Photography for site monitoring

I was contacted yesterday by one of the Wildlife Trusts, investigating the possibility of setting up a monitoring project for their sites by encouraging members to visit their sites and then posting the results on the UK Hoverflies Facebook page. It was an interesting proposal that got me thinking. My immediate response was that whilst I am always open to encouraging new participants in the Facebook group, it might not yield the sorts of results that would be useful for the Trust.

I don't want to discourage new initiatives, but I am concerned that such initiatives are heavily reliant upon a very small group of specialists who are willing to engage and provide reliable determinations. Such a system is not really sustainable as the current set-up is in need of significant change if it is to be sustainable in the future. More importantly, I think one has to give serious thought about the concept of monitoring: what are the principle objectives and can they be realised using a particular system?

Site monitoring arguably falls into a number of categories:

1. Development of a simple inventory of what is where and when was it seen.
2. Establishing ongoing surveillance of the continued presence of target organisms.
3. Evaluating responses of target organisms to particular management interventions.

There are doubtless other possibilities that elude me. The big question is, is photographic recording capable of delivering any of these objectives? In the case of hoverflies, I think we are starting to understand the practicalities of what might be delivered:

Site inventory and ongoing surveillance

This is probably achievable to some degree, but it needs to be borne in mind that ad-hoc photographing of what people see on a visit is mainly going to yield the most obvious species that fit the search patterns of the visitor. This is clearly shown by the composition of the photographic dataset generated by the Facebook group. These data show that casual recording reveals a relatively small number of species. Conversely, there is a group of contributors that make a serious effort to photograph everything that they see. This group generates a much broader spread of records and has shown that the FB page has been a valuable teaching tool - many of them have honed skills, developed a knowledge of their targets and now actively seek out species that have eluded them. In other words, they are developing the field skills required to generate comprehensive site lists.

Clearly, photographic recorders can and do provide important site-based data. The big question is therefore the degree to which species lists are comprehensive? I have been looking at this question for a fair while, comparing the composition of photographic lists with those that I generate through my own field work. This analysis shows that although there is considerable overlap, about 50% of the species that I record are either missing from the photographic database, or are represented by far fewer records than might be expected from comprehensive recording. One photographer has crossed over from strict photographic recording to a combination of photography and retaining specimens. His site lists are much more comprehensive and include many of the genera that would otherwise be absent from the photographic database.

Thus, I think that it is possible to develop site inventories if the photographer is experienced and knows what they are looking for. Lists of between 60 and 80 species ought to be possible in more southerly regions but will be harder to achieve further north. If the photographer is assiduous and has the time to sit and record everything they see for several hours on a regular basis, the lists could be even longer.

Species' abundance does vary from year to year, so site lists will grow over time. If the photographers are dedicated and continue to visit sites on a regular basis then the lists will grow and some sort of yearly comparison may be possible, but this will be heavily dependent upon maintaining sufficient active photographers. There is likely to be a long tail of single records and the presence or absence of these species may not be indicative of a change in the ecology of a site.

Responses of target organisms to management

Bearing in mind the natural fluctuation of invertebrate populations and the abundance of the most cosmopolitan species, the big question is whether there are species that can be used as indicators? Hoverflies are pretty good in this respect because they occupy a wide range of niches. Unfortunately, a lot of the specialists that might be useful indicators are difficult (or impossible) to identify from photographs. Genera such as Brachyopa, Cheilosia, Eumerus, Parhelophilus, Platycheirus, Pipiza (and other Pipizini) and Sphaerophoria comprise a significant proportion of the UK fauna but are very tricky to identify. Many require examination of male terminalia, wing microtrichia and characters that are obscured on the animal's underside.  Identifying these animals requires special skills that are built up over a long period of time, so even if a recorder retains specimens it will be several years before they have acquired the skills needed to arrive at reliable identifications. Consequently, photography is unlikely to deliver the sort of data needed to monitor targeted responses unless monitoring focuses on certain niches.

