Chemical senses involved in garter snake prey training

Journal of Comparative and Physiological Psychology 1979, Vol. 93, No. 4, 648-G67 Chemical Senses Involved in Garter Snake Prey Trailing John L. Kubie Mimi Halpern Program in Biological Psychology State University of New York Downstate Medical Center Department of Anatomy & Cell Biology and Program in Biological Psychology State University of New York Downstate Medical Center In a preliminary experiment, garter snakes were trained to follow earthwormextract trails in a multiple-choice maze and were subjected to either sham surgery (n = 4) or complete bilateral vomeronasal nerve transection (n = 4). Snakes with sham surgery trailed and ate at preoperative levels, whereas snakes lacking a functional vomeronasal system developed a feeding deficit and trailed at chance levels. In a second experiment, 16 snakes were preoperatively tested for their abilities to follow a battery of trails including a range of trail concentrations and two trail manipulations. After baseline testing, the snakes were subjected to sham surgery (n = 2), olfactory nerve transection (n = 7), or vomeronasal nerve transection (four partial; three complete). Snakes with vomeronasal nerve lesions demonstrated trailing and feeding deficits commensurate with the extent of nerve damage. In contrast, snakes with olfactory nerve cuts and sham surgery continued to follow all trails at preoperative levels and maintained high tongue-flick rates when following the stronger extract trails. Four of the snakes (two with sham lesions, one with olfactory nerve lesion, and one with a partial vomeronasal nerve lesion) were subsequently tested with the vomeronasal ducts sutured closed. These snakes were unable to follow any of the extract trails at better than chance levels, but in contrast to snakes with vomeronasal nerve cuts, most of these snakes continued to attack and ingest earthworm bits. These results suggest that garter snakes are heavily dependent on their vomeronasal system for following chemical prey trails. Garter snake feeding is guided by the chemical senses (Burghardt, 1970). Garter snakes readily follow odor trails to locate potential prey (Kubie & Halpern, 1975, 1978), and once in the vicinity of the prey, they attack objects coated with appropriate chemicals (Burghardt, 1966; Wilde, 1938). This research was part of a doctoral dissertation submitted by John L. Kubie to the State University of New York Downstate Medical Center. The research was supported by National Institutes of Health Grant NS11713. We wish to thank Sarah Winans and RuuTong Wang for their thoughtful advice, Vincent Tuonv for teaching us the duct-suture technique, Alice Vagvolgyi for data collection and historical processing, Jack Illari and Bruce Culver for the illustrations, and Rose Kraus for typing the manuscript. J.L. Kubie is now at the Department of Physiology, University of Pennsylvania Medical School. Requests for reprints should be sent to Mimi Halpern, Department of Anatomy and Cell Biology, State University of New York Downstate Medical Center, Brooklyn, New York 11203. The prey attack appears to be elicited by vomeronasal stimulation, since snakes without a functioning vomeronasal system do not exhibit the behavior (Burghardt & Hess, 1968; Burghardt & Pruitt, 1975; Wilde, 1938). It is not clear, however, which sense modalities garter snakes use when following prey-odor trails. In addition to garter snakes, a variety of snakes can use odor trails notpntial nrpv Vinprs and rohi f lorflf ?° .locate Potential prey. V ipers and CO1Ubn d snakes follow prey trails in an Open arena, and when doing so, they are described as exhibiting rapid tongue flicks and keeping their headg low to the groun(} (Baumann, -,c.r,n n UIK i \\i *^] • v v n -ian-t 1927 > ^ehlback, Watkms, & Kroll, 1971, Kahmann, 1932; Naulleau, 1965). Since snakes can use their tongues to deliver substrate-borne odorants to the vomeronasal organs (Kahmann, 1932; Kubie, -.gym th actjve tonffue flickinff observed , y '/'' tne actlve ™ngue IllCKing ODservea during trailing suggests that the vomeronasal system is in operation. In fact, several au- Copyright 1979 hy the American Psychological Association, Inc. Q021-9940/79/9304-0648I00.7S 648 CHEMICAL SENSES IN SNAKE PREY TRAILING thors have speculated that snake prey trailing may represent the archetypical function of the vertebrate vomeronasal organ (Broman, 1920; Gloor, Murphy, & Dreifuss, 1972). A few studies have investigated the role of the olfactory and vomeronasal systems in prey trailing. The olfactory and vomeronasal systems—both well developed in snakes—represent the snake's two major chemosensory systems since the taste system is much reduced in snakes and there are no taste buds on the tongue (Burns, 1969; Klauber, 1956; Oliver, 1955; Payne, 1945). When vipers have severe damage to their vomeronasal systems (e.g., cautery of the vomeronasal organs), they cannot follow prey trails (Kahmann, 1932; Naulleau, 1965). It has not been determined whether similar mutilation of the olfactory system would cause a similar deficit, although Naulleau (1965) reported that blocking the nostrils had no effect on prey trailing. The situation with colubrid snakes is less clear. Kahmann (1932) found that cauterizing the vomeronasal organs of a grass snake (Natrix natrix) or cutting off the tongue would disrupt trailing but that plugging the external nares had little effect. Noble and Clausen (1936), studying the prey trailing of two species closely related to Natrix (Storeria dekayi and Thamnophis sirtalis), found that plugging the external nares had a much more disruptive effect on prey trailing than any vomeronasal system lesion. The major objective of the following studies was to reinvestigate the relative contributions of the olfactory and vomeronasal systems to garter snake prey trailing. In contrast to earlier studies, these studies utilize training procedures to obtain stable, repeated measurements of individual performance (Kubie & Halpern, 1975), a multiple-choice maze to quantify trail discrimination, several concentrations of odor trails to assess thresholds (Kubie & Halpern, 1978), and experimentally controlled nerve-cut procedures, with histological verification of the lesions for interpretable results. In one study (Kubie & Halpern, 1975), we demonstrated that garter snakes would follow trails of earthworm extract accurately in a multiple-choice maze for 649 earthworm-bit rewards and that behavioral measures, such as tongue-flick rate, were stable from trial to trial for individual animals. In a second study (Kubie & Halpern, 1978), we found that garter snakes exhibit greater trailing accuracy, higher tongue-flick rates, and slower trailing speeds when following high concentration earthworm-extract trails than when following dilute extract or pure-water trails. We speculated that the elevated tongue-flick rates during odor trailing could be due either to olfactory stimulation, which aroused tongue flicking, or to vomeronasal stimulation, which had a "positive" feedback on tongue flicking (and would have a subsequent effect on odorant delivery to the vomeronasal organs). Therefore a second objective of this study was to determine whether any nerve-cut procedure would affect a snake's tongue-flick response to trail odorants. Finally, our previous studies demonstrated that garter snakes can follow dried trails but cannot follow trails removed from direct lingual contact (Kubie & Halpern, 1978). If snakes deprived of all olfactory information continued to follow these and other trails as they had before surgery, then the earlier findings could be taken to indicate the nature of adequate vomeronasal stimuli, i.e., the tongue or snout must directly touch the odorant source to deliver odorants to the vomeronasal organs. To answer these questions, we set out to determine the effects of olfactory and vomeronasal system disruptions on the earthworm-extract trailing behavior of garter snakes. Two types of test were employed: a short test series using one intense concentration of worm extract for 20 trials in a four-choice maze and a longer test series using 140 trials including seven concentrations of earthworm extract. The short test block was used to answer the question, Can this snake trail? The second test block was used to answer the question, If this snake can trail, are there any significant changes in its trailing profile? General Method Subjects The subjects were adult male and female garter 650 JOHN L. KUBIE AND MIMI HALPERN snakes of the species Thamnophis radix. All snakes. were purchased from either Chicago Zoological Supply or a local pet dealer and lived in our laboratory at least 3 mo prior to the beginning of an experiment. During this period they were fed chopped raw fish weekly. Snakes were housed in rooms lit for 12 hr a day and maintained between 23 and 25 °C. Snakes were selected from the common colony for pretraining on the basis of eating earthworms (Lumbricus terrestris) and exhibiting prey attacks at cotton swabs dipped in earthworm