P.S.Z.N. 1: Xlarinc Ecology. I1 (3): 263-275 (1990) Acccptcd: March 26, 1990 0 1990 Paul Parey Scicntific Publishers. Berlin and Hamburg ISSN 0173-9565
Observations on the Aggregative Behavior of Mysidium columbiae, the Mangrove Mysid
RICHARDF. MODLIN
Department of Biologiciil Scicnccs, ‘The University of Alabama in Huntsville. Huntsvillc, .Alab;imn 35899, U. S. A.
\Vith 3 figures
Kc? words: ,My.sitltrceu, My.sitliw~i col~rn1hc,shoal structure. diurnal dynamics. iii;in- groves. Cnribhcon Sca, light intensity.
Abstract. The mangrove mysicl My.sitfiurn cdirnihitre (Zihihiti<) occt~rsin aggrcgations of tens to millions of indiviclual>. Few indcpth studies on the nggrcgativc bchavior of this spccics havc cwiiiincd \tructiiral composition, I-csponsc to light intensity changes. and its ability to rctlucc predation. Snorkel arid SCUBA observations, as well ;is aquarium cxperimeiitation on shoals/ schools of this spccics in sh:111ow waters around a mangrove nrchipcl;igo were made at irrcgulnr intervals to dctcrminc their structurc. dynamics. and influcncc on predation. In stimnicr, aggrcga- tions occurred as large cylindrical shoals. in winter mall ovoid schools. In daylight. shoals/schools wcrc very sensitive to changes in light intensity. At night the smallest juveniles formed compact swarms. while largcr individuals bccainc solitary. At sunrise a dcfinitc rc-shoaling behavior cxistcd. Shoals/schools were stratified by life stages. A hypothesis stating that conspecific schools in winter did not mix was tested. Mysitliitrn shoalskchools sccrncd to discouragc predation by fishes.
Problem
The formation of aggregations is a common behavior of many pelagic (MAUCH- LINE, 1980; HARGREAVES,1985), deep water hyperbenthic (FOSSA,1985), and Coastal mysid spccies (CLUTTER, 1969; WITTMANN, 1976; O’BRIEN,1988). Attempts to interpret the role of this behavior have been made (CLUITER,1969; O’BRIEN,1988). Additionally, various physiological and environmental factors that influence the formation and maintenance of mysid aggregations have been investigated (STEVEN,1061; CLUTTER,1969; WITTMANN,1976; O’BRIEN,1988). These studies have provided schemes to classify the aggregations based on sociality (CLUTTER,1969), geometry (WI-ITMANN,1976), and intrinsic biological processes (ZELICKMAN,1974; O’BRIEN,1988). These schemes seem applicable to the swarming, schooling, or shoaling patterns of most mysid species. Mysidium columbiae (ZIMMER)is a ubiquitous mysid in the tropical and subtropical western Atlantic (MAUCHLINE,1980). It occurs almost exclusively in
U. S. Copyright Clearance Center Code Statcment: 0173-9565/90/1103-0263$02.50/O 264 MOIILIN large homotypic aggregations in the waters surrounding mangrove islands (SI'EVEN.1961; GootisuE, 1965; MODL~N,1987), but aggregations are also found near coral structurcs (EMERY,1968; MODLIN,1954, 1957) and just below the water surfacc in pclagic aggregations seaward of coral reefs (MODIJN,unpub- lished data). The only information on thc aggregative behavior of M. columbine is available in SEVEN(1961) and EMERY(1968). Mysiriium colunihiae aggregations range from tens to millions of individuals. Rarely are kl. colurnbiae individuals found solitary in the water column during daylight hours. The organization and response of these aggregations are very similar to those in fish species with obligatory schooling behavior (STEVEN, 1961). This report presents information obtained from field and laboratory observations on the structure, diurnal dynamics, and responses to light of M. coluinltitre aggregations in the western Caribbean Sea, specifically in the shoal waters in and around Twin Cays, Belize.