Saproxylic hoverflies are possible contenders as target organisms that may tell a story about woodland management, provided it is accepted that some species will not be reliably identified from photographs (e.g. Brachyopa and some Ferdinandea). Monitoring of single species may also be possible. For example, Microdon devius and Microdon analis have short emergence periods and are highly habitat specific so monitoring my be possible using volunteers who have been trained to recognise the species concerned.

Generating new capacity

The idea of using photographic recorders to gain a better understanding of wildlife presence on nature reserves is a good one because it is a new way of engaging with people and encouraging them to take an interest in the natural environment. The big challenge is doing this without placing greater strains on the existing cohort of specialists who do the bulk of identifications. In the case of hoverflies, we have just three specialists for adults and two specialists for larvae.

I've not kept a record of the actual numbers of photographs posted this year, but can provide a feel for the numbers of records generated in 2016: 27,247 to a firm identification and 7,065 to genus (or occasionally Tribe/Family) - a total of 34,312 records so far. In addition, the recorder who retains specimens as well as takes photographs has generated around 3,400 records and has a large number of specimens in the freezer that I will identify this winter. This phenomenal effort has been achieved by about 1,034 contributors with 65 people contributing 75% of the main photographic data.

This sort of record generation is totally reliant upon the presence of a small cohort of specialists, but there is a growing pool of people who are showing that they have the confidence to identify the more commonly seen species. In time, their skills will grow further. It is this growing skill-base that is essential to the future of hoverfly recording nationally and locally. It might be possible in some parts of the country to develop a partnership between Wildlife Trusts and recording schemes in which local groups of skilled people take on identification, with the Trust extracting records and passing on problem specimens /photographs to the national specialists. What I would not want to do is to generate vast numbers of photographs that have to be validated by national specialists - there is insufficient capacity and there is a danger that overload will lead to the loss of specialists who can no longer cope (I fall into that category if my workload is not significantly reduced).

I therefore think that the next stage is for Wildlife Trusts to consider setting up local systems to encourage recorders but to put in place local specialists to sift out the majority of the readily identified species and only pass on a small body of data for national specialists to determine. Crucially, they need to take on the data management - in the case of hoverflies I cannot cope with any further workload, either on Facebook or on iRecord.

Roadkill: an unquantified impact on invertebrate populations

Many readers will remember the RSPB's 'splatometer', which was abandoned because there were concerns that it could obscure number plates. Nevertheless, before it was abandoned, a trial was undertaken and its results were surprising: 324,814 insects were recorded at an average rate of one splat every five miles. Those figures look to be rather low and I suspect results may have been different had the device been fixed to another part of the vehicle.

The splatometer was a response to the need to find a way of monitoring changing insect populations.. It might have generated very interesting results over time, but sadly we shall never know. One of the possible outcomes might have been the start of an equally important idea - that road kill itself might be having a detrimental impact on invertebrate populations. This is a subject that has been in my thoughts for many years and I think needs a lot more attention. I am aware of an Indian study that demonstrated potentially serious impacts on butterfly populations but as yet I have found little else and nothing from the UK. I have therefore been looking for a possible way of investigating the problem.

In early May 2016 I noted a dead Anthophora plumipes on the roadside near Collyweston Quarry that got me thinking. The stretch of road I had just walked along might just be a useful study site. From then on, I made a serious effort to record insect roadkill along a 1 mile section of this road, the results of which are presented here.

The study site

An approximate 1 mile section of the A43 at Easton on the Hill (Figure 1) was chosen because it lies on the route of one of my regular walks. It is far from ideal because there are many places where the footpath is separated from the road by a grassy verge. Thus, records tended to be clumped around sections of path and tarmac

Figure 1. Study site and sections used to separate records

Methods

The eastbound carriageway was surveyed on a regular basis for any dead insects that might be attributed to collisions with vehicles. Although the gutter was inspected, most attention was given to adjacent footpaths and lay-bys because the gutter itself seemed to yield relatively few casualties. Incapacitated animals were removed and logged.

For much of the summer, it was not possible to visit the road on a daily basis and in several instances a period of a week or more elapsed between visits. The location of the casualties was logged according to the sections of the road shown on the map (Figure 1).