extract (Burghardt, 1966; Wilde, 1938). During the study, snakes were fed only during testing, as a reward for correct trailing. All snakes used in this study appeared to be adult animals. The females ranged in snout-vent length from 41 to 67 cm and in weight from 38 to 203 g. The males ranged in snout-vent length from 35 to 50 cm and in weight from 25 to 66 g. Apparatus The apparatus was a radial four-choice maze (previously described and illustrated, Kubie & Halpern, 1978) with wooden sides, a Plexiglas floor, and a removable Plexiglas roof. This maze had a runway (6.2 X 22 cm) which led to a choice point; from the choice point, four alleys (15 cm long) led to goal boxes. Drop gates separated the start box and the goal boxes from the runway and alleys. The maze was used in two configurations: (a) as a two-choice Y-maze and (b) as a four-choice star maze. When the maze was used in the Y configuration during training, the left and right alleys closest to the runway were blocked off with alley blocks. Earthworm extract was prepared by the method of Wilde (1938) and Burghardt (1966) at concentrations of 6 g of earthworm per 20 ml of water (Ix) and 18 g per 20 ml of water (3x). The extract was stored frozen in capped vials. On each test day vials of the appropriate extract were thawed. For testing in Experiment 2, dilutions of earthworm extract were made up from Ix extract daily (Kubie & Halpern, 1978). Snakes were tested 5 days a week. During pretraining, each snake was given 4 trials per test session and after pretraining as many as 14 trials per test session. When the spectacle of a snake's eye became opaque prior to shedding, the snake's behavior deteriorated, and testing was discontinued until the day after the snake shed. Trials were run as described in our previous studies (Kubie & Halpern, 1975,1978). Autoclave tape was cut, brushed with a suede brush to roughen the surface, and placed on the floor of the choice arms and the runway. Earthworm extract (Ix) was applied with a cotton swab to the tape in the runway and the arm selected for that trial as the "correct" arm. As a control trail, distilled water was applied to the tape in the "incorrect" alley(s). A small bit cut from a live earthworm was placed on the end of the tape in each goal box. The correct alley on four-choice trials was selected MOB Figure 1. Schemas of dorsal view of the brain, olfactory sacs, and vomeronasal organs as seen through transparent scales. (The position of the vomeronasal nerve transections is indicated in the left diagram [arrow]. The position of the olfactory nerve lesions is indicated by the arrows in the right diagram. Abbreviations: VNO = vomeronasal organ; VNN = vomeronasal nerves; AOB = accessory olfactory bulb; MOB = main olfactory bulb; OS = olfactory sac.) CHEMICAL SENSES IN SNAKE PREY TRAILING with a randomized series constructed with the following constraints: An alley could be correct on no more than two consecutive trials, and each alley was the correct alley five times in each 20-trial block. On each trial the snake was confined to the start box for 60 sec before the start gate was raised. A trial was scored as correct if the first stop-count mark (located 4 cm from the goal box gate) that the snake's head crossed was in the correct alley. On such trials the snake progressed to the goal box and ate the worm bit. If the snake first crossed a stop-count mark in an incorrect alley, the trial was scored as incorrect, the goal box gate of that alley was 651 lowered, and the snake was either removed from the maze (noncorrection procedure) or allowed to correct itself (correction procedure). For each trial the time taken by the snake to travel between the start-count mark (located 11 cm from the start box gate) and the stop-count mark was recorded, and the number of .tongue flicks (tongue excursions from the mouth) was counted. Stopwatches and hand talleys were used in recording these events. In preparation for surgical procedures, snakes were anesthetized with Sodium Brevital (.015 mg/g of body weight injected ip in a solution of .5% Brevital in saline; Figure 2. Photomicrographs of Bodian-stained horizontal sections through the olfactory and vomeronasal mucosae. (A: Intact olfactory mucosa. S = supporting cell nuclei; r = receptor cell nuclei; b = basal cell nuclei. B: Degenerate olfactory mucosa. Note absence of receptor cell layer [d]. C: Intact vomeronasal mucosa. Sc = supporting cell layer; re = receptor cell layer; Ud = undifferentiated cell layer. D: Degenerate vomeronasal mucosa. Note absence of receptor cell layer [d]. E: Partially degenerate vomeronasal mucosa. A sharp boundary exists between intact and degenerate areas of the mucosa. Bar = 100 /urn.) 652 JOHN L. KUBIE AND MIMI HALPERN Wang, Kubie, & Halpern, 1977a). The surgical procedures, performed under visual guidance by means of a dissecting microscope, were described in detail elsewhere (Kubie, Vagvolgyi, & Halpern, 1978). For complete olfactory nerve cuts, the rostral end of each olfactory bulb was exposed, and the bulbs were lifted by the dura with a fine forceps. The olfactory nerves were severed either by the tension of the lifting or by snipping with fine iris scissors (Figure 1). For complete vomeronasal nerve cuts, the vomeronasal nerves were exposed as they travel between the two main olfactory bulbs (Figure 1). The nerves were cut with iris scissors lowered to the base of the cranium, and the region of the nerve cut was cauterized with a fine (.005 in.; .127 mm) tungsten-wire cautery. For partial vomeronasal nerve cuts, the iris scissors were lowered approximately halfway to the floor of the cranium before cutting, and cautery was not used. After the partial nerve cuts there was little bleeding, and the dorsal fascicles of the vomeronasal nerve bundle could usually be seen pulling apart at the lesion site. Sham surgeries were performed by drilling holes in the skull over each olfactory bulb and tearing the dura. Following nerve-cut surgical procedures, a piece of Gelfoam was placed over the exposed nervous tissue. After all bleeding had stopped, the holes in the skull were sealed with dental cement or adhesive tape. For sacrifice and histology a snake was anesthetized with Sodium Brevital (.020 mg/g of body weight) and perfused transcardially with physiological saline followed by Bodian's fixative. The details of the histological procedures were previously described (Kubie et al., 1978). Ten-micrometer sections of the decalcified and paraffin-embedded heads were cut in a horizontal plane. A series of every fifth section was mounted, impregnated with 1% Protargol S (without copper) for 48 hr at either 37 °C or 23 °C and reduced, gold toned, and fixed according to the method of Bodian (1936, 1937). Two independent observers examined the olfactory and vomeronasal nerves through serial sections to determine the extent of the lesion. In addition, since previous studies in this laboratory demonstrated that following olfactory and vomeronasal nerve cuts the appropriate sensory mucosa degenerates (Kubie, Wang, & Halpern, 1977), it was possible to independently assess the extent of the nerve cuts by analyzing the extent of the degenerate olfactory and vomeronasal mucosae (Figure 2). A region of mucosa was considered degenerate when that portion of the epithelium normally occupied by differentiated receptor cells was replaced with an empty strip (Kubie et al., 1978; Wang, Kubie, & Halpern, 1977b). Serial sections separated by 240 //m were projected (12X for the nasal cavity and 35X for the vomeronasal organs) and traced. The amount of normal and degenerate chemosensory mucosa found in each nasal cavity and each vomeronasal organ was "measured" with a map measure (as in Winans & Powers, 1977) and calculated as a percentage of the entire epithelium. In our experience, nerve cuts followed by a 1- or 2-wk survival always result in degeneration. We think that the percentage of apparently degenerate mucosa at this survival is a useful index of the proportion of the nerve damaged during surgery. Experiment 1 This experiment uses 20-trial test blocks as pre- and postoperative measures of trailing performance. The question asked is whether snakes with complete vomeronasal nerve cuts continue to eat or trail after surgery. Method Ten adult Thamnophis radix were selected on the basis of eating earthworms and exhibiting prey attacks at cotton swabs dipped in earthworm extract. Snakes were pretrained for 3 days with the star maze in a twochoice configuration followed by 5 days with it in a four-choice configuration. A Ix earthworm-extract trail and correction procedures were used. Eight of these snakes were selected on the basis of consistency of trailing and eating. They were tested for five additional sessions (20 trials) in the four-choice maze and then subjected to either vomeronasal (n = 4) or sham (n = 4) nerve-cut surgeries. After 3-days recovery from surgery the snakes were tested for five additional sessions and sacrificed. Results Trailing. All eight animals trailed between 70% and 85% correct on