New unified list of birds - Avilist (1 Viewer)

[Paul Chapman]

One of the problems I sometimes have as an eBird reviewer is people entering a common species but using a subspecies group that doesn't occur here. This is fine if they understand that they're reporting something unusual and provide details to back up the subspecies ID,

In such circumstances, the issue is one of filters. Filters should trigger identification criteria if a subspecies is unusual.

The issue is one of where the default subspecies is in fact difficult to identify. Do you still allow a default subspecies to be recorded or do you add a filter when identifying to subspecific level even though the subspecies is expected & simply subspecific identification is difficult? The latter would seem counterintuitive as across the world when birding you are engaged in probability based identification to specific level so why not to subspecific level.

This is where I am unsure there is an answer or guidance. My local query was on Redwing subspecies. I got no answer. (I was amused that raised the Redwing query before the BB article which at least confirmed to me that I do not talk nonsense all the time.)

My frustrations boiled down to:-
(1) after several years, still unactioned changes to my recording areas;
(2) lack of guidance on the threshold of acceptance - are you mirroring an official record or adding something extra & more useful;
(3) lack of guidance & consistency on escapee, provisional, etc;
(4) proliferation of hotspots; &
(5) absence of a peer to peer community within reviewers to improve review & the quality of the overall dataset.

In the end, having done hard yards through hundreds of records, mainly in analysis & liaison rather than pressing the buttons, I was contacted about my inactivity & walked away from review... 😀 (Days of my life were spent breaking down hundreds of records in spreadsheets to source acceptance or otherwise because of the uncorrected geographical areas & the need to identify different local liaisons & reports.)

Enjoy.

All the best

Paul

 

[Mike Earp]

(5) absence of a pier to pier community within reviewers to improve review & the quality of the overall dataset.

I hear there's a good pier community in Clevedon. I guess you meant peer to peer ...

 

[Paul Chapman]

I hear there's a good pier community in Clevedon. I guess you meant peer to peer ...

Indeed. Corrected. I did record a Cory's Shearwater from the end of Clevedon Pier once which I did identify to subspecific level at the time so it withstands the split. 😀

(I frequently have to correct affect/effect & always get cast/caste wrong! 🤨)

Many thanks

Paul

 

[jts1882]

Looking at there website, looks like I was wrong. They suggest that they are now looking to publish an online version of their avian checklist.

They mention providing the checklist as free issue, but also mention subscriptions (and of course you can donate on their website). They mention an Excel version, but then also say '“Publishing on-line” is not a valid action in the eyes of the International Commission for Zoological Nomenclature because the html format is not protected. Where needed we will use e-publication with task-appropriate PDFs that are valid.' I only seem to be able to find on their website an online version of the 2014 checklist that cannot be downloaded and needs to be viewed family by family.

As I said previously, I think most parties have now settled on other taxonomies, and a new version of H&M (possibly in a difficult format to use as a database), will mean uptake will be limited.

They proposed to provide the checklist (family by family) free of charge, but to put the footnotes and reference lists behind a paywall (see here and search for "pay wall"). The paywall was to be introduced when they'd finished the updates, in the second quarter of 2024, but they've been silent since 2023.

They did start putting the updated families online. Iirc, they put up new versions for Struthionidae through Apodidae. But they pulled them and restored the old version with no explanation. It could be they have abandoned the update.

 

[Mysticete]

They proposed to provide the checklist (family by family) free of charge, but to put the footnotes and reference lists behind a paywall (see here and search for "pay wall"). The paywall was to be introduced when they'd finished the updates, in the second quarter of 2024, but they've been silent since 2023.

They did start putting the updated families online. Iirc, they put up new versions for Struthionidae through Apodidae. But they pulled them and restored the old version with no explanation. It could be they have abandoned the update.

putting footnotes and references lists behind a paywall is just dumb. The people who are most interested in that are more than capable of hunting those references down, and it would be competing with Birds of the World to some extent, which offers a lot more information not to mention images.

 

[TicoTyler]

This is interesting, but are you both saying that reviewers who are sent 'unusual' records, merely pass these on to local recorders? Steve, you mention your name popping up in the Norfolk report 'automatically', but you only get your name against a scarce or rare species, so your route into the reports could still be through reviewer access to your data.

