Lumping and splitting

by Gary Rosenberg
Authors are often characterized as lumpers or splitters. From one perspective, splitters recognize more species than really exist, and lumpers go around fixing the resulting damage. From another perspective, lumpers are indiscriminating bunglers incapable of appreciating the subtleties of nature. But it's not so simple. Some authors have introduced many names now sunken in synonymy yet have unjustly synonymized many names of previous workers. Others rarely name species and rarely synonymize them. Such types err respectively by jumping to or shying from conclusions, but they are neither splitters nor lumpers.

As scientific methods progress, one would expect the track records of authors to get better. I tried to evaluate this by taking those authors who have introduced 20 or more names for Western Atlantic gastropod species and determining the percentage of their names currently considered valid *. The resulting graph shows that track records have been steadily improving, from 30% accuracy around 1800, to almost 80% accuracy today. The overall average is about 52%: 8402 names, 4323 of them valid. (The points are plotted at the average of the years in which each author's species were named.) In the upper left hand corner is Linnaeus, at 96%. In the upper right are Dell, Ponder, and Warén at 100%, and Quinn and Jong & Coomans at 98%. Bottom of the barrel are Locard (4%) and Verkrüzen (5%).

By modern standards, Linnaeus was a lumper, many of his species being composite by modern standards; his high average merely reflects the advantage of going first. The averages for Dell, Ponder, Warén and Quinn will probably hold up since they publish thoroughly researched monographs. These workers are neither splitters nor lumpers. Jong & Coomans' work hasn't stood the test of time. Their book was published in 1988, and gave rather brief descriptions of a number of micromollusks from an understudied fauna. Whether they were splitters cannot yet be judged. Locard and Verkrüzen were clearly splitters, although many of the names they introduced were at the varietal level.

Who were the farthest from the norms of their day? The first American conchologist, Thomas Say (71%) did remarkably well without access to most of the European literature. Nowell-Usticke (22%) did poorly considering that he had access to major East coast collections and libraries. What about some of our favorite authors? Abbott 67%, C. B. Adams 54%, Dall 73%, E. H. Vokes 84%, Gmelin 55%, Lamarck 32%, Melvill 84%, Petuch 71%, Pilsbry 43%, Reeve 38%, Rehder 50%, Röding 14%, Watson 85%, d'Orbigny 61%. These numbers will surely change, especially as revisions of small, deep-sea, and recently named species proceed.

Assuming that species really do exist in nature, which is the greater sin, recognizing too many of them or too few? In practical terms, the difference between lumpers and splitters is that you can be relatively sure that you know what splitters are talking about, even if you don't agree with their classifications. Splitters are precise, even if they are not accurate. A lumper is neither precise nor accurate. It's much easier to recombine something that has been oversplit, than to tease apart what has been erroneously lumped together. This becomes apparent when you try to make a computerized database of names. It's easy to sum the geographic range of a species over all of its synonyms. It's hard to determine the geographic range of a species if several other species have been confused with it. Therefore, I encourage authors to include a statement such as "all the specimens I examined from Isla Utopia were variety x, which I considered to be a synonym of y." If in retrospect they were wrong in synonymizing the variety, their error could be easily corrected. (Even more desirable is deposition of specimens in public museums, to allow later authors to verify identifications.)

As computerized databases of names (and images of types) of mollusks are developed, knowing what has previously been discovered will become easier. And with the techniques available to modern biology, there is not really much excuse to be either a lumper or a splitter. Multivariate statistical analysis of shell morphology, anatomical studies and molecular genetics can solve most species level questions rigorously. Many questions argued unproductively for years can be solved in the laboratory in less than two weeks, given live collected specimens or frozen tissue to work with.

The technique is relatively simple: allozyme electrophoresis, which involves separating common metabolic enzymes by their electric charge. (There are similar DNA based techniques). If two taxa live together and do not interbreed, their enzymes will usually show fixed differences. A single fixed difference is sufficient to prove that they are not the same species. This can be a particularly effective technique for showing genetic similarity in species where the morphology of the individual can change in response to the environment. A good example of this kind of change is Turbo cornutus from Japan. Specimens living on wave-pounded coasts develop strong spines; specimens from sheltered areas are smooth. The animal apparently takes its cue from the seaweed rather than the surf; different species of seaweed grow on the rough and sheltered coasts. In the laboratory, one can manipulate spine development depending on the type of seaweed the animal is fed.

Proving that two species that live together are the same is harder than proving that they are different -- maybe the differences haven't been found yet. If two taxa live apart, proving that they are different species is more complicated -- maybe there is a gradient of intermediates in the geographic area between them -- although sometimes the genetic difference is great enough to be compelling evidence. But there are many cases where taxa live side-by-side, and no one has ever done the work to answer the question. Are Cassis madagascarensis and Cassis spinella one species or two? How about Phalium granulatum and Phalium cicatricosum? The problem of course is expense: collecting the material and paying for supplies and salaries for two weeks in the lab can easily total several thousand dollars. Anyone want to fund a study of his favorite species?

* Rosenberg, G. 1995. Malacolog Version 2.0. A database of Western Atlantic gastropods. Accessed via Internet, URL Currently contains more than 9800 records for Western Atlantic gastropod species and synonyms.

This article was published in the June 1996 issue of American Conchologist.
Dr. Gary Rosenberg is Associate Curator of Mollusks at the Academy of Natural Sciences of Philadelphia, Director of Grants for Conchologists of America, a member of The Lambis Group of COA, and Editorial Board member for American Conchologist. His column, "Conchatenations," appears regularly in American Conchologist. He can be reached at: Academy of Natural Sciences, 1900 Benjamin Franklin Parkway, Philadelphia, PA 19103-1195