BEHAVIOR AND PHYLOGENY OF OF THE COLISA AND THE FAMILY BELONTIIDAE

by

RUDOLPH J. MILLER') and AMBROSE JEARLD2) (Department of Zoology, Oklahoma State University, Stillwater, Oklahoma, U.S.A.) (With2 Figures) (Acc.25-VI-1982)

Introduction

In an earlier paper on the behavior and phylogeny of trichogasterine fishes (MILLER & RoBISON, 1974) we presented arguments supporting the use of behavioral characteristics in assessing phylogenetic relation- ships among . Though we agreed with ATZ (1970) that care must be taken in establishing "homologies of behavior", we concluded that behaviortaxonomy studies often prove useful at the family or genus group level (MAYR, 1958; CULLEN, 1959; ALEXANDER, 1962; WICKLER, 1967; HINDE, 1970). We were able to show that the four in the genus fall rather easily into two groups, based on numerous trench- ant behavioral characteristics. We also have been studying the behavior of the four species currently recognized in the genus Colisa (RESER, 1969; JANZOw, 1971; JEARLD, 1975). The purpose of this paper is to describe the reproductive behaviors of the Colisa spp., examine this information for insights on phylogenetic relationships of the group, and integrate this data with the earlier material on Trichogaster. Since LIEM (1963) placed both genera at the peak of adaptive radiation in the family Belontiidae, such a comparison may be of some value in assessing strengths and weaknesses of the comparative behavior technique. The are a group of over 50 species of perciform fishes in- habiting tropical and subtropical regions of Southeast Asia, , and Central Africa. The most recent revision of the group was done by LIEM (1963) who ordered the 15 genera into four families on the basis of

1) We are grateful to Drs H. C. MILLER,P. PARKER,T. CARTER,and J. SHAWfor reviewing the manuscript. The studies leading to this paper were supported by grants from the National Science Foundation (BMS 74-24197). 2) Present address: National Marine Fisheries Service, Northeast Fisheries Center, Woods Hole, MA 02543, U.S.A. 156

osteological characters. Colisa chuna (honey ), C. lalia (), C. labiosa (thick lip gourami), and C. fasciata () along with the Trichogaster spp. are considered by LIEM to be the most highly specialized forms in the family Belontiidae and subfamily Trichogasterinae. LIEM (1963) recognized several evolutionary trends such as increased depth and compression of the body, reduction of some bony elements (e.g. , 5 vs 6 branchiostegals) and reduction of dentition of the prevomer, palatine, and parasphenoid as indicative of increased specialization in the Trichogasterinae as a group. Recently VIERKE (1975) has criticized LrEni's arrangement of the subfamilies and genera in the Belontiidae, using mainly behavioral criteria. Although the present paper does not deal specifically with the phylogeny of the Belontiidae, several points at issue will be discussed. Published studies on the behavior of Colisa spp. include those of FoRSELIUS (1957), PiccioLo (1964), SOLE (1966), VIERKE (1969, 1972, 1975), MILLER & MILLER (1970), and Rossi (1969). The present paper utilizes primarily studies conducted in our laboratory (RESER, 1969; MILLER & MILLER, 1970; JANZOW, 1971; and JEARLD, 1975). Detailed descriptions of courtship, agonistic, and reproductive behaviors can be obtained from several of the above sources, though FORSELIUS (1957) and JEARLD (1975) are the most comprehensive. We agree with BARLOW (1968) that the "Fixed Action Pattern," the presumed behavioral unit of many earlier studies, is, in fact, difficult to verify short of careful experimental analysis: thus we will continue to use the most stereotyped behaviors for our comparison, though some are unlikely to fulfill all criteria for being "FAP's". Perhaps the majority of the behavioral patterns discussed here better fit BARROW' (1968) concept of the "modal action pattern". Non-social activities are similar among the species of Colisa so the pres- ent comparison will focus largely on threat, courtship, nest building, and spawning activities. Such behaviors tend to be stereotyped both intra- and inter-individually; and so we think that the effects of captivity have relatively little influence on the parameters being used to assess in- terspecific variability. We recognize, however, that the frequencies of some courtship and agonistic activities may vary appreciably from those found in natural conditions (MYRBERG, 1970). ' Methods

This research was done in the Behavior Laboratory of Oklahoma State University Life Sciences West Building. Uniform physical conditions were maintained in the laboratory throughout the study.