317.53 13 5.‘ r1: ' I: (I. L7 'iE 5:37:33 13:. flashy IIIIIILIIIIII IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII SSSSSS 010624384 _ [J LIBRARY Michlgar. Stan: Univcmty moi-u USE ONLY St? I 6 ‘99!) IL‘ , ABSTRACT SHAPE DISCRIMINATION IN THE SIAMESE FIGHTING FISH BETTA SPLENDENS by Veronica A. Cerny The experiments reported in this paper were designed to test the ability of the Siamese fighting fish Betta §plendens (Regan), to discriminate shape differences in floating forms in the nest-building situation. An effort was made to de- termine the possible basis for discrimination. Conditioning was not employed in these experiments. Rather, advantage was taken of the fact that all males will build bubble nests under flat floating forms. When the fish were given a choice of two forms differing only in shape, the form under which the larger nest was built was considered to be preferred. Preference implies the ability to dis- criminate, but the reverse is not necessarily true. The observations consisted of three parts designated A, B, and C. Forty-eight males were tested in part A. Part B in- volved the same number but not the same individuals. Six- teen were used in part C. The shapes tested were a circle, a square, and an equi- lateral triangle for part A; an elipse, an elongated Veronica A. Cerny rectangle, and an isosceles triangle for part B; and a right triangle and an isosceles triangle for part C. In part A three different pair-combinations of the shapes were pre- sented to the fish. These were circle-triangle, circle- square, and square-triangle. The three pair-combinations used in part B were: elipse-rectangle, elipse-triangle, and rectangle-triangle. In part C the one pair-combination used was: right triangle-isosceles triangle. The results were tested by a non-parametric sign test. Wherever discrimination was noted, rounded forms such as the circle or elipse were preferred over those with acute angles, e.g. equilateral triangle or isosceles triangle. The rounded forms were not favored when paired with the square or the elongated rectangle, forms having no acute angles. The rectangular form was preferred to the isosceles triangle, but the square was not favored over the equilat- eral triangle. It is suggested that the fish discriminated against "acuteness of angle" and that a difference greater than 300 was necessary for this to occur under the condi- tions of these experiments. When the data were pooled, significant preference was indicated for the circle and the elipse, while there was discrimination against the isosceles triangle. No Veronica A. Cerny significant preference was shown either for or against the square, the elongated rectangle, or the equilateral triangle. In the case of the latter, however, the results approached the level of significance (8.8%), and it is suggested that a larger sample might have indicated discrimination against this shape. In general, these results are in agreement with those cited above. lDixon, W. J. and F. J. Massey, Introduction t9_Statis- tical Analysis (New York, McGraw-Hill Book Co., 1957), pp. 280-302. SHAPE DISCRIMINATION IN THE SIAMESE FIGHTING FISH BETTA SPLENDENS BY Veronica A. Cerny A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Zoology 1965 ACKNOWLEDGEMENTS I wish to express my sincere thanks to Dr. J. C. Braddock for his guidance and kind assistance during the performance of the experimental work and the preparation of this manuscript. A debt of gratitude is also owed to the late Dr° P. J. Clark for his assistance in construct- ing the experimental design and to Dr. H. M. Slatis for his assistance in the statistical analysis. Special thanks are also due to Dr. J. A. King and Dr. W. D. Collings for their help and advice while serving as members of my guidance committee. I would also like to express my warmest thanks to my parents for moral support during this and other trials. ii INTRODUCTION MATERIALS ANDcMETHODS RESULTS . . DISCUSSION SUMMARY . . BIBLIOGRAPHY TABLE OF CONTENTS iii Page 15 29 36 38 Table II. III. IV. VI. VII. VIII. IX. Form Preferences (Part A) Form Preferences (Part B) Form Preferences (Part C) Combined Form Preferences (Part A) Combined Form Preferences (Part B) The Distribution of Area Differences LIST OF TABLES (Part A) . . . . . . . The Distribution of Area Differences (Parts B and C) . . . The for The for Distribution of Area Differences Combined Forms (Part A) Distribution of Area Differences Combined Forms (Part iv B) Page l6 17 18 18 19 22 23 24 26 LIST OF FIGURES Experimental Set Up . . . Dimensions of Forms Used Form Situations . . . . . Final Situations . . . . Rotation of Stimulus Fish Page 11 12 12 13 INTRODUCTION The Siamese fighting fish Betta splendens (Regan) be- longs to the order Labyrinthici, family Anabantidae. All members of this family possess an air breathing apparatus, the labyrinth, which consists of a pair of cavities lined with vascular epithelium. The fish gulp air at the surface of the water and pass it into the labyrinth (Smith, 1945; Forselius, 1957). Some species of this family are oral incubators while others are bubble nest builders. It is probable that the bubble nest building as part of the reproductive behavior evolved from the air breathing habit. The oral incubation which involves retaining the eggs in the buccal cavity until