development time. In the case of E. corollae there are reports that larvae will feed on fruit in addition to aphids. All three species are also known to be migratory.

Mass occurrences are not uncommon and in most years we see a short-lived peak of Episyrphus balteatus in the third or fourth week of July that often coincides with the start of the harvest of cereal fields. In most years the peak occurs across much of southern England and occasionally there are reports of vast numbers streaming in across the coast. Two reports are perhaps relevant:

• one observation by a friend of mine who witnessed such an event on the Suffolk coast and later told me that numbers were about a thousand per minute per linear metre; 
• the other came from somebody who operated a support vessel for cross-channel swimmers and told me that occasionally the boat and the swimmer would be smothered with hoverflies (mid channel).

Such events do cause local concern and I have previously been contacted by the press to answer questions about a 'swarm of wasps' arriving on the coast.

So, we can be sure that at least on some occasions mass-occurrence are of continental origin, whilst there may be times when the source is local. That would make sense because there must be some parallels between population dynamics on the continent of Europe and in the UK. Local climatic conditions are likely to dictate the degree to which there are differences on any one year. However, whilst we can be sure that there are mass movements from Europe to the UK, we can be less sure about movement the other way.

The mass occurrences in 2018 seem to be different from the norm. Unlike past events, they have occurred from East Yorkshire northwards to the east coast of Scotland, the Moray Firth and Orkney. This complicates matters because the three species in question tend to occur in far lower numbers in northern England and especially in Scotland. They are essentially southern species in the UK and, although there is a resident population of Episyrphus balteatus and Eupeodes corollae, it seems unlikely that there is much of a residual population in Scotland as far north as Orkney. So, the case for mass-occurrences to be of local provenance is less robust.

We must not forget, however, that 2018 has experienced some extremely unusual weather, with a serious heatwave in June and July at least as far as the north of Scotland. I experienced it first-hand on the west and north coasts of Scotland where there was glorious weather in early June. Hoverflies, however, were very scarce and aphidophagous species were extremely sparse (see my post of 4 July Square-bashing in Scotland - hoverfly records). Those of Eupeodes corollae were particularly sparse with just 4 records in about 100 hectads! I guess that this generation might have been responsible for the current generation, given that it has been a very warm summer, but I am sceptical.

I have less evidence for E. corollae numbers on the east coast of Scotland, but the phenology histogram for northern England and Scotland (Figure 1) in June looks to be consistent with those for more southerly regions at the same time. So, we have no real evidence to suggest that there were high numbers of E. corollae in June and therefore the starting point for a mass emergence is not exceptional. Furthermore, going on my own experience in The Trossachs in July, I am doubtful that aphidophagous species have done terribly well in Scotland. That trip was noteworthy for the overall absence of all hoverflies and also of midges. When were midges last absent from such an infested place?!

Figure 1. Phenology of Eupeodes corollae in 2018 represented as southern England, Wales and the Midlands, and northern England and Scotland. All records combined are also shown. This illustrates how numbers have climbed in the north whereas in the south numbers have not maintained their June peak.

Nevertheless, the case has been made that these mass occurences could be of local provenance, with the numbers accentuated by animals remaining in diapause until conditions were favourable. I don't think we have much to go on there but it is certainly a theory that is worthy of further investigation. I think there may be some truth that some species use diapause in response to droughts or extreme cold. As yet, we simply don't know what happens in many species. We do know that spring emergence times are often significantly affected by the severity of the winter. We can also see how numbers of hoverflies drop in periods of extreme heat. This effect is also obvious in the data for Eupeodes corollae in 2018 (Figure 2). Dis-aggregating the effects of migration on populations is more problematic.

Figure 2. Phenology of Eupeodes corollae in 2018 against the five-year average. Unlike previous years, 2018 depicts a sharp decline followed by a new peak as opposed to a gradual build-up of numbers. This decline coincides with the most extreme part of the 'heatwave', whilst the new peak(s) are displaced from the 'average'. This shift in peaks possibly corresponds to a response to the prolonged winter of 2017/18 and is also present in the June peak

So, in summary, I think the coastal locations of most of the reported mass-occurrences strongly suggests that the animals are reaching land and desperately searching for a refueling station before spreading landwards. As the swarm spreads its numbers will diminish but may remain high for some way inland. Elevated numbers are likely to be found far inland, but not in the extremes that are seen on the coast. Nevertheless, we simply don't have the data or and comparative data from Europe to determine where the swarms originated. Maybe it is time to develop a pan-European monitoring network!

How are hoverflies doing in 2018?

Initial impressions suggest that 2018 has been a very odd year for hoverflies. The season started very late, and there was precious little activity during the winter. Numerous observers have remarked upon the low numbers of species and the apparent absence of some of the most reliable species such as Eristalis pertinax and E. tenax. At this stage of the season it is quite difficult to be sure what is going on but I hope I have come up with a way of presenting the data in a way that tells its own story.

