Many aquatic and semiaquatic animals live near a land-water boundary where knowledge of the compass directions that lead either to water or land is important for daily movements and seasonal migrations (e.g., Ferguson, 1971; Ugolini et al., 1988; Salmon and Wyneken, 1994). The axis perpendicular to the shoreline has been called the y-axis (Ferguson and Landreth, 1966). Y-axis orientation by amphibians has been studied extensively in both their natural environment (e.g., Ferguson et al., 1967; Landreth and Ferguson, 1967a, 1968) and in the laboratory (e.g., Taylor, 1972; Adler and Taylor, 1973; Taylor and Auburn, 1978). In laboratory experiments, amphibians will learn the direction of the y-axis in a training tank with an artificial shore. When subsequently tested in a circular arena, amphibians exhibit y-axis orientation relative to directional cues, such as celestial cues (reviewed by Ferguson, 1966, 1971; Adler, 1976) and the geomagnetic field (Phillips, 1986a, 1986b, 1987).

Laboratory experiments investigating y-axis orientation in amphibians have typically used training times from several days (Taylor, 1972; Phillips, 1986a; Phillips and Borland, 1992) to weeks (e.g., Taylor and Ferguson, 1970; Adler and Taylor, 1973; Taylor and Adler, 1973) or even months (Taylor and Adler, 1978). In addition, in many of these training protocols the animals were moved daily from their preferred end of the y-axis (e.g., deep water for aquatic amphibians, shore for terrestrial amphibians) to the opposite end of the y-axis. The animals, therefore, had to move along the y-axis every day during training to return to their preferred habitat (e.g., Taylor, 1972; Taylor and Adler, 1973; Taylor and Auburn, 1978). These laboratory studies imply that amphibians require a significant amount of time and experience to learn the y-axis in their environment. In contrast, field experiments have demonstrated that anurans can learn the direction of a new natural shore, relative to celestial cues, within just a few days (Ferguson and Landreth, 1966; Landreth and Ferguson, 1966).

The goal of the laboratory experiments described here was to determine whether a salamander, the Eastern Red-Spotted Newt Notophthalmus viridescens, could learn the direction of the y-axis with respect to the direction of the geomagnetic field within one overnight period (i.e., 12–16 h). Newts exhibit migratory behavior and have well-developed spatial orientation capabilities (Twitty, 1959; Twitty et al., 1964; Gill, 1978). In addition, aquatic adult newts may move onto land temporarily to forage, avoid heat stress, rid themselves of parasitic leeches, or to find winter hibernacula (Hurlbert, 1969; Gill, 1978). The geomagnetic field is one cue that newts use for both y-axis orientation and homing (Phillips, 1986a, 1986b; Phillips et al., 1995). Newts also use celestial cues (e.g., the sun and linearly polarized light) for y-axis orientation (Landreth and Ferguson, 1967a; Taylor and Auburn, 1978). However, like y-axis studies of other amphibians, orientation in newts has either been tested in populations caught from their natural shorelines or in laboratory studies in which newts were trained for several days or weeks. Therefore, it is unknown how quickly newts are able to learn the direction of the y-axis with respect to various orientation cues. Because newts, and other amphibians, are often active on rainy nights when cloud cover would preclude the use of celestial cues, (Dole, 1965; Shoop, 1965; Hurlbert, 1969), it seems likely that the geomagnetic field would be a salient cue for determining the direction of the y-axis and, therefore, should be quickly learned.

Materials and Methods

Experiments were performed between August and January during 1994, 1995, and 1996. Adult, aquatic-phase newts were either collected from local ponds in the vicinity of Bloomington, Indiana (Crane Naval Base, with permission from the Department of Natural Resources) or purchased from a commercial supplier (Charles Sullivan, Inc., Nashville, TN). Locally collected newts were returned to their ponds after testing. Commercially purchased newts were returned to the supplier. Prior to training and testing, newts were held indoors in 120-liter aquaria at our testing facility and fed salmon chow (Rangen, Inc.) three times per week. At least five days prior to being placed into y-axis training tanks, the newts were put into new holding tanks in which the water depth was 1 cm or less. Exposure to shallow water causes newts to develop more terrestrial-like characteristics, such as a reduced tail fin and cornified skin, which permit greater mobility in our terrestrial testing arena (Phillips and Borland, 1994).