four-choice preoperative trials (Figure 3). This represents a performance significantly above chance for each animal (according to the binomial, if P correct = .25 for one trial, then the probability of an animal equaling or exceeding 45% correct after 20 trials is less than 5%). After surgery the four snakes with vomeronasal nerve cuts followed trails with accuracies ranging from 20% to 35% correct, all of which are in the range expected by chance. Preoperative performance of the sham-treated snakes ranged from 75% to 85% correct, and postoperative performance ranged from 50% to 95% correct, all above chance in a 20-trial block. Feeding. Preoperatively, all snakes ate on no fewer than 80% of the trials (Figure 3). During postoperative testing, all snakes with sham lesions ate on all trials. Of the four snakes with vomeronasal nerve lesions, all ate on the first postoperative trial, but none ate on more than 20% of the trials. Three of the snakes with vomeronasal nerve cuts ate on at least one of the first four postoperative trials and refused to eat thereafter. The CHEMICAL SENSES IN SNAKE PREY TRAILING SHAM GROUP 653 SHAM GROUP • DAYS Figure 3. Trailing performance and feeding behavior of snakes receiving vomeronasal and sham nerve lesions. (Trailing performance is depicted as percentage correct prior to and following surgery. Chance performance is 25% correct. Feeding behavior is reported as total number of worm bits eaten on each test day. Since four trials were run each day, the maximum number of worm bits available was four.) fourth, Snake 8C, ate during the first and last test session. This snake was subsequently tested for 5 days, and it continued to eat irregularly (not illustrated in Figure 3). Since three of the four snakes with vomeronasal nerve cuts ate on only the first postoperative test day, the postoperative feeding of those snakes can be described as decaying rapidly. Snake 1A struck at the worm bit on the first trial of the second postoperative day but spit the worm bit out and did not eat thereafter. Snakes IB and 67s tongue flicked the worm bit repeatedly on the first few trials after they had stopped eating but showed no interest in the worm bits thereafter. Histological verification of lesions. All four of the snakes with intended vomeronasal nerve cuts had completely transected vomeronasal nerves with only mild damage to the olfactory nerves. By tracing the course of vomeronasal and olfactory nerve fascicles in each animal, we determined that all, vomeronasal nerve fascicles were cut. The vomeronasal organs were severely degenerated throughout. In the four animals numerous intact nerve fascicles could be seen entering both main olfactory bulbs. The olfactory mucosae of three of the four snakes were not uniformly well preserved, but the well-preserved regions appeared to be intact. Snake 8c had degenerate olfactory epithelium in restricted areas covering anterior regions of the nasal septum bilaterally. These small degenerate regions of epithelium probably resulted from scissor-cut damage to medial fibers entering the main olfactory bulbs. Discussion The four snakes sustaining vomeronasal nerve cuts had successful and consistent bilateral vomeronasal denervation with only minimal damage to the olfactory nerves. There was no evidence of successful regeneration of the vomeronasal nerves. The immediate and total trailing deficit exhibited by snakes with complete vomeronasal nerve cuts is most likely due to an inability of these animals to perceive the location of the trail. If normal snakes rely solely on the tongue-flick delivery of odorant molecules to the vomeronasal organs for accurate trail following, then we would expect snakes with severed vomeronasal nerves to be unable to follow extract trails. The deficit we observed in snakes with vomero- 654 JOHN L. KUBIE AND MIMI HALPERN nasal nerve cuts compared with the normal trailing of sham animals is consistent with this hypothesis. An alternate hypothesis is that the trailing deficits were motivational, not sensory, since the snakes with vomeronasal nerve cuts stopped eating the worm-bit rewards. Possibly they ceased following the earthworm-extract trails not because they could no longer detect the trails but because the trails no longer led to attractive goals. This explanation, however, is not likely, since the trailing deficit was immediate whereas the feeding deficit developed over time. Although, admittedly, the sample was small, trailing accuracy on the first postoperative test day, the day when all snakes were still eating, was 31.25% correct, an accuracy within the range expected by chance alone. In short, it does not appear that the trailing deficits we observed after vomeronasal nerve cuts were due to decreased hunger or to the unpalatability of worm-bit rewards. The best guess from the results of this experiment is that the trailing deficit is primarily sensory, not motivational. Experiment 2 Although it is clear that snakes with their vomeronasal nerves completely severed do not trail, it is not totally clear whether the failure is due to a deficit in detection or motivation. Will vomeronasal system disruptions that do not affect feeding affect trailing? Will elimination of a snake's other major chemoreceptor system, the olfactory system, affect the snake's sensitivity to trails? A second study was designed to answer these questions, utilizing a daily test sequence (previously described in Kubie & Halpern, 1978) in which trials differed in trail extract concentration or trail type1 (wet, dry, and dry-rewet extract trails). The rationale for testing each snake with a variety of trail types and concentrations was to be able to assess both complete and subtle trailing deficits. It seemed particularly important to be able to determine whether olfactory nerve transections caused even slightly noticeable changes in sensitivity to trails. In a study of this sort, each animal's be- havior should be considered individually, and each animal's postoperative behavior compared with its behavior prior to surgery. Considerable variability is likely among animals in preoperative behavior, in the location and size of the lesion, and in each animal's response to its lesion. Although the design of the experiment was to put animals into "experimental groups," each animal represents an experiment in and of itself. Conclusions must be consistent with the results of each of these experiments. Method Subjects and procedure. Adult male and female Thamnophis radix were pretrained for four trials per test session for 16 test sessions. They were given three sessions (12 trials) with the maze in the two-choice configuration, five sessions (20 trials) with the maze in the four-choice configuration, and, finally, eight sessions (32 trials) with dry trails and with the maze in the two-choice configuration. Following pretraining, each snake was given 14 trials per daily test session. The first two "warm-up" trials of the session were with the standard Ix concentration of aqueous earthworm extract prepared by Burghardt's (1966) method. If a snake attacked the worm bit reward on a warm-up trial, the rest of the day's test sequence was presented. The remaining trials included seven concentrations of earthworm extract (3x, Ix, V3, Vg. '/27, Vsi, and V243), two trials with dry trails, two trials with dry-rewet trails, and a distilled water trail. The daily sequence of trials was determined by a 10 X 10 balanced Latin square (Williams, 1949) which created test sequences for a 10-day period.2 The experiment was organized into 10-day test blocks which were exact replications of the Latin square sequence. Snakes were also tested in an open-field apparatus before and following surgery. The apparatus and procedure used were slightly modified from the apparatus and procedure of Chiszar and Carter (1975). The 1 Dry trails were prepared by coating strips of autoclave tape with Ix extract and drying the tape two feet below a heat lamp for 3-4 hr. This tape was placed in the runway and the "correct" alley of the maze; untreated tape was placed in the "incorrect" alley. Dryrewet extract trails were prepared identically to the dry extract trails except that just prior to a trial, distilled water was applied to the dry tape in the runway and correct choice arm and to the untreated tape in the incorrect arms. 2 The 10 X 10 test sequence scheduled 10 trail types for each test day. The 12 trials (two drys, two dryrewets, one water, and seven concentrations of earthworm extract) were scheduled by inserting an "unscheduled" dry-rewet trial after the scheduled dry trial and an "unscheduled" dry trial after the scheduled dry-rewet trial. 