I think that with the BTOs BirdTrack it is different and that county recorders are granted a kind of superuser access, so that can pull all records for their area.

Honestly, I'm unsure what a "local recorder" is in your context beyond someone maintaining a region's checklist. I suspect it's a fairly UK-specific thing where you have enough birders that detailed local records are being kept.

 

[Jon.Bryant]

Honestly, I'm unsure what a "local recorder" is in your context beyond someone maintaining a region's checklist. I suspect it's a fairly UK-specific thing where you have enough birders that detailed local records are being kept.

You may well be right. In the UK County Recorders compile annual reports which present data on all the species in their area. So this may include all sightings for scarce species, high counts for commoner birds, or early, late and peak arrival dates for migrants. There may also be comments on whether it was a poor or good spring for a migrant, or comment on whether there were apparent major declines in certain species, perhaps following a harsh winter, or conversely whether it was a good breeding season for a certain species. For this to work, birdwatchers are encouraged to submit all their records to the County Recorders so that they have a good dataset to work with - so more than just info on scarce or unusual records (which would automatically pass by the EBird reviewer). When using EBird or BirdTrack, I tend to try and estimate counts for all species, as I am sure that '20' is a lot more meaningful than 'X'.

But something similar must be being completed in some countries... there are too many examples of atlas's, reports and illustrated checklists from around the globe, which must be based on large databases of observations. For example long before EBird, I bought the 'Oman Bird List Version 6'. This is far more than just a list, and has distribution maps, histograms showing seasonal presence etc. Inside the cover it states the book was written in collaboration with the Oman Bird Records Committee.

To be honest, I always feel a bit guilty when I travel abroad. I can spend several weeks of dawn to dusk (and sometimes later) birdwatching, but I do little afterwards to ensure my records get to a regional recorder. If Ebird data is gathered on a national or regional level, it would be heartening and I can feel a little less guilty... but alas, I know that in some countries like China (that I visit at least annually), they have their own recording systems (based on IOC), and I strongly suspect that they are not interrogating EBird for additional data.

Perhaps Avilist will lead to more unified systems, or at least greater simplicity in exporting and importing records to various databases around the globe. I would hope that data compliance and ownership is not a blocker to data flow, and that regional superusers can be created.

 

[Mysticete]

My guess is that even with unification of checklists you are still going to have different regional recording systems, because I don't think differences in taxonomy is a major reason why those differences exist. Use of different authorities seems to depend a lot on what system was established first in a given area: If Chinese researchers developed there own system independent of ebird earlier or at the same time as ebird began to pick up steam, then that would likely be the system observers would be most comfortable using and they would continue to use.

I imagine it's not really that difficult to incorporate data from other databases if you are researcher, although I would also guess most of the time the researcher will either be focused on one region or one species. If I am interested in Roadrunner population dynamics it doesn't matter how many other systems are present, because almost all the data i want is going to be in ebird.

 

[Paul Clapham]

But something similar must be being completed in some countries... there are too many examples of atlas's, reports and illustrated checklists from around the globe, which must be based on large databases of observations. For example long before EBird, I bought the 'Oman Bird List Version 6'. This is far more than just a list, and has distribution maps, histograms showing seasonal presence etc. Inside the cover it states the book was written in collaboration with the Oman Bird Records Committee.

I suspect that those collections of data were produced by a small group of "elite" birders, especially in places which don't have a tradition of birdwatching. Even here in British Columbia, back in the 1980s I used to write my records on little index cards and periodically ship them off to a government office in Victoria. As did many other people who were more elite than me. And so we had a 4-volume "The Birds of British Columbia" published in 1990. We do have a birdwatching tradition here, though, unlike in many places where it was mostly British expats.

But now eBird has overwhelmed those local initiatives. Our provincial checklist committee is now on hold, evaluating what their raison d'etre is given the existence of eBird. And the areas in which we have traditionally aggregated sightings have changed to match regional district boundaries, since that's what eBird uses to define "counties".

 

[johnallcock]

In such circumstances, the issue is one of filters. Filters should trigger identification criteria if a subspecies is unusual.