As this was a trial rather than a full-blown research study, the objective was to test the potential for investigating roadkill, rather than making an in-depth study. The lessons learned can be applied in subsequent years, either by me or by others.

Results

Data are presented in Table 1. Most of the records involved animals lying dead or dying on adjacent tarmac rather than on the road or in the gutter. Some were at a distance of at least 2 metres from the kerb, but the majority were in a zone about 1 meter from the kerb. 

Solitary bees	Bumblebees	Honey bees	Social wasps	Diptera
5	115	70	190	21
Coleoptera	Lepidoptera	Hemiptera	Orthoptera
7	6	1	1

Table 1. Total roadkill recorded between May and October 2016

Most of the casualties recorded were Hymenoptera, principally bumblebees during the summer months but in September and October social wasps and honeybees predominated (Figure 2). Lepidoptera were remarkable for their absence, although a small number of individuals were found. The relative lack of Diptera possibly reflects their tendency to break up on impact with vehicles and consequently they may be under-represented.

 

Figure 2. Casualties by month

Lessons

The distance from the kerb of many casualties suggests that ideal study sites may be difficult to find. In this case, no more than 25% of the route was anywhere close to 'ideal'. The numbers of casualties located was sufficiently high to suggest that detailed studies might be possible if sufficient replicates could be generated. My instincts are that sections of 50 or 100 metres would be viable. More detailed recording is needed to determine the width of tarmac required to highlight the full distance over which casualties may be spread.

It would not be wise to extrapolate from these data because they were collected on a relatively ad-hoc basis. No confidence can be placed in casualties remaining in situ for more than a day or two after impact. Some will crawl away and die in sheltered locations, others may be eaten by birds or small mammals.

Detailed recording of nearby nectaring sites may also be useful. My instincts were that the numbers of honey bee and social wasp casualties were quite closely related to nearby ivy patches. If this is the case, then apiarists might want to think carefully about where they position hives, because honeybee casualties do seem to be very high in the autumn.

The future direction of the Hoverfly Recording Scheme?

The HRS was established in 1976, at a time when we knew a great deal less about simple issues such as distribution. Recorders were reliant upon the RES key by Coe, which was a classic lesson in how to make identification difficult (a key by a specialist for specialists with access to museum collections). The complexity of the key meant that there were very few recorders and most Dipterists regarded hoverflies as 'too difficult'. Early on, many species that we now regard as quite 'run of the mill' were thought to be scarce and special. The development of two generations of guide books has changed perceptions (and reality) so that hoverflies are now a mainstream interest for many field naturalists - it is an amazing paradigm shift.

In a way, that lovely innocence of 1976 has been lost. The low-hanging fruits have been picked and the excitement of new discoveries now comes after much more effort. Modern maps tell us a great deal about species' current distribution and, apart from filling in a empty gaps, mapping at a Great Britain level alone is unlikely to provide the stimulus for recording that it once did. There is plenty still to do at a local level, with a good many under-recorded areas of Great Britain. Unlike other popular groups such as butterflies, moths and dragonflies, the numbers of recorders are still comparatively small and there is still scope for developing local groups and for local mapping projects to emerge (a challenge or opportunity whichever way you look at it).

With the basic task of mapping substantially established, what sort of niche should the HRS fill? Clearly, climate change and changes in countryside management mean that we still need to continue the basic recording so that changes in distribution can be detected. There are quite a few species marching northwards. The most obvious ones are Rhingia rostrata, Volucella zonaria and V. inanis, but others such as Epistrophe diaphana are quietly pushing boundaries too. Some northward expansion may be subtle but detectable over several decades. At the same time there are a few species spreading southwards: Callicera rufa, Sphegina sibirica and Xylota jakutorum immediately spring to mind. These seem most likely to be a late stage response to coniferisation and to the felling cycle. What about declines or contractions in range? These are MUCH harder to detect so we continue to need to generate the best dataset possible.