655 CHEMICAL SENSES IN SNAKE PREY TRAILING field was an 80 X 80 cm box with an underlit milkglass floor, 25-cm opaque walls, and a clear glass roof. The floor was marked off in 5-cm squares. A snake was placed in the box, allowed 1 min to adapt, and observed for the following 5 min. For each of the 5 test minutes the number of tongue flicks and the number of lines crossed were recorded, and it was observed whether the snake was consistently moving. Open-field trials were run on alternate test days so that a snake had five open-field trials during each test-block period. The experimental paradigm, organized in 10-day test blocks, is depicted in Figure 4. Each test block was administered over a 2-wk interval. All snakes were tested for one test block, given an injection of Sodium Brevital anesthesia (.015 mg/g), allowed 3 days to recover, and tested for a second test block. Following the second test block, each animal was given one of four intended surgical procedures: complete vomeronasal nerve cut, complete olfactory nerve cut, partial vomeronasal nerve cut, or sham nerve cut. Most snakes were sacrificed after three test blocks. Four snakes (one olfactory nerve cut, one partial vomeronasal nerve cut, and two sham nerve cut) had their vomeronasal ducts sutured closed prior to a fourth test block and the duct sutures removed prior to a fifth test block. Duct-suture procedures were also performed under Sodium Brevital anesthesia. A snake was placed on its back, and a wooden rod was placed in its mouth to keep the mouth open. Using 9-0 suture material and fine forceps, we inserted four to five individual sutures tieing together the two lateral ridges of the fenestra vomeronasalis (Figure 5). Behavior analysis. The purpose of presenting two preoperative test blocks was three-fold. First, it provided an initial training block to give a snake time to adapt to the daily testing schedule. Second, the effects of anesthesia alone on each animal's trailing performance could be determined. And third, it permitted a determination of which trails each snake was capable of following so that, without bias, a snake's performance on effective trails could be compared before and after surgery. The most sensitive measure of differences between a snake's trailing performance on two test blocks is a test for the significance of the difference between correlated proportions. If we match each trial in one test block with the corresponding trial in the second test block, we can treat the two test blocks as correlated proportions. The null hypothesis is that a statistically equivalent number of trials went from correct to incorrect (i.e., preopjeratively to postoperatively) as went from incorrect to correct. To increase the sensitivity of this statistic, we included in later comparisons only trail types the snake gave evidence of trailing in the first test block. Any trail types followed at or above 50% correct on the first test block and the strongest extract concentration trailed below 50% correct were considered "criterion" trail types and were included in the pre- and postoperative comparisons. In other words, if a snake followed trails of l/g, Vs, Ix, 3x, dry, and dry-rewet above 50% correct on the first test block, trials with these trail types, plus the l/27th, were used to compare trailing accuracy on the second (preoperative) test block with all subsequent (postoperative) test blocks. Tongue-flick rates (number of tongue flicks emitted divided by the time to traverse the maze segment) were calculated for each trial. For eight of the snakes (Males 63, 71, 57, 54, 71, 1, and 6, and Female XL) two tongue-flick rates were available for each trial: a tongue-flick rate for the time spent in the runway and a tongue-flick rate for the entire trial. For these snakes the tongue-flick rate for the time spent in the runway was used for all calculations, and tongue-flick rates on correct and incorrect trials were not separated. When snakes were tested with all trail concentrations, a t test was used to compare tongue-flick rates on distilled water trails with rates on Ix and 3x trails. When snakes did not eat after surgery, a t test was used to compare tongue-flick rates on pre- and postoperative Ix warm-up trials. For several subjects postoperative freezing behavior interfered with calculating a meaningful tongue-flick rate (the variance was too high to meet the constraint of homogeneity of variance), and tongue-flick rate analyses were not done. TEST BLOCK SEQUENCE GROUP IV (n-2) SHAM PARTIAL VOMERONASAL NERVE CUT Figure 4. Experimental paradigm. 656 JOHN L. KUBIE AND MIMIHALPERN Figure 5. Roof of a snake's mouth with sutures in place (left) and with sutures removed (right). Results 3 Preoperative behavior. The 16 snakes of this study exhibited similar and consistent preoperative trailing behaviors. Although criteria ranged from all seven trail concentrations (1/243 and greater) up to four concentrations (1/9 and greater), 13 of the 16 snakes achieved the same criterion of five concentrations (1/27 and greater). All snakes followed the Ix warm-up trails, the dry trails, and the dry-rewet trails above criterion levels. Each animal trailed correctly on more than 68% of all criterion trials, and all animals followed the intense trails (Ix and 3x) correctly 90%-100% of the time. As we reported previously (Kubie & Halpern, 1978), trailing accuracies increased as a function of extract concentration. We term the amount the mean tongueflick rate on Ix and 3x trails exceeds the mean tongue-flick rate on water trails the tongue-flick-rate elevation (Kubie & Halpern, 1978). All snakes exhibited statistically significant tongue-flick-rate elevations, ranging from 19 to 60 flicks per minute above the distilled water means. Tongue-flick rates .increased consistently with increasing extract concentrations. Open-field tongue-flick rates were dramatically below any of the tongue-flick rates recorded in the trailing maze. The snakes' mean open-field tongue-flick rates while moving ranged from 27.5 to 51.4 flicks per minute. With one exception, all snakes ate on all trials prior to surgery. This. one animal (Male I) failed to eat on four trials. Sham surgery. Anesthesia and sham surgery had no effect on trailing accuracy, tongue-flick-rate elevation, or feeding behavior. The behavior of snakes prior to these procedures was indistinguishable from their behavior following the procedures (Tables 1 and 2). Olfactory nerve cuts. Of the seven snakes subjected to olfactory nerve transection, six had damage to more than 94% of the olfactory nerve and less than 33% of the vomeronasal nerves (Table 3). Following surgery these snakes trailed at preoperative levels, demonstrated elevated tongue-flick rates, and ate on every opportunity (Tables 1 and 2; Figures 6 and 7). Male 63, in addition to having all olfactory nerve fibers severed, sustained complete disruption of its vomeronasal nerves and is considered separately below. Two snakes had more than 3 A thorough analysis of the preoperative behavior of these animals is included in a previous publication (Kubie & Halpern, 1978). 657 CHEMICAL SENSES IN SNAKE PREY TRAILING 95% of their olfactory nerve fascicles severed, with little or no damage to their vomeronasal nerves (Female 48 and Male 54). They exhibited no trailing deficits. The four snakes (Females 49 and XL; Males 1 and 6) that had their olfactory nerves totally severed bilaterally also sustained greater damage to their vomeronasal nerves, which caused up to 32% of the vomeronasal epithelium to degenerate, These animals continued to trail and trailed with accuracies not significantly below preoperative levels. The trailing, tongueflicking, and feeding behavior of snakes in the olfactory nerve cut group were indistin- Table 1 Trailing Accuracy on Criterion Trials Test block Criterion trial 2 3 Sham Female 37 1/27 and greater 77 Female 4A 1/27 and greater 74 77 (z = .00) 75.5 (z = .37) 24 (2 = 6.55*) 24.5 (z = 6.58*) 74 (2 = .68) 76 (z = .60) 1/27 and greater 86 17.3 (2 = 8.58*) 80 (2 = 1.37) 1/27 and greater 76 81 (2 = 1.27) 19.0 (2 = 7.10*) 59 (z = 2.34*) Subject Olfactory nerve cut Female 48 Female 49 75 ' 4 5 (2 = .22) Female XL 1/27 and greater 74 74 (2 = .00) Male 54 1/27 and greater 83 86 (2 = .62) Male 1 1/27 and greater 68 57 (2 = 1.24) Male 6 1/27 and greater 67 67 (z = .00) Partial vomeronasal nerve cut Female 38 1/27 and greater 73.6 Male 57 1/27 and greater 80.8 Male 74 1/243 and greater 78 Male 71 Complete vomeronasal nerve cut Female 45 Male 3A Female 145 Olfactory and vomeronasal nerve cut Male 63 1/27 and greater 87 1/27 and greater 72 95 76 95 74 90 1/27 and greater 1/81 and greater 1/9 and greater 71 80 59.8 (2 = 2.05* 50.4 (2 = 4.15*) 71 (2 = 1.26) 38.2° (2 = 6.50*) 24.2b 15° —C 10 25C 35C Note. Criterion trials include dry, dry-rewet, and Ix warm-up trials for all animals. After Test Block 2, nerves were cut; after Test Block 3, ducts were sutured; after Test Block 4, sutures were removed. Data are expressed as percentages. Surgery on Female 145 was originally an intended olfactory nerve cut, but histological examination revealed a complete vomeronasal nerve cut. Surgery on Male 63 was originally an intended olfactory nerve cut, but histological examination revealed complete bilateral destruction of both olfactory and vomeronasal nerves. For snakes that did not eat postoperatively, percentage correct on Ix warm-up trials is presented below those on criterion trials. a Accuracy above chance. b For 3 days (2 = 4.46, p <.01). 