', {"anonymize_ip":true});

Limits of using geography for species identification

A recent observation highlighted the limitation of using geography to determine species on iNaturalist. If you take a close look at the photograph in https://www.inaturalist.org/observations/41662271 you may note that animal exhibits the characteristics of a mule deer rather than a white-tailed deer. However, the location makes it nearly inconceivable that it is a mule deer.

This is a conundrum that scientists often run into when using citizen science data. What can be inferred from a given observation, and how do you ensure the accuracy of it? It is generally more accurate to base identifications on field marks (physical characteristics) than location. This is the essence of birding and why people get so excited about rare birds showing up in odd locations. If people just ignored field marks and just considered geography, most of these rare birds would never be documented. But why do we ignore this principle with mammals?

In the deer example, as noted by @sambiology, the habitat in the photo doesn't fit north Dallas. This makes it most likely it is a mule deer and the location is simply wrong. But what happens when we know the location is accurate? Should we look at field marks or just consider geography? Take a look at 2 other observations and see what you think.

1) https://www.inaturalist.org/observations/18767407
In this case, we have what appears to be an American marten (with known location). But American martens are endangered in Wisconsin, and are unknown to occur in this area (before this observation). The field marks suggest American marten. The geography suggests it is actually a mink or fisher or some other mustelid or maybe a squirrel. Based on this observation, @lincolndurey extended the range map for American martens. But what if we had just labeled this as a squirrel and moved on without considering the field marks?

2) https://www.inaturalist.org/observations/20912284
Similar to #1, here we have what appears to be a fisher (with known location). But fishers are endangered in California and the population at the southern edge of the Coast Range is considered basically non-existent (before this observation). Specifically, the last fisher sighting this far south was made by Joseph Grinnel nearly a century ago. The field marks suggest this is a fisher. The geography suggests it is a mink or squirrel. Based on this observation, fisher researchers have extended the known range of fishers in the Coast Range of California. But what if we had just labeled this as a squirrel and moved on without considering the field marks?

Posted by maxallen maxallen, April 17, 2020 16:32


Yep -- a conundrum no doubt. I know that when I worked in the herbarium, there's a level of trust that you put into the collectors of the specimens. For instance, I can only assume that a collector is honest when he/she makes a specimen, but the reality is that it could indeed be from his/her backyard or from a nursery or whatever.

I guess it further emphasizes that when dealing with citizen science data, an investigator needs to 'vet' the data, especially the anomalies. One of the beautiful things about iNaturalist is that there are a lot of eyes watching the data come in too. Errors are caught... eventually. :)

Posted by sambiology over 1 year ago (Flag)

I'm a generalist, I'm interested in being able to ID mammals, plants, insects, slime molds, etc. and this colors my opinion on this topic. If I specialized in just one thing with a fairly limited number of species (birds, for example), I know I would really try very hard to get it down to species.

In the two years I've been active on iNaturalist, I have learned so much that I can no longer identify some things to species with any certainty, whereas I would not have hesitated before such as: Evening Primroses, Dandelions, White-faced Meadowhawks, and most Sunfish to give just a few examples. These things have beat into me that going to genus is Just Fine.

I've also seen a lot of ID's going to subspecies based solely on geography. For birds especially, this seems silly. Over the past 6 months or so, I have decided that unless you there are morphological characteristics that distinguish them (Eurasian vs. American Green-winged Teal, and Eastern vs. Western Small Milkweed Bug as examples), I'm going to species, even if others have already put a subspecies ID.

So I guess that was a long way to say: Keep it at genus :-)

Posted by apgarm over 1 year ago (Flag)

This is a balance and comes up in 'professional' ecology too. For instance... if i encounter a pine tree in the woods of Vermont, I do not consider all ~111 pine species when making an ID, to do so would be absurd. On the other hand I don't add the subspecies Pinus strobus strobus, because the only other subspecise is an isolated population in Mexico thousands of miles away, and to do so seems absurd. In general I think geography is an important tool for species level identification and above, but I don't think it's usually a good idea to go to subspecies based on location alone because subspecies are by definition pretty tricky and or nondistinct. But i'm not a fan of adding subspecies IDs to iNat in general unless they are very clearly different in ways apparent in the observation photo/sound. I don't think there is any official rule though.

Posted by charlie over 1 year ago (Flag)

This happens a lot in tracking too. There are plenty of misidentified animal tracks and scats on iNat. Even after a tracker explains why the morphology of the track indicates a particular species, some observers still insist that the track is what they claimed. Often, this is one of the more "glamorous" species, like mountain lion or bear. It seems there is a desire to identify tracks as being from large, charismatic species rather than the more mundane domestic dog. Most of the tracks I am sent via email to identify are dog tracks that the photographer claims belong to a mountain lion. Trackers learn to rely on the morphology of the track and disregard the wishful thinking that wants to turn that dog track into something more "cool." This phenomenon leads to plenty of tracks on iNat that are not what they claim to be. And, non-trackers then just agree with the ID of the original observer, but not because they know what the track is. I think it's too easy to click "Agree" even on things that you shouldn't be agreeing with. This leads to a phenomenon where others chime in and confirm the incorrect ID, without any knowledge of track morphology.

Posted by beartracker over 1 year ago (Flag)


Here's another fisher out of known range. I caught this one with a trail camera. It was pretty exciting to see it.


Posted by beartracker over 1 year ago (Flag)

I think the strength of citizen science is its statistical power. Massive numbers of data points drown out the outliers. When drilling down to the individual observation, clearly field marks will trump geography. When a flock of a hundred Cedar Waxwings pass overhead, its certainly possible that among them is one Bohemian Waxwing. However, the data is good enough to identify trends that concern Cedar Waxwings.

Posted by scottmarnoy about 1 year ago (Flag)

Add a Comment

Sign In or Sign Up to add comments

Gracias al apoyo de:

¿Quiere apoyarnos? Pregúntenos cómo escribiendo a snib.guatemala@gmail.com