The basic training and testing protocols have been described in detail by Phillips and Borland (1994). Briefly, each training tank consisted of a 120-liter aquarium containing a sloping Plexiglas bottom to provide a shore at one end of the tank (Fig. 1). Two training tanks with perpendicular shore directions (i.e., west and south) were used for these experiments. The training tank with a west-facing shore was located outside, whereas the other training tank was located in a small greenhouse connected to our testing facility. Both training tanks were located at least 6.5 m from any ferrous metal or sources of electromagnetic disturbances (such as electric heaters, computers, etc.). An open-top, grey polyvinylchloride (PVC) chute (86 cm × 20 cm × 9 cm) was attached to the shore end of each tank so that it sloped down toward the water (Fig. 1). Thirty to 45 newts were individually placed on the chute so that they faced toward the water. The newts were allowed 30 min to walk down the chute into the training tank. Newts that turned and headed up the chute were turned around until they headed into the water. After all the newts entered the water, the chute was removed (note, in previous experiments a chute was not attached to the end of the tank and newts were placed directly into the water; Phillips 1986a, 1986b; Phillips and Borland 1992). The time of training varied slightly in each test but started between 1400 and 1800 h for all tests and was always completed at least 90 min before sunset. The water temperature in each tank was maintained between 15 and 20 C until the following morning, at which time the temperature was increased to approximately 31–32 C within 1.5 to 2 h. An increase in water temperature increases the tendency of newts to leave water (Gill, 1978) and orient toward shore (Phillips, 1987). After the water heated up, the newts were tested for y-axis orientation (just 12–16 h after they walked into the tank).

View Figure

• Download as Powerpoint
• Download Figure

For testing, each newt was individually removed by hand from the shallow end of the training tank, placed in a light-tight plastic container, and carried to the indoor testing arena. The newt was removed from the container in complete darkness and gently placed into a hydraulically controlled release device in the center of the testing arena. The experimenter then quietly went into an observation room adjacent to the testing room and opened a shutter, which allowed light to illuminate the arena from above. The 72-cm diameter circular testing arena consisted of a black surround (31.5 cm high) and a Pyrex floor. Under the Pyrex floor was a layer of Plexiglas radially marked every 5° so that the experimenter could accurately score the directional bearing of each newt. The light from above cast a silhouette of the newt onto the translucent arena floor which the experimenter could observe via a video monitoring system.

Sixty seconds after the arena light was turned on, the release device was lowered, and the newt was free to walk within the arena. Each newt's directional response was recorded when its head crossed a 20-cm radius criterion circle centered on the release device. As in previous experiments (Phillips, 1986a; Phillips and Borland, 1992, 1994), several behavioral and time criteria were used to score the directional responses of the newts. Newts that struggled excessively during any part of the transport to the arena, or while the release device was lowered, were not tested. Newts that crossed the 20-cm mark in less than 30 sec were not scored. Previous experiments have shown that newts that score in less than 30 sec exhibit a randomly oriented escape response (Phillips, 1986a; MED, unpubl. data). In addition, trials were discontinued if the newt did not move off the release device within 10 min or if the newt did not cross the 20-cm criterion within 15 min.

Each newt was tested only once in one of four symmetrical alignments of an earth-strength magnetic field [i.e., magnetic North (mN) = geographic North (gN), mN = gE, mN = gW, or mN = gS]. The horizontal direction of the magnetic field in the arena was altered using two orthogonally oriented, double-wrapped, 1 m Rubens coils (Rubens, 1945; Kirschvink, 1992) around the outside of the arena (for a complete description, see Phillips, 1986a; Phillips and Borland, 1994). Magnetic inclination and total intensity of the rotated magnetic fields were within 1° and 3%, respectively, of the ambient geomagnetic field. Magnetic field parameters were measured using a Develco model 105395 three-axis fluxgate magnetometer. In each test, a total of at least four newts were scored, one in each of the four alignments of the magnetic field. Data were pooled from six tests from each of the two training-tank alignments. By testing the directional response of an equal number of newts in the four different field alignments and pooling the results with respect to the magnetic field direction during testing, we could assess whether the newts were orienting with respect to the direction of the magnetic field during testing (Phillips, 1986a).

