Wednesday, June 8, 2016

Thoughts on studies using taxonomic data


tl;dr version: Your results are mostly wrong. Unless someone has revised (in a taxonomic sense) the species you are using in your study.

Longer version: People make mistakes. When we describe a new species, or a group of new species, we sometimes place them in a taxonomic rank (let’s say genus for argument’s sake) we think is correct but we cannot be absolutely sure, unless we have a very thorough phylogenetic analysis. Phylogenetic analyses are awesome, but in the age of genomics they cost a lot of money. And in many cases (incorrectly in my view), building that phylogeny is beyond the interest of the person describing these taxa.

Now consider that the majority of species were described a long time ago, long before people were thinking about phylogenetic relationships. Also, some of the early (we are talking 19-early 20th century here) taxonomists were not specialist per se, and would describe species among many different families of insects, thus not really knowing where those species belong. So, in many cases, if during the last 40 years nobody has taxonomically checked (=revise) the species you are are using, chances are that these species are: (a) synonyms of another species; (b) placed in the wrong genus or (c) placed in the wrong higher rank.

Example: in 2004 I published with several colleagues a study on when (day or night) rove beetles were active on Barro Colorado Island, Panama. If you want to read the details, the paper is here, but to save you from a lot of trouble, I just set up a flight intercept trap and for a period of 12 days I was sampling at dawn and dusk. Here is part of table 1:



My 2004 self made two three mistakes there. What we thought was Dysanellus ended up being a new genus of rove beetles described as Zackfalinus and Dysanellus is restricted to the southern part of South America. Likewise, what we thought was Philothalpus ended up being Oligotergus, because nobody had looked up before how messed up the generic limits were in Philothalpus. UPDATE: Adam Brunke correctly reminded me that all specimens identified as Quedius were indeed Cyrtoquedius based on this paper.

What I am saying is this: if you are pulling data out of a digitized collection of GBIF, good luck. I hope a taxonomist was interested in the taxa you want to use.

Of course, this idea is not new. Meier and Dikow (2004) have said this much more eloquently.