Just a bit of a mediocre observation but I uploaded a sighting of one of the most common and easily recognised birds in the UK today and checked the distribution map to see where the bird had been seen. This is a bird that is not afraid of people, often actively seeking out human activity in search of food and getting close enough for decent phone photos, but doesn’t concentrate large populations in urban areas like pigeons and corvids so does take some effort to record. I would expect any user who uses the website for more than a few casual Id requests would upload at least one observation of this bird at some point. It struck me though that the distribution map shows more about the population and habits of people who use the website than of the bird. The bird has a much wider distribution than shown but what is shown should show the favoured habitats of nature lovers. It’s an indicator species of iNatters.
Every observation is a sign of where iNatters are, and maybe how info about iNat is distributed across populations (and wealth of those populations, e.g. here it’s a straight correlation of wealth and number of obs per region).
But I also feel like you’re talking about Robin, as it’s a first UK bird that comes to my mind with your description.) It needs to be added, that this species behaves pretty different in different parts of Europe. I would say its range is quite nicely represented on iNat, could just be more observations in some areas. It reminds me a Dunnock that I feel is seen as garden bird in UK, and here you have the bigger chances to spot if you go to a dark forest and try to locate it by ear. Reading old book, I found out Blackbirds were considered secret far-from-human birds just 50 years ago. Wood pigeons live all over cities in Europe, in other parts of its range it’s not that, Stock doves also are known to be hating humans, but from certain observations I see it’s different in other places.
An easy to spot bias in data I see in my project, the further the spot is from the city, lesser amount of observations will be there. I know some regions that has 0 obs of particular birds, plants, insects, common ones, there’re just not enough big observers there.
Yes, I am sure that kind of thing (indicator species for iNatters) is a common phenomenon, although iNatters do vary quite a lot from one of them to the next in terms of whether they mostly photograph plants, mostly birds, or mostly insects, and not so many of the other organisms.
Funny that I was actually thinking about something similar recently: how would it be possible to model the activity of iNatters using uploads in order to use it as a null model for the distribution of species. If we compare the uploaded data for each species to this null model, we would get more meaningful, probabilistic distribution maps because few uploads from high-activity areas would really indicate low probability of occurrence, while few data from low-activity areas would mean either the opposite or nothing. Or we won’t. :)
Yes, I think some projects have done this. Generally, you could take a group that includes similar organisms as the null distribution (all snakes in a given area that you are studying as specific species of snake it, etc.).
When I work on “Unknowns” for a specific location, I find that the same half dozen or so species of cultivated plants are disproportionately observed. Whether this is an indication of which garden plants are most popular in that region, or which ones inexperienced naturalists are most likely to notice, that I don’t know.
There’s a few towns in my state where I’m the only person who has ever made observations there. It doesn’t really give an opportunity for an accurate portrayal of the biodiversity in those areas because it’s just me. That’s not exactly the same thing but another example of how the distributions can be altered or skewed by certain characteristics.
There’s a huge number of factors that will affect the iNatter patterns that show up in this kind of data. Cultural variations for instance. In the USA 26% of all animal observations are birds; in France that’s a mere 16%. The French (and similar in UK and Germany) record an impressive number of insects, though, making up more than 60% of all animals observed, while in USA that’s less than half. In UK and USA, observations of animals and plants are made almost equally (around 49% and 43% of all observations, respectively) but in France almost twice as many animals are recorded as plants (61% vs 34%).
I suspect these quite dramatic differences reflect national cultural interests and the communities among which iNat has happened to become popular rather than any great variation in the prevalence of plants, birds or insects in each of those countries. I dare say that, for a variety of reasons, similar patchiness occurs at all geographical scales (states, regions, towns, etc, not just countries) making it really tricky to disentangle the factors at play in these patterns in the data.
People use iNat in so many different ways, often very specific ways. Of all the millions of users who have recorded observations on iNat, 75% have not recorded any birds at all.
I don’t take a lot of bird pictures, because I’m not photographically equipped to make satisfying bird pictures that I enjoy as photography - so I’m not likely to bother at all with a common species. However, if I’m puzzled as to what’s going on, I might well take a picture and post it, and I make a post whenever I come across a probable window strike kill.
