Cane toads

Movement and dispersal in established and invading toad populations

Lin Schwarzkopf and Ross A. Alford

School of Tropical Biology, James Cook University , Townsville, Qld. 4811

Abstract

Dispersal ability is an important factor determining the success of invading organisms. Several independent pieces of evidence, from mark-recapture and genetics studies, suggested that adult cane toads disperse widely and may be nomadic. We measured individual movement in 64 cane toads using radio tracking, in two locations, one where toads had just arrived (Cape York Peninsula), and one where toad populations had been established for around 55 years (near Townsville). Our data indicate that adult toads move long distances and tend not to return to their place of origin, especially in the wet season. Toads crossed a variety of terrains, including steep embankments and wide, fast-flowing rivers, without difficulty. The movements of toads in the two populations were similar, although toads in the ‘new’ population were more likely to move in straight lines (away from their place of origin) than were toads in established populations. This difference may be due to differences in climate in the two locations, or to differences between new and established population characteristics. We used the individual movement parameters we measured to construct an individual-based, correlated-random-walk model that predicted toad movements well, especially in the wet season.

Introduction

Dispersal is a central part of the behavioural repertoire of most organisms, because it allows them to reach new and potentially benign environments that may enhance fitness. Dispersal ability determines the success of invading species, because successful invaders spread rapidly into new habitats (South & Kenward 2001, Trakhtenbrot et al. 2005). Similarly, dispersal ability in part determines the influence that invaders will have on local species, because it determines the number of habitats and the amount of area that will be influenced by the invaders.

Cane toads (Bufo marinus) were introduced to Australia in 1935, at several rural centres in coastal north Queensland (Mungomery 1935). Since then, they have spread into northern New South Wales , and western parts of the Northern Territory (Sutherst et al. 1995), as far as the Victoria River (I. Morris & G. Saywer, pers. comm.) The rapid spread of toads has been a feature of their highly successful invasion of Australia (Freeland & Kerin 1988) and episodes of rapid dispersal might be due to eggs and tadpoles washing downstream of breeding sites (Lever 2001, Van Dam et al. 2002). However, observations made when toads first arrive in an area suggest that it is adults, rather than juveniles, that first appear, and that adults may take more than a year to begin reproducing in a new location (Alford et al. 1995; pers. obs.). Thus, terrestrial adults may be the dispersal phase in cane toads.

Further evidence suggesting that adult toads may be important in the spread of toads comes from mark-recapture studies. Such studies suggest that the disappearance rate of adults is higher than the mortality rate (Alford et al. 1995). Parameterised population models of cane toad life-history suggest that adult survival rates are in the order of 30 to 70 % per year (Lampo & De Leo 1998), but the rate at which marked adult toads disappear from marked populations is much higher (>90% per year), suggesting that adults are dispersing out of the marked population (Lampo & De Leo 1998, Schwarzkopf & Alford 2002).

Observational and mark-recapture studies both suggest that adult cane toads are the important dispersal phase for this species, so we tested this by quantifying the movements of individual adult toads. Knowledge of individual movement patterns of introduced organisms is useful during implementation of control measures, such as trapping or disease. In addition, individual movement measures can be used to refine mathematical diffusion models, more effectively predicting habitat use and spread of introduced organisms (Turchin 1998). We used our empirical measures of movement to construct an individual-based model of toad movements (Schwarzkopf & Alford 2002). Our model can be used to predict the spread of toads into similar habitats in new areas.

Materials and Methods

Study sites

This study was conducted from January 1992 to March 1993 at two locations in north Queensland : near Townsville on “Table Top” cattle station (19° 23' S, 146° 26' E) and on Cape York Peninsula at Heathlands Ranger Station (11° 46' S 142° 36' E). These sites covered most of the available range of times since colonisation: Townsville colonised for about 55 years, and Heathlands newly invaded at the time of the study. Tracking periods were selected to cover seasonal variation in weather; they coincided with the wet season (January-February), late wet season (March–May), and dry season (July–November). Thus, three radio-tracking periods were conducted at each location (Schwarzkopf & Alford 2002). We measured daily rainfall on each tracking day at both study sites.

In general, the Townsville area is much drier and reaches cooler night temperatures than Heathlands, and this was true during our study (Schwarzkopf & Alford 2002). The vegetation at our study site near Townsville consisted of open, grassy fields with scattered trees, interspersed with eucalypt woodland. The vegetation in creek beds consists mainly of introduced Lantana bushes, with emergent trees. Creeks near Townsville hold water for short periods, but are usually dry, and water is often available only at artificial sources, such as dams.

