UN> "^RS»TY OF AT URBANACHAMPAIQN NATURAL HIST. SURVEY I '4 NATKRAl HISTORY SURVEY M •3> ' '^'-'^"^'''^ ,jy,jQ6 1985 Zoology Published by Field Museum of Natural History LIBRARY \ FIELDIANA Zoology New Series, No. 14 SPATIAL AND TROPHIC INTERACTIONS IN WET AND DRY SEASONS BETWEEN GAM.BUS\A LUMA AND GAMBUSIA SEXRADIATA (PISCES: POECILIIDAE) IN BELIZE, CENTRAL AMERICA DAVID W. GREENFIELD CHESTER P. RAKOCINSKI TERESA A. GREENFIELD January 13, 1983 Publication 1340 SPATIAL AND TROPHIC INTERACTIONS IN WET AND DRY SEASONS BETWEEN GAMBUSIA LUMA AND GAMBUSIA SEXRADIATA (PISCES: POECILIIDAE) IN BELIZE, CENTRAL AMERICA FIELDIANA Zoology Published by Field Museum of Natural History New Series, No. 14 SPATIAL AND TROPHIC INTERACTIONS IN WET AND DRY SEASONS BETWEEN GAMBUSIA LUMA AND GAMBUSIA SEXRADIATA (PISCES: POECILIIDAE) IN BELIZE, CENTRAL AMERICA DAVID W. GREENFIELD Department of Biological Sciences Northern Illinois University, DeKalb Research Associate Field Museum of Natural History CHESTER F. RAKOCINSKI Department of Biological Sciences Northern Illinois University, DeKalb Present address: Department of Biology University of Southern Mississippi, Hattiesburg TERESA A. GREENFIELD Department of Biological Sciences Northern Illinois University, DeKalb Accepted for publication January 19, 1981 January 13, 1983 Publication 1340 Library of Congress Catalog Card No.: 82-83326 ISSN 0015-0754 PRINTED IN THE UNITED STATES OF AMERICA CONTENTS Abstract 1 Introduction 1 Materials and Methods 2 Results 4 Spatial Interactions in Sympatry 4 Trophic Interactions in Sympatry 5 AUopatric Cambusia sexradiata 9 AUopatric Cambusia luma 9 Discussion 11 Acknowledgments 16 Literature Cited 16 LIST OF ILLUSTRATIONS 1. Drainage map of Belize showing study sites 3 2. Composition of food of Cambusia luma at Spanish Creek during April 6 3. Composition of food of Cambusia sexradiata at Spanish Creek during April 7 4. Comparison of composition of food for Cambusia luma and G. sexradiata at Span- ish Creek during April 8 5. Composition of food of Cambusia luma at Spanish Creek during May 10 6. Composition of food of Cambusia sexradiata at Spanish Creek during May 11 7. Comparison of composition of food for Cambusia luma and G. sexradiata at Span- ish Creek during May 12 8. Composition of food of Cambusia sexradiata at the Rio Hondo during December 13 LIST OF TABLES 1. Food categories for Cambusia luma and G. sexradiata from Spanish Creek dur- ing April 5 2. Food categories for Cambusia sexradiata and G. luma from Spanish Creek dur- ing May 9 3. Food categories for Cambusia sexradiata from Rio Hondo during December 13 ABSTRACT Gambusia luma and G. sexradiata are sister species that are sympatric in the large rivers of northern Belize, but are also allopatric at various localities. The spatial and trophic interactions of these species were studied in both the wet and dry seasons at a locality where they occur together. Food for freshwater fishes is thought to be more abundant in the wet than in the dry season in the tropics, and thus, if lower food abundance results in more intense competition among fishes for food, the spatial and trophic partitioning between tropical river fishes should be greatest in the dry season. Field observations and feeding studies have demonstrated increased spatial separation (inshore-offshore) during the dry season for these two species when sympatric, and dietary dif- ferences support this separation: overlap coefficient (CX) changed from .90 to .76 between the wet and dry seasons. No evidence for niche expansion was found for an allopatric population of G. sexradiata; however, where G. luma is allopatric, it will utilize all habitats occupied by both species while in sympatry. Therefore, it is suggested, in sympatric situations, the presence of G. sexradiata close to shore in the vegetation may prevent G. luma from utilizing that micro- habitat during times of low water when potential competition for food may be greater. Although untested, the role of predation may be a factor in the ability of G. luma to live farther offshore than does G. sexradiata and in the failure of G. sexradiata to move offshore in the absence