How does a hotspot work on Ebird? (1 Viewer)

[THE_FERN]

I would also like to see eBird separate sightings from hotspots and lists. Log the sighting where it was if the person is in "real time" mode but still have it in their list. Then roll those sightings into hotspots but also have the detailed info there. Hotspots should be hierarchical as well. But I'm not holding my breath on that. Massive changes and they're pretty entrenched.

However, adding some editable contextual information to Hotspots is nowhere near as hard and I still think eBird is the best location for this info. It will be hard to keep it up to date but having a world full of hotspot editors already and the raw number of contributors makes it a lot more likely to be successful and useful there.

Yes I tend to agree. It's why I've put more effort into the other USP items I listed

 

[jurek]

Lists are how birders have been keeping track for decades

Ebird tries to work precisely how birders have NOT been keeping their observations. Vast majority of birders do not record complete list and record only rarer and more interesting birds. Ornithology surveys worldwide have been trying to force them otherwise for decades and failed.

This way Ebird introduces another bias: most birders don't know all birds so 'absence' means nothing. People miss species they don't know very well, especially calls. Which becomes obvious abroad. This bias created fake when ebird was used to estimate populations of all birds of the world. Garden birds of North America and Britain appeared most common birds of the world - because birders knew them well.

For the hotspot, an obvious bias is how much of the hotspot was actually visited - if this is a vast area or a long road, then birds in more remote part will be under-recorded. And birders will be at a loss which parts of the hotspots are actally worth visiting. This bias does not exist if recorded by a precise location or a 1km square.

I still remember the frustration of trying to find birds in an ebird 'spot' of 30 km road in California, which passed through all habitats from villages through creeks, forest and desert and held all possible species from woodland songbirds to the Roadrunner.

 

[pbjosh]

Ebird tries to work precisely how birders have NOT been keeping their observations. Vast majority of birders do not record complete list and record only rarer and more interesting birds. Ornithology surveys worldwide have been trying to force them otherwise for decades and failed.

Probably at least 1/2 the birders I know record full lists of the birds they see. Actually, among Americans, it's almost all. Perhaps it's a cultural divide. eBird being US based and full of US employees and all...

This way Ebird introduces another bias: most birders don't know all birds so 'absence' means nothing. People miss species they don't know very well, especially calls. Which becomes obvious abroad. This bias created fake when ebird was used to estimate populations of all birds of the world. Garden birds of North America and Britain appeared most common birds of the world - because birders knew them well.

Of course this bias is there. I've also watched "researchers" miss bird calls and mis-ID bird song in the neotropics with somewhat shocking frequency. This will always be an issue with all data. The issue is in understanding limits when analyzing the data. Yeah, the data is messy, but there's a lot of it and it has led to some really cool things in cases where researchers understand and work within the limitations.

For the hotspot, an obvious bias is how much of the hotspot was actually visited - if this is a vast area or a long road, then birds in more remote part will be under-recorded. And birders will be at a loss which parts of the hotspots are actally worth visiting. This bias does not exist if recorded by a precise location or a 1km square.

In my experience, 1x1km squares are patently worse. Ornitho.ch and other 1x1 km systems are pretty useless for someone who doesn't already know how to bird in Switzerland and doesn't know where the parkings are, where the trails are, isn't in the whatsapp groups, etc. With hotspots at least you basically know what the birding site is. When I see something interesting on ornitho I invariably just ask my Swiss friends where the bird actually is or what the site is actually and where to park and where the trails are / who to talk to for access /etc. That info is pretty hard to sort out otherwise, and i speak German fairly fluently and can read French and Italian well enough, but the info isn't there if you're not plugged into WhatsApp or not already up to speed. Another thing I frequently do is to go to eBird and see if it's recorded there or if there's a hotspot in the same 1km square that will clue me in. But eBird has the same problem of course as you mention - I also frequently contact local people I know to see where things actually are... I just think it's actually worse with 1x1km squares.

I still remember the frustration of trying to find birds in an ebird 'spot' of 30 km road in California, which passed through all habitats from villages through creeks, forest and desert and held all possible species from woodland songbirds to the Roadrunner.

