JOURNAL OP COMPARATIVE AND PHYSIOLOGICAL PSYCHOLOGY Copyright © 1975 by the American Psychological Association, Inc.
Volume 89 Number 7
September 1975
Laboratory Observations of Trailing Behavior in Garter Snakes John Kubie Program in Biological Psychology, Downstate Medical Center, Brooklyn, New York
Mimi Halpern Department of Anatomy?, and Program in Biological Psychology, Downstate Medical Center, Brooklyn, New York
Using a Y-maze apparatus, nine garter snakes were trained to follow an earthworm-extract trail. Correct trailing improved significantly as training progressed. All snakes were able to attain a trailing criterion of 8 out of 10 correct trials by the termination of training. For each trial, the number of tongue flicks each snake made was recorded and tongue flick rate calculated. Tongue flick rate was found to be highly characteristic and constant for individual subjects. In addition, running speed and tongue flick rate were found to be significantly correlated.
vomeronasal system. The vomeronasal sense is especially well developed in snakes (Parsons, 1970). Gaseous molecules reach the snake's olfactory mucosa by passing through the nostrils, whereas the vomeronasal receptors lie in two blind sacs that open into the roof of the mouth. It is often suggested (Stidworthy, 1971) that flicks of the forked tongue deliver the odorants from the external world into the pouches that house the vomeronasal organ. The tongue flick delivery of odorants to the vomeronasal organ appears to be an effective mode by which snakes can follow an odor trail. In a recent series of studies Burghardt and his colleague This research was supported by U.S. Public (1966, 1969; Burghardt & Hess, 1968) used Health Service Grant NS-11713 and by a State tongue flick rate as an indirect measure of University of New York grant-in-aid to M. Hal- vomeronasal stimulation. Initial observapern. The authors thank Barry Feinberg for his help in the conduct of the experiment, Rolf Dun- tions in our laboratory indicated that some heim for building the maze, and Carolyn B. Ware measure of tongue flick activity might proand Richard M. Klein for the helpful criticism vide insight into the more subtle aspects of of the manuscript. sensory control of trailing behavior. Requests for reprints should be sent to Mimi Past investigators (Baumann, 1927; KahHalpern, Department of Anatomy, Downstate Medical Center, 450 Clarkson Avenue, Brooklyn, mann, 1932; Noble, 1937; Noble & Clausen, New York 11.203. 1936) interested in studying sensory control 667 There is good evidence that for some snakes odor is a primary sensory cue in the complex chain of behaviors that terminates in food ingestion (Fox, 1952; Kauffeld, 1953, 1954). Food substances repeatedly rejected by captive snakes may be made acceptable by adulterating the food with a scent associated with a normal food source (Kauffeld, 1954). The precise sensory events that govern recognition by snakes of appropriate food sources are as yet poorly understood (Burghardt, 1970). Two odor-sensitive systems exist in most vertebrates: an olfactory system and a
668
JOHN KUBIE AND MIMI HALPERN
of trailing behavior in snakes used a variety of procedures. In genera], these studies use an open field with a trail placed on the field surface either by allowing the odor-releasing animal (mouse, worm, another snake) to traverse the field or by having the experimenter lay down a trail of appropriate chemical extract. The experimental subject is then released into the field and its behavior with respect to the trail is observed. Individual performance is measured in terms of time to locate the odor source, congruence between route taken by the subject and the trail, or success rate in locating the odor source. These procedures, however, do not permit repeated measurements on the same animal under controlled conditions, nor do they lend themselves to quantifiable measures other than time to locate the odor source. A more serious deficiency in the procedures, however, is that subtle behavioral changes over time or after experimental manipulation are not easily observed (see Burghardt, 1970, for an excellent review). The following study was designed to develop a procedure for studying trailing behavior in snakes that
would provide repeated measurement, quantifiability, and sensitivity to experimental manipulation. METHOD Subjects The experimental subjects were nine juvenile garter snakes, Thamnophis sirtalis sirialis, purchased from a local vendor and housed in our laboratory for 1 yr prior to the commencement of the experiment. The snakes were 8.25-13 cm in length. One week prior to the initiation of testing, the snakes were removed from the laboratory's common colony and housed in individual plastic cages, 30.48 X 15.24 X 15.24 cm. The snakes were maintained on a 12:12 light/dark cycle. The temperature was kept between 23° C and 25° C. During the testing period the snakes were fed only in the experimental apparatus.
Apparatus The testing apparatus was a Y-maze (see Figure 1) with wood bottom and sides. A hinged roof and a removable floor were made of clear Plexiglas. Masonite doors separated a start box and two goal boxes from the body of the maze. The doors could be raised by strings running through overhead pulleys.
Start box
8.9cm
95cm
FIGURE 1. Schematic diagram of Y-maze used in trailing study.