At Heathlands Ranger Station, there were grassy fields with few emergent trees, tall and short Acacia heath on ridges, and vine thicket with an understorey of ferns (Pteridium sp.) in creek beds. Water at Heathlands is available throughout the year in creeks, and additional water becomes available in gullies during the wet season.

Tracking

One trip was conducted during each season at each location, and 9–12 toads were tracked each trip. Toads were captured by hand and returned to the laboratory for transmitter attachment. Radio-transmitters (Sirtrack NZ 3.5 g, 100 mm trailing whip antenna) were tied around the toads’ inguinal area using surgical gauze, except for 12 toads, into which transmitters, equipped with loop antennas, were surgically implanted. External attachment was quicker and easier, provided a greater detection radius, and did not result in chafing or other discomfort for the toads, so we used this method of attachment for most toads. We tracked toads with a Telonics TM TR-4 receiver for up to 30 days, determined by the battery life of the transmitters and the movements of the toads. Toads were located once during daylight hours ( 09:00-16:00 ) and twice per night ( 19:00-20:00 , and 23:00-1:00 ). Shelter sites were described (substrate, cover, distance from nearest water, shade) for every toad every day, and behaviour of each toad was recorded every evening when they were sighted. More details on field sites, field trip dates, weather, tracking and transmitter attachment are available in Schwarzkopf and Alford (2002).

Movement data

Toad paths were described using total displacement over the period of tracking (standardised to displacement per day), and ‘straightness’, the quotient of total displacement divided by total distance travelled (Sinsch 1988). A value of straightness near one indicates that the path was straight, whereas a value near zero indicates that little net movement occurred.

Because it was likely that toads would move shorter distances in the 4 h between nocturnal tracking locations than in the 8-12 hs available to them to move before the following diurnal sample, we used only the nocturnal point at which the toad was located farthest from the previous daytime location in analyses of distance moved and in creating empirical distributions of distances moved for use in our model.

Movement was extremely variable, and movement variables were not normally distributed; therefore we log-transformed all distance measures, and arcsine-square-root transformed straightness (a ratio) to normalise the distributions before comparing seasons (Sokal & Rohlf 1981).

To examine how weather affected movement, we calculated mean values for each transformed movement variable for each sampling period and correlated these values with total mm of rainfall.

Random Movement model

To determine the extent to which movement appeared to be nomadic in each season and location, we constructed a computer simulation model of toads that had the same distributions of daily movement parameters as the real toads for each locality and season, but made movement decisions at random. We used the model to simulate 30-d movement patterns of 10,000 toads for each location in each season. We then compared the summary movement parameters of real radio-tracked toads to the distributions of summary parameters for 10,000 toads making decisions at random generated by our model, to determine whether real toads appeared to be moving nomadically or not.

The model incorporated random decisions regarding three aspects of movement:

• the probability of moving from a point;
• the probability of returning to a point after moving from it;
• and the distance moved and angle of each movement relative to the previous movement.

We modelled movement behaviour as a three-stage process. The first stage was the decision to move at all. At each iteration of the model, toads “decided” whether to move with a probability equal to the measured mean probability of movement of toads for the location and season that was being simulated. Once a simulated toad “decided” to move, it then decided whether the movement should be a return to the point it had most recently moved from. Again, the probability of this decision was set to equal the observed probability for toads at the location, season, and time of day being simulated. Finally, if the toad did not return to its previous location, it chose an angle to move at and a distance to move (the third stage). It chose these randomly from empirical distributions measured for the actual toads we had tracked at that locality and season (Schwarzkopf and Alford 2002).

To summarise :

each model toad had empirically determined probabilities of leaving a site and returning to the same site. On any given iteration of the model, if the toad emerged, it either returned to its previous site or it moved some distance at some angle, both randomly chosen from measured distributions appropriate to location, season, and (for some parameters) time of day. The model was iterated over 30 days to determine expected movement parameters over 5-30 days of tracking, mimicking measurements made during tracking.

Movement parameters generated by the model for simulated toads were total distance moved and straightness. We compared the movement parameters of real toads to those of simulated toads by using the means, and upper and lower 68 th percentiles (= 1 standard deviation from the mean if distributions are normal) for simulated toads on each day of the simulation to convert the final parameters for each real toad into parameters standardized to the distribution generated by the simulation. Boxplots of these standardized movements of real toads appear in Fig 1A&B. If the movements of real and simulated toads matched perfectly, each boxplot would be symmetrical, with the median at zero, the upper and lower quartiles just outside +/- 1 standard deviation, and the tips of the range bars near +/- 2 standard deviations. We used this as the standard of comparison when discussing how the movement patterns of real toads depart from those of our simulated animals.