of G. luma. INTRODUCTION Northern Belize has distinct wet cind dry seasons, resulting in large seasonal fluctuations in fish habitats and presumably also in the type and amounts of available food resources. Zaret & Rand (1971) reviewed the literature and pre- sented their own evidence supporting the conclusion food was more abundant in the wet season in the tropics, as did Lowe-McConnell (1975). Assuming food is limiting in the dry season, competitive interactions should be greater, re- sulting in increased spatial and trophic partitioning. This corresponds with the findings of Zaret & Rand (1971) for a fish community in a small stream in Panama. Whereas Zaret & Rand (1971) compared various fish species belonging to different genera and families, our study compares two closely related (sister) species. Due to the similarities in morphology and general habitat utilization, potential competition should be greater between a pair of sister species than between species belonging to different genera or families (Dunham et al., 1979). Our study also differs from that of Zaret & Rand in that it compares the species in both allopatric and sympatric situations, thus providing information on 2 FIELDIANA: ZOOLOGY habitat utilization in wet and dry seasons for each species untiffected by poten- tial competitive interactions with its sister species. The closely related poeciliid fishes Gambusia luma and G. sexradiata each have distinctive distributions in Belize but are sympatric in several areas. Gambusia luma ranges from the southern border of Belize north to the Rio Hondo and New River drainages, but is most abundant in southern Belize where it is often the only species oi Gambusia present. Gambusia sexradiata occurs primarily in north- em Belize away from coastal areas; south of the Sibun River it is rare (it was collected at only four localities) and is generally replaced by G. luma (Greenfield et al., 1982). Both species, however, are broadly sympatric in the large rivers of northern Belize. MATERIALS AND METHODS For stomach analysis, samples of both species were taken from Spanish Creek at Rancho Dolores (Belize District) during high water on April 1 and during low water on May 1, 1976. As this was the period of the most rapidly dropping water levels, it therefore provided samples from wet and dry periods separated by the least amount of time. In addition, an allopatric population of G. sexradiata was sampled on December 19, 1975 (wet season) from the Rio Hondo at San Antonio (Orangewalk District) (fig. 1). All samples were taken between 0800 and 1100 hours, the predetermined period of peak feeding. The April 1 Spanish Creek sample consisted of 29 G. luma (22.7-30.6 mm standard length [SL]) and 27 G. sexradiata (19.7-30.1 mm SL); the May sample consisted of 25 G. luma (20.6-31.6 mm SL) and 36 G. sexradiata (15.0-29.0 mm SL). The December 19 Rio Hondo sample consisted of 45 G. sexradiata (15.2-30.8 mm SL). All individuals had food in their digestive tracts. Observations on behavior and spatial preferences of sympatric G. luma and G. sexradiata were made on eight occasions (total of 14 hours) between March and June 1976 and on one day (three hours) in March 1977 at Rancho Dolores. (This locality cannot be reached prior to March due to extensive flooding of the dirt roads during the wet season.) Gambusia luma and G. sexradiata are easily sepa- rated in the field due to the visibility (from shore) of black falcate anal fin on G. sexradiata. Observations of allopatric G. sexradiata were made on two days (eight hours) in December 1975 at the Rio Hondo. Allopatric G. luma were observed for two days (12 hours) in December 1975 and one day (two hours) in March 1976 at North Stann Creek at the Melinda Forestry Station. At Rancho Dolores, Spanish Creek is relatively wide (48 m) and deep, and the shoreline is covered with heavy terrestrial vegetation that is completely sub- merged during periods of high water. The Rio Hondo at San Antonio is also wide (80 m) and deep, with a shoreline similar to that at Spanish Creek. North Stann Creek at the Melinda Forestry Station is smaller than either Spanish Creek or the Rio Hondo (15 