TRAILING BEHAVIOR IN GARTER SNAKES Procedure Prior to the start of training, each snake was handled by the experimenter and permitted to explore the maze daily for a minimum of 1 wk. Testing sessions were held five afternoons a week; each snake was tested two or three times each week. Two snakes, YS6 and YS9, shed during the training period. Testing of these two animals was discontinued when their eyes became cloudy and was reinstituted after shedding was complete. Results on the trials conducted during the week preceding shedding were discarded. During training, each snake was given two trials in the apparatus on each test day. In preparation for each trial, autoclave tape was laid in the runway and in each choice arm of the maze. A "correct" arm was selected for each trial, using a Gellermann series (1933). Aqueous earthworm extract, prepared by the method of Wilde (1938) and Burghardt (1966) at a concentration of 6 g of earthworm to 20 cc of water, was applied with a cotton swab to the tape in the runway and in the arm selected for that trial as the correct arm. Water was applied with a cotton swab to the tape in the "incorrect"arm. A small bit cut from a live earthworm was placed on the end of the tape in each goal box. Before the snake was placed in the start box, the start box gate was closed and the goal box gates were raised. On each trial the snake was confined to the start box for 60 sec before the start gate was raised and timing of the trial commenced. One observer recorded elapsed time with a stopwatch. Recordings were made (a) when the snake's head passed the start box gate; (b) when the snake's head reached the junction; (c) when the snake's entire body entered one of the choice arms; and (d) when the snake's head entered the correct arm goal box. A trial was scored as correct if the first arm the snake entered with its entire body was the correct arm. On such a trial the snake progressed to the goal box and ate the worm bit. If the first arm the snake entered with its entire body was the incorrect arm, the trial was scored as incorrect, the goal box gate of the incorrect arm was lowered, and the snake was allowed to retrace its path until it entered the correct arm goal box and ate the worm bit. If the snake spent more than 3 min in the start box or runway of the maze, the trial was scored as "null" and was not tabulated with the other trials. A second observer using a manual event recorder counted the number of tongue flicks on each trial. One tongue flick was registered each time the snake's tongue emerged from its mouth. Tongue flick counting was begun when the snake's head passed the start box gate. A record was made of the number of tongue flicks made between the start box gate and the junction and, on correct trials, the number of tongue flicks made between the junction and the goal box. The following measures were available for each trial: (a) correct choice, incorrect choice, or null
669
trial; (b) time out of start box: interval between lifting the start box door and the subject's head passing the start box gate; (c) time to traverse runway: interval between subject's head passing the start box gate and arrival at junction; (d) rate of tongue flicks during runway travel: number of tongue flicks made during runway travel divided by the time to traverse the runway; (e) on correct trials, time to traverse correct arm; and (f) on correct trials, rate of tongue flicks during correct arm travel. RESULTS All nine snakes trailed above the 50% level (Table 1). When all trails arc considered, the mean overall trailing performance was 75.0 % correct (p = < .005) with a range of 62.2 % correct for the poorest snakes to 93.9 % for the most accurate trailing. Correct trailing improved significantly as training progressed. The nine snakes trailed an average of 6.6 correct trials in the first 10 trials as compared with an average of 8.8 correct trials in the last block of 10 trails, t(lQ) = 3.67, p < .01. Not all animals demonstrated improvement. Some commenced training with very high correct trailing scores (note YS6 and YS7, Table 1) from which no improvement could be seen. Snakes initially trailing at close to chance levels (YS1, 2, 4, and 5) showed marked improvement in correct trailing. With a criterion of 8 correct trials in a block of 10, two snakes reached criterion within the first 10 trials, an additional four reached criterion in the first 20 trials, and the remaining three met the criterion within TABLE 1 PERFORMANCE OF INDIVIDUAL SNAKES IN THE Y-MAZK Animal
Overall % Correct
No. of trials correct in first 10 trials
YS1 YS2 YS3 YS4 YS5 YS6 YS7 YS8 YS9
86.6 68.7 71.4 62.2 62.8 93.9 81.8 71.8 76.0
6 5
M
75.02
6.6
7 5 5 9 9 6 7
No. of trials correct in last 10 trials
10 9 10 8 9 9 8 8 8
8.8
670
JOHN KUBIE AND MIMI HALPERN 300 -i
200-
100-
1-5
6-10
11-15
16-20
21-25
Blocks of Five Trials
FIGURE 2. Mean time in seconds (±1 SD) for all animals to traverse maze from start box to goal box in five trial blocks.