Understanding trends

Modern data management and statistical modelling have enabled researchers to look at recording scheme data in ways that could only be dreamed of ten years ago. Stuart Ball started to look at HRS data in this sort of detail a bit more than a decade ago when he and I published an analysis of the distribution of Volucella zonaria. He also did some similar work on coincidence analysis back in the 1990s and was a key pioneer in this approach, doing some great work on climate change predictions and occupancy modelling. More recently, modellers at CEH have taken up the challenge and various groups are now using HRS and other data to investigate animal and plant distribution and trends. Their work is taking the ideas forward and demonstrates just how important it is to maintain and grow biological recording across the country. So much more can now be achieved with 'ad hoc' data.

Although the dataset of the HRS (and other schemes) is somewhat 'ad hoc', there are elements within it that are more structured. This structuring of data collection is something that I believe we need to pursue. Butterfly and dragonfly recording have the transect approach. Moth recording has the Rothampstead trap programme (and garden moth scheme). I think hoverflies also need something. Alan Stubbs proposed a garden monitoring scheme back in 1991 and various approaches have been trialled in recent years. Those trials have not been frightfully successful because of the vagaries of weather and low numbers of people involved. However, we are now seeing a significant growth in people who make a daily record of what they see in their garden or local nature reserve. This sort of daily recording is really very valuable and analogous to the garden monitoring scheme so I have great hopes for the future.

The big issue I think is not only to grow the data flow, but also to improve feedback to contributors. If people can see that their efforts are both valued and used, they will hopefully be stimulated to continue.

Feedback on yearly and longer trends is an obvious progression for the scheme, and is something that we are already doing to some degree. I think we need to take it further, and I know that Stuart has this in mind in the design of the new HRS website. We should soon be able to provide on-line access to such analyses. That will be a major step forward.

Making links to biology

The other crucial stage in the process is trying to make sense of what trends tell us. For that, we really need a much better understanding of hoverfly ecology and larval biology. That is where we have needed to make advances within the scheme. A great deal of work was done by Graham Rotheray and the Malloch Society. They have been the pace-setters for hoverfly ecology. More recently Graham's daughter Ellie has started to engage with a much wider constituency and, together with Geoff Wilkinson, has really got things moving with the UK Hoverfly Larvae Facebook page. That is a great advance that I think needs to start to be the nucleus of the way forward into a more holistic approach. There is plenty more to do, as the morphology and biology of about 60-70% of the British hoverfly larvae is known to some degree. Investigating larvae is something that can be pursued by lots of people if they have the inclination. You don't have to be an academic with lab facilities - basic field skills, tenacity and inquisitiveness are the most relevant skills.

Live animal taxonomy

Most keys were developed using preserved specimens. They are subtly different to living animals and until very recently live animal taxonomy was very tricky. Modern digital photography has revolutionised the imaging process (especially with stacking techniques) but identification of living animals has lagged behind. Many experienced specialists can discriminate between species in the field but actually explaining why two species differ is often a great deal harder to put into a structured approach to identification.

We need to develop is aspect of identification – something that a relatively small group of people have started to develop, with Ian Andrews and Joan Childs taking on the bulk of this challenge. Should we be creating better links with overseas specialists to try to develop a pan-European network of specialists working on new ways of describing how to identify live animals? The WILDGuide that Stuart and I wrote is a first stage in the process, but I am the first to admit that my approach would be a bit different today after experience of the last five years dealing with photographic records. That is not to say that I find fault with the WILDGuide - it is simply that with many more years practical experience of what people see and how they record, I think we might present some elements differently and focus on slightly different issues. Maybe this is something beyond what can be done by the HRS but it points to a direction that we might want to go.

Organisation for the future

When Stuart and I took on the HRS in 1991 it was really focused on dot maps. I'd like to think that we have taken it several stages beyond those early objectives and that the sorts of outputs that happen today are a logical progression from basic mapping. I think we probably have a little way to go before the scheme is seen as anything more than a mapping and data assembly project, but progress has been made. The next step is to embed those wider principles of data assembly AND analysis. For this, I think we will have to start to broaden the team further, so we have greater analytical depth. It is probably too soon to actually recruit further analysts – I am hopeful that in time Ellie will develop part of this portfolio, but we will doubtless need others. I have my eyes open for potential recruits!

