Bigger is Not Better
The smartphone fallacy – when spatial data are reported at spatial scales finer than the organisms themselves (2018) Meiri, S., Frontiers of Biogeography, DOI: https://escholarship.org/uc/item/2n3349jg
One of the greatest annoyances when using museum specimens, old datasets, or large occurrence databases (such as GBIF) is when the locality of an occurrence is only vaguely described, and the coordinate uncertainty is high; “Norway” or “Indochina” doesn’t really tell you much about where that specific animal or plant was seen. Luckily, the days where such vague descriptions were the best you could get are long gone, as most of us now walk around with a GPS in our pockets, and even community science data can be reported very accurately, and more or less in real-time.
However, we have now encountered the opposite problem: the reported coordinates of organisms are often too precise to be realistic, and in the worst-case scenario, they might be borderline meaningless. The author of this study wanted to highlight how this advance in technology coupled with our eagerness to get more accurate data and results have made us too bold in our positional claims.
What They Did
This is an opinion paper written by a museum curator at the Steinhardt Museum of Natural History, Tel Aviv University. Based on examples from his experience and from the scientific literature, he reviews the new issues we might be facing during this age of technology. He shows the measured distances when using different levels of coordinate precision (i.e. how many decimals in a set of coordinates), and illustrates why too much accuracy in reported coordinates can be meaningless.
Did You Know?
Reporting coordinates can be a messy business – in short, this is due to the general issues with trying to translate the Earth’s (almost) round surface into planar coordinates. This is why we have several different ways of reporting coordinates (we talk about Coordinate Reference Systems (CRS) and projections); in short, they represent different ways of making this translation. Every way of doing it comes with pros and cons – no CRS and projection can convey both distance, direction, area and shape correctly at the same time. Different systems work better for certain areas and for different purposes.
What Did They Find Out?
Some journals seem to have their gold standard of reported decimals set at six – this is equivalent to an accuracy of 13.3 cm at the equator (even shorter as we move toward the poles). This is such a small distance that by standing up, sitting down, or moving your phone to have a better look at the screen would change the coordinates. If e.g. the animal you’re measuring is larger than this (and especially if it is mobile too), you would technically need to report whether you measured the position of the head, midpoint or the tail!
Similarly, our smartphones frequently report position with a precision down to 6-7 decimals. However, a reasonable estimate is that such a handheld GPS can only be precise down to the 4th decimal – everything beyond that is uncertain. Conversion between different coordinate reference systems can also occasionally lead to this kind of error, in some cases resulting in reporting coordinates down to 10 decimal places.
Not only is such “precision” false, it is useless – practically no studies need that kind of accuracy; ecologists are generally working on much larger spatial scales, and using too many decimal places might overly complicate analyses.
As this is a relatively short opinion paper, there is not much of a method to point out problems in. Indeed, the whole paper is about pointing out the problems in the general field of ecology and biogeography. However, as there was no formal study and thus no systematic investigation of the problem, it cannot be interpreted as more than that: an opinionated claim (regardless of how much one might agree with the author).
There are still plenty of issues with poorly reported coordinates in older datasets (only a locality name, or a vaguely described area, which might have changed beyond recognition) and wrong coordinates (mixing up longitude and latitude), and compared to these issues, having coordinates that are too precise is potentially a minor problem.
Regardless, this paper suggests that we need to stow away our constant longing for better, more precise data, and take it as a “nice exercise in humility” to not try and add precision, where it is unwarranted and/or unnecessary.