Material and Methods
In this papci-. aggregations :ire classified cithcr as shoals. school. or swill-nis. Definition?,of the tlircc Arc lll0d~i'i~~lti~)IlSOf tht>Sc USCd by WII-IhlANN (1976) Lllld O'!
of surfacc light. A QSI-140 lntcgrating Quantum Scalar lrmdiancc Mctcr was used to mcasurc undcrwatcr light intcnsitics dircctly during Junc 19HH. Surface watcr salinity was mcasurcd with an hydromctcr. Salinity at dcpth was dctcrmincd from sigma-t values of a watcr samplc collected in a plastic bottlc at thc dcsircd depth. Watcr tcmpcraturc w3s mcasurcd simultancously in situ with a mcrcury thcrrnomctcr.
M 250m South Point Fig. 1. 'Twin Cays mangrovc (TC) and Carric Bow Cay (CBC) barricr rccf systcms off central Bclize. Ccntral America. Thc Twin Bays study sitc in Twin Cays is stippled. Blackcncd circlcs at TC and CBC are additional collecting and obscrvation sites whcrc Mysidiurn colurnbitre was present. Open circles arc sites lacking M. colurnbiae. Stipplcd arrows indicatc pelagic M. columbiae shoals. Insets show thc relationship of the two study areas. thc location of CBC and TC in thc central portion of the Carribcan Barricr Kccf System, and thc location of thc barricr rccf in Central America. 266 MOVLIN Prior to observing diurnal aggrcgative behavior in the field, two to three schools of M.colurnbiae were simultaneously maintained in plcxiglas, rectangular 301 aquaria (0.25 in wide x 0.5 m long x 0.3 m deep). Aquaria werc placed on a rectangular table designed to temporarily $.rap ovcrtlow aquarium water (wet table); water depth on this table was about 0.15m. About 100 to 150 individuals comprised an aquarium school and :ill within an aquarium were collcctcd from the same shoal at Twin Cays. These schools were held for a maximum of 30 h and then released because thcy wcrc not fed while in captivity. Observations were niadc irregularly throughout thc day and night. Most intensive observations were madc during sunrise and sunset. Aquarium specimen? were used to tcst interactions between conspccific schools. Spccimcns from two widely separated. non-communicating M.colurnbiae schools were placed in the sainc aquarium. However, the individuals from one school wcre stained with neutral red dye (How~~ \ I \ 1015 1 2 3 L 5 6 7 8 9 10 11 12 13 1L 15 16 17 18 19 20 2'1- 22 23 2L Hours Fig. 2. Light intensity under the mangrove canopy (black dots) during the first week of June, 1988. along the shoreline of the Twin Rays study site of Twin Cays, Belize. The intensity curve (solid line) was estimated from linear rcgression analyses of SC~Sof light measurements obtained during sunrise. when sun was near or at its zenith. and at sunset. Arrows indicate time and light intensity at sunrise and sunset. Aggrcgation in Mysitfiurn columhinr 267 largest separated by a north-south channel with an average width and depth of 75m and 2Sm, respectively (Fig. I). Additional descriptive information on this mangrove community can be found in RU'rzLm & FELLER(1%7/88). hlysicfium colum6iue aggregations were observed in many areas in and around Twin Cays (Fig. 1). However, the most extensive observations and collcctions were made along the southern pcrimetcr of an indentation on the south shore of the wcstcrnmost island (Fig. I). Although this area is not in the strict sense a bay, it was called Twin Bays. It is shallow (maximum depth 2.5 m) with soft silty sediments. The substrate is covered in June with an extensive growth of turtle grass, Tllulussia resrudinum. Along the studied perimeter the bank inclines steeply from about the midtide level to a depth of 1.5 m. This depth occurs at a horizontal distance of about 1 .O to 2.5 m from the top of the bank. At high title, water overflows the bank and inundates the shoreline mangrove community. The tidal range is about 0.3 m. Above the w:iter the perimeter of the study area is canopicd by overhanging branches of the red mangrove, R1iizophora mungle. This mangrove canopy extends over open water for one to two meters. Beneath the canopy the underwater environment is vertically dissected by the trees' proproots. somc of which extend into the bottom substrates and others which remain suspended. Seaward of thc perimeter enormous schools of red-carrcd sardincs. Harcngulu humerrrlis, scnlc sardines. /-I. jtrguuna, dusky anchovies, Ancliorr lyolepis, and round herrings. Jenkinsia sp. fill the hay during daylight hours. Landward of thc inysid shoals arc swarms of the cyclopoid copcpod Oitllotlrl (Ir)ioit/lorlrr) oculatrr. Salinity and temperature in thc bay averaged 37.9% and 31.9"C in June and 36% and 24.S'C in Dccembcr. Horizontal and vertical variations in salinity and temperature were ncgligiblc. Light intensity during June in unshadcd areas at midday was 3.6 X 10'" quanta. ern-?. s-I (Fig. 2). During December. light intensity was of the samc order of magnitude. but daylength was contracted by about one hour (J. AbliJi,tii~,pers. comm.). 