0 All below chance and significantly below preoperative levels (p <.01). * p <.01. 658 JOHN L. KUBIE AND MIMI HALPERN Table 2 Tongue-Flick-Rate Elevation for Snakes Prior to and Following Operative Procedures Test block 1 2 3 51.6 60.68 50.84 55.17 50.36 59.62 5.85a 6.39" 35.31 39.95 43.75 59.85 18.46 31.56 37.25 34.3 58.48 39.93 21.33 26.93 36.03 28.66 47.34 50.25 21.09 31.32 22.88 31.97 2.8" 47.4 50.36 41.32 18.88 26.32 35.31 46.45 33.73 26.23 33.11 40.79 37.68 6.52a -4.34a 31.8 37.90 30.90 54.58 33.50 21.97 60.35 2.1 39.1 40.5 Subject Sham Female 37 Female 4A Olfactory nerve cut Female 48 Female 49 Female XL Male 54 Malel Male 6 Partial vomeronasal nerve cut Female 38 Male 57 Male 74 Female 71 Complete vomeronasal nerve cut Female 45 Male 3A Female 145 Olfactory and vomeronasal nerve cut Male 63 4 5 Note. Tongue-flick-rate elevation is the mean amount that tongue-flick rates on Ix and 3x trails exceed mean tongue-flick rates on water trails. After Test Block 1, snakes were given anesthesia; after Test Block 2, they had surgery; after Test Block 3, ducts were sutured; after Test Block 4, sutures were removed. * Not significantly different from tongue-flick rate on water trails. guishable from their preoperative behavior and the behavior of control animals. Male 63 whose olfactory and vomeronasal nerves were bilaterally severed did not eat or trail after surgery (Figure 8). Since this snake never ate following surgery, it was tested only on Ix warm-up trails. It followed Ix trails at chance levels (40% correct) which was significantly below preoperative trailing accuracy (2 = 2.01, p < .05). Tongue-flick rates on these trails (Af = 62.5) were significantly below preoperative tongue-flick rates on Ix warm-up trials (M = 89.0 flicks per minute), £(39) = 8.40, p < .01. Postoperative open-field tongue-flick rates (M = 23 flicks/min) were not significantly below preoperative open-field rates (M = 28 flicks/min), £(23) = 1.43, p > .05. Table 3 Summary of Nerve and Mucosa Damage for Olfactory Nerve Cut Group % destruction in olfactory mucosa Visible olfactory nerve fibers Subject Left Right Left Right % destruction ore vomeronasal1 mucosa Malel Male 6 Male 54 Female XL Female 48 Female 49 100 100 100 100 99 100 100 no no no no yes no no no yes no yes no 33 31 18 13 0 19 Male 63 100 100 100 95 100 94 100 100 659 CHEMICAL SENSES IN SNAKE PREY TRAILING TEST BLOCK: i 1/243 1/9 3X 1/243 1/9 3X TRAIL 1/243 1/9 3X 1/243 1/9 3X 1/243 1/9 3X CONCENTRATION Figure 6. Trailing accuracy (percentage correct) as a function of trail extract concentration for one snake in each experimental group. (For Female 45 the percent correct on IX warm up trials is depicted for Block 2 to provide a comparison with Block 3. Since this animal ate only on three test days, the concentration profile is for three test days only. VN = vomeronasal.) Complete vomeronasal nerve cuts. We attempted complete vomeronasal nerve-cut surgery on three snakes. All three sustained lesions that appeared complete and caused total degeneration in the vomeronasal organs but left more than 78% of the main olfactory system apparently intact (Table 4). Fol- lowing surgery all three snakes exhibited severe feeding deficits (Figure 8). Two of these three ate one or fewer times and were tested only on Ix warm-up trails after surgery. They followed these trails in the chance range (Table 1). The third snake (Female 45) continued to eat sporadically for nr TEST BLOCK 6 SNAKE z °~ I 40FEMAUE48 £ . l<oJ FEMALE 45 w F E M A L E 4A w 20- I 40- 1/243 1/9 3X 1/243 1/9 3X 1/243 1/9 3X 1/243 1/9 3X 1/243 TRAIL CONCENTRATION Figure 7. Tongue-flick-rate elevation for one snake in each experimental group. (Each point is the mean tongue-flick-rate elevation the snake exhibited during correct trailing of one extract concentration over the 10-day period. VN = vomeronasal.) JOHN L. KUBIE AND MIMI HALPERN 660 COMPLETE VN CUT 14-: o Female 3A D U C T SUTURE 14-|TllillllliltlllH Female4A t t t t t t t t t PARTIAL VN CUT Female ifo tn i— S U0 tr 0LJ co 14- OIF + VN CUT 14 Female 45 0 Male 63 T T T T t t r t t t -DAYS DAYS - Figure 8. Postoperative prey attack of snakes exhibiting feeding deficits. (VN = vomeronasal.) 3 test days during which time it was tested on all trail types. During this period the snake followed criterion trials below chance levels (24.2% correct) and exhibited no tongue-flick-rate elevation (Figures 6 and 7). Tongue-flick rates during trailing of snakes with complete vomeronasal nerve cuts dropped toward the preoperative tongue-flick rates exhibited by these animals in the open field. For all three snakes, postoperative tongue-flick rates during trailing were significantly below the rates exhibited to any trail type (including distilled water) prior to surgery, but in the open field they remained unchanged. Male 3A's preoperative mean tongue-flick rate, for example, ranged from 90.3 (distilled water) to 118.1 (Ix) flicks per minute during trailing and was 38.6 flicks per minute in the open field. Following vomeronasal nerve transection this snake's mean tongue-flick rate on Ix warm-up trials fell to 64.9 flicks per minute—compared with preoperative Ix Table 4 Summary of Nerve and Mucosa Damage for Vomeronasal (VN) Nerve Cut Groups % destruction in VN mucosa Subject Complete nerve cut Female 45 Male 3A Female 145 Partial nerve cut Female 38 Male 71 Male 57 Male 74 Visible VN nerve fibers % destruction of olfactory mucosa Left Right Left Right 100 100 100 100 100 100 no no no no no no 16 22 5 13 96 56 38 11 94 81 34 yes yes yes yes yes no yes yes 7 5 0 4 CHEMICAL SENSES IN SNAKE PREY TRAILING warm-up, £(38) = 23, p < .01—whereas the open-field mean tongue-flick rate (36.8 flicks/min) was not significantly depressed, £(36) = .57,p>.l. Partial uomeronasal nerve cuts. The four snakes that sustained partial vomeronasal nerve cuts had either insignificant (Male 74) or significant (Female 38, Male 57, and Male 71) trailing deficits. The histological appearance of the accessory olfactory bulbs of these snakes indicated that for all four snakes there was scar formation dorsally between the bulbs but ventrally vomeronasal nerve fibers could be traced traveling unsevered into one or both accessory olfactory bulbs (Table 4). In general, the degree of the trailing deficit correlated with the severity of the nerve cut. The surgery inflicted on Male 74 spared the olfactory epithelium and caused bilateral degeneration in about 35% of the vomeronasal epithelium. This snake's trailing accuracy dropped insignificantly from 78% correct to 71% (z = 1.26). Male 57 apparently sustained damage to more than half of the nerves emanating from each vomeronasal organ, and its trailing accuracy dropped significantly from 80.8% correct to 50.4% correct (z = 4.15) but remained well above chance levels. Female 38 was allowed to survive for three postoperative test blocks (47 days), and there was minor histological evidence of vomeronasal nerve damage or degeneration in the vomeronasal epithelium. Since vomeronasal nerve fibers were clearly visualized separating during surgery, it may well be that this snake's epithelium regenerated (Wang, Guida, & Halpern, 1978) and therefore the amount of degeneration may not indicate the extent of the nerve cut. After surgery this snake's trailing accuracy dropped from 73.6% correct to 59.8% correct (z = 2.05, p < .01). Finally, Male 71 had the most complete vomeronasal nerve damage in this group—about 95% degeneration bilaterally. This snake's trailing accuracy dropped dramatically from 87% correct to 38.2% correct (2 = 6.50) which for the 110 criterion trials was significantly above chance. Male 71 was the only snake in the partial vomeronasal nerve cut group to refuse worm bits on any trials after surgery. This snake ate on all trials for nine test sessions, but on the 10th test session ate only on the first 661 trial, spit out the worm bit on the second trial, and refused to eat thereafter. This pattern of feeding decay is considerably more gradual than that seen in any of the snakes with complete vomeronasal nerve cuts, but it is similar to the feeding decay seen in Female 4A when it had its vomeronasal ducts sutured closed (see below). On the strongest trails three of the four snakes continued to exhibit levels of tongue-flick-rate elevation that were comparable with those seen before surgery. After surgery these snakes exhibited smaller tongue-flick-rate elevations to the trails with intermediate concentrations of earthworm extract. After Male 71 had its partial nerve-cut surgery, it exhibited the most severe trailing deficit of the group and exhibited a statistically insignificant tongueflick-rate elevation of 5.76 flicks per minute on Ix and 3x trails. Vomeronasal duct suture. Four snakes (Females 37, 4A, 48, and 38) had their vomeronasal ducts sutured closed after completion of Block 3 testing. Prior to Block 3 testing, Female 37 and Female 4A had been subjected to sham surgeries, Female 48 to olfactory nerve cut, and Female 38 to partial vomeronasal nerve