Yeah, we've all been there. It's happened to me tons and tons of time. At some point, whether 1x1km squares or hot spots, you still have to understand habitat and know how to bird, and even then local knowledge is key and that details can be missing in both systems. Really, none of the systems are perfect. Observado can provide some very precise locations but it's just random dots for good birds over time, it's not nearly as flush / complete as eBird despite all of eBird's limitations. I'm not defending eBird here, I would love more precise data when available. Again, not to really defend hot spots (I have lots of complaints) but slowly but surely, hotspots are breaking up / getting more granular and 30km long hotspots are less common than they used to be. It still doesn't mean it's a great solution but from an end user's perspective it's a bit better over time

 

[Mysticete]

Probably at least 1/2 the birders I know record full lists of the birds they see. Actually, among Americans, it's almost all. Perhaps it's a cultural divide. eBird being US based and full of US employees and all...






Yeah, we've all been there. It's happened to me tons and tons of time. At some point, whether 1x1km squares or hot spots, you still have to understand habitat and know how to bird, and even then local knowledge is key and that details can be missing in both systems. Really, none of the systems are perfect. Observado can provide some very precise locations but it's just random dots for good birds over time, it's not nearly as flush / complete as eBird despite all of eBird's limitations. I'm not defending eBird here, I would love more precise data when available. Again, not to really defend hot spots (I have lots of complaints) but slowly but surely, hotspots are breaking up / getting more granular and 30km long hotspots are less common than they used to be. It still doesn't mean it's a great solution but from an end user's perspective it's a bit better over time

On the first point, most everyone I know who uses ebird will put in everything they are seeing. About the only exception is when people enter in historical data, for instance people who want a rare bird they saw pre-ebirding to register in their ebird list for that county/state/etc. But there is a specific protocol for that. The other exception of course is stuff seen while doing something else, but again...ebird has a specific protocol for that.

On the second point, it's not like that issue didn't exist pre-ebird. Areas with poor birder coverage are always going to have uncertainties like this. I've certainly ran into this issue using trip reports, ebird, and bird-finding guides.

 

[Paul Chapman]

Chiming in, I never record my ebird list in the field. I prefer to write in a physical list than transcribe later. It's usually done back at my apartment, usually with lunch or breakfast. If I am traveling, it might be back in my hotel room at the end of the day (or even at the end of my trip). I think quite a few people do something similar.

Exactly how I record stuff. Amusing to see various biases, misrepresentations, etc in this thread. A remarkable resource that of interrogated properly can help greatly in researching for your future birding and if you choose, it can be a great repository for your sightings, photos and sound recordings.

Today I input seven complete checklists along a road that is not on Google Maps....

All the best

Paul

 

[Jim M.]

Ebird tries to work precisely how birders have NOT been keeping their observations. Vast majority of birders do not record complete list and record only rarer and more interesting birds. Ornithology surveys worldwide have been trying to force them otherwise for decades and failed.

This way Ebird introduces another bias: most birders don't know all birds so 'absence' means nothing. People miss species they don't know very well, especially calls. Which becomes obvious abroad.

People love to overstate things on this forum and especially about eBird--perhaps because of their own biases?

Regardless, I've been birding since 1966 when I was very young--obviously well before eBird started. From the beginning, it was popular to bird with paper checklists that allowed you to tick off all the species you saw for a day or a field trip. You wanted to see how high your species list could get. So claiming that birders rarely make complete checklists is just wrong. In any event, eBird accomodates however you want to record--you can make complete checklists or use the many other modes, such as incidental.

Second, to say that absence means "nothing" because not all birders know all the birds is nonsense. Obviously, researchers are aware of this possibility, and there are ways to accomodate for it statistically. Moreover, if I don't know the birds in an area I am visiting well, I don't check the complete checklist box, and I expect that is the case for many birders. In any event, if you spend any time researching areas using eBird, you will quickly find that, for areas with a significant number of checklists, the bar charts almost always accurately record how likely you are to observe a particular species and you will rarely observe any bird not on the chart.