'Thc density of turtle grass and macroalgae was considcr:ibly reduced during this winter intcrval. Results At Twin Cays, M. columbine was found in shoals, schools, and occasionally swarms. During June, M. colurnhim shoals predominated and were found in most Twin Cays sites (Fig. 1). In the Twin Bays study area the population was comprised of two to four shoals located adjacent to the shoreline, These shoals were nearly continuous along the southern perimeter of the Twin Bays study site. In horizontal aspect the shoals were somewhat cylindrical, with the upper surface slightly flattened and about 15 to 20cm below the water surface. The lower boundary was about 0.5m above the substrate. These shoals varied in diameter from about 0.3 m in the vicinity of submerged mangrove roots to 1.2 m in more open water locations. Few schools were observed in isolated waters and between clumps of mangrove roots. The latter usually occurred shoreward of the shoal. The schools seemed to be small aggregations that had broken from the shoal. Swarms were observed only after the shoal or school was naturally or artificially disturbed; they occurred either separate from the parent shoal or as clusters within a shoal. M. colurnbiae individuals were clustered into schools rather than shoals during December. These schools varied in size from about 0.2m to 0.7m in diameter and about 0.5m to 2.0m in length; shape varied from spherical to fusiform. Volumetrically, the schools ranged from 0.03 to about 1.5 m3. They occurred at depths of 10 to 30cm below the water surface, with their lower boundaries higher in the water column than in June. In the Twin Bays study site, 20 to 30 schools were located next to the shoreline. Here they concentrated in 268 MOVLIN regions illuminated by beams of light. Schools were found in most sunlit regions along the shorelines of Twin Cays in December. They were absent in areas where light intensity was < 60 % of the surface value. Adjacent to patch coral formations behind the barrier reef at Carrie Bow Cay (Fig. l), M. columhiae schools occurred throughout the year near to, 01: contigu- ous with, schools of M. integrurn. These schools were not numerous; they were homotypic and < 0.5 m in diameter or length. Such schools seemed to prefer crevices, depressions, and undercuts located on the leeward sides of coral formations. M. itzregrum schools were higher in the water column and preferred water currents. No M. colurnbiue schools or shoals were observed in benthic locations seaward of the barrier reef (Fig. 1). However, large pelagic shoals were seen and sampled 100 m to 200 m seaward of the south point of the Carrie Bow Cay barrier reef (Fig. 1). These discoid shoals were about 20cm below the surface: they were about 0.3 m thick and about 2.0 m in diameter. All individ- uals were oriented into the current (Fig. 1). The pelagic shoals were larger in June than in December. Enormous M. columbiae shoals were observed along the bases of pinnacle rcefs off the west shore of Grand Cayman Island (reef base about 17 m). The shoals extended vertically along the reef wall from just above the subsi:rate to a depth of about 13 m. They extcndcd horizontally away from thc reefs approxi- mately 1 .0 to 2.Om and from 6.0 to 10.0m along the reef's length. These shoals were composed of Inrgc juvcnile and adult niysids. From ;I distance they I-cscmblcd rcsuspcndcd clouds or sedimcnt. Normally, at these depths water currents ;ire ncgligible. Howcvcr, Mysirliriin shoals were able to reorganize rapidly cvcn after being subjected to extremely strong turbulence. On 15 Scptcmbcr 1988, two days aftcr I-Iurricane Gilbert passed over Grand Cayman Island, the mysid shoals were well organized and in the same locations as prior to the hurricnne. Translocation of large amounts of bottom material from the pinnaclc reef base indicated that the wind generated an enormous amount of turbulence at 17m. Individuals comprisiiig the shoals and schools at Twin Cays were always oriented in thc same direction except when disturbed. Spacing betwcen indi- viduals - the inter-individual distance - ranged from < 0.5 cm to 