cut. All four snakes trailed at chance levels following duct suture and failed to exhibit tongue-flick-rate elevations. Three of the four snakes (Females 37,48, and 38) ate throughout this postsurgical period. During the second and fifth test days of Block 4, Female 4A spat out worm bits on the final trials of the day; on the sixth, seventh, eighth, and ninth days it spat out the worm bit on the first trial and did not prey attack thereafter; finally, on the last day it did not prey attack at all. Following completion of Block 4 testing, the sutures closing the vomeronasal ducts were removed. During test Block 5, Female 4A ate on all trials, and all four snakes assumed patterns of trailing and tongue flicking that were indistinguishable from presuture patterns. Discussion Trailing. The results of these experiments support the conclusion of Experiment 1 that vomeronasal nerve cuts in garter snakes disrupt trailing. They also demon- 662 JOHN L. KUBIE AND MIMI HALPERN strate that the garter snake's vomeronasal system acts as the critical sensory system in earthworm-extract trailing. The behavior of two snakes (Male 3A and Female 145) with complete vomeronasal nerve cuts essentially replicated the results of the first experiment: These snakes did not follow earthworm-extract trails, and their feeding decayed on the first postoperative test day. Female 45, however, with a complete vomeronasal nerve cut, did continue to eat for 3 test days but gave no evidence of trail following during this period. Apparently, during the period when Female 45 was motivated by worm bits, it could not follow earthworm-extract trails. Partial vomeronasal nerve cuts had little or no effect on feeding behavior, yet three of the snakes with these lesions trailed significantly below preoperative levels but above chance. Partial nerve cuts appear to cause a decrease in sensitivity to earthworm-extract trails, resulting in loss of accurate trail following, especially of intermediate trail concentrations. Again, the preponderance of evidence suggests that the deficit is primarily sensory and not motivational. Finally, we come to the four snakes that had their vomeronasal ducts sutured closed. This technique of sensory deprivation was not used in Experiment 1. Its advantages over the nerve-cut procedure are that it is reversible and that it does not damage neural structures. One problem with this procedure is that in a number of animals (two of four) the duct sutures appeared to interfere with the mechanics of tongue flicking. Three of the four snakes with vomeronasal duct sutures ate on every trial, the fourth struck at worm bits until the final test session, but none gave any evidence of trail following. When these snakes had their duct sutures removed, their trailing accuracies returned to presuture levels. Both the partial vomeronasal nerve cut procedure and the vomeronasal duct suture procedure produce trailing deficits without comparable feeding deficits. We conclude, therefore, that a loss of sensitivity to earthworm-extract trails, not a motivational disruption, produces the trailing deficit seen in animals with damaged vomeronasal systems. Garter snakes with their vomeronasal nerves cut or their vomeronasal ducts su- tured closed still have a functioning olfactory system—in fact, these snakes can be thought of as olfactory preparations—yet this system alone is unable to sustain trailing. The olfactory apparatus does not appear to be an important receptor of earthworm-extracttrail-odor information. This conclusion is supported by the observation that garter snakes with lesions restricted to their olfactory nerves showed no trailing deficits. None of the six snakes with 95%-100% of their olfactory nerves severed exhibited trailing deficits. Four of these snakes had lesions that appeared to sever all olfactory nerve fibers, and they, therefore, could not have received any olfactory information during trail following. Snakes with olfactory nerves severed can be thought of as vomeronasal preparations. It appears that these animals rely solely on their vomeronasal organs for detecting trail odorants. Since snakes without a functioning olfactory system trail normally and snakes without a functioning vomeronasal system cannot detect trail odorants, we conclude that garter snakes are relying on their vomeronasal organs when they follow earthworm-extract trails. Tongue-flick patterns. We previously reported that garter snakes flick their tongues more rapidly when following intense trails of earthworm extract than when following weak trails (Kubie & Halpern, 1978). High-speed film analysis revealed that when garter snakes were following intense trails and exhibiting high tongue-flick rates, their individual tongue flicks were of extremely short excursion length and duration (Kubie, 1977). We were interested in whether vomeronasal or olfactory stimulation produced these tongue-flick responses. The two extreme alternatives are that during trailing, olfactory stimulation arouses the snake and causes increased tongue-flick exploration or that vomeronasal stimulation feeds back on tongue-flick motor mechanisms and leads to an increased rate of tongue flicking which affects subsequent stimulation of the vomeronasal organs. The present results support the latter speculation. Snakes without a functioning vomeronasal system (nerve cut or duct sutured) failed to exhibit elevated tongue-flick rates when following intense odor trails, and CHEMICAL SENSES IN SNAKE PREY TRAILING snakes without a functioning olfactory system (olfactory nerve cut) exhibited normally elevated tongue-flick rates while trailing. We have other evidence from film analysis that normal snakes and snakes with their olfactory nerves cut exhibit short-excursion and short-duration tongue flicks during trailing (Kubie, 1977). Apparently, stimulation of the vomeronasal organ produces the tongue-flick behaviors characteristic of trailing. Olfactory stimulation from potential prey may play an important role in activation of a quiescent snake, leading to locomotion and tongue flicking in search of the source of the odorant (Cowles & Phelan, 1958; Fox, 1952). Such a role for olfaction was not adequately tested in this experiment since the snakes after training were virtually always active in the maze, even when there was no trail to arouse them. This high level of activity may have been due to handling, to the structure of the maze, or to a conditioned expectation of food. Our results, therefore, should not be interpreted as evidence against the hypothesis that olfactory stimulation can arouse a quiescent snake to explore for food. Feeding. Although the primary purpose of this experiment was to investigate prey trailing, perplexing changes in feeding frequently accompanied vomeronasal system disruptions. A number of snakes with vomeronasal system disruptions showed a postoperative decay of feeding behavior. These snakes ate on the first postoperative trial, but over a period of time ranging from 1 to 10 test sessions, they stopped eating. A few snakes prey attacked and spit out worm bits on several trials before they stopped eating altogether. These changes suggest that the garter snake vomeronasal system, in addition to its critical role in prey trailing, plays an important role in eliciting prey attack and swallowing. These changes also suggest that the vomeronasal system may help to determine the reinforcing value of food. There is strong evidence that the vomeronasal organ is important in the feeding reactions of garter snakes. Wilde (1938) demonstrated that garter snakes make open-mouthed prey attacks at cotton swabs dipped in earthworm extract. Garter snakes do not make prey attacks at swabs dipped in 663 earthworm extract if their vomeronasal nerves are cut (Halpern & Frumin, in press; Wilde, 1938) or their tongues are removed (Burghardt & Pruitt, 1975; Wilde, 1938). Halpern (Note 1), on the other hand, demonstrated that when garter snakes are preoperatively trained to eat earthworm bits from a dish, they continue to eat earthworm bits from this dish after their vomeronasal nerves are severed. There are several alternative explanations for the decay of feeding we observed. One is that vomeronasal system disruptions cause a slow physiological change in the snake which decreases the snake's response to food. One k'nown change that takes place after vomeronasal nerve cuts is the degeneration of the vomeronasal mucosa (Kubie et al., 1977). One could speculate that a vomeronasal nerve lesion that was incomplete at surgery caused a scattered degeneration within the vomeronasal mucosa that slowly destroyed the functioning of the remaining bipolar neurons. The weaknesses in this hypothesis are that (a) it does not explain the spitting out behavior, (b) the snakes with complete vomeronasal nerve cuts that extinguished feeding appeared to have complete nerve cuts at the site of the lesion, and (c) Female 4A extinguished feeding when it had its vomeronasal duct sutures in place and resumed eating as soon as the duct sutures were removed. At present, we