Patterns in iNaturalist Data

General
1

You may have heard of the species-area relationship. It states that the number of species doubles if you increase the area by a factor of ten. This can be seen in iNaturalist data at medium scales (in between property size to country size). If you are in an area with a very uneven distribution of observers or unproductive habitats (such as a large lake), you can modify the radius until you get the observations to 10 times the first area. Then the species should be roughly double.

The species area relationship holds for specific taxonomic groups, but the area increase to double species varies. For example, birds are highly mobile. Many migrate so you have summer, winter and migration ranges plus birds sometimes wander far out of range. For birds, expect to increase the area by a factor of 100 to double the species. Molluscs are highly endemic and/or habitat specific so require less than a 10 times increase in area to double the number species.

Another pattern that can be seen in iNaturalist data is most of the time, the person at the top of a leaderboard in terms of species for a specific area tend to have seen about a third of the species for that area. Now it is not terribly difficult for a person to get over 3/4 of the species for many specific taxonomic groups, say birds, butterflies, and odonata. But getting over 1/3 across all taxonomic groups is extraordinarily difficult due mostly due to the a bunch of extremely diverse insects orders and fungi. For the diverse insect orders (flies, beetles, true bugs and hymenoptera (bees, wasps, ants, sawflies)) finding even half of them requires extraordinary effort. This effort involves time hunting down obscure field guides and primary literature, absorbing terminology, utilizing keys often without illustrations, using a variety of traps, examining genitalia under a microscope and so on. Fungi is also diverse and requires a compound microscope for examining spores for a large fraction of species plus you need to spend hundreds of dollars on field guides to cover all the taxa you may find.

I think most people would rather go further afield to find more easy taxa than to chug away at obscure taxa more locally.

Any other interesting patterns you can find in iNaturalist data?

2

Wouldn’t you expect the iNaturalist species/area graph to rise more steeply, because the natural species/area relationship should be magnified by the number of iNaturalist observers also increasing with increasing area?

3

I have found that using a Comparative Distribution Map is a very effective way to spot observations of a species outside of an expected range or geographic area (alpine, lowlands, coastal)

I’ve been surveying one specific valley over the last 9 months daily and I’m still finding new species there (almost at 550 total), it would be a lot easier to travel around a bit more and record heaps of new species that way.

4

Benford’s Law, which could be summarised that most numbers start with 1.

5

Here is some examples. The first link is a 20 km radius circle in the heart of southern Ontario. The next is 72 km then 200 km and finally 720 km.

Results
20k obs, 2913 species
419k obs, 9476 species
4.6 million obs, 18704 species
42 million obs, 45443.

The fifth circle at 27.8 km in radius
44 k obs, 4192 species.

While step 1 to step 2 there is a greater than 10 increase in the number of observations, but the next three steps the observations do increase by a factor of ten. If you replace step 1 with the slightly larger fifth circle that has approximately one tenth the observations of step 2 than you see a rough doubling of species until the last step. The last step the species more than double because the circle is so large that a latitude gradient is having an noticeable effort.

https://www.inaturalist.ca/observations?lat=45.045199002000004&lng=-77.74762284299999&radius=20&subview=map

https://www.inaturalist.ca/observations?lat=45.045199002000004&lng=-77.74762284299999&radius=72&subview=map

https://www.inaturalist.ca/observations?lat=45.045199002000004&lng=-77.74762284299999&radius=200&subview=map

https://www.inaturalist.ca/observations?lat=45.045199002000004&lng=-77.74762284299999&radius=720&subview=map

https://www.inaturalist.ca/observations?lat=45.045199002000004&lng=-77.74762284299999&radius=27.77692024085836&subview=map

I did a second set starting with a 10 km radius circle in Austria, but adjusted the circle size so I get a 10 time increase in observations each time.
6696 obs, 2062 species, 10 km radius
65k obs, 5927, 25.3 km radius
659k obs, 13066, 54.0 km radius
6.6 million obs, 29866, 308 km radius

Here it appears to increase by a more than a factor of 2 with each jump but not by much except for the first jump.

6

i’m not sure if it’s exactly what you’re looking for, or if it’s interesting exactly, but i think the thing that most people wanting to use iNaturalist data should know is that it very much reflects the behavior of the humans that generate that data.

most obviously, observations tend to occur where there are more humans, not necessarily where there are more of the non-human subjects. and more specifically, observations tend to occur where there are humans that have the time, interest, and other resources needed to make such observations.

but it’s not just the observers that influence the data. because identification and identifiers are critical to iNaturalist data, too, you get artifacts related to identification, as well. for example, i’ve noted in the past how some countries seem to have distinctly different research grade ratios vs other countries: https://forum.inaturalist.org/t/identifriday-is-the-happiest-day-of-the-week/26908/1341 (also see: https://jumear.github.io/stirfry/iNat_map?view=subsetratio&verifiable=true&quality_grade=research&compare_exclude_param=quality_grade).

finally, i think it’s also worth recognizing the temporal patterns in the data. it’s well known that observations spike during events like the City Nature Challenge, the Great Southern Bioblitz, etc., but observations also reflect when people have free time to observe. At a global level, you can see more observations happening on weekends and during big bioblitzes: https://jumear.github.io/stirfry/iNat_calendar_heatmap?verifiable=true. but then you can also see local differences. for example, Israeli weekends happen on Friday and Saturday: https://jumear.github.io/stirfry/iNat_calendar_heatmap?place_id=6815&verifiable=true, and even though Canada and Russia both have a lot of observations in summer, Russian observations still mostly happen on weekends, bioblitzes, and holidays: https://jumear.github.io/stirfry/iNat_calendar_heatmap?place_id=7161&verifiable=true, whereas Canadian observations seem to happen during weekdays, too: https://jumear.github.io/stirfry/iNat_calendar_heatmap?place_id=6712&verifiable=true.

7

You can also see that huge range across RG within the CNC projects - Graz and San Antonio flying high above the average. With some projects still waiting for ANY identifiers to notice them.
Would be interesting to see the CNC project list sorted by RG.

Consider the 10 most observed sp for your local CNC project - then consider WHY those 10 ? Eye candy ? Cute ? They are everywhere ! Vehement observer bias.

For Cape Town this year our top 10 includes 4 birds. Guineafowl, Egyptian goose, coot, and the hadeda ibis in my garden each day. The everywhere plant is tall yellow daisy bush (planted 3 in my garden). The other 5 are Protea repens (big flowers), Pelargonium capitatum (pink flowers), Chironia baccifera (bright pink flowers and red berries) and Leucadendron salignum (sunshine bush). With an unexpected tenth the Portuguese millipede.

Turning my curiosity to the umbrella project’s top 10. Again birds but mallard and rock dove. Followed by honey bee. And 7 plants I learnt to recognise in Ostrava led by dandelion and lawn daisies. The global emphasis is Northern hemisphere. Which is why we get a second turn with GSB.