The issue is one of where the default subspecies is in fact difficult to identify. Do you still allow a default subspecies to be recorded or do you add a filter when identifying to subspecific level even though the subspecies is expected & simply subspecific identification is difficult? The latter would seem counterintuitive as across the world when birding you are engaged in probability based identification to specific level so why not to subspecific level.

Yes, rarer subspecies are triggered by the filter. The common subspecies are all allowed in the filter, so would not trigger a review if reported. Where a subspecies is seasonal it will only trigger the filter at times when it is not expected (eg a wintering subspecies reported in summer). My comment is not about when someone reports a rare subspecies because they've knowingly ID'ed it as such.

My point was that some people always put in a subspecies ID, even when they haven't thought about subspecies at the time of the observation. I'm in Hong Kong and I've had to deal with reports of the Australian subspecies of Little Egret, Indian subspecies of Scaly-breasted Munia, Cape Verde subspecies of Purple Heron, etc. The species are not unusual but none of these subspecies are known here, so the filter is triggered by any report. Because the filter requires a description the observer may have sometimes entered a comment such as "This species is common". Sometimes when I contact the observer, they reply that they haven't actually thought about the subspecies at all.

If you haven't consciously identified the subspecies, you don't really need to report it. There is no loss of information if I report Night Heron without confirming it is the Eurasian rather than American subspecies. There is usually no need to submit a subspecies ID if you're basing it only on range - but if you do, definitely check that you choose the subspecies that is in range!

 

[Jon.Bryant]

I imagine it's not really that difficult to incorporate data from other databases if you are researcher

I would agree that if you are researching a particular species, then incorporating data from disparate recording systems is not going to thwart you. Saying that it is interesting that I recently read an article about the search for Streaked Reed Warbler, with listed the recent records of this possibly extinct species. As far as I can see, it didn't include 'alleged' records from the Chinese recording system. The top lister in China (who is based in Shenzhen) was surprised when I said there had been no authenticated records in recent years - his view is that it remains a very rare migrant along the east coast of China, which is reported occasionally by the Chinese birdwatching community. Were the authors of the paper missing important data - I would hope not.

If a researcher is looking at the decline of passerine birds in Indonesia and the impact of the cage bird trade, then I think extracting data from disparate recording systems is a bit more complex. And if you are not a researcher, and just want data to plan your next trip, of something less important, collating data from various systems can be difficult. My wife and I find this now when planning trips to China - a bit of data from Ebird, and more from the Chinese system.

I know that we have discussed before how simple it is to combine data and how important a single taxonomy is, and probably have different views. So I will have one last go to try and explain why from a technical/database point of view, multiple versions of taxonomies are bad.

Relational databases, will be behind apps such as EBird, or websites such as the BirdLife datazone. These databases will be comprised of multiple tables linked by primary keys. The actual taxonomic tree fits quite well into a relational database structure - we could have a table for Order, linked to a table for Family and so on for Genus, Species, Subspecies. But the keys linking tables need to be immutable otherwise we will likely throw referential integrity errors when if we try to alter a key. This means that if we want the ability to change the Latin, we cannot use it as a key, but need to have a structure of immutable keys that then reference the Latin. We can now simply change say the genus name in the genus table and any corresponding changes to the species name in the species table. The database is now up to date and we can do this without breaking the key system. In general keys are part of the hidden working of a database, that end user do not see.

If data was recorded in each database to subspecies level, I could export data for Anas crecca carolinensis from a Clements based database and 'convert' and combine this with Anas carolinensis data exported from an IOC based database. However, databases in general do not force submission of records to subspecies level. We therefore get into the more complex territory of having to do regional filters of Anas crecca observations, assuming that they are all Anas crecca carolinensis records in the Americas, but potentially also having to excluding any records reported as Anas crecca crecca. Even the regional filters may be messy to implement, if we consider trying to define geographical limits in the Aleutians and the Pribolof's.

To really make this simple, each database needs to index taxa (and enforce data entry) to a level sufficient to enable taxa to be mapped to other systems. This is where Avibase uses ZooKeys. But to follow my Teal example EBird now has to enforce data entry to 'Common Teal (Green-winged)' & 'Common Teal (Eurasian)', whereas in IOC the two species are already mapable.