hatching, has probably evolved from the bubble nest building behavior. Betta pugnax (Cantor) and Betta picta (Cuvier and Valenciennces) are two oral incubating species which are closely related to Betta splendens. The male Betta splendens build bubble nests. A bubble of air is taken into the mouth, covered with mucus, and deposited on the surface of the water. This is repeated until a nest is accumulated. In nature the bubbles are usually deposited under a floating leaf (Smith, 1945). The shape and size of the nest are not consistent either among the individuals nor in any one individual. The shape is usually adapted to the shape of the floating object (Braddock and Braddock, 1959), and the size may depend on such species specific factors as the intensity of nest building activity and/or availability of nesting material (Forselius, 1957). In nature the males will not build bubble nests in the pres- ence of the female, but will drive her away. In the labora- tory however, they will build nests while in visual contact with a female or even another male. They will also build nests when visually isolated (Braddock and Braddock, 1959). Mating occurs directly underneath the nest. After mating the male catches the eggs in his mouth and deposits them in the nest where they remain until hatching. The eggs which fall from the nest are caught and returned to the nest. The male also replaces bubbles which have burst, and fre- quently enlarges the nest by adding a new layer of bubbles (Braddock and Braddock, 1959). This addition of bubbles to the nest increases the firmness of the nest and raises the eggs slightly above the water where the oxygen is plentiful (Forselius, 1957). B, splendens has been cultivated for its fighting quality in Thailand for at least a hundred years and thus the literature concerning it varies from folklore to scien- tific studies. Among the earliest scientific studies are those of Regan (1909), who identified the species, and Lissman (1932) who presented information concerning its stimulus-response system. Smith (1937, 1945) published information concerning certain aspects of the behavior and ecology of the species. Forselius (1957) published an extensive monograph covering the behavior, ecology and certain aspects of the endocrinology of Anabantid fishes in general. Braddock and Braddock (1955) published a study of the aggressive behavior of the female B, splendens, and (1959) presented information concerning the development of nesting behavior. Several discrimination studies of B, splendens have also been carried out. Herter (1953) showed that many fishes, including Betta splendens can be trained to make visual discrimination. Braddock, Braddock and Kowalk (1960) published studies concerning size discrimination in the species. This was later extended by Childs (1963). Gude (1965) has carried out studies on color discrimination, and Picciolo (1964) has published information on the importance of color and other factors in sex discrimination in Ana- bantids. Braddock, Braddock and Richter (1960) presented information concerning shape discrimination in Betta splen- gggg, They found that the fish exhibited a marked preference for three compact* forms, a circle, a square, and an equi- lateral triangle, over a rectangle. However they did not find discrimination or preference at a significant level for any form among the three compact forms. This work repre- sents an attempt to learn whether shape discrimination exists among the compact forms and, if possible what aspect of the shape is preferred by the species. *In this thesis the description "compact form" is used to signify a form which has the shortest possible axes for a particular area and shape. MATERIALS AND METHODS* The fish were housed, and the experiments took place, in a laboratory on the third floor of the Natural Science Build- ing at Michigan State University, East Lansing, Michigan. This room contained steam heating apparatus plus an auto- matic air conditioner, which kept the room temperature quite uniform (800-8lOF). However, on five days the temperature fell as low as 78°F, and on three days it rose as high as 82°F. The three windows, facing north, were covered with venetian blinds to reduce the amount of natural light en- tering the laboratory which was lighted with sixteen 40-watt fluorescent bulbs. These provided 12 hours of light (8:00 AM - 8:00 PM), and 12 hours of darkness, and were controlled by an automatic timer. In addition, the experimental aquaria, which were located on a bench next to the windows, each received 24 hours of light from a 25-watt bulb in a goose- neck lamp. A lamp was centrally placed on one side of each aquarium (Fig. l). The purpose was to provide enough light *The eXperimental design used here was approved by the late Dr. P. J. Clark, and the final statistical analysis was approved by Dr. H. M. Slatis. for the fish to see their nests, since in total darkness, they tend to destroy them. The fish were kept in wide-mouthed, gallon (3.78 L) jars, filled to the depth of 17.5-18.7 cm. with aged water,1 and arranged in double rows on wooden racks. They were visually isolated from each other by partitions made of white index cards placed between the bottles where they touched side to side, and by brown—paper partitions where they met back to back. They were not