Observe locally, identify globally?
Given the success of national platforms like Artportalen.se, Naturgucker.de, and Waarneming.nl (all of which generate more observations than iNaturalist), it’s fair to ask what the benefits are of platforms like iNaturalist that aim to be more globally representative. iNaturalist has a North American bias, but it’s unique in that around 40% of its observations come from outside of its principal country (in this case the US).
Are there any benefits to platforms with global communities like iNaturalist that would allow them to compete with local platforms and compensate for the costs associated with less cohesive community cultures and identities? A huge part of the educational and scientific value of platforms like iNaturalist come from the quality and volume of identifications on observations. Identifications from the community allow novices to participate while simultaneously vetting data quality. If the geographic distribution of identifications is different from the geographic distribution of observations, then global platforms may better facilitate collaboration between observers and identifiers towards getting observations identified.
The animation below shows a sample of identifications on iNaturalist from a 24-hour period. The lines originate at the location of the identifier (approximated as the centroid of their observations) to the location of the observation they identified. Note that while many identifications originate from the same region as the observations (e.g. the United States, New Zealand, or South Africa), many identifications also originate from users in very different parts of the globe than the observations. If all identification expertise was geographically focused we'd expect to see only short lines, but because identification expertise has a different distribution we see longer lines.
There’s plenty of anecdotal reasons to think that identifications might be less geographically localized than observations. Aside from a handful of world travelers, most observers seem to be fairly geographically focused. In contrast, many identifiers appear to be geographically indiscriminate. Many of these identifiers are taxonomic specialists with broad geographic expertise. For example, the figure below shows the locations of a recent sample of 100 observations identified by each of @wongun in South Korea, @susanhewitt in New York, @jwidness in Connecticut, and @borisb in Germany within their respective specialty taxa.
One interesting study system to explore the impact of a global community of identifiers further is to compare the City Nature Challenge (CNC) 2019 bioblitzes from Ecuador. Of the two cities from Ecuador that competed in CNC 2019, Tena used the global iNaturalist platform for their bioblitz while Quito used a separate fork of the iNaturalist software localized for Spain called Natusfera. The Natusfera community is completely separate from the iNaturalist community and is almost entirely Spanish. Nonetheless, both Bioblitzes produced about the same number of observations (~37k) from the same number of observers (~1.2k). It’s also convenient that Natusfera and iNaturalist are otherwise nearly identical (since they are running roughly the same software).
We took a sample of 10,000 observations with photos from each Bioblitz and calculated the percentage of these observations that had an identification made by someone other than the observer. In the Natusfera Quito Bioblitz there were many fewer identifications per observation (0.1) compared with the iNaturalist Tena Bioblitz (2.0). And in Quito, only 8% of the identifications came from outside of Ecuador (the orange lines connect the origin of the identifications to the observations) while in Tena 37% came from outside of Ecuador.
In this analysis, there were many more identifications contributed to observations on the global iNaturalist platform compared with the local Natusfera platform. There are certainly several factors that account for this discrepancy. One is likely the hard work and local expertise contributed by local identifiers which included members of the INABIO staff such as @inclandj who sponsored the Tena event. Likewise the size of the Ecuadorian community on iNaturalist (~200k observations) is larger than the Ecuadorian Natusfera community (~40,000 observations).
But if identifications had identical geographic distributions as observations, we’d expect similar, small proportions identifications coming from outside of Ecuador on both Natusfera and iNaturalist. The fact that percentage of identifications coming from outside of Ecuador is nearly 4 times larger on iNaturalist supports the hypothesis laid out above that identifications have a less geographically focused distribution than observations.
If this pattern is general, then the ability for global communities like iNaturalist to lend identification expertise from around the globe to observations from any particular locale gives them a clear benefit over local communities that can only draw on local identification expertise. But whether these benefits can compensate for the costs associated with moderating global communities and their many disparate cultures, languages, and identities remains to be seen. The iNaturalist Network is an attempt to maintain these benefits of a global community while mitigating the issues associated with moderating a global platform by collaborating with local institutions to localize iNaturalist at the national level.
Comments
That's exactly the reason why a global platform like iNat should exist. Human has country boundaries but animals and plants don't. The regional observation/identification procedure will create more bias towards clusters of distribution following the geopolitical borders.
Glad I could find myself on that gif map; nobody else really uses iNat out here in the Palouse!