The following graphs are based on the following processes:

• Records extracted from Facebook and other media such as Flickr have been used. The reason for this choice is that it is one block of data that I know has been assembled consistently. Also, it is the only block of data that I have ready access to (other data go on the HRS database that Stuart manages).
• I created a baseline by generating a set of tables for each species in the years 2013 to 2017 inclusive and then used these data to create an average for each week for each species. This is the background data used for comparative analysis and shows up as a blue line on the charts.
• I then overlaid weekly data for each species up to the end of week 30 in 2018; i.e. up to Saturday 28^th July.

There are charts for nine species, all of which are relatively straightforward to identify from photographs and are regularly recorded by contributors. This choice is ideal because it comprises species for which there are good numbers of records and which therefore make the outputs as statistically robust as possible.

Interpreting graphs is always tricky and in this case we have to think about the shapes of the charts and not the absolute numbers, although there will be places where numbers of records are strikingly different.

The reason for concentrating on patterns is that the graph of average numbers is based on a five-year period in which we have seen recording activity grow dramatically but in that time there has also been a shift towards members maintaining their own spreadsheets (a big thank you to all who have). It means that no two years are made up of the same group of recorders, so, in addition to seasonal and yearly variation, there are also demographic changes too!

The critical consistencies are that the data have been checked/ extracted by a tight nucleus of specialists and extracted by just three of us (Geoff Wilkinson, Ian Andrews and me). In addition, we see a growing throughput of novices who gradually turn into more experienced recorders whose ability to recognise hoverflies increases over the years. Some may drop out, but there will also be new recruits.

Results

These nine species tell a significant story.

Firstly, the numbers of hoverflies recorded were low during the winter lower than average) and the spring burst of activity was delayed for about two weeks. This is amply demonstrated by both Episyrphus balteatus (Figure 1) and Eristalis pertinax (Figure 4).

The impact of the drought in July is clearly demonstrated in Volucella pellucens (Figure 8) and Eupeodes corollae (Figure 2). There is possibly also a bigger impact on Rhingia campestris (Figure  6),. I expect the summer brood to fail in south-eastern England if experience in other droughts holds true; but the impact will probably only really become clear when the summer generation emerges or fails to emerge.

One of the more perplexing issues is whether we can detect the impact of the long wet winter? I think this may be apparent in Volucella zonaria (Figure 9). I think there is probably enough evidence to suggest that numbers are substantially down this year. We cannot discount the impact of drought because there was an uncharacteristic dip in the emergence at a point when it might have been expected that numbers would be rising rapidly; nevertheless, the overall numbers of this highly noticeable species appear to be heavily down on the average. I suspect there has been a winter impact and we might be able to detect this using occupancy models at the end of the year

I am also somewhat surprised by the degree to which the chart for Volucella inanis (Figure 7) seems to be mirroring previous years. In broad terms, the first generations of Eristalis tenax (Figure 5) and Rhingia campestris (Figure 6) are also  pretty close matches against the long-term average.

Finally, there always have to be winners in this sort of situation and I think this is amply demonstrated by the graph for Eupeodes latifasciatus (Figure 3), which has been far more abundant than in the last 5 years.

Figure 1. Episyrphus balteatus

Figure 2. Eupeodes corollae

Figure 3. Eupeodes latifasciatus

Figure 4. Eristalis pertinax

Figure 5. Eristalis tenax

Figure 6. Rhingia campestris

Figure 7. Volucella inanis

Figure 8. Vlucella pellucens

Figure 9. Volucella zonaria

Recorder activity - a possible proxy for looking at the impact of weather on datasets?

I have had the feeling that we had lost a lot of recorders from the UK Hoverflies Facebook Group and that activity on this page was down on 2017. However, upon closer inspection I think activity on this page rather more accurately reflects the prevailing conditions and we might even be able to use past years' activity as an indication of the conditions that prevailed then.


I attach two graphs. The first comprises the numbers of contributors in each week in 2017 and 2018. The second is the numbers of records generated. The results come with a health warning because in 2017 we started to encourage members to keep their own spreadsheets. This trend has continued into 2018 and we now have about 95 people maintaining their own spreadsheet or putting records onto iRecord. So, the graphs are not entirely comparable. Nevertheless, I think there are sufficient consistencies to make something of the results.

Figure 1. Numbers of contributors in each week

Numbers of records in each week

What is immediately apparent is that we had a much later winter this year, and that is clearly shown in both graphs. A clear drop in the numbers of records since the start of June (week 23) is not matched with recorder activity, whose decline is far more marked about 3 weeks later. So it looks like the numbers of records is not completely related to the numbers of recorders. It is, of course, possible that a small number of  very active recorders have switched to spreadsheets, but I am unconvinced that this is the case (I can think of one). To test this theory I split the recorder data (500 recorders) into a series of classes. The most active recorders this year range from one who cas contributed nearly 500 records to a longer tail of between 200 and 50 records. These I have placed in a single class because they form a suitably sized group (37). The gradated scale is then shown in Figure 3 with increasing numbers in each class.