visually isolated from the experimenter during feeding. When evaporation reduced the water level 1.3 cm, it was raised to its original level by the addition of a mixture of aged and tap water, in a ratio of 5 to 1. Five test aquaria were used. Each had a total capacity of 75.6 L and the dimensions of 76.2 x 33 x 35 cm. They were filled to the depth of l7.5~18.7 cm, and thus contained approximately 40.7-43.5 L of water. The floors of the aquaria were covered with gravel to the depth of 0.64 cm. The sides were covered with a layer of 0.001" white opaque plastic sheeting, and the tops were covered with clear glass on which rested a sheet of 0.0075" transluscent laminating vinyl of the color of standard white waxpaper. This allowed 1Aged water, as used here, means tap water aged for not less than three days. some light to enter the aquaria while preventing the fish from being disturbed by visual events occurring outside. A stimulus aquarium was centrally placed against one side of each test aquarium opposite to the side where the gooseneck lamp was located (Fig. 1). Where the stimulus aquaria were in contact with the test aquaria the sheeting was omitted in order that the two fish involved might see each other. This was done to encourage nesting activity, since two males in visual, but not in physical, contact tend to exhibit displacement nest-building. The increased amount of bubbles enabled more accurate measurements to be taken. The total capacity of the stimulus aquaria was 9.5 L. They were placed upon supports 5 cm high, and were filled to a depth of 12.5-13.7 cm which brought the surfaces of the water in both test and stimulus aquaria to the same level, and thus they contained approximately 8.3 L of water. Their floors were covered with gravel 0.64 cm deep. The sides and tops were covered in the same manner and with the same material as those of the test aquaria. Forms, constructed from 0.001" transluscent laminating vinyl, were suspended in the test aquaria in contact with the water surface. When each form was cut, a small tab was left in the center of each side. These tabs were turned up, perforated with a needle and threaded with a white thread. The threads were knotted above the center of the form and were attached to a copper wire suspended from the central plate of glass covering the aquarium. This arrangement prevented the forms from moving about on the surface of the water. /5“~ cm HEWQEZZZVZI Fig. 1. Experimental Set Up The forms were separated by a space of 5 cm where they were closest to each other (Fig. l) and were centrally located in the aquarium. The fish were fed ground frozen shrimp once a day, with a supplement of live brine shrimp nauplii once every two weeks. Uneaten food was removed from the bottles daily, The fish were fed just prior to being placed in the test situ- ation. The temperature of the water in the aquaria was measured and recorded. The fish were then gently netted and placed in the centers of the test aquaria between the two forms, and the covers were put in place. Twenty-four hours later, the covers were removed and the areas of any bubble- nests present were measured in cmz. In actual practice the area of the nests under any particular form was measured with the aid of a form which was identical in shape and size to the form under which the nest was constructed. These forms were made of clear 0.01" acetate on which a grid consisting of 1 cm squares had been drawn. This form was held above the floating disc, and the number of square centimeters, which the nest covered, was read off the grid. The number of layers of bubbles in the nest was considered, and a nest with 2 layers of bubbles was recorded as twice as large as one with only one layer. After all measurements were taken the temperature of the water in the aquaria was measured and recorded, and the fish were returned to their home-jars. The forms were removed and dried, and all bubbles were removed from the aquaria. The forms for another set of observations were then placed in the aquaria, thus preparing 10 the aquaria for the next trial. Tests were conducted daily from April 30 to June 17, 1964; from July 21 to September 9, 1964; and from August 25 to September 9, 1964; for parts A, B, and C of the experiments respectively. A total of 150 adult male Betta splendens (Regan) were used in the observations reported here. These experiments consisted of three parts which were designated as A, B, and C. A total of 48 test fish and 6 stimulus fish each were used for parts A and B, while 16 test fish and 4 stimulus fish were used for part C. These were purchased from a New York firm on two separate dates, approximately four months apart with the exception of 8 fish purchased at a local pet shop. Nineteen individuals died, eleven during the period of experimentation, and were replaced from a group of 26 fish held in reserve. All forms used in parts A, B, and C of the experiments had an area of 200 cm2. Their dimensions are indicated in Fig. 2. Three forms were used in part A of the experiment. The shapes were: an equilateral triangle, a circle and a square. These were presented to the fish in pairs, thus giving three 0Z1 A pair combinations 0E3 which were designated as B and and respectively. 345 C I 11 25 cm - @7.97 cm ‘ 8 cm 14.4 cm ' J 25 cm 0 A. O O 35 30' .