think that a classical conditioning hypothesis effectively explains the observed feeding decay. It postulates that for worm bits to maintain their reinforcing value, they must be accompanied by vomeronasal stimulation—vomeronasal stimulation is the unconditioned stimulus. According to this hypothesis, after vomeronasal deafferentation snakes exhibit a prey attack conditioned to the visual, tactile, and olfactory properties of the worm bits, but the conditioned prey attack extinguishes in the absence of reinforcing vomeronasal stimulation. The decay of feeding described above, especially the "spitting out" of worm bits, is evidence for a third function of the vomeronasal organ in the feeding behavior of snakes, the function of analyzing food that is in the mouth. The spitting-out behavior could be due to an expectation of sensation from food in the mouth which is not met. 664 JOHN L. KUBIE AND MIMI HALPERN Normal garter snakes prey attack worms that have pieces of dirt or twigs stuck to them; the snakes frequently swallow the worm bit and spit out the dirt or twig. Although there is no direct evidence, it may well be that the vomeronasal organ is screening food in the mouth and detecting its reinforcing value before the final act of swallowing. We should note that although all seven snakes in these experiments with complete vomeronasal nerve cuts eventually stopped eating, six other snakes with disruptions of their vomeronasal systems continued eating throughout postoperative testing. Three of the snakes that continued to eat had partial vomeronasal nerve cuts, and the absence of a feeding deficit is easily accounted for on the basis of the remaining intact vomeronasal nerve fibers. The other three snakes had their vomeronasal ducts sutured closed. The difference between their behavior and the behavior of snakes with complete vomeronasal nerve cuts is more difficult to explain. One possible explanation is that the vomeronasal duct suture does not eliminate all odorant access to the vomeronasal organs. If a duct-sutured animal can still detect food in its mouth with its vomeronasal organs, it should continue to eat. The evidence against this hypothesis is that ductsutured snakes cannot detect earth worm extract trails and that duct-sutured snakes cannot tongue flick radioactive proline into their vomeronasal organs (Kubie, 1977). The difference in the extinction pattern between vomeronasal nerve cut and ductsutured animals remains an enigma. General Discussion We have found that garter snakes deprived of vomeronasal sensation cannot follow earthworm-extract trails whereas garter snakes deprived of olfactory sensation trail normally, and we conclude that garter snakes are uniquely reliant on vomeronasal sensation for such prey trailing. Both the results and the conclusions of this study are in sharp contrast to those of Noble and Clausen (1936) in their study on sensory control of prey trailing in snakes. Using intervention techniques different from ours and studying the ability of snakes to follow crushed earthworm trails in an open field, they concluded that the olfactory system was more important for prey trailing than the vomeronasal system. There are four general differences between the Noble and Clausen study and our own. First, the two studies used different species of snakes. Noble and Clausen used Thamnophis sirtalis, whereas we used Thamnophis radix. Yet these two species of garter snakes are closely related and when both species were tested previously in this laboratory (Kubie & Halpern, 1978), they behaved similarly in the test situation employed here. The species differences, therefore, probably do not account for the differences observed. Second, Noble and Clausen used trails of water extracts of crushed earthworms, whereas we used trails of water extract of whole earthworms. Perhaps only crushing can release the olfactory stimulating components. Third, the procedures of sensory elimination used in these two studies are not comparable. We have previously commented on Noble and Clausen's use of nose plugs to block olfaction (Kubie et al., 1978). We found that when we plugged the nostrils of two garter snakes, they were forced to resort to open-mouthed breathing and one snake's breathing became labored. We speculated that a snake forced to resort to open-mouthed breathing was stressed and probably could not adequately tongue flick odorants to its vomeronasal organs. Noble and Clausen cut off the tongues of snakes and observed partial trailing deficits, but Wilde (1938) and Burghardt and Pruitt (1975) showed that tongueless snakes can respond to earthworm extract (presumably with their vomeronasal organs) if their lips contact the extract. Noble and Clausen also cauterized the vomeronasal organs and found partial trailing deficits. It is hard to know exactly what they meant by "cauterizing the Jacobson's organs." No detail is given. Getting a cautery inside the bony capsule of the vomeronasal organ would in fact be quite difficult and mutilating. We have attempted to cauterize the vomeronasal ducts of several snakes and found that many snakes forced open the ducts after several days (unpublished ob- CHEMICAL SENSES IN SNAKE PREY TRAILING 665 servations, 1975). Fourth, Noble and nonvolatile components of the prey trail may Clausen used a very different testing proce- be critical for prey trailing. We and others have observed that during dure than the one described here. Burghardt (1970) criticized the Noble and prey trailing, a snake keeps its head low to Clausen study for its qualitative data pre- the ground, exhibits pendular head movesentation, for its lack of statistical treatment, ments, and touches the substrate on virtually and for its failure to compare each individual every tongue flick (Kahmann, 1932; Kubie subject's trailing before and after surgery. & Halpern, 1975; Naulleau, 1965; Weideman, We think that our testing and surgical pro- 1931; Brock & Means, Note 2). In addition, cedures have demonstrated the garter its tongue-flick patterns change: The snake's critical dependence on the vomero- tongue-flick rate is very fast, and the tongue nasal system for earthworm-extract flicks are of a short-excursion, short-duratrailing. tion type. Snakes with vomeronasal nerve transection and normal snakes in the open field rarely exhibit these "trailing" tongue Role of the Tongue in Prey Trailing flicks but instead produce long-excursion, long-duration "exploratory" tongue flicks The tongue presumably plays two im- (Kubie, 1977). The trailing tongue-flick portant sensory roles in snake behavior. patterns that snakes exhibit in response to First, it serves as a tactile organ to "feel" the vomeronasal stimulation may be an opporsnake's environment during tongue flicks, tunistic trailing strategy. The short, quick and second, it serves to pick up odorants tongue flicks of trailing could provide precise from the external environment and deliver information on trail location, set a relatively them to the vomeronasal organs (Burghardt high threshold for detection, and deliver a & Pruitt, 1975; Kahmann, 1932; Kubie, maximum number of data points within a 1977). The evidence reported above given time frame. It appears as if, after the suggests that the tongue-flick delivery of tongue flick delivers odorants to the voodorants to the vomeronasal organs is critical meronasal organs, the snake uses the memfor prey trailing. We previously reported ory of previous tongue excursions to deter(Kubie & Halpern, 1978) that garter snakes mine the position of the odor trail. Whether have great difficulty following trails if the a snake can compare odorant concentration trails are removed from direct lingual access on the two tips of its tongue to determine a by a perforated floor, that garter snakes can trail location is not known but is experireadily follow dry trails, and that they follow mentally testable. Since tongue-flicking these dry trails more accurately if they are behavior samples the environments for vore-wet with water. These results suggest meronasal stimuli, careful analysis of this that the snake's tongue must have direct behavior can demonstrate a snake's strategy access to the chemical trail to follow it ac- for seeking odorants. curately and that a water medium aids the tongue in delivering odorants to the vomero- Prey Trailing as a Representative nasal organs. Sheffield, Law, and Bur- Vertebrate Vomeronasal Function ghardt (1968) demonstrated that at least Broman (1920), using snake prey trailing some of the components of earthworm extract that are effective in eliciting prey at- as an archetype, suggested that the functions tacks from garter snakes are nonvolatile. of the vertebrate vomeronasal organ could Tongue flicking is a behavioral mechanism best be summarized by the German word capable of delivering nonvolatile chemicals, Spursinn which means, roughly, the tracksuch as carbon black or proline, to the vo- ing of odors. Although our data support the meronasal organs (Kahmann, 1932; Kubie, contention that snakes do rely on their 1977), but no comparable mechanism for vomeronasal organs for prey trailing, the delivering nonvolatile odorants to the ol- vertebrate behaviors for which the vomerofactory mucosa exists. For these reasons, we nasal organ have been shown to be of imsuspect that the tongue's ability to pick up portance are not well described by Spursinn. 