m), shallower, and with a stronger current. The shoreline varies from sand bars lacking vegetation to areas with heavy terrestrial vegeta- tion. Aquatic vegetation is lacking at all three localities. Gambusia specimens were collected with a 3-m seine. The body cavities were injected with 10% formalin immediately upon capture and the specimens pre- served in 10% formalin; they were later washed in water and transferred to 40% isopropyl alcohol. Standard length and sex were recorded for each specimen. The digestive tract was severed at the esophagus and anus and teased from the Fig. 1. Drainage map of Belize showing localities of study sites. 3 4 RELDIANA: ZOOLOGY body cavity. All contents were removed from the digestive tract and the lining scraped, hydrated with a few drops of isopropyl alcohol in a petri dish, and the food items sorted and identified into taxonomic categories using a dissecting microscope. Depending upon the numerical importance of the food items, identifications were made to various levels ranging from family to order. Food category volume determinations were made using a squash-plate grid tech- nique modified by Ross (1974, 1977) from Hellawell & Abel (1971). Although food items were not actually squashed, approximations were based on a mea- sure of the calibrated two-dimensional area covered (volumes previously de- termined for squashed and unsquashed samples revealed that, due to the small size of the food items, the differences between the two methods were negligi- ble). The number of food items in each category was counted. Graphs were prepared of the number of prey taxa found in the digestive tracts vs. the cumulative number of Gambusia individuals (in the order in which they were examined) (Pontius & Parker, 1973); in all cases the curves became hori- zontal, indicating the sample sizes for the feeding study are adequate for de- scription of prey taxa. The overlap coefficient (CX) of Horn (1966) was calculated for diet between the different species as: CX = 2S (Xi)(YJ The proportional values used for this coefficient were calulated following Zaret & Rand (1971), where the proportional contribution (% volume) of each prey kind was determined for each stomach before summing and averaging for each prey category. The CX varies from 0, when two species have no food categories in common, to 1, when the proportions of each food category are identical; s = total number of food categories, Xj is the proportion of the total diet of species X, Y, is the proportion of the total diet of species Y, for food category i. The relative importance of each food category was determined using the index of relative importance (IRI) (Pinkas, 1971): IRI = F (N -I- V), where N = numerical percentage, V = volumetric percentage, and F = frequency of occurrence percentage. This value integrates the measurements of numbers, volume, and frequency of occurrence usually used in evaluating fish stomach contents. In presenting proportions of particular food categories, the values refer to the percentage of total IRl for that category. To clarify feeding dif- ferences, prey organisms were designated, where possible, as allochthonous vs. autochthonous, aquatic vs. terrestrial, surface vs. submerged and/or shoreline, weedy vs. open, flowing. When making comparisons, qualitative category de- scriptions were considered independently, although strong correlations could exist between them. RESULTS Spatial Interactions in Sympathy During f>eriods of high water (March-April), when much of the terrestrial vegetation along the shoreline was submerged, G. sexradiata and G. luma oc- GREENFIELD ET AL. : GAMBUSIA INTERACTIONS 5 curred in the same general area at Spanish Creek and could be seen swimming together (i.e., they were syntopic); however, G. sexradiata was most abundant close to the shore in the vegetation (61% G. sexradiata, 39% G. lutna). In May, when the water level was at its lowest just prior to the rains that would end the dry season, G. lunta became more difficult to capture, and habitat differences became more pronounced. Gambusia luma occurred mostly in the open waters and, when pursued, moved out toward the middle