Newts are extremely sensitive to small changes in the slope of the arena floor (Phillips and Borland, 1994). Before each test, the arena was leveled using a Lucas Shaevitz LSO inclinometer. In addition, the directional responses of two to four pretest newts were recorded in different alignments of the magnetic field. If the pretest newts exhibited a consistent topographic bias, the arena was releveled, and directional responses of two to four more pretest newts were observed. Once the pretest newts did not show a topographic bias, we began trials to test the newts' response to the magnetic field.

All data were normalized with respect to the direction of magnetic north during testing. The data were analyzed using standard circular statistics (Batschelet, 1981). Statistics for bimodal distributions were calculated by doubling the magnetic bearings. Mean vectors were calculated and tested for significance using the Rayleigh test. Ninety-five percent confidence intervals were used to determine whether the mean for the distribution included the shoreward direction (see Batschelet, 1981). A Watson U²-test was used to test for differences between the distributions of magnetic bearings obtained from newts trained in tanks with different shore alignments.

Results

Newts trained in the tank with the shore to the west exhibited significant bimodal orientation (79–259°, r = 0.35, n = 34, P < 0.03) relative to the direction of the magnetic field (Fig. 2A,D). The 95% confidence interval for the axis of orientation includes the direction of the y-axis (270° for shore, 90° for deep water) indicating that the newts oriented along a magnetic axis coinciding with the shoreward/waterward axis in the training tank. Newts trained in the tank with a south shore also exhibited significant bimodal orientation (178–358°, r = 0.40, n = 33, P < 0.007) relative to the direction of the magnetic field (Fig. 2B,E). Again, the 95% confidence interval for the mean vector includes the direction of the y-axis (180–360°). The difference between the distributions of magnetic bearings exhibited by newts trained to a west shore and a south shore is highly significant (Watson U² = 0.436, P < 0.001, on doubled angles), and the mean axes of orientation for the two distributions differ by approximately 90°. Furthermore, pooling the bearings of all the newts relative to the magnetic direction of shore in training results in a highly significant bimodal distribution (173–353°, r = 0.37, n = 67, P < 0.001) coinciding with the y-axis (Fig. 2C,F).

View Figure

• Download as Powerpoint
• Download Figure

Discussion

Evidence for geomagnetic orientation in some animals has been criticized because of insufficient replicability and the small magnitude of the effects reported (Griffin, 1975, 1982). However, the evidence for geomagnetic orientation in the Eastern red-spotted newt is particularly robust. Newts use geomagnetic information for two orientation tasks: y-axis orientation and homing (Phillips, 1986a, 1986b, 1987), and the effect of the magnetic field on these behaviors has been replicated in several experiments (Phillips and Borland, 1992, 1994; Phillips et al., 1995) as well as here. In addition, in our studies on newts, the magnitude of the effect of the geomagnetic field is relatively large in comparison to studies of other animals. In experiments testing magnetic orientation in migratory birds, for example, the directional tendency of individual birds is assessed over several trials, and a mean vector is calculated for each bird. Often the means for individual birds are not significant, and statistical significance is achieved only by analysis of the second-order distribution of means (e.g., Wiltschko and Wiltschko, 1972, 1995a; Able 1994; but, for exceptions, see Wiltschko and Wiltschko, 1995b; Munro et al., 1997). In our tests of newt magnetic orientation, individuals were tested only once, and significance is determined based on the first-order distribution of magnetic bearings.

Two advantages of our laboratory studies in newts have contributed to the strong case for magnetic sensitivity. First, Phillips (1987) has shown that manipulating water temperature just prior to testing significantly increases the tendency of the newts for both homing and shoreward orientation, thereby increasing the consistency of their directional choices. Second, newts have a relatively small body size compared to the size of our testing arena and are not restrained once the release device is lowered. In contrast, laboratory experiments designed to test magnetic orientation in migratory birds and other animals often confine the subjects preventing normal movements associated with migration. Restricting an animal's movement may influence the accuracy of their directional choice. For example, free-flying birds, such as homing pigeons, appear to exhibit stronger magnetic orientation than captive migrants (e.g., Keeton, 1969, 1971).