__E 90 ' p.90 25 cm ’4 O O 60 35 30' f o A, 16 cm 72 15, "21.48 cm 25 cm ‘- 72° 15' 16 cm Fig. 2. Dimensions of Forms Used The test aquaria were designated as T-I, T-II, and T-IIII and when randomized with the above three pair-combinations, gave rise to 6 possible form situations (Fig. 3). Since an aquarium has right and left sides there were 8 permutations of each situation (Fig. 4). This then resulted . . . . . 3 in 48 poss1ble final Situations. A random number was 2Their companion stimulus aquaria were S-I, S-II, and S-III. 3Dixon, W. J., and F. J. Massey, "Table A-I Random Num- bers," Introduction §2_Statistical Analysis (New York: McGraw-Hill Book Co., 1957), p. 336. 12 2:53;:I2n T-I T-II T-III 1 A B c 2 A C B 3 B A c 4 B c z 5 c A B 6 c B A. Fig. 3. Form Situations Aquar%°m T-I T-II T-III Situation I 1 A(RL) B(RL) C(RL) 2 A(RL) B(RL) C(LR) 3 A(RL) B(LR) C(RL) 4 A(RL) B(LR) C(LR) 5 A(LR) B(RL) C(RL) 6 A(LR) B(RL) C(LR) 7 A(LR) B(LR) C(RL) 8 A(LR) B(LR) C(LR) etc. Fig. 4. Final Situations (For example since A was designated as CDAH A(RL) means, that the (D is on the right side of the aquarium and the z: is on the left. A(LR) means that the Axis on the right side and the C>is on the left.) assigned to each final situation and these were then run, one per day in the order of increasing magnitude of their random numbers. 13 The test fish were labeled T-I, T-2, T-3, . . . T-48, and were used three per day in this order. Each individual was thus used three times at intervals of 16 days. The stimulus fish were labeled S-l, S-2, . . . S-6, and.were used every other day. To equalize the possibility of one fish having more stimulus value than another, the stimulus fish were rotated (Fig. 5). Trials Aquaria S—I S-II S-III 1. S-1 S-2 S-3 2. S—4 S-S S-6 3. S-3 S-l S-2 4. S-6 S-4 S-5 5. S-2 S—3 S-1 6. S-5 S-6 S-4 7. S-1 S-2 S-3 etc. Fig. 5. Rotation of Stimulus Fish The same procedure was used in part B but the forms were changed to an elipse, an isosceles triangle and a rectangle. The aquaria used were the ones used in part A, but new fish, labeled T'-l, T'-2, T'-3, . . . T'-48, and S'-l, S'-2 . . . S'-6, were used. Part C was conducted in a similar manner but only two shapes, an isosceles triangle 14 and a right triangle were used. This gave rise to only four final situations with regard to the shapes tested. However, by also considering the position of the right angle.of the right trianble with regard to the other form and the side of the aquarium, it was possible to get eight situations. Each of these was used four times, thus giving 32 trials. These were randomized by the method using random numbers as de- scribed previously (p. ll). Two test aquaria, T-IV and T-V, and two stimulus aquaria, S-IV and S-V, were used. New fish were used and were labeled T"-l, T"-2, T“-3, . . . T"-l6, and S"-1, S"-2, . . . S"-4. RESULTS Forty-eight individuals were tested in part A of the experiments. The purpose of this part was to determine whether or not the fish were able to discriminate among the shapes; namely circle, square, and equilateral triangle when these were presented in pair-combinations. Preference was measured in terms of the areas of the bubble nests constructed under each form and the preferred form was considered to be that one under which the larger had been constructed. Each fish was used for three trials at sixteen-day intervals. 0f the 48 fish tested, 10 were presented with all three pair-combinations, 31 with two, and seven with only one. This was due to the method of randomization involved in the experimental design. A sign testl indi- cated that the forms presented in preceding trials did not influence the choice in subsequent trials. The same test was also used to determine whether preference was shown for any of the forms. Results which gave a probability for chance occurrence (d) of 5% or less were considered significant. 1Dixon, W. J. and F. J. Massey, Introduction §2_Statis— tical AnaLysis (New York: McGraw-Hill Book Co., 1957), pp. 15 16 Pair-combinations Number of choices made 2 presented 2(a) Total + — ”I O 37 29 8 (.008 A (—) ”I O 34 19 15 >.392 :1 (-) + ( ) D 39 16 23 . .236 A; (-) Table I. Form Preferences (Part A) The results obtained in part A (Table I) indicate that the fish were able to discriminate between a circle and a triangle and preferred the circle with a departure from chance of less than 0.8%. They were either unable to discriminate between, or had no preference for, either shape in circle-square and square-triangle combinations. The experimental procedures and method of analysis of the data used in parts B and C were identical with those used in part A, with the exception that in part B the forms tested were an elipse, an isosceles triangle, and an 2Dixon, W. J. and F. J. Massey, "Table A-lOB: Distri- bution for the Sign Test," Intgoduction §9_Statistical Analysis (New York: McGraw—Hill Book Co., 1957), pp. 418- 420. l7 elongated rectangle while those used in part C consisted of a right triangle and an isosceles triangle. These results are presented in Tables II and III. The information pre- sented in Table II indicates that the fish were able to discriminate between an elipse and an isosceles triangle and between a rectangle and an isosceles triangle at significant levels (both 2.8%). The elipse and the rectangle were both preferred over the