It's a small world after all. A few of the plants I knew from the Appalachians also grow in Iceland. Some plants native to South Korea grow well in South Carolina. Sometimes I will pick a country I've never been to just to see what kind of plants and animals live there. Occasionally I recognize a favorite ornamental. Fun!!
The benefits of international versus local must surely be predicated by two major factors:
The level of endemism.
The Cape Flora contains 20% of the Flora of Africa (i.e. 1 in 5 plant species in Africa occurs in the Cape Floral Kingdom), which is 0.6% of the area of the continent. That is not as significant as the level of endemism: 67% of the species occur only in the CFK (and it approaches 90% only occur in southern Africa).
As a consequence, the global community is useless for CFK plant identifications. Almost all our identifications and identifiers are local. We would not expect it to be otherwise.
This contrasts with our macrofungi, most of which are alien invasive brought in association with European and American trees. It would be nice to have the statistics (it is not that easy to extract the hometown or of users on iNat in bulk!), but I would bet that the bulk of our identifications come from the northern hemisphere (it did on iSpot, but the UK observers did not migrate, so those links are lost). Here the international community are priceless, not only making identifications, but also in pointing out where our national checklists are out of date with recent advances in taxonomy.
Similarly, I would predict that birds and mammal IDs would be well balanced internationally, but that lizard and frog IDs would be predominantly local, based mainly on the degree of endemism in these groups.
The home base of experts.
iNat does not credit specialist taxonomists with any recognition (iSpot did, and a reputation too), but it soon becomes obvious when there are experts in a group in other countries.
@borisb (Germany) is a good example for southern African beetles (although he ranks only #3 behind two locals, and is not in the top 10 for Scarab Beetles).
@damionjp (France) and @mattmatt (London - both home bases from iSpot : could not find their bases on iNaturalist) contribute immensely to the IDs of our southern African Pelargoniums (ranking 1 and 2, and significantly above the local in position 3).
@jurga_li (Lithuania) overwhelms southern African Lichen IDs, possibly because there are no active lichenologists in the subcontinent, and without her contribution over 95% of our lichen observations would still be at Class rank.
Similarly amongst succulent groups there are experts in Europe who dominate the identifications in some taxa (and seem to know South Africa better than the locals too).
If this sounds contradictory, it is not. Although the two Pelargonium experts account for a large number of the IDs in Pelargonium, the vast majority of regular identifiers are local.
It should work the other way too. The Cape has contributed immensely to horticulture, with Gladiolus and "Geranium" significant examples. And yet locals focus on "pure" species and I would venture almost none (but this is speculation) explore the northern hemisphere hybrids and garden observations. Certainly, for Proteaceae (non-Australian) I am happy to help out with identifications, but the task is a nightmare of weird cultivar selections and intermediate hybrid crosses (with a significant number of 4-way crosses in Leucospermum cultivars), and one soon discovers how important location (location location! location!!) is when making an ID: outside of their home ranges, ID takes significantly more effort, and the uncertainty is far greater.
Only the best resourced countries will have national experts in most of their plant, fungal and animal groups. For most of the world there are significant European and American experts who contribute disproportionately to identification. For this interaction a global website is essential. Second and Third World countries will probably benefit most from an international site.
I think this topic deserves more than just a blog. It should be explored further and be the seed of a proper analyses and evaluation in a top scientific journal.
Nice little analysis. In the second paragraph, do you mean the following: "If the geographic distribution of identifications is different from the geographic distribution of observations..."?
Interesting analysis!
Another way global identifiers contribute is by doing a high-level categorization of the "Unknown" observations to put them in front of subject matter experts. I don't know which Asteraceae (sunflowers, etc) or insects live in Africa, but I (usually) know a daisy or a beetle when I see it.
That is another interesting twist.
Are the generalist (high-level) identifiers more or less localized or international than the fine- (or final-) level identifiers?
For the Cape Town City Nature Challenge, we had non-Capetonians contributing to the general IDs, but for plants most of the fine-level IDs were done by locals (but outsider IDs to species level were really appreciated!!). It would be nice to put figures to these generalizations.
Specifically, did very high level IDs of plants (e.g. Plant, Dicot, Monocot) help at all: few specialists work above family level and are these courser identifications available to them? Is an intermediate-tier identifier (one working to family level) crucial to the process of getting (rapid for the CNC) species-level IDs?
Thanks @crellow - typo fixed!