2.0ci-n during daylight hours. This distance remained uniform but varied directly with mysid size and inversely with light intensity. Shoals contained all life stages organized vertically by size, with the smallest on top and the largest on the bottom (Fig. 3A). Life stages were stratified into roughly four layers: juveniles, immature males and females, mature males and females, and gravid females with the largest males. Stratification was very stable. The ap&xmch of a diver did not disrupt stability; instead, the shoal moved away as a unit. If the approaching object was a fish, a constant distance of about 0.5m was maintained between the boundary of the shoal and the intruder. If the fish moved closer the boundary nearest the fish seemed to fold inward to maintain this constant distance. Disruption occurred when an object swam or was thrust into the shoal. After such perturbation the shoal reorganized in approximately the same location in less than 30 seconds. Perturbation causes the shoals to fragment into swarms. These quickly organize into small schools, each school being composed of a particular life Aggregation in My.dium colunibiue 269 stage or age class. The individuals comprising these schools remain as strong cohesive units and each moves in a defined direction away from the perturbation (Fig. 3 F). Fig. 3. Schematic rcprcscntation of the behavior of Mysidium columbiue shoals and schools relativc to changes in light intensity and prcsencc of predators under thc mangrove canopy at Twin Cays, Belizc, during thc month of June: A. full sunlight, light intcnsity = I > 10l6 quanta .cm-2.scc-I; B. cloud cover, I = 1Ols; C. during the night, I < lot3;D. just before sunrise, I = 2x 10'3; E. about 30min after sunrise. I = 6x loL3;and F. when prcdatory fish (thick white arrow) penctrate shoal/ school boundaries. Thin black arrows indicatc dispcrsal of Mysidium life stages to avoid predators. 270 MODLIN None of the M. columbiue shoals sampled or observed were composed of only a single life stage. In contrast, schools usually contained individuals of a single life stage, although larger schools did contain several different stages. In the latter, these stages were stratified. A school or swarm composed of it single life stage was usually located at a depth roughly corresponding to its level when it was part of a shoal, i. e., sniallest juveniles located just below the surface, gravid females near the bottom, immature females and males at mid-depths. During daylight hours, M. columbiue individuals were photopositive (Fig.3A). However, the degree of sensitivity to light appeared to be size- dcpendcnt: the smallest juveniles showed the greatest phototaxis, the largest individuals were less sensitive. After dark, individuals were generally photo- negative to white light. Although the intensity of the underwater light varied even those intlivicliials farthest from the light source swam away. Only 1 I % of the 334 mysicis collected in light-traps were M. columbiae. Interestingly, the fewest M. colritnhiue specimens were taken when thc light-traps (n = 3) were placccl in the Twin Bay study site where the abundance of this species was the greatest. My.sidiiim colunzbim did riot respond to red light: they swarn neither away from nor toward the flashlight fitted with a red filter. An inverse relationship existed between compactness of the M. columbiuc shoals and light intensity. At high light intensity, shoals were very co,npact and well defined; inter-individual distance was reduced to about 0.5 to 1.0cm (Fig. 3 A). The responsc to light intensity is very sensitive. Under low light intensities (cloudy days and two ruin events) the swarms were loosely organized (Fig. 3 B); inter-individual distances liere varied from 2.0 to 5.0 crn. Shoals rcmaincd polarized. On partly cloucly days, sho:ils expanded and contracted as clouds passed. Shoals also moved under the mangrove leaf canopy when a cloud shaded the study area. With reestablishment of full light intensity, the shoal moved into the more open water at the margin of the mangrove canopy. Responses to light were more subtle in December because of lower underwater light intensities and fewer individuals than in June. Shoals and schools were smaller in December and spent more time in the canopied regions, emulating the summer cloudy day response (Fig. 3R). At sunset, when light intensity decreased rapidly (Fig. 2), inter-individual distance in M. columbrtrc increased; in some cases schools or shoals could no longer be