Our experiments indicate that newts can learn the direction of the y-axis with respect to the geomagnetic field within 12 to 16 h. Allowing newts to actively walk into the tank from the shoreward direction on the PVC chute may have facilitated learning of the y-axis. Experiments in which newts were placed directly into the water of the y-axis training tanks resulted in significant unimodal y-axis orientation after five days in the training tanks (Phillips, 1986a, 1986b; Phillips and Borland, 1992). In contrast, the short training protocol used here resulted in bimodal magnetic orientation along the y-axis. Bimodal y-axis orientation has been previously reported for a variety of amphibian species, directional cues, and testing models (e.g., Ferguson et al., 1968; Landreth and Ferguson, 1967b; Jordan et al., 1968). Ferguson (1971) suggested that bimodal orientation along the y-axis might occur because some individuals orient toward shore and others toward deep water. Developmental differences in y-axis orientation within ambystomatid salamanders support Ferguson's idea (Taylor, 1972; Tomson and Ferguson, 1972; Adler and Taylor, 1980). As would be expected of a primarily aquatic amphibian, larval Ambystoma sp. tested before the beginning of gill reabsorption oriented predominantly toward deep water. Larvae in the gill reabsorption stage, however, oriented bimodally along the y-axis. After the gills were completely reabsorbed, juveniles oriented predominantly toward the shore.

There are at least two possible explanations for the bimodal magnetic orientation in newts after the short training time. One is that the short training time (12–16 h) may only have allowed the newts enough time to “learn” the y-axis but not on which end of the y-axis the shore was located. Hence, when tested in the arena, the newts oriented toward either end of the y-axis in an attempt to go toward shore. Alternatively, newts may have learned the direction of the shore, but the short training period may not have elicited the proper “motivational,” or physiological, state to result in unimodal orientation toward shore. Similar to Ferguson's (1971) hypothesis for bimodality, some newts might have preferred to orient toward shore and others toward deep water.

The environmental physiology of habitat preference in newts, as well as differences between the five-day and short (less than one-day) training protocols, suggests that the polarity of newt y-axis orientation may be influenced by a hormonally induced physiological state. Habitat preference in adult newts is influenced by the hormones thyroxin and prolactin (Tassava and Kuenzli, 1979; Dent, 1985; Moriya and Dent, 1986). Preference for a terrestrial habitat (so-called land drive) results from an increase in thyroxin and a corresponding decrease in prolactin. “Water drive” results from an increase in prolactin and a complementary decrease in thyroxin. Besides the difference in training time, conspecific density and food availability, both of which can influence thyroxin levels in amphibians (Denver, 1997a, 1997b), differed between the five-day and short (less than one-day) training experiments. In five-day experiments (Phillips, 1986a, 1987; Phillips and Borland, 1992), approximately 60 newts were in each training tank, whereas only 30–45 newts were in each tank for the short training experiments. In addition, during both training periods, newts were not fed. Five days of food deprivation in conjunction with a high density of conspecifics may have caused an increase in thyroxin and promoted land drive (Phillips, 1986a; Phillips and Borland, 1992, 1994). Short training (and hence a shorter time of food deprivation), combined with the lower conspecific density, may not have been sufficient to elicit the hormonal changes resulting in land drive in all the newts. A mix of newts in land drive and water drive condition would result in a bimodal distribution, with some newts oriented toward shore and others toward deep water.

Experiments in which thyroxin or prolactin is administered to newts prior to training may provide evidence which would support or refute the “hormonal hypothesis” for bimodality (but see Adler and Taylor, 1980). If hormonal treatment were to have no effect on the polarity of newt y-axis orientation, it would suggest that, after less than one day of training, newts are unable to distinguish the direction of shore along the y-axis using only magnetic cues (i.e., supporting a learning basis for the bimodality of the newts' response).