For the hotspot, an obvious bias is how much of the hotspot was actually visited - if this is a vast area or a long road, then birds in more remote part will be under-recorded.

Which is exactly why eBird always encourages birders to use smaller areas to report sightings by including a notice to select a more finely grained location if a hotspot covers a large area. And you will see in eBird that hotspots covering large areas will generally also have other hotspots covering smaller areas within the same location. In any event, it is not a "bias" to have such a hotspot--it simply reflects how some birders record their sightings.

 

[Kirk Roth]

There seems to be a misconception among some in this thread that an eBird hotspot refers to a single point - that is not true and not the intent. Just as in a birding guide or any other checklist resource, if one is using it for bird finding, all the necessary birding skills (e.g. knowing habitats, time of day, and how to actually find the bird) are still necessary. If someone is looking for an exact location of a bird in a point in time, then the standard stationary points created by observers are the way to go - albeit even then the target species isn't just going to show up on cue!

The hotspots are a data collection resource and while I agree that it would be wonderful if each one were associated with a discrete polygon, the amount of time it takes to produce such a thing for the hundreds of thousands of hotspots is an unreasonable thing to ask. Whether someone is using them for data or birding, it does require some common sense and basic research knowledge (just like any other resource) and that doesn't seem to be a problem for the majority of users.

 

[qwerty5]

Moreover, if I don't know the birds in an area I am visiting well, I don't check the complete checklist box, and I expect that is the case for many birders

This actually isn't how eBird intends it though. It is supposed to be a complete list of the birds you are able to identify.
Anyway, I do agree with you that this doesn't make complete checklists useless.

 

[Jim M.]

This actually isn't how eBird intends it though. It is supposed to be a complete list of the birds you are able to identify.
Anyway, I do agree with you that this doesn't make complete checklists useless.

I agree that eBird standards permit you to do a complete checklist when you can't identify all the birds, and that is how it should be because even experienced birders can't identify all the birds they see or hear on every outing. But I don't think this means ebird's intent or preference is that you do a complete checklist when you can't identify many of the birds or recognize many of the calls--though read literally the wording of the standards do permit this.

 

[THE_FERN]

There seems to be a misconception among some in this thread that an eBird hotspot refers to a single point - that is not true and not the intent.

? To be clear, ebird collapses what is always an area (could be a track through an area---i.e. a line---still fundamentally an area) to a "representative" point. As with other such summaries (GB ordnance survey in the past, at least), it's unclear how this point is chosen. It seems very unlikely that it follows any rules (see my previous posts), except (presumably) that the point intersects with the area. We certainly cannot predict anything about the area from the point, except that it's "there".

As hotspots are arbitrary in extent and you have no way of knowing what that extent is, they're of very limited use (for anything much). Contrast them with a list for a national park, for example. Since you know its boundaries, you have its extent. Knowing that, you can guess or find out about habitat types, find major routes, tracks and trails etc. All of that makes it easier to find targets---if that's what you want to do.

a discrete polygon, the amount of time it takes to produce such a thing for the hundreds of thousands of hotspots is an unreasonable thing to ask.

Well the way everyone else does this is to give people basic editing tools to let them draw the shape. Works fine... ...And ebird already has some basic geometry editing capability.

Edit: geometry already exists for many sites like urban parks... Just need to download it

 

[jurek]

The hotspots are a data collection resource and while I agree that it would be wonderful if each one were associated with a discrete polygon, the amount of time it takes to produce such a thing for the hundreds of thousands of hotspots is an unreasonable thing to ask.

Actually, it was already done in observation.org. Which makes one curious, how much Ebird and Observation hotspots match each other.

It would be easy to cover the Earth with 1 km squares like ornitho and ebird. It would be even easier for Ebird to make an agreement with Ornitho and Observation and import their system. I imagine the problem may be that scientists are possessive about data for future publications.

 

[jurek]

In my experience, 1x1km squares are patently worse. Ornitho.ch and other 1x1 km systems are pretty useless for someone who doesn't already know how to bird in Switzerland and doesn't know where the parkings are, where the trails are, isn't in the whatsapp groups, etc. With hotspots at least you basically know what the birding site is.