triangle. The fish were either unable to discriminate between, or showed no preference for, either an elipse or a rectangle when these shapes were tested against each other. Pair-combinations Number of choices made presented 2(a) Total + - . F (+) - ' 36 25 11' .028 (+) 39 23 16 .336 ‘ DH (+) 36 25 11 .028 AH Table II. Form Preferences (Part B) 18 From Table III it can be determined that the fish were either unable to discriminate between, or indicated no preference for, a right triangle and an isosceles triangle, when these two forms were presented in a pair-combination. . . . Number of choices made Pair-combinations presented Total + — I (+)E::;7 28 18 10 .184 A <-> Table III. Form Preferences (Part C) 2(a) I’The data were then pooled to indicate whether a pref- erence existed for or against any particular shape. This information is presented in Tables IV and V.3 Number of choices made ,, 2(a) Form tested Total Eggrihe igzlgz:m For Against 0 71 47 24 .008 .992 D 73 33 40 >.350 (.650 A. 78 31 47 .912 3.088 Table IV. Combined Form Preferences (Part A) Since only two forms were used in part C, the data are already arranged in this manner (Table III). 19 Number of choices made . 2(a) Form tested Total For the Against F . form the form or Against 75 48 27 .020 .980 72 22 50 (.994 >.006 0 [j 75 40 35 >.358 (.642 A Table V. Combined Form Preferences (Part B) The information presented in Table IV indicates that a preference was demonstrated for the circle, and against the triangle at significant levels 0.8% and 8.8% respectively. Although the discrimination against the equilateral triangle cannot be considered significant since its confidence limits are above 5%, it does approach significance quite closely and there is possibly a need for a more adequate sample. No preference was demonstrated for or against the square. The information presented in Table V demonstrates that a pref- erence was indicated for an elipse at the 2% level and against an isosceles triangle at the 0.6% level. No pref- erence was indicated either for or against the rectangle. The differences between bubble nest areas under each form of all pair-combinations is indicated in Tables VI and VII. The medians, and the upper and lower quartiles are 20 included in these tables to inform the reader concerning the distribution of the area differences of the nests. The quartile deviations are also included for this purpose. The information concerning part A (Table VI) indicates that the ranges were +59 to -26, +69 to —58, and +68 to —58, for the circle-triangle, circle-square, and square-triangle com- binations respectively. The upper quartile, the median, the lower quartile, and the quartile deviation, for the circle- triangle combination were +15, +2, +0.5, and 7.75 in that order. The circle-square combination had an upper quartile of +12, a median of +0.5, and lower quartile of -6, and a quartile deviation of 93. For the square—triangle com— bination, the upper quartile, the median, and the lower quartile were +4, -1, and —7 respectively. The quartile deviation was 5.5. This information for parts B and C (Table VII) indicates that the ranges for elipse-isosceles triangle, elipse- rectangle, rectangle-isosceles triangle, and right triangle- isosceles triangle combinations were +43 to -36, +33 to -27, +85 to -51, and +32 to -26 respectively. The upper quartiles, the medians, and the lower quartiles for elipse—isosceles triangle, elipse-rectangle, rectangle-isosceles triangle, and right triangle-isosceles triangle combinations were +2, 21 +2, and -2; +14.5, +1, and -6.5; +11, +2, and -2; and +17.5, +2.5, and -9.5 respectively. The quartile deviations for the above mentioned combinations were 7, 10.5, 6.5 and 13.5 in that order. As previously mentioned the data collected were pooled in order to learn whether or not a preference existed for or against a particular shape. The upper quartiles, the medians, the lower quartiles, and the quartile deviations for the pooled data are given in Tables VIII and IX.4 The information for part A (Table VIII) indicates that the ranges for the circle, the square, and the triangle respectively, were +69 to -58, +68 to -59, and +58 to —68. The upper quartiles were +15, +4, and +14. The medians were +1.5, -0.5, and -1; while the lower quartiles were -4, —915, and -11 in that order. The quartile deviations were 9.5, 6.75 and 12.5. From Table IX (part B) it can be seen that the ranges for the elipse, the rectangle, and the isosceles triangle were +43 to —36, +85 to -51, and +51 to -85, respectively. The upper quartiles were, in the above mentioned order, +14, +8.5, and +2. Again in the same order the medians and the 4Since only two forms were used in part C the data already are in this form (Table VII). 22 Pair-combinations 2 2 o (+) cm c>(+) cm D (+) cm £1(-) E](-) - IA (—) +59 + 1 +69 - 1 +68 - 1 +42 + 1 +38 - 2 +39 — 2 +41 + 1 +37 — 3.5 +37 - 3 +37 + 1 +29 — 4 +36 — 4 +34 + 0.5 +18 - 4 +20 - 4 +33.5 + 0.5 +18 - 6 +19 - 4 +24 + 0.5 +15 - 6 +17 — 5 +22 + 0.5 +15 - 8 + 6 - 6 +21 - 2 +12 -19 + 4 - 6 +15 - 2 +10 —22 + 4 - 7 +15 - 3 + 9 -23 + 3 - 9 +14 - 7.5 + 4.5 .34, + 1 -10 +10 -12 + 4 -57 + 1 -10 + 8 -15 + 3 -58 + 0.5 -11 + 5 -25 + 2 + 0.5 -15 + 3.5 -26 + 1.5 + 0.5 -16 + 3 + l O -26 + 3 + 0.5 0 -34 + 2 + 0.5 - 0.5 -57 + 1.5 0 - 0.5 -58 + 1.5 - 1 - 0.5 § = +15, * = +2 §= +12, *=-+.5 §==+4, *2 —1 *= +.5, 0]: =7.75 *= - 6, 00: 9 *: -7, OD: 5.5 Table VI. The Distribution of Area Differences in Part '(§ = upper quartile, * = median,*r= lower quartile, and QD = quartile deviation.) 