I’m also occasionally finding something to observe to record my own presence at a location, but I’m more likely to choose insect/fungus/plant for that.
Additionally, I have years of stored photographs, and sometimes go through and upload them - understanding that my interests have changed.
If you want to model the activity of iNatters as a spatial probability to correct for sampling bias, I’d recommend using the global pool of observation points, and calculate a probability map with a Kernel Density Estimation.
Such indicator species will vary from country to country without doubt and in countries that don’t suffer from the same level of nature depletion as us it might not be as obvious. The robin is the third most observed species in the UK and for the reasons I gave it’s likely to have a more even distribution across the country compared to the other nearest top species. Being a highly territorial species that sings its presence loudly to the world means that you don’t even need a photo to get an ID. Granted only 4149 users out of 91,578 in the UK have recorded it so only a small percentage in reality but that’s still a decent data set. You’ll have a big chunk of registered users on one side that only signed up for help with a few IDs and another chunk on the other side who use the website for a special purpose. The combined distribution map for all species will do a better job of representing all that but it will also soften the edges of the data that I think the map of robin sightings shows.
I wrote a blog post around the Bioblitz last year. The top ten, most observed, most visible, most seen - and yes I have used my own pictures of them.
That is humans reacting to bling it on. Not about a biologist mapping actual diversity.
I’m completely fascinated by the patterns of inatters and the way it affects the distribution maps of various species. It’s also cool to compare it to some of the older biodiversity databases and see the differences.
For instance, I’ve long noticed with Calflora that plants are much more likely to be recorded in areas where they are uncommon. Looking at some of the species maps there, you ended up with nearly the inverse of the actual distribution - plants have very few records where they are most common, but that one tiny out-of-range location with a disjunct population of 5 specimens will have a hundred records.
With iNat, there are more less-experienced botany enthusiasts who photograph common plants as they run across them, just because they’re pretty, or still “new” to that person, or they’re trying to up their number of species for the day. Ironically, a certain lack of expertise actually leads to more accurate distribution maps.
And then there are those common yet obscure species that only a few people observe, and it’s always fun to run across the maps for on of them, because it’s just a map of where the people who recognize that organism have happened to go. As an example, as far as I can tell, this moth is pretty common in Douglas Fir, but I’m the only one who records it. Well, @norikonbu has a couple observations now too, but I pointed them out to her :D
There is one hiking path I know of that, based on the maps, has an absolutely stunning amount of insect diversity. All kinds of rare things, more species packed into this one short trail than in most of the rest of the county. I went there once because it was the only spot around where a certain butterfly I wanted to see had been recorded, and happened to run into the lone insect-obsessed inatter who walks that trail every single morning taking photos! I’m sure any other spot she chose would end up showing the same kind of diversity eventually.
This happens for phenology too. I help run an anole project in schools and it runs during winter. We totally messed up the phenology graph for anoles for a while and it showed them being easiest to find during cold months (when it’s just school kids making several thousand observations). So it’s worth thinking about specific iNatters or events that may be driving patterns, especially for species with lower sample sizes.
That’s the case with many species, genera or even families. Since most people live in cities, that’s also where most observations are being made, although most species should be much more abundant in rural and natural areas. For the group that I’m mostly dealing with, jumping spiders, you see a similar pattern. Most observations are in big cities, but the greatest number and diversity you obviously find in natural habitats.
This extreme sampling bias is one of the reasons why iNat data cannot be considered scientific in any way. It still gives us a good understanding of the boundaries of a species range, though.
Being not acquired by scientific method doesn’t make data non-scientific, each (normal) observation is as scientific data point as herbarium page or collected animal.
I suppose it would have been more accurate to say, “…iNat data cannot be considered systematic in any way.”
Well I’ve set myself a challenge to fill in some of the gaps in the Robin map this winter. It started well enough but I just have to remember to check the map before I go anywhere and take the long lens just in case. I missed out on two because I couldn’t get close enough. I’ve only logged ten obs so far but they’re pretty spread out. Hopefully the Christmas break will give me some Robin hunting time.
I may have underestimated how easy it is to log a Robin observation I’ve only managed 13 now and missed quite a few.
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.