The real problem is that the 'level' of identification and data validation, and and code developed to export and import data to databases using different taxonomies has to constantly change. Yes we may be able to predict a new split, not currently recognized in any taxonomy and plan for this, but the export and import functions have to cater for even minor changes such as variations in Genus or species name spelling... to a binary computer a one letter difference in spelling is not a equal, although a human brain can see that the two are probably variations of the same. As you can imagine, with annual or biannual updates to various taxonomies, keeping mapping levels and export and import functions current becomes a large task. And with EBird as they don't force data entry to 'group' level (hopefully of sufficient granularity to allow mapping) we already have problems.

If we want to do what the Scythbill database does, which caters for two different taxonomies, things get more complex still. I have been developing a database catering for three taxonomies, and you basically have to create a core list of mapable taxa, and then create mapable taxa tables in each taxonomy, so that each taxonomy can be linked in - you basically end up with three sets of tables connected via a central mapable tax. Maintaining and updating this structure is proving laborious - it is very difficult to fully automate, and requires a level of input to say that X has changed to Y (it is pretty hard to write code that can spot this automatically). Furthermore, the size of the database, and the code to display the data in different taxonomies becomes larger and more complex. If things were that simple, then I am sure that EBird or BirdTrack would have already catered for users, wishing to use different taxonomies.

In the electronic age, if we could settle of a single taxonomic concept, then all the above problems melt away, and we can export and import information purely on a common Binomial Nomenclature.

I suspect that when Carl Linnaeus created the Binomial Nomenclature, his intent was to have a unique system to index all species. If he knew that various authorities were developing their own variations, and in a way corrupting the indexing system, I am sure he would be turning in his grave. If we never changed or had variations in Latin Order, Family, Genus, Species and Subspecies names, but merely added and deleted names, then his system would be a good primary key system for any modern database!

I suppose at the end of the day, the debate over having differing taxonomies comes down to whether you are a fan of 'freedom of opinion' or a fan of 'functionality over freedom of opinion'. I am the latter and therefore strongly believe there should only be one indexing system.

 

[Jon.Bryant]

The issue is one of where the default subspecies is in fact difficult to identify. Do you still allow a default subspecies to be recorded or do you add a filter when identifying to subspecific level even though the subspecies is expected & simply subspecific identification is difficult? The latter would seem counterintuitive as across the world when birding you are engaged in probability based identification to specific level so why not to subspecific level.

If (as I suspect you do) you wish to convert your list to IOC, then presently you need to record all birds to sufficient granularity to allow 'conversion'.

In a perfect world EBird would enforce recording to this level, so that records can be simply ported between systems. Even if the groups are simply a means of future proofing against possible changes to the Clements taxonomy, then they still should enforce data entry to this level.

Of course this causes the problems you mention. Was the female teal you saw a Green-winged or Eurasian, and there seems to be no option but to use probability.

But I personally think that this is where EBird needs improvement. If I am in SE Ecuador, does the system give me 'Black-faced Dacnis (Yellow-tufted)' as the default? I am not confident that it does this as accurately as for species. I recall birding in Britain and using EBird many years ago, and it gave me options for Common Merganser, Common Merganser (American) and Common Merganser (Eurasian), although of course the American subspecies has not been recorded in Europe. Issues like this may have been resolved, but I do not have 100% confidence that EBird always gets it right, and so I tend to be caution and only record to group level when it is unequivocally the case. This then potentially screws up future proofing (say if Yellow-tufted Dacnis is split), and portability between taxonomies. Yes, EBird may later assign my records to Yellow-tufted and Black-faced Dacnis, based on location, but I don't find this very satisfying. I would much prefer I was 'told' at time of observation that it was probably a Yellow-tufted, and then I get the opportunity to look more closely at at the bird and work out why (or why not) it belongs to that group.

If there are options to submit more than one group for a particular location, without an indication of probability, and if observers guess, then I can see this screws up the system and potentially creates additional workload for reviewers. My wife often says 'I am going to click X group', and I normally retort 'your guessing, just record to species level'. But if records can be flagged to reviewers as something odd, why can't the software also highlight this to the user. The EBird database either knows that Black-faced Dacnis (Yellow-tufted) is in the east of the country only, or it doesn't. Perhaps some improvement in the filtering and the smiley face symbols would allow the 'probably' correct entry to sufficient granularity for future proofing, portability and to stop giving reviewers headaches.