23 Pair-combinations O (+) cm2 0“ cm2 DH cm2 m0 cm2 D(-) E] (-) AH AH +43 + 1 ‘+33 + 1 +85 + 2 +32 ~10 +37 + 1 +28 + 0.5 +47 + 1.5 +31 ~10 +33 + 1 +25: 0 +42 + 1 +28 ~11 +26 + l . +25 ~ 1 +40 + 0.5 +26' ~11 +26 + 1 +23 ~ 3 I +40 0 +25 ~13 +20 0 +21- ~ 4 +27 0 +22 ~21 +19 0 +18 ‘- 5 +25 0 +18 ~26 +19 ~ 1.5 +16 ~ 6 +23 0 +17 +16 . ~ 2 +16 - 6 +21 ~ 1 +12 +11 ~ 2 +15 ~ 7 +13 — 3 + 8 +10 ~ 5 ‘ +14 - 9 4+ 9 ~ 3 -+.8 + 8 — 9 + 8 ~10 + 8 ~ 6 + 7 _ + 5 ~ 9 + 4 ~12 + 8 — 6 + 7 + 4 ~11 + 4 ~12 + 7 ~ 9 + 3 + 3 ~13 + 4 ~15 + 5 -ll + 2 + 2 ~14 + 3 ~17.5 + 4 ~14 + 1 -+ 2 ~14 -+ 2 ~18 -+ 3 ~20 -+ 0.5 + 2 ~36 + 2 ~25 + 3 ~36 -+ 0.5 + 2 + 1 ~27 + 2 ~51 - 6 + 2 + l + 2 - 8 + l + l + 2 ~ 9 § = +12, * = +2 § = +14.5,*=+1 § = +11,*= +2 § =-17.5,*= +2.5 * = ~2, QD = 7 ‘k = ~6.5,QD=10.5* = ~2,QD=6.5 *= 9.5,QD=13.5 Table VII. The Distribution of Area Differences in Parts B and C (§ = upper quartile, * = median, *= lower quartile, and QD = quartile deviation.) 24 Form 2- 2 'tested cm D cm +69 +10 + 0.5 ~15 +68 + 3.5 +59 1 +10 + 0.5 -19 +58 + 3 +42 ‘ + 9 + 0.5 ~22 +57 + 2 +41 + 8 + 0.5 ~23 +39 + 1 +48 + 5 + 0.5 ~25 +37 + 1 +37 + 4.5 + 0.5 ~26 +36 + 1 +37' + 4 0 -34 +34 + 1 +34 + 3.5 - 1. ~57 +23 + 0.5 +33.5 + 3 - 1 ~58 +22 + 0.5 +29 + 3 - 2 +20 + 0.5 +24 + 3 - 2‘ +19 0 +22 + 2 - 2 +19 0 +21 + 2 _ 3 +17 0 +18 + 1.5 - 3.5 + 8 - 0.5 +18 + 1.5 - 4 + 6 - 0-5 +15 + 1.5 - 4 + 6 - 0-5 +15 + 1 ~~6 + 6 ~ 0.5 +15 + 1 _ 6 + 4 - 0.5 +15 + 1 ~ 7.5 + 4 - 1 +14 + 1 - 8 + 4 ‘ 1 +12 + 1 —12 + 4 - 2 § = +15, * .g»..+1.5 1k = ~4, QD = 9.5 § _= +4. * = 0-5 Table VIII. The Distribution of Area Differences for (§ = upper quartile, * = median, * = lower quartile, and 25 A; cmz — 2 -15 +58 + 4 - 1 —19 - 3 ~16 +57 + 3 - 1 —20 - 3 ~18 +34 + 3 - 1 -21 - 4 ~18 +26 + 2 - 1 -22 - 4 ~26 +26 + 2 - 1.5 —24 — 4 —29 +25 + 2 - 1.5 —33.5 - 4 -34 +16 + 1 - 2 —34 — 4.5 -37 +15 + 0.5 - 3 ~36 - 5 ~38 +15 + 0.5 - 3 -37 - 6 -57 +12 + 0.5 - 3 -37 — 6 ~58 +11 0 - 3.5 -39 - 7 ~69 +10 0 - 4 -41 - 9 +10 - 0.5 - 4 -42 - 9 + 9 - 0.5 - 5 -59 -10 + 7.5 - 0.5 - 6 ~68 -10 + 7 - 0.5 - 8 -10 + 6 - 0.5 -10 -11 + 6 - 0.5 -14 -12 + 5 - 0.5 -15 -15 + 4 - 1 -15 -15 + 4 - 1 -17 u“: = -9.5, QD = 6.75 § = +14, * = -1 'k = -11,QD=12.5 Combined Forms in Part A OD = quartile deviation.) 26 Form 2 2 tested 0 cm [I cm +43 +10 + 1 ~ 9 +85 + 8 +37 +8 + 1 — 9 +47 + 7 +33 + 8 + 1 ~10 +42 + 7 +33 + 5 + 1 ~11 +40 + 6 +28 + 4- + l -12 +40 + 6 +26 + 4 + 0.5 ~12 +27 + 5 +26 + 4 0 ~13 +27 + 5 +25 + 4 0 ~14 +25 + 4 +25 + 3 0 ~14 +25 + 4 +23 + 3 ~ 1 ~15 +23 + 3 +21 + 2 ~ 1.5 —l7.5 +21 + 3 +20 + 2 - 2 ~18 +18 + 3 +19 + 2 -‘2 ~25 +17.5 + 2 +19 + 2 ~ 3 ~27 +15 + 2 +18 + 2 - 4 ~36 +13 + 2 +16 + 2 ~ 5 +12 + 2 +16 + 2 - 5 +12 + 1.5 +16 + 1 - 6 +10 + 1 +15 + 1 ~ 6 + 9 + 1 +14 + 1 - 7 + 9 + 0.5 +11 + 1 ~ 9 + 8 0 9: +14, * — +1 * = -5, OD = 9.5 9 = +8.5, * = +0.5 Table IX. The Distribution of Area Differences for (§ = upper quartile, * = median, *'= lower quartile, 27 A .m2 o — 9 +51 + 1 ~ 2 ~21 0 ~11 +36 0 ~ 2 ~23 0 ~14 +36 0 - 2 ~25 0 ~14 +20 0 ~ 3 ~26 ~ 0.5 ~15 +14 0 - 3 ~26 ~ 1 ~16 +14 0 ~ 3 ~27 ~ 1 ~16 +14 0 ~ 4 ~33 ~ 1 ~18 +13 — 0.5 ~ 4 ~37 — 1 ~20 +11 ~ 1 ~ 5 ~40 ~ 1 ~21 +11 ~ 1 ~ 5 ~40 ~ 2 ~23 + 9 ~ 1 ~ 7 ~42 ~ 2 ~25 + 9 ~ 1 ~ 8 ~43 - 3 ~25 + 9 ~ 1 ~ 8 ~47 ~ 3 ~28 + 6 ~ 1 ~ 9 ~85 ~ 3 ~33 + 6 ~ 1.5 ~10 - 4 ~36 + 5 ~ 2 ~11 _ 4 ~51 + 3 ~ 2 ~13 - 4 + 3 ~ 2 ~16 _ 6 + 2 ~ 2 ~19 - 6 + 2 - 2 ~19 _ 8 + 1.5 ~ 2 ~20 *.= -5, QD = 6.75 § = +2, * = -2 = ~11, QD =36.5 Combined Forms in Part B and OD = quartile deviation.) 28 lower quartiles were +1 and ~5, +0.5 and ~5, and -2 and ~11. The quartile deviations were 9.5 (elipse), 6.75 (rectan- gle) and 6.5 (isosceles triangle). DISCUSSION Studies involving the ability to discriminate among various shapes have been performed by a number of investi- gators and have covered a large variety of species. Thus, Young (1958) reported discrimination between a compact form (circle) and an elongated form (rectangle) in Octopus vul- qggig (Lamarck). Sutherland (1960), working with the same species, demonstrated discrimination between a square and an elongated rectangle. He also reported that the octopus can distinguish between an elongated rectangle and a diamond— shaped form. This animal also distinguishes a vertical from a horizontal rectangle (Sutherland, Mackintosh, and Mackin- tosh, 1963). Again, using this same species, Boycott (1965) determined that the ability existed to distinguish between the presence and absence of a square form in feeding situa- tions. Typical examples taken from studies of vertebrate species are Dodwell (1957), who discovered that male hooded rats can distinguish a square from a circle and Rensch (1957) who reported elaborate discriminatory ability in the Indian elephant. Perhaps the discriminatory abilities of birds have received more attention than those of any com- parable group of animals. Tinbergen's work with the 29 30 European robin ('Turdus merula L.) is famous (Tinbergen, 1958). The least common denominator of all of these studies is the fact that conditioning is used as the criterion for ability to discriminate. Studies of shape discrimination among fish are less numerous than those concerning birds and