The next important step is to develop a more secure data management structure. I have taken this as far as I realistically can, and am jolly pleased that Geoff Wilkinson has joined in and is tackling the data from the larval group. I hope that by Christmas Geoff and I will have developed a new protocol and system that allows a broader group of people to do the data extraction.

The bigger question is whether we should be actively promoting the development of local groups? My instincts are that there are probably too few people who might act as local 'shakers and movers' for there to be a significant network, but I do see potential for some growing nodes of activity, possibly as local Diptera groups rather than simply for hoverflies. This sort of local organisation is the place where I think we might look to recent graduates who want to establish credentials as potential shakers and movers for the future. I have previously said that I would like to gradually reduce my involvement so that there is a seamless transition from the 'old guard' to a new an vibrant generation. Geoff and Ellie are obvious successors, but I suspect we will need a much bigger team because there is so much more to be done than in the simple days of record cards and the microscope!

In setting out these thoughts, it is my intention to put down a marker that I have recognised a need to evolve the structure of the HRS in order to give it the resilience to survive beyond the Ball & Morris partnership. We need to be gradually taking more of a back seat so that new blood takes over and gives the scheme the impetus for a further generation.

Data for data's sake

Aaron M. Ellison's article in Nature (11 October) gave considerable food for thought. It got me thinking about the issue of data and what we use it for. Is there a tendency for researchers to call for more and better data even though the real power of data may not be at the level of minutiae? It is much the same as the usual caveats about monitoring after a development project.

I have seen far too many cases where there has been a 'monitoring' package that is far too woolly and unfocussed to generate any real enlightenment about the impact (or otherwise) of a particular project. In development cases, what is proposed is often more closely aligned to what the developer is willing/able to pay for and not what is needed to pick up causal links. I concluded a long while ago that much of this monitoring was of very limited value and that conservationists rarely kept much of an eye on its outputs. In my experience, an agreement is made, the development takes place and a lot of expensive 'monitoring' is carried out. The resulting reports are rarely scrutinised by anybody in sufficient detail to determine whether they have actually served their purpose, and any critical lessons are rarely, if ever, learned.

As an example, I will use the monitoring for South Humber Bank Power Station back in the mid-1990s. ERM did an excellent piece of work that clearly showed that shorebirds were not affected by pile-driving – they habituated very rapidly. This same piece of work drew attention the the daft anomaly whereby a developer was expected to put in place millions of pounds of measures to protect birds from noise and human disturbance, whilst those same birds were constantly disturbed by dog walkers (and dogs), fishermen (beachcasting) and even by birders. The list of 'other' disturbances was shocking.

I certainly drew a lot of conclusions from this and regularly cited the report. I also marked it as a critical report when I left EN's Wakefield office. Two years later I wanted access to the report but the local team could not locate it. I went back to Wakefield to look for it and found it: when I left, I had carefully laid out all the the critical documentation with carefully compiled notes so that my successor had an idea of what mattered. The following day the they had simply piled all the documents up, put them in boxes and dumped them in a corner of a store cupboard. Nobody knew a thing and that explained why I was regularly rung up with queries. What a waste of three weekends on my part – I should not have bothered!

Do we need more data?

In the early days of English Nature many of us were shocked by the message from above that we had sufficient data and that there was no need to collect any more! That saw the demise of survey teams such as the 'England Field Unit'. In fairness, we were in a very difficult financial state but that does not justify the statement that we don't need new data. Subsequent events have shown that there is an ongoing need for new and better data, but that it must be collected and USED, rather than simply collected to sit in unread and forgotten reports.

That brings me on to what sort of data should we collect? In terms of biological recording, data on all taxa are needed. BUT, I would argue that there is a crying need for better data on more obscure taxa and, more particularly, people who actually understand what they mean. Thus, those who wish to pursue a career in conservation I think the critical issue is to be both a user and a contributor of data. That means developing field skills and then pursuing a course of investigation that takes you beyond the comfort zone of popular groups. If we don't see a significant expansion in skills in less popular taxa they will always be Cinderella subjects and will not gain traction when used to justify the case for conservation.