 

[Jon.Bryant]

My point was that some people always put in a subspecies ID, even when they haven't thought about subspecies at the time of the observation. I'm in Hong Kong and I've had to deal with reports of the Australian subspecies of Little Egret, Indian subspecies of Scaly-breasted Munia, Cape Verde subspecies of Purple Heron, etc. The species are not unusual but none of these subspecies are known here, so the filter is triggered by any report.

So my concern that EBird does not filter groups properly for a location seems correct (see post above). If the Ebird phone app can provide a list of probably and scarce species by location why can't it do this for groups. It would make sense that when I am in Hong Kong, I am not even given the option for the Cape Verde subspecies of Purple Heron, and that I would have to deliberately search for this subspecies, if by some miracle I was confident I found one there.

All I can think is that the back end regional filters in the online database are more robust than the filters in the phone app - there must be filters in the phone software, as you can use the app without signal and with gps only. But there are not that many groups, and I cannot believe that improving the filters on the phone software is impossible.

 

[TicoTyler]

Yes, rarer subspecies are triggered by the filter. The common subspecies are all allowed in the filter, so would not trigger a review if reported. Where a subspecies is seasonal it will only trigger the filter at times when it is not expected (eg a wintering subspecies reported in summer). My comment is not about when someone reports a rare subspecies because they've knowingly ID'ed it as such.

That also relies on the reviewers configuring the subspecific groups in the filter. At least in Costa Rica, that is not the case. Almost none of the subspecies are set up in the filters here, so they are all flagged as rare with a handful of exceptions. Conversely what benefit is there at recording to subspecies level when the subspecies can be assumed by locality unless there's an area where multiple subspecies overlap (ie Yellow-rumped Warbler)? I was reporting to ssp level for a bit but was tired of cluttering up the rare bird alert emails and figured I wasn't really adding any value in doing it.

 

[Paul Chapman]

That also relies on the reviewers configuring the subspecific groups in the filter. At least in Costa Rica, that is not the case. Almost none of the subspecies are set up in the filters here, so they are all flagged as rare with a handful of exceptions. Conversely what benefit is there at recording to subspecies level when the subspecies can be assumed by locality unless there's an area where multiple subspecies overlap (ie Yellow-rumped Warbler)? I was reporting to ssp level for a bit but was tired of cluttering up the rare bird alert emails and figured I wasn't really adding any value in doing it.

The advantage from a recorder's perspective is to pre-empt taxonomic change. Otherwise, you may need to spend time reallocating your records for instance African/European Red-rumped Swallow & Cory's/Scopoli's Shearwater where I was surprised by some of the decisions made allocating the splits when I considered that there were defaults for locations and/or periods that were allocated to either/or following the taxonomic changes.

If you had carried on, then maybe the filters would (eventually) have been sorted out?

So it is future-proofing.

All the best

Paul

 

[Jon.Bryant]

when the subspecies can be assumed by locality

You are probably right that the majority can be assumed, but there must be a few more examples from Costa Rica where this is not the case - Yellow Warbler (Northern) and Yellow Warbler (Mangrove)? There was an recent article in British Birds that made the case for us all being more interested in subspecies - a subspecies at a strange location or at a strange time of year, of no less impressive than a species - just we can't tick it. And I thought the whole point of EBird Groups (we are not really recording to subspecies level) was based on identifiable taxa, so is there a reason we should abandon the use?

I am surprised that the reviewers configure the filters. So how does this get passed back up into the filters in the phone app, or is this why spurious races can be selected at time of input?

I would suggest that reviewers should be encourages to properly set the filters - they will be be helping themselves in the long run, and protecting themselves from being swamped by records from annoying people like me that try to use the groups, even when common and expected.

 

[l_raty]

I suppose at the end of the day, the debate over having differing taxonomies comes down to whether you are a fan of 'freedom of opinion' or a fan of 'functionality over freedom of opinion'. I am the latter and therefore strongly believe there should only be one indexing system.

As I see it, the debate rather comes down to whether you regard taxonomy as a scientific endeavour or, as you put it, as a mere "indexing system".