Using the model to predict movement of toads

We used the movement rates predicted by our model to predict an annual rate of dispersal for cane toads by using our model’s predictions for monthly displacement in each of the three seasons to predict cumulative displacement for toads moving through a full year.

We examined two possible scenarios; in the first, individual toads had biases towards greater or lesser degrees of displacement, so that toads that tend to choose movement parameters that lead to greater displacement always made such choices (shown in Figure 1A), and in the second, individuals always choose movement parameters completely at random (Figure 1B).

In both models, we assumed that toads move using wet-season parameters for four months during the summer, using late wet parameters for the succeeding two months, and using dry season parameters for the next six months.

Figure 1. Simulations of cumulative displacement (dispersal) of toads over one year (4 mo wet season, 2 mo late wet season, 6 mo dry season) using parameters measured at Heathlands: A. Results if toads show individual biases, so that individuals that disperse at median rates always do so, ones that disperse at 99 th percentile rates always do so, etc; B. Results if there are no individual biases, so that individuals that happen to disperse long distances in one time period are no more likely than other individuals to continue to disperse in the next time period.

To model the first scenario, we simulated 10,000 toads moving for one month in each season using the movement parameters we measured in the population near the front of range expansion at Heathlands. We used the results of these simulations to determine predicted monthly displacements for toads with median and 1%, 5%, 95%, and 99% of maximum displacements per month for each season. We assumed that toads that move at each of these rates in one month will continue to move at these rates for a full year, so the cumulative displacement for the 99 th percentile toad is simply the sum of 99 th percentile monthly displacements.

To model the second scenario, in which all toads always choose movement parameters at random, we ran our simulation over a full year for 10,000 toads, and varied the sets of parameters from which they chose seasonally, as described above.

Results

Tracking periods and behaviour of tracked toads

We radio-tracked a total of 66 toads for periods ranging from 5 to 33 days (Schwarzkopf & Alford 2002). We assumed that toads that disappeared suddenly after one or several long movements, and were never recovered, moved out of receiver range rather than experiencing transmitter failure, although it is possible that transmitter failure may be responsible for some “missing” toads. There were no significant differences between movement parameters (distances moved, probabilities of leaving and returning to sites, and turning angles) of male and female toads (Schwarzkopf & Alford 2002), so we did not distinguish between sexes when constructing the random movement model.

Although there were no statistically significant differences between the sexes in the measured movement parameters, it appeared to us that male and female toads moved differently in some respects. Males tended to restrict their movements to wetter areas (creekbeds and low lying areas), and to move between calling sites (Figure 2), whereas females tended to avoid water and to move across country (Figure 3). Females only remained at water bodies if they were gravid. Females sometimes carried amplexed males on their backs for periods up to 3 days before depositing eggs.

Toads were able to move across all the habitats found at our study sites, including those that were difficult for us, such as dense heath and steep embankments (pers. obs). They also easily crossed rivers in flood and open dry areas, such as paddocks in the dry season (pers. obs.) Toads were able to move long distances (at least 700 m per night) even under the driest conditions in Townsville (Figure 4A).

Movements and environmental conditions

Both temperature and rainfall may influence toad movements (e.g. Carey 1978, Smits 1984, Adler & Taylor 1981, Preest & Pough 1989, Sinsch 1989, Jørgensen 1991, Cohen & Alford 1996). On a seasonal scale, there were important differences among tracking periods both in terms of weather and of movement by toads.

Figure 2. Movements of 2 male toads at Heathlands Ranger Station in the wet season. Code: ▬ roads; ▬ water courses; ▬ movements of one male over 27 days; ▬ movements of another male over 32 days; ˜ calling sites at the time of tracking.

Figure 3. Movements of 2 female toads at Heathlands Ranger Station in the wet season Code: ▬ roads; ▬ water courses; ▬ movements of one female over 30 days; ▬ movements of another female over 28 days.

A

B

Figure 4 . Frequency distributions, presented by season and study locations: A. Movements per night by individual toads. Most toads only moved short distances per night, although occasionally long distance movements occurred; B. Turning angles (between successive mapped movements) used by toads. Toads commonly made sharp turns, at or close to 180°, but shallower turning angles, resulting in straighter lines of travel were common in the wet season at both locations.