of the river over deeper water. Gambusia sexradiata, however, occurred near the terrestrial vegetation close to shore and when pursued moved back into the vegetation; seine hauls made exclusively in the vegetation yielded 100% individuals of G. sexradiata, whereas seine hauls in open water yielded 100% individuals of G. luma. There was thus increased spatial segregation during periods of low water. During periods of high water, specimens of both G. sexradiata and G. luma were col- lected for feeding studies from all areas; during periods of low water, G. sexradiata was collected from areas with vegetation, and G. luma, from open water. Trophic Interactions in Sympatry Differences in spatial preference between these two closely related species are reflected in their feeding habits. For April's high-water period, Horn's CX gave a value of .90 (almost complete diet overlap); the diet of both species consisted mainly of insects, including various beetles, true bugs, flies, ants, and wasps (figs. 2-4; table 1). The two main resources, Hydraenidae and bryozoan floato- blasts, made up 59.4% and 69.3% of the total IRI values for G. luma and G. sexradiata, respectively. These prey are both autochthonous and constituted the greatest difference between the species: Gambusia luma utilized the Hydra- enidae almost twice as much (42%/22%) as did G. sexradiata; however, G. sexradiata utilized a much greater proportion of the bryozoan floatoblasts Table 1. Food categories by taxonomic groups presented in Figures 2-4 for Gambusia sexradiata and G. luma from Spanish Creek during April (IRI = index of relative im- portance). G. luma G. sexradiata Categoiy IRI % Total IRI % Total A. Hydraenidae 3596 42.0 1993 22.0 B. Bryozoan floatoblasts 1493 17.4 4284 47.3 C. Acalyptrate muscoids 923 10.7 772 8.5 D. Apocrita (winged) 453 5.2 99 1.0 E. Oligochaeta 430 5.0 27 0.2 F. Corixidae 308 3.5 522 5.7 G. Nematocera (adults) 297 3.4 176 1.9 H. Gerridae 224 2.6 45 0.4 I. Chrysomelidae 192 2.2 12 0.1 J. Staphylinidae 144 1.6 2 0.02 K. Labidognatha 113 1.3 61 0.6 L. Hydrophilidae 106 1.2 51 0.5 M. Hydracarina 76 0.8 131 1.4 N. Noteridae 67 0.7 107 1.1 O. Chironomidae (larvae) 68 0.7 491 5.4 P. Coleoptera 36 0.4 261 2.8 Q. Nematocera (larvae) 35 0.4 17 0.1 HELDIANA: ZOOLOGY K UJ CD 2 r> z 10 ■ 5 ■ 5 ■ 2 § 10 15 23 ^ 24%^ '^^ Gombusio lump- APRIL I 41% 52% F 6 M 41% 35% 35% 41% J K 28% cAfo 21% 24% 20% 21% 48% 76% % FREQUENCY OF OCCURRENCE Fig. 2. Percent composition of food of Gambusia luma by taxonomic groups in number, volume, and frequency of occurrence at Spanish Creek during April (see Table 1 for abbreviations). (47.3%/17.4%) than did G. luma. Allochthonous, terrestrial insects were also utilized to a greater extent by G. luma (22%/12%). Food resources considered allochthonous made up 25% of the diet of G. luma and 13% of the diet of G. sexradiata, whereas autochthonous resources made up 75% and 85% of these diets, respectively (percentages may not total 100 because it was not possible to assign all prey organisms to these categories). Some major dietary shifts occurred between April 1 and May 1 (figs. 5-7; Table 2), resulting in a decrease in dietary overlap to .76 (Horn's CX). The proportions of hydraenid beetles and Acalyptrate muscoids decreased. Some allochthonous terrestrial sources increased, especially Terebrantia (thrips) and Nematocera adults. Seven categories dropped from the diets of both species, but these con- stituted only a very minor portion of their total diets: 0.75% for G. luma and 0.5% for G. sexradiata. Both species experienced a reduction of April food re- sources (based on the summed percent reduction of IRI resource area for each GREENFIELD ET AL.: GAMBUSIA INTTERACTIONS % FREQUENCY OF OCCURRENCE Fig. 3. Percent composition of food of Gambusia sexradiata by taxonomic groups in number, volume, and frequency of occurrence at Spanish Creek during April (see Table 1 for abbreviations). prey item) that was much more dramatic forG. luma (67%) than forG. sexradiata (36%). Both species added proportionately to their diets from other categories present among the April food items: G. luma added 59% and G. sexradiata