mammals, but sig- nificant examples exist. Thus, it has been demonstrated that the common goldfish, Carassius auratu§,can distinguish between horizontal and vertical rectangles (Mackintosh and Sutherland, 1963). Hemingway and.Mathews (1963) discov- ered that the Egyptian Mouth breeder, Tilapia macrocephala, is able to discriminate between a circle and a rectangle, a square and a rectangle, a circle and a triangle, and.a tri- angle and a square. Once again, as was the case with the studies involving birds and mammals, all of these analyses of discrimination by fish were based upon criteria of conditioning. There is a built-in disadvantage inherent in all dis- crimination studies based upon criteria of conditioning. This is the fact that, in the absence of suitable rein- forcement, conditioning may not occur and thus ability to discriminate may not be made manifest. Thus, an animal may actually learn not to discriminate. Studies based upon the 31 inherent behavior of all members of a species, or of all members of the same sex of that species are largely free from this defect when restricted to individuals naive to the particular experimental situation used. Thus, the experi- ments of Tinbergen with the 3-spined stickleback (Gastero- stelus oculeatus) demonstrated the importance of certain semi-abstract characteristics of form to successful mating. Braddock, Braddock, and Kowalk (1960) employed differential nesting activity of Betta splendens under circular forms of a varying size as a criterion of size discrimination, and Braddock, Braddock, and Richter (1961) adapted this method to the first such studies of shape preference in the same species. It is important to note that studies of discrimi- nation based upon instinctive behavior are not certain indicators of discriminatory ability under all circum- stances. Under circumstances where choice plays no role in the ecology of the species, the animals may be able to dis- criminate but may not indicate this in terms of preference for a particular choice offered them. This particular study is primarily an attempt to deter- mine whether B, splendens discriminates among the compact forms: circle, square, and equilateral triangle. An attempt was also made to determine, if possible, what aspect 32 or aspects of these forms were preferred. To a human ob— server the obvious differences among the forms are the round- ness or absence of angles in the circle, the presence of 900 angles in the square, and the presence of more acute angles (600) in the equilateral triangle. It was postulated that if the presence or absence and acuteness of angles was the basis for preference and hence discrimination, similar preferences should exist among the elongated forms: elipse, rectangle, and isosceles triangle. The fish should also be able to discriminate between a right triangle and an isosceles triangle. It was found that male B, splendens were able to dis~ criminate between a circle and an equilateral triangle at significant levels but indicated no preference for a circle paired with a square, nor between a similar pairing of a square with a triangle. Among the elongated forms signifi- cant discrimination was found between elipse and isosceles triangle and between a rectangle and an isosceles triangle, although no preference was shown for either shape when an elipse was paired with a rectangle. It thus seems probable that the fish either are not able to discriminate or have no preference under the con- ditions of these experiments for "roundness" or the presence 33 of 900 angles, as presented in the circle versus square, elipse versus rectangle combinations. A similar situation exists also between equilateral triangle and square pairings. It should be noted that the difference between the smallest angle of the triangle (60°) and the largest angle of the square (90°) is only 300. On the other hand this angular difference (540 30') is much greater between an isosceles triangle (smallest angle 350 30') paired with a rectangle (largest angle 900). Since in all cases where preferences were demonstrated, they involved absence of angles or the less acute angles, it seems probable that the fish dis- criminate against acuteness of angles. They exhibit such preference, however, only when there is a sufficient minimum difference in the size of the angles. Thus, it is suggested that discrimination between angles is on the basis of rela— tive size. Childs (1963) found that size discrimination of circular discs in B, splendens was made on a relative basis such as this. The fish failed to show preference when a right triangle was paired with an isosceles triangle. In this particular instance the most acute angle of the right triangle had a value of 320 40', while that of the isosceles triangle was 350 40'. If, as has been previously postulated, the quality 34 of acuteness of angle is the basis for preferencer.this result is consistent with the rest of the observations reported here. Thus, if a right triangle with two 450 angles had been used, it is possible that it would have been pre- ferred over the isosceles triangle actually used. When the pooled data are reviewed, one notes significant preferences for the circle and the elipse and discrimination against the isosceles triangle. Thus further supports the hypothesis that the basis of discrimination