distinguished. Small, very compact swarms of the smallest juveniles (1.4 to 2.8 mm) remained intact throughout the night. These swarms aggregated within 2-3 cm of mangrove roots, which were heavily encrusted with biogenic growth (Fig. 3 C). They remained about 10 to 20cm below the water surface. Larger juveniles and some adults were observed swimming in very loosely organized schools in the water column between the root systems (Fig. 3 C). Inter-individual distances ranged from about 10 to 30 cm. Gravid females swam just above the bottom or next to the bank of the cay (Fig. 3 C). They showed no tendency toward aggregative behavior. When the sky began to lighten, about 20 min before sunrise, M. columbiae shoals were defined. However, stratifica- tion was not apparent, Gravid females and mature males were intermixed with juveniles at the top of the shoal (Fig. 3 D). The schools and swarms of smallest juveniles remained next to the mangrove roots and did not move into the shoal until shortly after sunrise (Fig. 3 E). Stratification was established abmt 15 min Aggregation in Mysidiurn colurnbiue 27 I after sunrise. The respective life stages were generally segregated in the shoal at their particular depth level (Fig. 3 A). No seasonal d rences in the diurnal dynamics of the shoal were observed. The density of individuals within a shoal showed a significant day/night change (P < 0.20). Shoals during daylight hours contained an average of 24,215 indivs. . m-3 (standard error of the mean (s. e.) = 2 7030 indivs. . m-3, n = 10), while those sampled during the night averaged 11,559 indivs:m-' (s.e. = k 2386, n = 10). A significance level of 20 % was chosen because of the semiquantitative nature of the samples. Aquarium studics supported the observations on in situ diurnal dynamics of shoals. The aquarium school remained intact and stratified during the day; individuals dispersed after dark and schools were reorganized at sunrise. In the morning the school concentratcd in the sunbeams passing through the aquarium. At other times during the day, when direct sunlight did not enter the laboratory, the school expanded horizontally but remained in a layer that extended from just below the surface to a depth of about 6.0cm. Individuals that escaped from the aquarium through the overtlow tubing remained in the water contained on the wet table. Thesc congregated near the aquarium sidcs and remained contiguous with the school confined inside. Field observations in December suggested that conspecific schools of M. colurnbiue remained segregated and did not coalesce with adjacent schools. A hypothesis stating that conspecific schools did not mix was tested in an aquarium and falsified. Individuals from the two experimental schools mixccl within less than 2 rnin after the aquarium partition was pulled, even though one swarm was heavily stained with neutral rcd dyc. Direct fish prcdation on M. columhkie shoals at Twin Rays was not observed. Fish such as bluehead wrasses (Thulussorna sp.) and small snappers (Lutjcnus spp.) attacked only when a shoal or school was artifically disrupted. Individual niysids in the open water were preycd upon. During such a disturbance most individuals swam to the bottom or to the bank of the cay (Fig. 3 F). The smallest juveniles moved into the very shallow water between the mangrove roots atop the bank. Direct predation on a Mysidium shoal was observed at the base of the pinnacle reefs off Grand Cayman Island. A group of margates, Haemulon album, suspended under an overhanging ledge, periodically plucked individual mysids from the inner edge of a shoal adjacent to the reef base. Discussion The work of CLUTTER(1967, 1969), WITTMANN(1976), and O'BKIEN(1988) indicated that mysids closely associated with the benthos prefer specific sub- strates near which they aggregate, while holoplanktonic species are not sub- strate-specific in their clustering behavior. Mysidium spp. are predominantly holoplanktonic (EMERY,1968; MAUCHLINE,1980). Of the three species common to the western Atlantic only M. gracile associates with specific substrates. It prefers to cluster around the urchin Dindemu antillarum and nests of damsel- fishes Eupomacentrus sp. (EMERY,1968; HAHN& ITZKKOWITZ,1986). During daylight hours, M. colurnbiae and M. inregrum (MODLIN,unpublished data) did 272 MOIILIN not appear to favor a particular substrate. Mysidiiim columhiae did, however, seem to prefer shoaling in quiet, protected waters. Unlike its congenitor M. integnrm, which is typically found around coral