added 36%. Ten new categories were added, including some important ones such as crustaceans, Collembola, and chironomid pupae: 16% of the diet of G. sexradiata , but only 8.5% of that of G. luma, came from new categories. The diet of Gambusia luma was expanded in the use of thrips, an allochthonous source that composed a large proportion of its diet in May (16% vs. 6% for G. sexradiata). Bryozoan floatoblasts were important elements of the diets in May as well as in April, but were more equalized in May, with only a 5% proportional difference GREENFIELD ET AL.: CAMBUSIA INTERACTIONS 9 Table 2. Food categories by taxonomic groups presented in Figures 5-7 for Gambusia sexradiata and G. luma from Spanish Creek during May (IRI = index of relative im- port cince). G. luma G. sexradiata Category IRI % Total IRI % Total A. Bryozoan floatoblasts 5693 49.9 4279 54.8 B. Terebrantia 1872 16.4 504 6.4 C. Nematocera (adult) 902 7.9 533 6.8 D. Chironomidae (pupae) 531 4.6 279 3.5 E. Oligochaeta 332 2.9 57 0.7 F. Coleoptera (minute) 307 2.6 37 0.4 G. Apocrita (winged) 2% 2.5 23 0.2 H. Oribatei 238 2.0 30 0.3 I. CoUembola 223 1.9 318 4.0 J. Hydracarina 211 1.8 247 3.1 K. Scolytidae 206 1.8 2 0.02 L. Formicidae 178 1.5 11 0.1 M. Hydraenidae 140 1.2 68 0.8 N. Staphylinidae 139 1.2 5 0.06 O. Chironomidae (larvae) 76 0.6 364 4.6 P. Noteridae 54 0.4 70 0.8 Q. Corixidae 4 0.03 300 3.8 R. Ostracoda 2 0.02 365 4.6 S. Cladocera 1 0.01 101 1.2 T. Copepoda 1 0.01 215 2.7 between the two species. AUochthonous food sources made up 32% and 14% of the total diets of G. luma and G. sexradiata, respectively, whereas autochthonous food sources provided 65% and 86%, respectively (percentages may not total 100 because it was not possible to assign all prey organisms to these categories). Allopatric Gambusia sexradiata The allopatric population of G. sexradiata at the Rio Hondo during December 1975 (wet season) utilized food categories typical of G. sexradiata at Spanish Creek where it was sympatric with G. luma (fig. 8; table 3). Chironomid larvae, CoUembola, Hydracarina, and crustaceans made up 82% of the diet. Categories typical of G. luma, including terrestrial insects such as ants, thrips, wasps, and flies, composed only 4% of the diet of G. sexradiata, and aquatic beetles, Hemiptera, and oligochaetes made up 1%. Visual observations showed G. sexradiata occurred close to shore (within 2 m). Seine hauls in open water more than 2 m from the shoreline did not yield any G. sexradiata. Allopatric Gambusia luma The allopatric population of G. luma from North Stann Creek showed a shift in habitat utilization. In contrast to the open-water habitat of the sympatric population, it was most abundant close to shore. The greatest concentrations of fish were close to branches of bamboo or other terrestrial vegetation that ex- tended into the water. Gambusia luma individuals were seen foraging in the open water several meters from the shoreline, but when pursued by a seine they moved toward the shoreline. In fact, the easiest method of collecting samples of G. luma from North Stann Creek was to seine toward the flooded terrestrial a. < IS i CO 1 VJ o __ z H > s| < 2 -I ^ 1 ^ CM E _3 -3 i * C • i E c ^ ^ r i Oi o u. / » g * UJ o 3 m < ?. — , — 1 )( I m 8381^ ON % swmoA % lO o o i- . s u c 10 GREENFIELD ET AL.: GAMBUSIA INTERACTIONS 11 % FREQUENCY OF OCCURRENCE Fig. 6. Percent composition of food of Gambusia sexradiata by taxonomic groups in number, volume, and frequency of occurrence at Spanish Creek during May (see Table 2 for abbreviations). vegetation, trapping the fish along the shoreline. Because the concentrations were so obvious, counts of individuals taken in seines at various distances from the shoreline wrere not made. Numerous collections made throughout the southern portion of the country, w^here G. luma is usually the only species of Gambusia present (Greenfield et al., 1982), supported these observations. In southern Belize, v^here G. luma is the dominant species of Gambusia, G. sexradiata was taken four times and then in low numbers and in marginal habitats. For example, at the Swasey Branch of the Monkey River, G. luma was abundant in the main