Figure 3. Recorder activity in 2018 split into 5 groups depending upon the volume of records submitted.

We can see that the same broad pattern emerges in all of the groupings of recorder activity, which suggests that the fall in the numbers submitted is not simply a change in recording method but that it actually reflects a change in recorder activity. In reality, it looks as though recorder activity was probably greater in 2018 until the start of the drought! We will have a far clearer picture at the end of the year, but I think the trends are sufficiently clear from this sample of data.

What is also clear, is that the numbers of contributors has dropped during the drought. In other words, recorder behaviour may be a useful proxy for hoverfly abundance. We should, of course, also recognise that it may be that the hovers are about but the weather is discouraging recording! My standard ground-truth of graphs is to ask whether I have detected any changes during the course of my own field work? Have I changed my recording behaviour and why? Answer, yes I have and this is because it is darned hard work and unproductive on many days! I suspect a lot of datasets will show similar trends, but it will be interesting to see whether this is the case for all invertebrates? If press reports are true, then the trends for butterflies may be different.

Phenology histograms - are they an over-simplification?

I had quite a lot of correspondence yesterday on interpreting phenology histograms with two separate research teams. It got me thinking about the HRS data and how we present such data. In the developing HRS website, it is possible to change the date range and lattitude range to interrogate the phenology of particular species. This should make life a lot easier but, even then, I have spotted problems.

Cheilosia albipila is a classic example. The phenology histogram (Figure 1) suggests that it is double brooded, but we know that this is not the case. The adults fly relatively early in the spring and by June the larvae are big enough to record by splitting thistle stems. So, a representation of all records will give the impression that it is multi-brooded. This is clearly wrong, so what should we make of the histogram?

Figure 1. Cheilosia albipila - all records

I think a number of interpretations are needed.

Firstly, we need to interrogate the database and work out which records are of larvae and which are of adults. Where no stage is given and the record falls outside the main flight time I think we need to be somewhat sceptical about it unless we know the recorder and the sort of fieldwork they do. We might well have to go back to the original recorders too.

This is a species that is relatively infrequently recorded, so the histogram is composed of a comparatively small number of data points. The coarser the data, the more care must be taken in interpreting the outputs, but it seems that flight times have not changed markedly. Most records will be of adults at spring flowers but there may be occasional ones of adult females sitting on marsh thistle rosettes.

We get data from a lot of sources, including Mapmate synchs, so it is not possible to scan every record before it is uploaded. That means we have got to do some retrospective analysis and adjustment to the data. It looks like it could be a very big job!

Secondly, there are species where we know comparatively little about the larvae and can be fairly sure that larval records don't make up much of the data. The phenology histogram for Cheilosia pubera seems to confirm this (Figure 2). There are a couple of questionable outliers that might be larvae, but they might equally be misidentifications. In my experience C. pubera has a pretty short emergence period and is gone by the end of June.

Figure 2. Cheilosia pubera - all records

The above are two quite simple cases because the larvae are plant miners and coincide their development with optimum plant growth and nutrient mobilisation. It becomes a lot more complicated when investigating species whose larvae live in nutrient-poor rot holes or within decaying timber. In these cases it is quite likely that larvae may take two or more years to develop. What can we make of the histograms in these cases?

Callicera aurata provides a nice example because we know quite a lot about it from people who investigate rot holes. It seems that larvae in Callicera pass at least one winter as a larva and possibly more than one winter as such. So, we can say that it is almost certainly not multi-brooded or even with a partial second brood. Yet, its phenology (Figure 3) certainly implies that there are two emergence peaks. I don't think this involves full or partial second generations, but simply means that emergence is staggered. The dip in late July might simply reflect regular hot periods that reduce hoverfly activity? The lack of winter records of larvae might also beg the question whether we have got all of the data - clearly we don't!

Figure 3. Callicera aurata - all records

Callicera rufa is a species that has been studied in far more detail and we know quite a lot about larval colonisation of freshly created artificial rot holes. So what can we make of the phenology histogram (Figure 4)?

Figure 4. Callicera rufa - all records

This histogram is very misleading because most of the records are of larvae found in Scotland. It is often possible to find several different age clasess in the same rot hole, and clearly they can be found at most times of the year. In the last ten years, however, new populations have been found in England, with lots of data coming from several sites in Shrophsire and Norfolk.

The histogram of data from England (Figure 4) is largely free of larval records (not entirely) and the histogram reflects this. It also tells me that there must be a data glitch because I know there are records from the Saddleworth area later in the year but these don't show up because I have set the lattitude too far south!

Figure 4. Callicera rufa English Midlands only

These few examples highlight some of the challenges that exist in presenting and interpreting data. As the dataset increases in volume and complexity, we will need to provide detailed interpretations because the majority of readers will not have access to raw data and many will not know much about the biology of the animals in question.