reefs, M. cofumbiac schools are usually observed in areas of coral reef communities protected from strong currents (EMERY,1968). The underwater area beneath the mangrove canopy is characterized by a confusion of light beams and shadows. The resulting irregular light intensity patterns may strongly influence the structure and integrity of M. columhiae shoals. Holoplanktonic mysids have been shown to use visual cues as primary stimuli in the maintenance of their shoals and schools (Ct.ulTrn, 1969; MAC- QU ART-MOU 121N, 1973; O’BIUEN, 1988). M. coliitnhitrc shoals in Twin Bays were very sensitive to changer, in light intensity (Fig. 3A-E). Shoal integrity began to break down as light intensity decreased. During December, when the aggregations were small, shadows may have functioned as barriers effective enough to keep individual M. c,diimhiae schools separated. O’BIIIEN(1988) found that when the light reg’lime was uneven, niysids concentrated in areas of greatest intensity. Observations made at Twin Rays and in aquaria supported this contention. Even when disturbed, i n d i vi d u a 1 s w a r in s avoided d a I’ k sh nded regions be tie a t h t he mangrovc: ca tiop y . Thcsc swarms quickly reorganized within the same bright area the parent school originally occupied. Consequently, the assumption that conspecific sc hools did not mix was generated, tested, and I’alsified. In June, M. colunibiae individuals were so numerous that the shoals were nearly continuous around the study site. Since the sun was almost dircctly overhead, the narrowcr angles of i ncidcnce produced stronger undcrw:iter light intensities than in December (GAl’ES, 1962; Wirrm., 1975). Thc shoals were more compact and extended seaward of the mangrove canopy in regions of greatest intensity. In the shaded portions of the shoal the inter-individual distance greatly increased, but shoal coliesivcness was in a i n t a i n ecl . . Shoals and schools of M. coliirnhiae are visually stimulated, social organiza- tions that contain all life stages (Fig. 3A). The aggregations at Twin Cays are unique becausc of their highly ordered structure; life stages or age classes are distinctly stratified during daylight hours and remain so cven when mildly perturbated. Stratification is even maintained within swarms or schools that break from the parent shoal. Previous studies have not reported mysid aggrega- tions to be strongly ordered. Instead these aggregations are composed! of either admixtures of life stages (CLUTTER,1969) or their composition is liniited to a single life stage (O’BRIEN,1988). Cohesion of M. cofnmbiue shoals is maintained by visual cues (Srevm, 1961; CI.U.ITER,1969; MACQUART-MOULIN, 1973; O’BRIEN,1988). Field (Fig. 3 A-E) and aquarium observations supported this contention. Cohesion is strongest during daylight hours. However, as light intensity decreases, shoals bcxome less cohesive, but reorganize as light intensities increase. Similar responses were observed in other mysid species that supposedly use visual cues to maintain their aggregations (CLUTTEK,1969; ZELICKMAN,1974; WITTMANN,1976). In their paper on heterotypic shoals, MCFARLAND& KOTCHIAN(1982) reported that M. columhiae individuals completely dispersed after dark, leaving the incorpor- ated school of postlarval french grunts, Haemulon flavolinetrtLlm, exposed. The Aggregation in Mysicliurn columhircc 273 school of grunts remained intact and in contact with the bottom throughout the night. Just before sunrise the Mysirfiiim shoals reorganized (as shown in Fig. 3 D) and reintegrated with the postlarval fish. Predator avoidance exhibited by shoals and schools of M. colrrmbim in the Twin Cays and Carrie Bow Cay study areas seem to be 1) a combination of two responses described by O’BRIEN& RITZ(I988), i. e., splitting and flash expan- sion, and 2) a coordinated avoidance response. If the predator, real or artificial, moved to within 0.5 m of the shoal, that group of mysids closest to the predator moved in concert to maintain some critical distance. This group did not split from the shoal, but the response produced a depression or furrow in the shoal. If the potential predator movcd away the shoal quickly returned to its original shape. When a predator approached too close to, or actually penetrated the shoal, the shoal split very rapidly into smaller compact schools and swarms which. except those of the smallest juveniles, quickly moved downward (Fig. 3 F). Division of the shoal and expansion of the individual swarms and schools away from