river and G. sexradiata was found only in an isolated overflow pond alongside the river. These southern populations of G. sexradiata are believed to be related to populations in southern Guatemala and probably have had a much longer history of interaction with G. luma (Greenfield et al., 1982). DISCUSSION In sympatry, Gambusia sexradiata andG. luma utilized different microhabitats, with G. luma increasing in abundance away from shore. Spatial separation is g •o o 6 ^ 12 Table 3. Food categories by tcixonomic groups presented in Figure 8 for Gambusia sexradiata from the Rio Hondo during December (IRI = index of relative importance). Category A. Cyclopoida B. Cladocera C. Chironomidae (larvae) D. Chironomidae (pupae) E. Collembola F. Hydracarina G. Nematocera (adult) H. Oligochaeta I. Protozoa (tested) J- Apocrita (winged) K. Homoptera L. Helodidae (larvae) M Ephemeroptera (larvae) N. Amphipoda O. Harpactacoida P. Staphylinidae 0. Noteridae R. Conchostraca s. Formicidae T. Oribatei U. Ostracoda V. Insect eggs w Veliidae IRI % Total 3833 34.1 2821 25.1 1392 12.3 837 7.4 584 5.1 556 4.9 374 3.3 183 1.6 134 1.1 84 0.7 70 0.6 69 0.6 67 0.5 65 0.5 59 0.5 31 0.2 25 0.2 19 0.1 14 0.1 8 0.07 4 0.03 1 0.01 1 0.01 30 S 20 CD 3 10 20 30 1 50% MAIN RIVER 64% 87% 71% 64% 6HII jj^L^ 0 42% 47% 91% 62% % FREQUENCY OF OCCURRENCE Fig. 8. Percent composition of food of Gambusia sexradiata by taxonomic groups in number, volume, and frequency of occurrence at the Rio Hondo during December (see Table 3 for abbreviations). 13 14 HELDIANA: ZOOLOGY supported by dietary differences: G. luma contained 87.5% surface-terrestrial organisms and only 12.5% shoreline and cover-inhabiting organisms, whereas G. sexradiata contained 37.5% surface-terrestrial organisms and 62.5% shoreline and cover- inhabiting organisms. Segregation increased as the water level dropped: CX changed from .90 to .76. These results are similar to those of Zaret Si Rand (1971) for stream fishes in Panama. No evidence for niche expansion was found for an allopatric G. sexradiata population. Data from the population at the Rio Hondo indicate G. sexradiata occupied essentially the same microhabitat and fed on the same major food categories as it did in sympatry with G. luma; however, in southern Belize, where G. luma was often the only Gambusia species in a river, it occurred from the shoreline (with vegetation) out into the river during periods of both high and low water. Unfortunately, information on feeding habits of G. luma from these allopatric situations is not available. Gambusia luma is typically a river and stream species, seldom occurring in ponds or standing water, whereas G. sexradiata is typically found in ponds throughout Belize and is most abundant in ponds with a heavy cover of sedges and rushes (Greenfield et al., in press). The adaptation of G. luma to river environments is supported by its streamlined body and by the tendency of individuals to school and remain very close to the surface. Additional evidence suggesting an adaptation to rapidly flowing waters is provided by the gonopo- dial structure of G. luma which, in contrast to that of G. sexradiata, has large terminal hooks (Greenfield et al., 1982). Studies of other genera within the family Poeciliidae, particularly the swordtails in the genus Xiphophorus, showed that species with larger holdfasts inhabit more swiftly flowing waters (Rosen, 1%0). The ability of G. luma to utilize more open surface waters with a greater current may be a consequence of the fish having evolved in the rapidly flowing streams of the emergent mountainous areas of southern Belize when higher sea levels covered the coastal plain (Greenfield et al., 1982). Although conclusive evidence for microhabitat exclusion has not been pre- sented, our field observations show that where G. luma is allopatric it will utilize all habitats from the shoreline vegetation out into the more open water during both high- and low-water periods; Gambusia sexradiata, however, utilizes the near-shore, vegetation habitat regardless of whether G. luma is present. Therefore, in sympatric situations, the presence of G. sexradiata close to shore in