I just don't get it. Ornitho.ch uses hotspots, too, but much smaller ones. And if you cannot navigate on 1 km square, how can you navigate better in a much larger area? And in ornitho.ch, the squares are overlaid automatically with the extremely detailed map showing tiniest details - where the problem is? That is why I generally think that precise observation points are the best.

 

[qwerty5]

Edit: geometry already exists for many sites like urban parks... Just need to download it

Assuming it's accurate, and knowing if it's accurate would take knowledgeable local reviewers who actually care, something that I find is very hit-or-miss.
I still think it's a great idea, but it certainly would take a lot of effort.

As hotspots are arbitrary in extent and you have no way of knowing what that extent is, they're of very limited use (for anything much). Contrast them with a list for a national park, for example. Since you know its boundaries, you have its extent. Knowing that, you can guess or find out about habitat types, find major routes, tracks and trails etc. All of that makes it easier to find targets---if that's what you want to do.

In most cases, the hotspot name indicates the extent (e.g. a state park, trail, beach, campground, wetland, etc.). And while it is sometimes ambiguous, one can (at least in the U.S.) usually get a good idea of what the hotspot includes. The problem is of course eBirders reporting to the wrong hotspot, which is not uncommon at all, and this could be mostly remedied with defined areas.

 

[Paul Chapman]

I just don't get it. Ornitho.ch uses hotspots, too, but much smaller ones. And if you cannot navigate on 1 km square, how can you navigate better in a much larger area? And in ornitho.ch, the squares are overlaid automatically with the extremely detailed map showing tiniest details - where the problem is? That is why I generally think that precise observation points are the best.

I am not interested in recording my sightings in 1km squares. Each birding event of mine is normally 2-3 hours and covers a larger area. I anticipate that I am most representative in that regard...

I anticipate that is the problem.

All the best

Paul

 

[THE_FERN]

Just to note: never assign your incidental sighting to a hotspot. Your record is associated with a point location (coordinates where you were). If you assign the sighting to a hotspot, the system will substitute the hotspot's coords for the actual ones which will then be lost forever.

 

[Kirk Roth]

Actually, it was already done in observation.org. Which makes one curious, how much Ebird and Observation hotspots match each other.

It would be easy to cover the Earth with 1 km squares like ornitho and ebird. It would be even easier for Ebird to make an agreement with Ornitho and Observation and import their system. I imagine the problem may be that scientists are possessive about data for future publications.

I'm admittedly not that familiar with observation.org. However, in eBird the important unit for Cornell's analysis is the tertiary political boundary - e.g. a "county" in the U.K. or USA. As others here have mentioned, the point of the Hotspot "point location" is to simply have a spot where one can click on it, with the understanding (to most!) that it represents an area, not a point, and that area has borders that correspond with the borders of the named entity, be it a park, a lake, a section of roadway, or whatever the "birding wisdom" declares it to be. Yes, the eBird system also allows for point counts and discrete area counts, and just as in a breeding bird survey it is up to the reporter to communicate the critical details as per their discretion. But if you poke around you might note that if a hotspot straddles county lines, each county gets its own hotspot for data compilation. It is worth remembering that eBird was in large part born of the BBS and Christmas Count traditions, and in my opinion opened up to common birders with features like checklists, top 100, listkeeping capabilities, "needs lists" etc. as an incentive for data collection.

If one is less interested in county trends, statistics for particular hotspots, observer data and so on - if one is interested in distributions in 1 km squares for example, then the formats you mention seem more in line with that sort of goal. Bird distributions provide particular difficulty with regard to vagrancy and rapid distribution change - they are just more mobile than reptiles and rodents - so more dimension is needed than simple geographic data and I hope obsercation.org etc. provide that for flying taxa. If eBird were to adopt an additional global mapping function (in addition to the multiple existing ones - coordinates, political boundaries, observation symbols, and the purple/orange/gray square system) that is another major undertaking that would again be a huge amount of effort. Its okay for people on this forum not to fully understand the nuts and bolts of the computing that goes into eBird (I don't!), but if we want to understand the system, we need to understand that there is a cost to making these changes. Or let me put it this way - if it were easy to implement such obviously beneficial things such as polygons for each hotspot, a small scale worldwide grid, or other things suggested in this thread, don't you think that would already have been done?