There was a significant, positive correlation between mean total displacement and total rainfall (Figure 5A), and between straightness and total rainfall (Figure 5B). In general, these data suggest that toad paths are straighter when it is wetter and warmer, leading to greater total displacements.

Figure 5. The relationship between: A. Mean straightness in each tracking period and total rainfall for each period, and B. mean total displacement in each tracking period and total rainfall (mm per trip). Code: ¢ means for Townsville; £ means for Heathlands Ranger Station. Total rainfall and displacement are plotted on logarithmic scales to show data as they were analysed.

Movement characteristics of toads

Toads usually emerged from their diurnal shelter by 20:00 , or failed to emerge for that night. After emergence, toads often spent short periods hydrating before foraging, or remained at water to call or mate (especially males). Distances moved by individual toads within single nights were extremely variable, ranging from 0 m - 1.3 km (Figure 4A).

There were significant differences in movement parameter (Figures 4A & B) distributions among seasons and between locations (MANOVAs on distributions for individuals, p > 0.05, details in Schwarzkopf & Alford 2002). Most individual movements between locations were less than 100 m, and typically distributions of distances moved were uni-modal and left-skewed. Modal movements at Townsville tended to be longer than those at Heathlands, especially in the dry season, because toads moved a long distance between shelter sites (mostly hollow logs) and water in the dam. At Heathlands, water was more abundant in creeks even in the dry season, and toads tended to be distributed along riparian areas that were relatively moist, flattening the movement distribution at Heathlands in the dry season (Figure 4A). There were individual long movements in all seasons, although the longest move was made by a toad at Heathlands in the wet season.

Turning angles were highly biased: even after removal of events in which animals returned to previous shelter sites, toads tended to turn at large (close to 180°) angles (Figure 4B). Large turning angles indicate that toads tended to repeatedly move between local areas for shelter and activity, without returning to exactly the same spots in each area. A very high proportion of large turning angles are evident in all seasons in Townsville, and at Heathlands in the late wet and dry seasons (Figure 4B). In the wet season, Heathlands had strikingly fewer near 180° turns, indicating that toads were least likely to return to the same general locality for feeding or shelter at Heathlands in this season.

Random movement model

In the wet season, generalizing over both locations and all movement parameters, the boxplots (Figure 6) suggest that the movements of real toads were similar to those predicted by the model. The median of the standardized final values for each movement parameter is close to the expected mean of zero; the upper and lower quartiles are close to +/- 1 standard deviation from zero, and the ranges approximate the range of approximately +/- 2 SDs that might be expected from samples sizes of approximately 10 data points. On the other hand, examining the late wet and dry seasons, straightness, and final displacement, measured for toads at both locations show poorer fits to the predicted data. In the late wet, and especially the dry seasons, real toads have lower straightness and final displacement than predicted (i.e. the mean is displaced below zero, and the quartiles and ranges do not extend as far upwards as expected, Figure 6). In the dry season data, the medians for the real data are typically at approximately 1 SD or more below the mean of the simulated toads.

These results indicate that real toad movements are generally similar to those predicted by the model during the wetter times of year, but that they tend to be less linear and extensive than those predicted during the late wet, and are almost always less during the dry season. Similarly, movements for Townsville tend to be less similar to the model, especially in the late wet and the dry season, compared to Heathlands.

Figure 6. Boxplots of movement data for real toads standardised to the data for the random movement model. The mean for each box is indicated by the horizontal bar. When the data for the model and real movements are similar, the mean and ‘box’ portion of the boxplots fall within the grey portion of each graph. Individual points represent toads with parameter values that are outliers or extremes, which are more than 1.5 times the interquartile interval from the 75% quartiles. A. Total distance moved; B. Straightness.

What kind of movements does the model predict over a year?

If individual toads differ, such that there are some individuals that tend to select long distance movements, then, the 99 th percentile toad would disperse more than 20 km per year, and that the median dispersal distance would be approximately 5 km per year (Figure 1A). On the other hand, if toads are simply choosing dispersal distances at random, the dispersal distances predicted by this model were lower (Figure 1 B), since toads that happen to disperse large distances in any one time period are no more likely than any other to disperse large distances in the next.