the predator appeared to be a coordinated response. Under these conditions mysids did not appear to scatter as individuals as EMERY(1968) described. The schools coalesced rapidly after the predator left the area. These responses were not observed in the deep waters off Grand Cayman Island. Here M. coliirnhinc shoals did not show any response to the approach of predators or diver. Only penetration and rapid agitation disrupted shoal integrity. Lower light levels here may have impaired visual cues. Predation on M. coirrmhiue shoals has not been quantitatively examined. However, MCFARIAND& KOTCHIAN (1982) showed tli:it postlarval t’rench grunts that mingled in M. columbiae shoals did occasionally feed on the mysicls. Various species of grunts foraged in the study area during the night, but they were never observed to feed on M. columbiae. More voracious predators on solitary mysids may be caridean shrimps (KNEIB,1988) that move from Thrrlris- sir1 beds into the study area after dark. Aggregations of different mysid species differ in origin, form, and function (MAUCHLINE,1980). EHRLICHSr EHRIJCH(1973) proposed the features which schooling behavior provides fishes. These essentially parallel those suggested for mysids, i. e., increased reproductive and feeding efficiency, sociality, and protection from predators (CLUTTER,1969; WITTMANN,1976; O’BRIEN,1988). In iM. coliimbiae shoals and schools sociality and protection seem to be of greatest importance. Aquarium experiments, especially on the behavior of those individ- uals that escaped from the aquarium, strongly supported the evidence for gregarious behavior of this mysid. In the study area solitary M. columbine were very rare. Shoals and schools were rarely preyed upon even though potential predatory fishes were present: it occurred only when a shoal or school was disrupted artificially. Since all life stages are found in M. columbine shoals the advantage to reproductive success is a natural corollary of aggregative behavior. Feeding benefits of the shoaling behavior were not observed. More field and laboratory experiments will be done on future trips to obtain additional information, using for example stained individuals. A quantitative paper describing the M. columbine life cycle and population dynamics at Twin Cays is forthcoming. 274 MOUI-IN Summary I. Shoalskhools of M. colirmbiac are most common in waters around nian- grove islands. 2. Summer aggregations of M. columhiae occur as cylindrical shoals > 2.0 m in length and 0.5, or larger in diameter, while in winter they occur as ovoid schools < 2.0 m in length or diameter. 3. Individuals that comprise schools/shoals were oriented in the same dircc- tion, with inter-individual distances ranging from < 0.5 cm to 2.0c:m during claylight hours. 4. Shoals contain a11 life stages and are vertically stratified by size with the smallest juveniles on top and mature maledgravid females on the bottom. 5. During daylight the inter-individual distance increases and deszreases as light intensity, respectively, decreases and increases. 6. Conspccific schools do not mix during winter if separated by a shaded area. 7. As long as the school/shoal remains intact, predation by fishes is greatly reduced or non-existent. Acknowledgements I gr:~tcl'ully:ickriowlc:llgc llic collahor;iliori ancl !icld iissist;incc provided by FRANKFIXI<.\IU+ PAI I~ICIA A. I~l~i~ic~s,and Jui IP. AMIII.I~R:I apprcciLitc their cfforts ancl contributions. Additionally, DK. FI:I~IL\IUmadc niany uscl'ul comments that grciitly iinprovcd this manuscript. I would also likc to th;ink C. 1.nvtii.i Shiiiii and .IIMTYi.L!ic for allowing nic to rcmovc thc mysids from their light-trap and ichthyoplmkton samplcs. and for providing mc much useful knowledge on thc schoolirig of fislics. I cspcci;illy likc tu tli;iiik Die. KI.ALJSIIU.ULEK foi- giving me the opportunity to work at tlic Sniithsonian Institution's Field Station on Carrie Bow Cay. Belize. Funding was provickd by a Mini- Grant from the l References CI.UITPR.R. I., 1967: Zonation of ncorshorc mysicls. Ecology. 48: 200-208. --, 1969: Thc microdistribution and social bchavior of some pelagic mysid shrimps. J. Exp. Mar. Biol. Ecol., 3: 125-155. Eiiicr irii. P. R. B A. [-I. EHI~LICH.1973: Coevolution: Hctcrotypic schooling in Caribhcan reef fishes. Am. Nat.. 107: 157-160. EMERY.A. R., 1968: Preliminary ohscrvations on cord reef plankton. Limnol. Occanogr.. 13: 293-303. FOSSA, J. H., 1985: Near-bottom vcrtic;iI zonation during claytimc of dccp-living hyperbenthic iiiy