30 trials. No snakes were discarded for failure to perform the task; all snakes were able to reach the criterion established for correct trailing. In addition to the increased reliability of performance as a result of training, the snakes as a group required less time to traverse the maze as training progressed (see Figure 2). Average total running time for the group during Trials 1-5 was 246.7 sec compared with an average of 125.7 sec on Trials 21-25. Total running time decreased significantly over the five blocks of trials, F(4, 156) - 7.63, p =
Volume 89 Number 7
September 1975
Laboratory Observations of Trailing Behavior in Garter Snakes John Kubie Program in Biological Psychology, Downstate Medical Center, Brooklyn, New York
Mimi Halpern Department of Anatomy?, and Program in Biological Psychology, Downstate Medical Center, Brooklyn, New York
Using a Y-maze apparatus, nine garter snakes were trained to follow an earthworm-extract trail. Correct trailing improved significantly as training progressed. All snakes were able to attain a trailing criterion of 8 out of 10 correct trials by the termination of training. For each trial, the number of tongue flicks each snake made was recorded and tongue flick rate calculated. Tongue flick rate was found to be highly characteristic and constant for individual subjects. In addition, running speed and tongue flick rate were found to be significantly correlated.
vomeronasal system. The vomeronasal sense is especially well developed in snakes (Parsons, 1970). Gaseous molecules reach the snake's olfactory mucosa by passing through the nostrils, whereas the vomeronasal receptors lie in two blind sacs that open into the roof of the mouth. It is often suggested (Stidworthy, 1971) that flicks of the forked tongue deliver the odorants from the external world into the pouches that house the vomeronasal organ. The tongue flick delivery of odorants to the vomeronasal organ appears to be an effective mode by which snakes can follow an odor trail. In a recent series of studies Burghardt and his colleague This research was supported by U.S. Public (1966, 1969; Burghardt & Hess, 1968) used Health Service Grant NS-11713 and by a State tongue flick rate as an indirect measure of University of New York grant-in-aid to M. Hal- vomeronasal stimulation. Initial observapern. The authors thank Barry Feinberg for his help in the conduct of the experiment, Rolf Dun- tions in our laboratory indicated that some heim for building the maze, and Carolyn B. Ware measure of tongue flick activity might proand Richard M. Klein for the helpful criticism vide insight into the more subtle aspects of of the manuscript. sensory control of trailing behavior. Requests for reprints should be sent to Mimi Past investigators (Baumann, 1927; KahHalpern, Department of Anatomy, Downstate Medical Center, 450 Clarkson Avenue, Brooklyn, mann, 1932; Noble, 1937; Noble & Clausen, New York 11.203. 1936) interested in studying sensory control 667 There is good evidence that for some snakes odor is a primary sensory cue in the complex chain of behaviors that terminates in food ingestion (Fox, 1952; Kauffeld, 1953, 1954). Food substances repeatedly rejected by captive snakes may be made acceptable by adulterating the food with a scent associated with a normal food source (Kauffeld, 1954). The precise sensory events that govern recognition by snakes of appropriate food sources are as yet poorly understood (Burghardt, 1970). Two odor-sensitive systems exist in most vertebrates: an olfactory system and a
668
JOHN KUBIE AND MIMI HALPERN
of trailing behavior in snakes used a variety of procedures. In genera], these studies use an open field with a trail placed on the field surface either by allowing the odor-releasing animal (mouse, worm, another snake) to traverse the field or by having the experimenter lay down a trail of appropriate chemical extract. The experimental subject is then released into the field and its behavior with respect to the trail is observed. Individual performance is measured in terms of time to locate the odor source, congruence between route taken by the subject and the trail, or success rate in locating the odor source. These procedures, however, do not permit repeated measurements on the same animal under controlled conditions, nor do they lend themselves to quantifiable measures other than time to locate the odor source. A more serious deficiency in the procedures, however, is that subtle behavioral changes over time or after experimental manipulation are not easily observed (see Burghardt, 1970, for an excellent review). The following study was designed to develop a procedure for studying trailing behavior in snakes that
would provide repeated measurement, quantifiability, and sensitivity to experimental manipulation. METHOD Subjects The experimental subjects were nine juvenile garter snakes, Thamnophis sirtalis sirialis, purchased from a local vendor and housed in our laboratory for 1 yr prior to the commencement of the experiment. The snakes were 8.25-13 cm in length. One week prior to the initiation of testing, the snakes were removed from the laboratory's common colony and housed in individual plastic cages, 30.48 X 15.24 X 15.24 cm. The snakes were maintained on a 12:12 light/dark cycle. The temperature was kept between 23° C and 25° C. During the testing period the snakes were fed only in the experimental apparatus.
Apparatus The testing apparatus was a Y-maze (see Figure 1) with wood bottom and sides. A hinged roof and a removable floor were made of clear Plexiglas. Masonite doors separated a start box and two goal boxes from the body of the maze. The doors could be raised by strings running through overhead pulleys.
Start box
8.9cm
95cm
FIGURE 1. Schematic diagram of Y-maze used in trailing study.
