8 



ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM 



has a mandate to guarantee long-term archival storage and easy 

 access. Several such institutions that presently house larval fish- 

 es or are willing to do so are the Zoological Museum of the 

 University of Copenhagen, which maintains the extensive 

 worldwide collections taken during the Dana Expeditions, as 

 well as ones documenting the earlier classical studies on larval 

 fishes by Johannes Schmidt and his students, the Harvard Mu- 

 seum of Comparative Zoology, the Smithsonian Institution, and 

 the Natural History Museum of Los Angeles County. If collec- 

 tions of ELH stages are to realize their full potential in system- 

 atics, then it is timely for ichthyoplankton specialists to offer 

 good developmental series, especially illustrated ones, and for 

 museum curators to accept them. 



Fossils have been studied for clues to the major classification 

 of fishes since the days of Louis Agassiz (Patterson, 1981a) and 

 to the extent that they were available have been widely consid- 

 ered as important adjuncts or indeed prerequisites to compre- 

 hending the phylogeny of particular groups. Although this view 

 is now receiving heavy criticism (Patterson, 1981b), the fact 

 remains that it did exist for many years and may have detracted 

 from the potential contribution of the non-fossil suites of char- 

 acters carried by early life history stages. Even so, students of 

 fossils and of larvae share a preoccupation with the caudal fin 

 skeleton, a structure that is often well preserved in fossils and 

 can be studied in two dimensions and which, during the course 

 of ontogeny, exposes a wealth of information of great value to 

 the systematist. 



Because adult stages have been the chief source of characters 

 used in fish systematics, a perception has arisen that these char- 

 acters are in some way more useful or more indicative of a 

 phylogenetic classification than are the characters of early life 

 history stages. How did such a view arise? For many years, 

 systematists tended to concentrate on the search for conserva- 

 tive, "non-adaptive" characters (labeled the Darwin Principle 

 by Mayr, 1969). They discarded not only ones that they believed 

 were directly affected by the environment but also ones that 

 appeared to smack of convergence. It seemed reasonable and 

 proper, for example, to group together for phylogenetic purposes 

 fishes with one spine and five soft rays in the pelvic fin because 

 the character was apparently conservative, non-adaptive, and 

 non-convergent. On the other hand, it seemed wrong to group 

 together all fishes with canine teeth because the character was 

 apparently non-conservative, adaptive, and surely convergent. 

 With regard to larval fishes, Moser (1981) recently discussed 

 the occurrence of a large number of apparently highly adaptive 

 larval characters distributed across a broad taxonomic spec- 

 trum. He states, "Marine teleost larvae have evolved an enor- 

 mous array of morphological specializations, such that it seems 

 to me we are looking at a distinct evolutionary domain quite 

 separate from that of the adults. It is reasonable to assume that 

 these remarkable structural specializations are adaptive and re- 

 flect each species' solution to the challenge of survival in a 

 complex and demanding environment." My point here is that 

 if a systematist rejected adaptive characters (and many did), 

 then he would have been unlikely to use ELH stages, and this 

 may be another reason why they have not received sufficient 

 attention. 



How Systematists Do Their Work 



Even if systematists agreed among themselves about their 

 immediate goals and how best to achieve them, the task of this 



Symposium would be daunting. But contemporary systematists 

 do not agree on either objectives or methodology. The concepts 

 that purport to link systematics to phylogeny are being actively 

 reassessed, and it is within the context of rapidly changing ideas 

 in systematics that our presentations and discussions will occur. 



There are basically three conceptual methods now being used 

 by systematists, and although the bare bones of these methods 

 are easily comprehended, in practice they become more complex 

 and their independence from each other less clear. The interested 

 reader who is as yet unaware of the intense debate both between 

 and within the several schools of systematic classification is 

 referred to the pages of the journal Syslonatic Zoology for many 

 articles and references as well as ones cited in this section. A 

 recent description and comparison of the three methods is given 

 by Mayr (1981), who lists many important references. Although 

 1 do not propose to use very much space here on a redundant 

 treatment, 1 will briefly describe each method and comment on 

 its strengths and weaknesses. 



The theoretically simplest method (or methods— there is more 

 than one algorithm, and there is disagreement on which is best) 

 is called phenetics or numerical taxonomy and is described in 

 detail by Sokal and Sneath (1963) and Sneath and Sokal (1973). 

 It is based on overall similarity. Many unweighted characters 

 are used to generate clusters of OTUs (operational taxonomic 

 units), which may be anything from individuals, populations, 

 or species to orders, classes, or phyla. The hierarchically ar- 

 ranged clusters, which lack a time dimension, are called phe- 

 nograms. Neither homology nor the fossil record are considered 

 in selecting characters. Each member of a cluster bears a closer 

 resemblance, although not necessarily genealogical relationship, 

 to other members of its cluster than it does to members of other 

 clusters. Some pheneticists claim that if a sufficient number of 

 characters is analyzed, any influence of convergence becomes 

 dampened and the phenogram will express phylogenetic rela- 

 tionships. Unfortunately, there seems to be no good way to 

 ascertain how many characters are needed. Other pheneticists 

 do not ascribe phylogenetic significance to their clusters and 

 merely claim to be representing overall similarity. Replicability 

 of results is the chief objective. Many classifications that purport 

 to be based on the methods of cladistics or evolutionary clas- 

 sification, upon close scrutiny appear to be basically phenetic. 

 There are apparently few fish classifications using ELH char- 

 acters, which are explicitly based on phenetic methods. One 

 example is a paper on Northeast Pacific cottid genera (Rich- 

 ardson, 1981a) which, according to the author, was not entirely 

 satisfactory for phyletic purposes. Ichthyologists who restrict 

 their data sources for a phenetic analysis to a single life history 

 stage should consider a study by Michener (1977), who gener- 

 ated four different phenetic classifications of a group of bees 

 based on different life history stages or character suites. 



A second method is called cladistics or phylogenetic system- 

 atics, and although it has been more or less on the scene for 

 many years, it is only since the revision and translation into 

 English of its original presentation (Hennig, 1950, 1966) that it 

 has gained wide currency and is now used, either explicitly or 

 implicitly, by many systematic ichthyologists all around the 

 world but particularly in North America and western Europe. 

 A recent guide to the method is a book by Wiley (1981), and 

 the reader is advised to consult also Brundin ( 1 966) for a notably 

 lucid interpretation. Cladistics requires a stringent evaluation 

 of characters. Primitive or generalized ones (called plesiomor- 



