32 



ONTOGENY AND SYSTEMATICS OF FISHES -AHLSTROM SYMPOSIUM 



sented. Novel sorts of characters or ways of manipulating data 

 are sometimes needed to identify larvae and the data required 

 may not be retrievable from "standard" descriptive accounts. 

 Synthesis and elimination is the normal procedure used by tax- 

 onomists to identify adult fishes. It has been called the "look- 

 alike" system when applied to larval fishes (Leiby, 1981). It is 

 basically a simple procedure but the pitfalls are numerous and 

 subtle. As with some early adult fish taxonomy, premature syn- 

 thesis may often be based on the wrong characters (e.g. con- 

 vergent characters) and lead to spurious identifications. 



General references on larval fish identification include Berry 

 and Richards (1973), Ahlstrom and Moser (1976) and Moser 

 (1981). Some recent works which provide exposure to a wide 

 range of larval forms and literature are Ahlstrom and Moser 

 (1981) and Fahay (1983) for marine taxa, and Auer (1982) and 

 Balon (1975a, 1981a) for freshwater taxa. 



The purpose of the following is to describe the tools— pref- 

 erably sharpened, polished and comfortable to use— which should 

 be at hand when the ichthyologist sits down to identify larval 

 fishes. Our emphasis is on three main factors: 1 ) the larval fish — 

 its anatomy, ontogeny, and phyletic relationships; 2) the study 

 area— its ecology and zoogeography and 3) the investigator— 

 his experience, knowledge and ingenuity. 



Systematics, Ontogeny and Anatomy 



Perhaps the most important type of character for identifica- 

 tion of larvae is meristic, as counts usually do not increase or 

 decrease once established. All meristic characters can be im- 

 portant, but vertebra/myomere counts and fin element counts 

 are of particular value. Meristic variables are useful at different 

 taxonomic levels, e.g., principal caudal fin ray and pelvic fin 

 element counts at the family or order level, median fin elements 

 at the genus/species level, pectoral fin ray counts at the species 

 level. Frequency distributions of meristic counts are extremely 

 important (particularly when it is uncertain whether develop- 

 ment of a character is complete) but often are not given in 

 published literature. Some important characters may not be 

 included in published studies (e.g., pectoral fin rays, procurrent 

 caudal rays). Differences in methodology and variable attention 

 to detail may also affect the quality of published meristic data. 

 Thus, published studies must be treated with caution and one 

 must be prepared to collect and compile one's own information 

 when opportunities arise. Despite potential problems with pub- 

 lished works, these are the obvious place to start with compi- 

 lations. Few "regional" meristic publications as exemplified by 

 Miller and Jorgensen (1973) exist, but many publications on 

 larval fishes include extensive tabulations of meristic infor- 

 mation. 



Various ways exist for facilitating use of meristic compila- 

 tions. A simple taxonomic listing (e.g.. Miller and Jorgensen, 

 1973) can be time-consuming to use, while a "gazetteer" format, 

 with species arrayed in order of counts (e.g., Fahay, 1983) may 

 be more practical. X-Y plots of two meristic variables (e.g.. 

 Berry, 1959b) can include frequency distributions and be very 

 useful for separating closely-related forms. 



A second suite of characters of broad use is specialized larval 

 characters which may characterize whole groups. These include 

 but are not limited to: characteristic shapes (e.g., Anguilli- 

 formes/Elopiformes, Pleuronectiformes), spination (Acanthur- 

 idae, Holocentridae), fin development patterns (argentinoids), 

 fin element development (Pleuronectiformes, epinepheline Ser- 



ranidae), fin placement (pelvic fin placement in Pleuronecti- 

 formes), eye shape (myctophid subfamilies, salmoniform 

 groups), and phoiophore development pattern (Gonostomati- 

 dae). The elucidation of such characters is a focus of this volume, 

 and reference should be made to specific chapters for further 

 detail. The important point is that a broad knowledge of larval 

 fishes is frequently necessary for accurate, efficient identification 

 of larvae. 



Finally, identification of larvae depends on a suite of dynamic 

 characters (pigmentation, body form, spination, fin develop- 

 ment pattern, etc.), which may change rapidly and differentially 

 over a small size range. Generally, a combination of such char- 

 acters is required for accurate identification; this is particularly 

 true in early stages. These characters can vary extensively, even 

 within a species, due to regional differences; method, time or 

 area of collection; preservation method or duration. Develop- 

 mental changes can be extremely rapid (e.g., changes in mela- 

 nophore distribution from some yolk-sac to post-yolk-sac lar- 

 vae). Again, no extensive treatment of these characters is possible 

 here, but the important point is that detailed, disciplined ob- 

 servations of larvae are essential for accurate identification. 



The importance of osteological characters for larval identi- 

 fication is increasingly recognized (Dunn, this volume). Use of 

 these depends on clearing and staining techniques (PotthofT, this 

 volume) or X-ray techniques (Tucker and Laroche, this vol- 

 ume). As with meristics, osteological characters may be useful 

 at different taxonomic levels. Caudal osteology has been widely 

 used because of its early development and relative simplicity, 

 but cranial osteology and pterygiophore patterns are also useful. 

 Recent application of cartilage-staining techniques has permit- 

 ted use of cartilaginous structures in identifying larvae (e.g., 

 Fritzsche and Johnson, 1980). Other internal characters such as 

 gut shape (Ahlstrom and Moser, 1976; Govoni, 1980) may also 

 be useful. 



Keys have not generally been used in larval fish identification 

 because of the dynamic nature of characters (a separate key 

 would be required for each size class or development stage) and 

 because of "incompleteness" of information (i.e., it has usually 

 been impossible to completely cover a defined region or sys- 

 tematic group with a key). Generally, much more information 

 is required to identify a larva than an adult, and summarizing 

 this in a key has been impractical (the information-organizing 

 capacity of computers may eventually help to permit this). Ex- 

 ceptions, such as Bertelsen's (1951) key to larval Ceratioidea, 

 Johnson's ( 1 974b) key to genera of larval scopelarchids, and the 

 key of Bertelsen et al. (1976) to notosudids do exist. 



Because of the complexity of identification of larvae, a wide 

 ichthyological background is important. A good knowledge of 

 fish anatomy is essential, particularly when (as often occurs) 

 damaged specimens must be identified. Published descnptions 

 exist, for example, which interpret broken branchiostegal rays 

 as jugular pelvic fin rays. A general knowledge of suspected 

 phylogenies and inter-relationships (e.g.. Greenwood et al., 1966; 

 Nelson, 1976) is essential if attempting to identify by synthesis 

 or elimination. This should at least cover those groups to be 

 expected in a given area, but wider knowledge is desirable, par- 

 ticularly in the marine environment where exotic larvae may 

 be transported great distances (e.g., Markle et al., 1 980). Finally, 

 thorough familiarity with the ontogenetic continuum is neces- 

 sary to place unknown specimens in perspective. Absorption of 

 the yolk sac, flexion of the notochord in the caudal region, 

 development of median fins, and transformation from larval to 