I think you're right on @tonyrebelo with the importance of endemism and the level of resources across countries. Also agree that this would be neat to explore further in journal article format
As an example of this phenomenon, I live in California, US, but try to ID as many as possible of the observations arising from Ladakh, India. There doesn't seem to be any local ID'ers, and I have spent the past two summers there, spending a good part of my time studying the local flora. Between my study and a Flora of Ladakh published by a group of Poles, I'm able to ID most of what pops up if the photos are decent. The other occasional plant identifier is in Europe, and there are some Indians from elsewhere that occasionally identify as well.
Love this analysis!
Another feature of the global IDs is that remote ID'ers may be in better position to spot a new invasive than are locals (serendipitous example of Ludwigia peruviana in Australia https://www.inaturalist.org/observations/17952516).
@ tonyrebelo - regarding the endemism argument, where I work in SE Asia it's often the international experts who have their main base outside the region that are the only people who an accurately ID the endemic species, not locals.
This is super good! Thank you for sharing with us! A bit more background when trying to push people towards iNat :-)
I love this post and I look forward to seeing the more in-depth assessment in a paper!
6920 observations (17.87%) from the Tena CNC were identified to RG status is there a way to see the proportions of Quito observations which were IDed to RG?
What would not be as useful for Tena is to have a lot of international folk telling them they have plants in the Amazon. This is not the case as identifications to Plants or Dicots were mostly done by locals.
How do the pie charts work for the 'IDs per Observation'?
I think another advantage of iNaturatist is that travellers are not always aware of local networks or identifiers so to have a site like iNaturatist is great. I have used the site both in preparing to travel to get some idea of what is to be found in the area I plan to visit and then to add my observations.
Interesting
Very interesting take on things. Lately, I have been focusing more and more on international IDs as I get more confident/learn more about birds. Currently focusing on Central and South American birds, as a lot of those sneak by without getting research grade (and a surprising number of mis-IDs get research grade).
Where do you get this number from ' but it’s unique in that 60% of its observations come from outside of its principal country (in this case the US).
I get US = 15.5 mln records https://www.inaturalist.org/observations?place_id=1&subview=table
Outside US = 11.6 mln records https://www.inaturalist.org/observations?not_in_place=1&place_id=any&subview=table
That's not 60% from outside the US.
hm - good call I was going off of very rounded numbers 1-(10M/25M) from here. You're right that at the moment it looks like 11650115 / 27165013 = 43%. I changed it to 'around 40%'. Still so much US bias!
Ok, I thought maybe it was an imperial to metric conversion issue.
Hi @loarie, that's a neat analysis and interesting look at a major component of citizen science projects - the geographic extent and focus of their respective communities, and how this aspect translates into user engagement as well as resulting data quality.
From a user perspective, I'm completely sold to iNat's global platform for various reasons, but if you want to ask how iNat performs in terms of identifications vis-a-vis more local platform such as the Swedish, German and Dutch examples linked to in the intro, I think the proper analysis would be to compare a Swedish, German or Dutch data set taken from iNat, and compare it with the corresponding data set from these nationally focussed projects.
I have no doubts that the many experts sharing their expertise here on iNat are contributing to “islands of excellence“, that is where certain taxa receive very good coverage (both in terms of # of observations reviewed and IDs given) across the globe. The 4 users highlighted above are good examples.
However, there are pretty vast seas in between these islands, both geographically and taxonomically, where “ID coverage“ is pretty low. This certainly applies to many megadiverse, developing countries (and I'd argue that South Africa is megadiverse, but highly developed in the science sector).
I'm not very familiar with the 3 European projects but would suspect that they've bound a lot of local or regional expertise, which is often lacking here on iNat for the respective regions. I would further expect that the comparison made for the 2 cities in Ecuador would look pretty different if applied to comparable data sets in iNat vs those in the respective European projects (or from any other developed region for that matter).
It pretty much boils down to the question that expertise is usually limited, and how to spread this limited resource in an efficient and effective way. iNat has taken the global avenue and is serving its community in a state-of-the-art way. I believe that every effort to serve its diverse members, cultures and regions is worth it, and I'm curious to see how aspects of this post develop with exponentially growing observations.
@loarie - Have you considered setting up language-specific iNat Forums so that non-English communities can more easily communicate with each other? Might be worth testing out to see if folks would use it or not.
@zygy I've spearheaded the Forum. I've definitely considered adding language or region specific categories, but I would like to have a native speaker in that language who can serve as a moderator before moving ahead with it.
Really good article, I was always saying that Natusfera was an enormous CSIC error bcs of that. I like very much how u described this fact in this great article. Thank you very much for sharing this information @loarie because it is an important step for the good manner of make science.
Add a Comment