the vegetation may prevent G. luma from utilizing that habitat during times of low water when potential competition for food may be greater. Possi- bility of exclusion from shoreline vegetation obviously did not prevent G. luma from moving northward into the range of G. sexradiata in northern Belize after that area became emergent. If one assumes that the environment is near satura- tion (in regard to mosquitofishes), because the May sample was taken at a time of lowest water level and presumably greatest habitat separation, then the overlap values for May at Spanish Creek (CX = .76) may represent a reasonable estimate of maximal overlap along the gradient of prey kind for two sister species of Gambusia in a large open river. Although untested, the role of predation may be a factor in the ability of G. luma to live farther offshore than does G. sexradiata and in the failure of G. sexradiata to move offshore in the absence of G. luma. Anderson (1980) showed GREENFIELD ET AL.: GAMBUSIA INTERACTIONS 15 that there are two major predators of Gambusia species in Belize, the cichlid Petenia splendida and the poeciliid Belonesox belizanus, both of which are rela- tively abundant at Spanish Creek. Thus, in the large northern rivers where G. luma and G. sexradiata are sympatric and where G. luma is found farther out from shore where predators tend to be more abundant, it is tempting to specu- late that G. luma is more successful at escaping predation than is G. sexradiata, which remains near cover along the shoreline. The more streamlined body shape and schooling behavior of Gambusia luma may aid in escaping predation. Chamov et al. (1976) demonstrated the presence of a predator can result in changes in the microhabitat and behavior of the prey, perhaps explaining the failure of G. sexradiata to utilize the more offshore, open waters occupied by G. luma. If G. sexradiata is indeed limited to the shoreline microhabitat by pred- ators, then one would expect selection to have favored behavioral adaptations in feeding methods that maximize efficiency in taking shoreline and cover- inhabiting organisms. Observations in the field and also in aquaria have shown G. sexradiata will feed from the bottom as well as throughout the water column. Conversely, G. luma is a schooling species that tends to remain near the surface and seldom feeds from the bottom. Werner's (1977) ideas on the importance of differences between morpho- logically similar species in foraging efficiency as it relates to microhabitat utili- zation can be illustrated by these two species. Gambusia sexradiata generally eats smaller organisms than G. luma does during both periods of high and low water, mainly due to the greater numbers of bryozoan floatoblasts, crustaceans, and insect eggs in its diet. This is inferred by comparative average values of 6.875 X l(r^ ml/organism for G. luma vs. 5.079 X 10"' ml/org. for G. sexradiata feeding in April, and 5.212 x 10"^ ml/org. for G. luma vs. 3.339 x lOr* ml/org. for G. sexradiata in May. The ability to utilize smaller prey would seem to be a more appropriate feeding method in vegetation; however, G. luma could be a more efficient predator, forcing G. sexradiata to feed more frequently on smaller organisms. If G. sexradiata is more efficient at foraging in the covered vegetative portion of the shoreline area, it may be excluding G. luma from this area (during periods of low water), or G. luma may simply be more efficient at taking open- water submerged and surface-terrestrial organisms. Obviously this is a very dynamic system, as evidenced by the dramatic major shifts demonstrated in the diets of these species over just one month between the wet and dry seasons. These shifts presumably involve either changes in the abundance (i.e., availability) of prey organisms and/or are the result of in- creased habitat separation. Distinct differences between these species in the extent of utilization of few resources such as corixids, chironomid larvae, os- tracods, cladocera, and copepods (fig. 7) point to a latent potential for increased nonoverlap at other times, such as the Terebrantia (thrips) provided between April and May. It has been recently emphasized that tropical environments and their insect populations are no more stable than are temperate ones (Wolda, 1978). The nature of this dynamic system may involve either regular (cyclical) or erratic fluctuations of the prey populations. These predators are adjusted to each other in a finely tuned manner through complementary differences in the use of most of the same food resources. This kind of system may be particularly characteristic of a tropical community, such as this one, allowing for better species-packing potential. 