is the degree of acuteness of the angles of the forms used. While the equilateral triangle was neither preferred nor discriminated against, the confidence level was only 8.8%. It is sug- gested that a larger sample might add this shape to the list of those discriminated against. Further work is required to determine whether or not limits exist above and below which the degree of acuteness of angle cannot be discriminated. It is postulated here that the upper limit may be close to 900 since this was not dif- ferentiated from an absence of angles as in those instances where a circle was paired with a square and an elipse with an elongated rectangle. Nothing in these data suggests the lower limit except that it must lie at or below 320 40', the most acute angle involved in any of the forms used. 35 The preference for compact, rather than elongated, shapes as revealed in this study can well have adaptive significance in the nesting situation. A bubble nest is easily disrupted by water movements, and this is much more likely to occur when a nest is deposited under a long narrow leaf than if it is built under a broad one. S UMMARY 1. Male Siamese Fighting Fish were presented with pair combinations of floating plastic forms under which they could construct bubble nests. The actual combinations tested were: circle-equilateral triangle, circle-square, square— equilateral triangle, elipse-isosceles triangle, elipse- elongated rectangle, elongated rectangle-isosceles triangle, and right triangle—isosceles triangle. Preference, and thus ability to discriminate, was indicated when larger nests were consistently placed under one form in a pair combination. 2. The fish were able to discriminate between a circle and a triangle, but did not discriminate between a circle and a square, or between a square and a triangle. They also dis- criminated between an elipse and an isosceles triangle and between a rectangle and an isosceles triangle. They did not discriminate between an elipse and a rectangle nor between a right triangle and an isosceles triangle. 3. The data were pooled to indicate whether any of the shapes offered were preferred or discriminated against, and significant preferences were shown for the circle and the elipse. The isosceles triangle was discriminated against, and some discrimination was also found against an equilateral 36 37 triangle. No preference for or against was indicated with regard to the square and the rectangle. 5. It is suggested that the fish may discriminate on the basis of ”roundness" or the presence or absence of acute angles. BIBLIOGRAPHY Boycott, B. B., 1965. "Learning in the Octopus," Scientific American 212:42-47. Braddock, J. C. and Z. I. Braddock, 1955. "Aggressive Behavior Among Females of the Siamese Fighting Fish, Betta Splendens," Physiological Zoology 28:152-172. Braddock, J. C. and Z. I. Braddock, 1959. "The Development of Nesting Behaviour in the Siamese Fighting Fish, Betta Splendens," Animal Behaviour 7:222-232. Braddock, J. C., Z. I. Braddock and G. Kowalk, 1960. "Size Discrimination in the Siamese Fighting Fish, Betta Splendens," Bull. Ecol. Soc. Amer. 41:82. Braddock, J. C., Z. I. Braddock and H. Richter, 1961. "Form Discrimination in the Siamese Fighting Fish, Betta Splendens," Amer. Zool. 3:345. Childs, M. 0., 1963. "Size Discrimination in the Siamese Fighting Fish Betta Splendens," M.S. Thesis, Michigan State University. Dodwell, P. C., 1957. "Shape Recognition in Rats," Brit. g, Psychol. 48:221-229. Forselius, S., 1957. ”Studies in Anabantid Fishes." ZoBliqiska Bidraq_Fr§n Uppsala 32:93-599. Gude, R., 1965. "Color Discrimination in the Siamese Fighting Fish Betta Splendens," Ph.D. Thesis, Michigan State University. Goodrich, H. B. and H. C. Taylor, 1934. "Breeding Reactions in Betta Splendens," Copeia 4:165-166. Hemingway, G. and W. A. Mathews, 1963. "Shape Discrimina~ tion in Tropical Fish,” Quart. g, Exp. Psych. 15:272-278. Herter, K., 1953. Die Fischdressuren und Ihre Sinnesphysio- logischen Grundlaqen, Berlin: Akademie Verlag. 38 39 Lissman, H., 1932. "Die Umwelt des Kampffishes Bgtta Splendens," Zeitsch. Verlg. Physiol. 18:65-111. Mackintosh, J. and N. S. Sutherland, 1963. "Visual Dis~ crimination by Goldfish: The Orientation of Rectangles," Animal Behaviour 9:135-141. Picciolo, A. R., 1964. "Sex and Nest Discrimination in Anabantids," Ecol. Mono. 34:53-76. Regan, C. T., 1909. "Asiatic Fish of Family Anabantidae," Proc. 2001. Soc. Lond. 767—787. Rensch, B., 1957. ”The Intelligence of Elephants,“ Scien- tific American l96:44~50. Smith, H. M., 1937. "The Fighting Fish of Siam,” Natural History 39:265—271. , 1945. "The Fresh-Water Fishes of Siam or Thia- land,” U.S. Nat. Mus. Bull. 1883456~461. Sutherland, N. S., 1960. ”Visual Discrimination of Shape by Octopus: Squares and Rectangles," g, Comp. Physiol. Psychol. 53:93-103. Sutherland, N. S., J. Mackintosh and N. J. Mackintosh, 1963. "The Visual Discrimination of Reduplicated Patterns by Octopus," Animal Behaviour 9:106-110. Tinbergen, N., 1958. B_Study _§_Insting§, Oxford: Clarendon Press. yYoung, J. 2., 1958. "Responses of Untrained Octopus to Various Figures and the Effect of Removal of the Vertical Lobe," Proc. Roy. Soc. Lond. B, 149:463—483. 5T9 E33 MICHIGAN TE UNIV. LIBRARIES IIIIIIIIIIIIIII IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII 312 010624884