Of course, a birder with a target species has a different geographic interest altogether - it is the exact point where a bird is observable at a given moment! All of these are useful tools for accomplishing that goal - but they are only tools, and as such must be used with a certain amount of skill for those means.

 

[Kirk Roth]

Just to note: never assign your incidental sighting to a hotspot. Your record is associated with a point location (coordinates where you were). If you assign the sighting to a hotspot, the system will substitute the hotspot's coords for the actual ones which will then be lost forever.

It needs to be noted that this is not in accordance with eBird's data gathering guidelines. Incidental observations are by definition those with imperfect data, including location data, but also timing and effort data. If you know you saw some bird on a particular data at a particular park, but don't recall the exact location, it is absolutely appropriate to assign your incidental sighting to a hotspot. Not to do so would mean there is no place to put it - then not only is the precise location data lost forever, but so is the "rough" location, the date, the sighting, the species.... everything!

If one has a discreet location but not the timing and effort data, (or if it was a sighting that was just plain not birding - e.g. no effort!), that should be assigned to a particular geographic location, but still remain incidental. In that case, I agree the location bubble is better than the hotspot.

Note that a hotspot's coordinates are never substituted for a point count location in Cornell's system. If a checklist is attributed to a hotspot then it is understood that the location is somewhere within the commonly understood hotspot. Obviously, every bird sighting at a particular park does not occur at one coordinate!

On the other hand, if one is doing an intentional point count or a discreet area count (meaning actual science and not "just" birding!) then absolutely it is far better to assign it to a particular custom-located spot that corresponds to your survey. However, this is NOT an incidental sighting and if you have at least the minimal effort data (time, distance and number of observers), it should be recorded as a checklist. If it has all observed birds, it should be a complete checklist.

 

[pppppvvvv]

I am not interested in recording my sightings in 1km squares. Each birding event of mine is normally 2-3 hours and covers a larger area. I anticipate that I am most representative in that regard...

I anticipate that is the problem.

All the best

Paul

It's how we work isn't it. From local patches to gathering info for trips.. go to site X: hope to see list of species.
Or we zoom out and look at the global distribution of a species.

Someone out there must be interested in which tree the bird is at which time of day, but they probably accept that they need to conduct their own personal study for that.

 

[THE_FERN]

It needs to be noted that this is not in accordance with eBird's data gathering guidelines. Incidental observations are by definition those with imperfect data, including location data, but also timing and effort data. If you know you saw some bird on a particular data at a particular park, but don't recall the exact location, it is absolutely appropriate to assign your incidental sighting to a hotspot. Not to do so would mean there is no place to put it - then not only is the precise location data lost forever, but so is the "rough" location, the date, the sighting, the species.... everything!

If one has a discreet location but not the timing and effort data, (or if it was a sighting that was just plain not birding - e.g. no effort!), that should be assigned to a particular geographic location, but still remain incidental. In that case, I agree the location bubble is better than the hotspot.

This is to misunderstand the value and use of incidental observations. And also perhaps to not understand the continuum of effort which exists between an incidental (supposedly "zero" time expended) and a birding trip (non-zero). To take the first: most people record rare or unusual birds as incidentals to augment their personal lists. These observations have scientific value too—as all presence records do [including museum specimens (which also don't conform to ebird's recording protocols !)]

As an example. Let's imagine I'm on Halmahera on a scuba diving holiday. I look up and see a black-naped oriole (Clements taxonomy). I decide to record it as an incidental. Does this have scientific value? Of course, since that species is not known from Halmahera.

To repeat the warning: if it's an incidental don't assign it to a hotspot or you'll lose valuable (scientific) information about the precise location.