16 FIELDIANA: ZOOLOGY ACKNOWLEDGMENTS We are indebted to the government of Belize, and especially to Mr. G. Win- ston Miller, Fisheries Administrator, for granting permission to collect fish specimens. The Rev. Leonard E. Dieckman, S.J., BrotherGlyde Croy, and all the members of the Jesuit Community at Saint John's College, Belize City, provided invaluable assistance in arranging fieldwork in Belize and offered support in many ways during our stays in Belize. The follov^ing students provided assis- tance in the field: Robert S. Anderson and Herbert Jacque Carter. Ms. Jane Glaser, Northern Illinois University, prepared the figures. Dr. Stephen T. Ross reviewed the manuscript and made many valuable suggestions, as did Dr. Carl von Ende and Dr. Donald J. Stewart. We would also like to thank all the people of Belize, too numerous to mention, who have assisted us while collecting fishes in their country. This research was supported by NSF Grant BM75-08684 to D. W. Greenfield. LITERATURE CITED Anderson, R. S. 1980. Geographic variation and aspects of the life history of Belonesox belizanous (Pisces: Poeciliidae) from Central America. Unpubl. M.S. Thesis, Northern Illinois Univ., DeKalb, 76 pp. Charnov, E. L., G. H. Orians, and K. Hyatt. 1979. Ecological implications of resource depression. American Naturalist, 110(972): 247-259. Dunham, A. E., G. R. Smith, and J. H. Taylor. 1976. Evidence for ecological character displacement in Western American catostomid fishes. Evolution, 33(3): 877-8%. Greenfield, D. W., T. A. Greenfield, and D. M. Wildrick. 1982. The taxonomy and distribution of the species of Cambusia (Pisces: Poeciliidae) in Belize, Central America. Copeia, 1982(1): 128-147. Greenfield, D. W., T. A. Greenfield, and S. L. Brinton. In press. Spatial and trophic interactions between Cambusia sexradiata and Cambusia puncticulata yucatana (Pisces: Poeciliidae) in Belize, Central America. Copeia, 1983. Hellawell, J. M., AND R. Abel. 1971. A rapid volumetric method for the analysis of the food of fishes. Journal of Fish Biology, 3(1): 29-37. Horn, H. S. 1966. Measurement of "overlap" in comparative ecological studies. American Naturalist, 100(914): 419-424. LowE-McCoNNELL, R. H. 1975. Fish communities in tropical freshwaters. Longman Press, London, 337 pp. PiNKAS, L. 1971. In Pinkas, L., J. S. Oliphant, I. L. K. Iverson, Food habits of albacore, bluefin tuna, and bonito in California waters. California Fish and Game Fish Bulletin No. 152, 105 pp. Pontius, R. W., and M. Parker. 1973. Food habits of the mountain whitefish, Prosopium williamsoni (Girard). Transactions of the American Fisheries Society, 102(4): 764-773. Rosen, D. E. 1960. Middle- American poeciliid fishes of the genus Xiphophorus. Bulletin of the Florida State Museum Biological Sciences, 4(4): 57-242. Ross, S. T. 1974. Resource partitioning in searobins (Pisces, Triglidae) on the west Florida shelf. Unpubl. Ph.D. Dissertation, Univ. of South Florida, Tampa, 205 pp. . 1977. Patterns of resource partitioning in searobins (Pisces: Triglidae). Copeia, 1977(3): 561-571. Werner, E. E. 1977. Species packing and niche complementarity in three sunfishes. American Naturalist, 111(979): 553-578. WoLDA, H. 1978. Fluctuations in abundance of tropical insects. American Naturalist, 112(988): 1017-1045. Zaret, T. M., AND A. S. Rand. 1971. Competition in tropical stream fishes: Support for the competitive exclusion principle. Ecology, 52(2): 336-342. Field Museum of Natural History Roosevelt Road at Lake Shore Drive Chicago, Illinois 60605-2496 Telephone: (312)922-9410 UNIVERSITY Of ILLINOIS UHHANA 3 0112 018406972