TRAILING BEHAVIOR IN GARTER SNAKES Procedure Prior to the start of training, each snake was handled by the experimenter and permitted to explore the maze daily for a minimum of 1 wk. Testing sessions were held five afternoons a week; each snake was tested two or three times each week. Two snakes, YS6 and YS9, shed during the training period. Testing of these two animals was discontinued when their eyes became cloudy and was reinstituted after shedding was complete. Results on the trials conducted during the week preceding shedding were discarded. During training, each snake was given two trials in the apparatus on each test day. In preparation for each trial, autoclave tape was laid in the runway and in each choice arm of the maze. A "correct" arm was selected for each trial, using a Gellermann series (1933). Aqueous earthworm extract, prepared by the method of Wilde (1938) and Burghardt (1966) at a concentration of 6 g of earthworm to 20 cc of water, was applied with a cotton swab to the tape in the runway and in the arm selected for that trial as the correct arm. Water was applied with a cotton swab to the tape in the "incorrect"arm. A small bit cut from a live earthworm was placed on the end of the tape in each goal box. Before the snake was placed in the start box, the start box gate was closed and the goal box gates were raised. On each trial the snake was confined to the start box for 60 sec before the start gate was raised and timing of the trial commenced. One observer recorded elapsed time with a stopwatch. Recordings were made (a) when the snake's head passed the start box gate; (b) when the snake's head reached the junction; (c) when the snake's entire body entered one of the choice arms; and (d) when the snake's head entered the correct arm goal box. A trial was scored as correct if the first arm the snake entered with its entire body was the correct arm. On such a trial the snake progressed to the goal box and ate the worm bit. If the first arm the snake entered with its entire body was the incorrect arm, the trial was scored as incorrect, the goal box gate of the incorrect arm was lowered, and the snake was allowed to retrace its path until it entered the correct arm goal box and ate the worm bit. If the snake spent more than 3 min in the start box or runway of the maze, the trial was scored as "null" and was not tabulated with the other trials. A second observer using a manual event recorder counted the number of tongue flicks on each trial. One tongue flick was registered each time the snake's tongue emerged from its mouth. Tongue flick counting was begun when the snake's head passed the start box gate. A record was made of the number of tongue flicks made between the start box gate and the junction and, on correct trials, the number of tongue flicks made between the junction and the goal box. The following measures were available for each trial: (a) correct choice, incorrect choice, or null
669
trial; (b) time out of start box: interval between lifting the start box door and the subject's head passing the start box gate; (c) time to traverse runway: interval between subject's head passing the start box gate and arrival at junction; (d) rate of tongue flicks during runway travel: number of tongue flicks made during runway travel divided by the time to traverse the runway; (e) on correct trials, time to traverse correct arm; and (f) on correct trials, rate of tongue flicks during correct arm travel. RESULTS All nine snakes trailed above the 50% level (Table 1). When all trails arc considered, the mean overall trailing performance was 75.0 % correct (p = < .005) with a range of 62.2 % correct for the poorest snakes to 93.9 % for the most accurate trailing. Correct trailing improved significantly as training progressed. The nine snakes trailed an average of 6.6 correct trials in the first 10 trials as compared with an average of 8.8 correct trials in the last block of 10 trails, t(lQ) = 3.67, p < .01. Not all animals demonstrated improvement. Some commenced training with very high correct trailing scores (note YS6 and YS7, Table 1) from which no improvement could be seen. Snakes initially trailing at close to chance levels (YS1, 2, 4, and 5) showed marked improvement in correct trailing. With a criterion of 8 correct trials in a block of 10, two snakes reached criterion within the first 10 trials, an additional four reached criterion in the first 20 trials, and the remaining three met the criterion within TABLE 1 PERFORMANCE OF INDIVIDUAL SNAKES IN THE Y-MAZK Animal
Overall % Correct
No. of trials correct in first 10 trials
YS1 YS2 YS3 YS4 YS5 YS6 YS7 YS8 YS9
86.6 68.7 71.4 62.2 62.8 93.9 81.8 71.8 76.0
6 5
M
75.02
6.6
7 5 5 9 9 6 7
No. of trials correct in last 10 trials
10 9 10 8 9 9 8 8 8
8.8
670
JOHN KUBIE AND MIMI HALPERN 300 -i
200-
100-
1-5
6-10
11-15
16-20
21-25
Blocks of Five Trials
FIGURE 2. Mean time in seconds (±1 SD) for all animals to traverse maze from start box to goal box in five trial blocks.
30 trials. No snakes were discarded for failure to perform the task; all snakes were able to reach the criterion established for correct trailing. In addition to the increased reliability of performance as a result of training, the snakes as a group required less time to traverse the maze as training progressed (see Figure 2). Average total running time for the group during Trials 1-5 was 246.7 sec compared with an average of 125.7 sec on Trials 21-25. Total running time decreased significantly over the five blocks of trials, F(4, 156) - 7.63, p =