(For unknown, unfathomable and entirely spurious reasons, ebird forces you to assign a location before you submit all lists. (Why is unfathomable etc? because if it's tracking your GPS it knows exactly where you went. It makes sense to ask you if you switched tracking off but not in any other case. 'Ahh, but it needs to know whether your checklist pertains to a "hotspot" ' you cry. "Only if it's not properly recording hotspots" I reply—if it were it'd just use a geoquery to see if your track and its hotspot intersect.)

Note that a hotspot's coordinates are never substituted for a point count location in Cornell's system. If a checklist is attributed to a hotspot then it is understood that the location is somewhere within the commonly understood hotspot.

The nub of the matter is that they're not "commonly understood". I've discussed this above...

...Even if they were, the effort expended varies with checklist. Analysis is appropriately done at checklist level and not at hotspot level. What would you meaningfully do? Compare the number of birds recorded over one unknown area and effort category with those in another? Compare the number of hotspots of (unknown area) in a region with those in another?

Burgess Park in S London has about 6 hotspots. One covers the whole park, the others segments of it. Only the whole park one receives regular updates. How do we know what the extent of each hotspot there is? We don't. Is there any comprehensive metadata to help us understand? There is not. If I'm walking down the street that borders the park am I in the hotspot or not (dunno). If I'm in the park and I see a bird flying slightly outside it should it go with the hotspot? Who "commonly understands" this? [and municipal parks are the easy ones with clear boundaries]

Obviously, every bird sighting at a particular park does not occur at one coordinate!

Since we have no idea of the extent of the hotspot and no idea of the area covered by a trip list, then it conceivably might. Even if it were a large area and a long walk, a single encounter with a flock of birds makes it possible !

 

[raymie]

This is to misunderstand the value and use of incidental observations. And also perhaps to not understand the continuum of effort which exists between an incidental (supposedly "zero" time expended) and a birding trip (non-zero). To take the first: most people record rare or unusual birds as incidentals to augment their personal lists. These observations have scientific value too—as all presence records do [including museum specimens (which also don't conform to ebird's recording protocols !)]

As an example. Let's imagine I'm on Halmahera on a scuba diving holiday. I look up and see a black-naped oriole (Clements taxonomy). I decide to record it as an incidental. Does this have scientific value? Of course, since that species is not known from Halmahera.

To repeat the warning: if it's an incidental don't assign it to a hotspot or you'll lose valuable (scientific) information about the precise location.

(For unknown, unfathomable and entirely spurious reasons, ebird forces you to assign a location before you submit all lists. (Why is unfathomable etc? because if it's tracking your GPS it knows exactly where you went. It makes sense to ask you if you switched tracking off but not in any other case. 'Ahh, but it needs to know whether your checklist pertains to a "hotspot" ' you cry. "Only if it's not properly recording hotspots" I reply—if it were it'd just use a geoquery to see if your track and its hotspot intersect.)


The nub of the matter is that they're not "commonly understood". I've discussed this above...

...Even if they were, the effort expended varies with checklist. Analysis is appropriately done at checklist level and not at hotspot level. What would you meaningfully do? Compare the number of birds recorded over one unknown area and effort category with those in another? Compare the number of hotspots of (unknown area) in a region with those in another?

Burgess Park in S London has about 6 hotspots. One covers the whole park, the others segments of it. Only the whole park one receives regular updates. How do we know what the extent of each hotspot there is? We don't. Is there any comprehensive metadata to help us understand? There is not. If I'm walking down the street that borders the park am I in the hotspot or not (dunno). If I'm in the park and I see a bird flying slightly outside it should it go with the hotspot? Who "commonly understands" this? [and municipal parks are the easy ones with clear boundaries]


Since we have no idea of the extent of the hotspot and no idea of the area covered by a trip list, then it conceivably might. Even if it were a large area and a long walk, a single encounter with a flock of birds makes it possible !

To use your oriole example - what do you lose by using a hotspot? You haven't explained it?

Also, not all checklists have GPS data - only those submitted on the mobile app. Even among those that do, a single point is needed for viewing the maps.