666 JOHN L. KUBIE AND MIMI HALPERN Anatomical Record, 1937, 69, 153-162. Broman, J. Das Organ vomeronasale Jacobsoni; ein Wassergeruchs-organ. Anatomische Hefte: Abt. 1. Arbeiten aus Anatomischen Instituten, 1920,58, 137-192. Burghardt, G. M. Stimulus control of the prey attack response in naive garter snakes. Psychonomic Science, 1966,4, 37-38. Burghardt, G. M. Chemical perception in reptiles. In J. W. Johnston, Jr., D. R. Moulton, & A. Turk (Eds.), Advances in chemoreception: Vol. 1. Communication by chemical signals. New York: AppletonCentury-Crofts, 1970. Burghardt, G. M., & Hess, E. H. Factors influencing the chemical release of prey attack in newborn snakes. Journal of Comparative and Physiological Psychology, 1968, 66, 289-295. Burghardt, G. M., & Pruitt, G. H. The role of the tongue and senses in feeding of naive and experienced garter snakes. Physiology and Behavior, 1975, 14, 185-194. Burns, B. Oral sensory papilla in sea snakes. Copeia, 1969, 617-619. Chiszar, D., & Carter, T. Reliability of individual differences between garter snakes (Thamophis radix) during repeated exposure to an open field. Bulletin of the Psychonomics Society, 1975, 5, 507-509. Cowles, R. B., & Phelan, R. L. Olfaction in rattlesnakes. Copeia, 1958, 77-83. Emson, P. C., Jessell, T., Paxinos, G., & Cuello, A. C. Substance P in the amygdaloid complex, bed nucleus and stria terminalis of the rat brain. Brain Research, 1978, 149, 97-105. Estes, R. D. The role of the vomeronasal organ in mammalian reproduction. Mammalia, 1972, 36, 313-341. Fox, W. Notes on the feeding habits of Pacific coast garter snakes. Herpetologica, 1952,8,4-8. Gehlbach, R. F., Watkins, J. F., & Kroll, J. C. Pheromone trail-following of typhlopid and colubrid snakes. Behaviour, 1971, 40, 282-294. Gloor, P., Murphy, J. T., & Dreifuss, J. J. Anatomical Reference Notes and physiological characteristics of the two amygdaloid projection systems to the ventromedial hypothalamus. In C. H. Hochman (Ed.), Limbic sys1. Halpern, M. Conditioned prey attack: sensory tem mechanisms and autonomic function. considerations. Paper presented at the meeting of Springfield, 111.: Charles C Thomas, 1972. the American Society of Icthyologists & HerpetoloHalpern, M., & Frumin, N. Roles of the vomeronasal gists, Williamsburg, Virginia, 1975. and olfactory systems in prey attack and feeding in 2. Brock, 0. G., & Means, D. B. Preliminary obseradult garter snakes. Physiology and Behavior, in vations on the prey trailing behavior of the eastern press. diamondback rattlesnake Crotalus adamanteus. Paper presented at the meeting of the American Herrick, C. J. The connections of the vomeronasal nerve, accessory olfactory bulb and amygdala in Society of Icthyologists & Herpetologists, 1977. amphibia. Journal of Comparative Neurology, 1921, 33, 213-280. References Hokfelt, T., Elde, R., Johansson, O., Terenius, L., & Stein, L. The distribution of enkephalin-immunoBaumann, F. Experimente iiber den Geruschinn der reactive cell bodies in the rat central nervous system. Viper. Revue Suisse de Zoologie, 1927, 34, 173Neuroscience Letters, 1977, 5, 25-31. 184. Hokfelt, T., Kellerth, J. O., Nilsson, G., & Pernow, B. Bodian, D. A method for staining nerve fibers and Substance P: Localization in the central nervous nerve endings in mounted paraffin sections. Anasystem and in some primary sensory neurons. tomical Record, 1936, 65, 89-97. Science, 1975, 190, 889-890. Bodian, D. The stain of paraffin sections of nervous Kahmann, H. Sennesphysiologische Studien an Reptissue with activated protargol: The role of fixatives. tilien: I. Experimentelle Untersuchungen fiber das The vomeronasal organs are of importance in garter snake mating behavior (Kubie et al, 1978), hamster mating behavior (Winans & Powers, 1977), and snake feeding reactions (Burghardt & Pruitt, 1975; Wilde, 1938). Alternate hypotheses that the vomeronasal organs serve as sex-scent detectors (Estes, 1972), as water smellers (Broman, 1920), or as smellers of food in the mouth (Herrick, 1921) are also too limited. We are not ready to propose a comprehensive theory of vomeronasal function. There is, however, evidence that suggests that the vertebrate vomeronasal organs may be sensitive to nonvolatile odorants (Sheffield et al., 1968), that they may require an active behavior of the animal, such as Flehmen, nose rubbing, or tongue-flicking, for odorant delivery (Estes, 1972; Kubie, 1977; Meredith, 1976), and that vomeronasal projection sites in the brain may be especially sensitive to hormones and humors (Emson, Jessell, Paxinos, & Cuello, 1978; Hokfelt, Elde, Johansson, Terenius, & Stein, 1977; Hokfelt, Kellerth, Nilsson, & Pernow, 1975; LaSalle & Ben-Ari, 1977; Pfaff, 1968; Sar & Stumpf, 1973). Although snake prey trailing may not be the representative vertebrate vomeronasal function, a more thorough understanding of how snakes use their well-developed vomeronasal apparatus may help us understand the functioning of this organ in other vertebrate groups. CHEMICAL SENSES IN SNAKE PREY TRAILING Jakobsonische Organ der Eidenschsen und Schlangen. Zoologische Jahrbiicher, Abteilungen filr Allgemeine Zoologie und Physiologie der Tiere, 1932, 51, 173-238. Klauber, L. M. Rattlesnakes (Vol. 2). Los Angeles: University of California Press, 1956. Kubie, J. L. The role of the vomeronasal organ in garter snake prey trailing and courtship. Unpublished doctoral dissertation, State University of New York, Downstate Medical Center, 1977. Kubie, J. L., & Halpern, M. Laboratory observations of trailing behavior in garter snakes. Journal of Comparative and Physiological Psychology, 1975, 89, 667-674. Kubie, J. L., & Halpern, M. Garter snake trailing behavior: Effects of varying prey-extract concentration and mode of prey-extract presentation. Journal of Comparative and Physiological Psychology, 1978, 92, 362-373. Kubie, J. L., Vagvolgyi, A., & Halpern, M. Roles of the vomeronasal and olfactory systems in courtship behavior of male garter snakes. Journal of Comparative and Physiological Psychology, 1978, 92, 627641. Kubie, J. L., Wang, R.-T., & Halpern, M. Observations on morphological changes in the vomeronasal organ of garter snakes following vomeronasal nerve section. Neuroscience Abstracts, 1977, 3, 80. LaSalle, G. L., & Ben-Ari, Y. Microiontophoretic effects of substance P on neurons of the medial amygdala and putamen of the rat. Brain Research, 1977, 735, 174-179. Meredith, M. Efferent control of stimulus access to the vomeronasal organ (VNO). Neuroscience Abstracts, 1976, 2,157. Naulleau, G. La biologie et le comportement predateur de Vipera aspis au laboratorie et dans la nature. Bulletin Biologique de la France et de la Belgique, 1965, 99, 395-524. Noble, G. K., & Clausen, H. J. The aggregative behavior of Storeria dekayi and other snakes, with especial reference to the sense organs involved. Ecological Monographs, 1936,6, 269-316. Oliver, J. A. The natural history of North American amphibians and reptiles. Princeton: Van Nos- 667 trand, 1955. Payne, A. The sense of smell in snakes. Journal of the Bombay Natural History Society, 1945, 45, 507515. Pfaff, D. W. Uptake of estradiol-17 0-H3 in the female rat brain: An autoradiographic study. Endocrinology, 1968,82, 1149-1155. Sar, M., & Stumpf, W. E. Autoradiographic localization of radioactivity in the rat brain after the injection of 1,2-3H testosterone. Endocrinology, 1973, 92, 251-256. Sheffield, L. P., Law, J. M., & Burghardt, G. M. On the nature of chemical food sign stimuli for newborn garter snakes. Communications in Behavioral Biology, 1968,2,7-12. Wang, R.-T., Guida, L., & Halpern, M. Postnatal neurogenesis and regeneration of the vomeronasal epithelium following axotomy in garter snakes. Neuroscience Abstracts, 1978,4,93. Wang, R.-T., Kubie, J. L., & Halpern, M. Brevital sodium: An effective anesthetic agent for performing surgery on small reptiles. Copeia, 1977, 738-743. (a) Wang, R.-T., Kubie, J. L., & Halpern, M. Observations on the normal structure of the vomeronasal apparatus of garter snakes. Neuroscience Abstracts, 1977, 3, 85. (b) Weideman, E. Zur biologie der Nahrungs-aufnahme eurapaischer Schlangen. Zoologische Jahrbiicher, Abteilungen fur Systematik, Okologie und Geographie der Tiere, 1931,61, 621-636. Wilde, W. S. The role of Jacobson's organ in the feeding reaction of the common garter snake, Thamnophis sirtalis sirtalis (Linn.). Journal of Experimental Zoology, 1938, 77, 445-465. Williams, E. J. Experimental designs balanced for the residual effects of treatments. Australian Journal of Scientific Research, 1949, 2,149-168. Winans, S. S., & Powers, J. B. Olfactory and vomeronasal deafferentation of male hamsters: Histological and behavioral analyses. Brain Research, 1977, 726, 325-344. Received September 7,1978