 

[Jon.Bryant]

I wasn't really adding any value in doing it

Thinking again, I think you are perhaps on to something. A well designed database should consider data validity. If a subspecies group is resident and isolated from other subspecies groups, and the subspecies group can be determined entirely by location, then why have the group as an input option at all? Isn't this just opening up the possibility of erroneous data entry? Perhaps the EBird subspecies groups should be pared back to only those that do or may overlap geographically. Paring back the groups may also mean that they have more meaning and impact, when displayed on the input screen.

 

[Jon.Bryant]

As I see it, the debate rather comes down to whether you regard taxonomy as a scientific endeavour or, as you put it, as a mere "indexing system".

I will probably ruffle some feathers, but I would argue that taxonomic lists are based on science, but that the outcome is an opinion - normally a judgement of a panel of people. If a taxonomic list could be defined by scientific fact, then why have a panel to cast a majority view?

The definition of a species is imprecise, and any definition ends up having exceptions. And as I have discussed on the forum before, any definition becomes absurd when we think about the the point in time when a subspecies blossoms into a species.

I have also been advised that to a taxonomist, the indexing system, is the science. A web article describes a taxonomist thus

'A taxonomist is an information scientist who studies how different ideas and entities can be categorized....Taxonomists use a complex methodology to sort and classify information so that others can easily access and understand concepts in an intuitive, user-friendly way.'

A definition of Taxonomy from Wikipedia states 'the branch of science concerned with classification, especially of organisms; systematics'.

This is why I think we have frequent updates to Genus names and reshuffling of taxonomic sequences. These changes make little impact on Ornithology, but are driven by the desire to classify and order things correctly. Yes Ornithologists might gain some insights from reassigned Genus names, but I think cases are rare - as I have said before did the unique Genus assignment for Baikal Teal lead to any further understanding of the actual bird - or just a change of the Latin Name? The interests and science of Ornithologists and Taxonomists can be a different as chalk and cheese.

As I have argued already, I consider the the 'science' of classification (and particularly the competing views) are now detracting from use of the classification system... and the top taxonomic organisations must share this view, or why strive through the WGAC to complete a voluntary process of harmonization, and commit to transition to this system? Many moons ago, when I first wrote to IOC, Cornell and BirdLife with a plea for harmonization and the reasons why, I was not met with a barrage of counter arguments as to why I was wrong, but with agreement and confirmation that the process was in fact already in hand. So why when I present the arguments on this forum is there so much contrary opinion? Are most of us not just mere end users of the classification system. We may not personally gain from from the unification process and may loose a few ticks, but surely this is small fry, particularly if others see a benefit.

-----------------------------------------------------------------------------------------------------------------------------------------------------------------------

Back to the start of this thread - we know that the AviList is due out in early 2025, but is there any news on where this will appear? The WGAC webpage has not been updated for ages, and does not seem perhaps the best place from which to launch a new list. A search for the AviList on line leads me to a company selling helicopter and airplane parts. Perhaps the WGAC should have chosen a name based on an available domain name!

 

[Paul Clapham]

So my concern that EBird does not filter groups properly for a location seems correct (see post above). If the Ebird phone app can provide a list of probably and scarce species by location why can't it do this for groups. It would make sense that when I am in Hong Kong, I am not even given the option for the Cape Verde subspecies of Purple Heron, and that I would have to deliberately search for this subspecies, if by some miracle I was confident I found one there.

All I can think is that the back end regional filters in the online database are more robust than the filters in the phone app - there must be filters in the phone software, as you can use the app without signal and with gps only. But there are not that many groups, and I cannot believe that improving the filters on the phone software is impossible.

I don't really understand what you're talking about. The reviewers configure a list of species for the region ("county" in Britain), including recognizable forms, which can reasonably expect to be identified there. Some of them are flagged "rare", which results in the app asking you for a description. And what they configure in eBird is what you see in your app.

(Although maybe you didn't know about "Show subspecies for data entry" in the app's settings. Turn that on.)

And yes, it's possible that you may see a bird of a species or form which isn't in the eBird list for the region. That happened to me when a Whooper Swan showed up in a local park. Type enough of its name and you get "Can't find your bird?" which, when clicked on, will give you a list of absolutely anything.