Discussion

For relatively small vertebrates, cane toads are able to move surprisingly long distances within short periods of time. They are able to move through grassland, open eucalypt woodland, dense heath, and to climb steep terrain and to cross flooding rivers. In addition, they travelled well away from water (approximately 1km per night) in both dry and wet conditions, and although they are more likely to return to shelter sites with specific properties under dry conditions (Schwarzkopf & Alford 1996), they do not always do so. Toads appear to be nomadic at certain times of year. In the wet season at both locations, toads moved in straighter lines, causing greater final displacement, and their movements were more similar to those generated by the random movement model. In the dry season at both locations, toads were more likely to remain within a smaller area, moving between a shelter site and water.

Movements were more or less similar at both locations, although they were much more extensive during the wet season at Heathlands than at any other time. These extensive movements may have been due to environmental conditions, i.e. Heathlands is wetter during the wet season than is Townsville, or they may have occurred because toads were in the process of invading the area around Heathlands at the time of the study. It is not clear from our study if the movement behaviour of ‘invading’ toads is fundamentally different from that of ‘established’ toads. In all seasons, toads at Townsville were less likely to move in straight lines than predicted by our random movement model (Figure 6B), and, similarly, turning angles at Townsville were more likely to approach 180° in all seasons (Figure 4B). Even under similar environmental conditions, toads at Townsville tended to show less final displacement than toads at Heathlands (Figure 5A), although this difference was not significant (Schwarzkopf & Alford 2002). A slightly increased tendency to return to both shelter and feeding sites in Townsville may indicate that established toad populations are slightly less nomadic than are invading populations, although this needs to be tested with replicate populations, preferably in habitats with more similar climates.

Our simulation model predictions of cane toad movements in the long term (Figure 1A & B) were consistent with mark recapture studies which suggest that, over one year, only about 10% of adult toads remain in the same local area (Alford et al. 1995). The models also suggested that nomadic movements by individual adult toads, in patterns consistent with those we measured during tracking, predict range expansion at rates that accord well with other estimates of toad movement rates based on mapping (e.g. 15 km per year, van Beurden 1981; 27 km per year, Freeland & Martin 1985), but only if some toads are consistently prone to move longer distances while others move shorter distances (Figure 1A). Individual toads moving randomly, with no individual tendency to be “movers”, moved only about half as far as predicted by mapping (approximately 7 km per year, Figure 1B). Further studies of individual movements in different habitats, such as wet sclerophyll, rainforest and open dry country, are necessary to help predict rates of spread in habitats that we did not examine.

References

Adler K. & Taylor D.H. (1981). Toad orientation: variability of response and its relationship to individuality and environmental parameters. Journal of Comparative PhysiologyA144: 45-51.

Alford R.A., Cohen M.P., Crossland M.R., Hearnden M.N. & Schwarzkopf L. (1995). Population biology of Bufo marinus in northern Australia . Pp. 173-181. [In] M. Finlayson (ed) Office of the Supervising Scientist Report 101, Canberra , Australia .

Carey C. (1978). Factors affecting body temperatures of toads. Oecologia35: 197-219.

Cohen M.P. & Alford R.A. (1996). Factors affecting diurnal shelter use by the cane toad, Bufo marinus. Herpetologica52: 172-181.

Freeland W.J. & Kerin S.H. (1988). Within-habitat relationships between invading Bufo marinus and Australian species of frog during the dry season. Australian Wildlife Research 15: 293-305.

Freeland W.J. & Martin K.C. (1985). The rate of range expansion by Bufo marinus in northern Australia . Australian Wildlife Research12: 555-559.

Jørgensen B.C. (1991). Water economy in the life of the terrestrial anuran, the toad Bufo marinus. Biologiske Skrifter39: 5-30.

Lampo M., DeLeo G.A. (1998). The invasion ecology of the toad, Bufo marinus: From South America to Australia . Ecological Applications8: 388-396.

Lever C. (2001). “The Cane Toad: The History and Ecology of a Successful Colonist.” Westbury Publishing, Yorkshire , 230p.

Mungomery R.W. (1936). A survey of the feeding habits of the giant toad, (Bufo marinus L), and notes on its progress since its introduction to Queensland . Proceedings of the Queensland Society of Sugar Cane Technologists7: 63-74.

Preest M.R. & Pough F.H. (1989). Interaction of temperature and hydration on locomotion of toads. Functional Ecology3: 693-699.

Schwarzkopf L. & Alford R.A. (1996). Desiccation and shelter-site use in a tropical amphibian: Comparing toads with physical models. Functional Ecology10: 193-200.

Schwarzkopf L. & Alford R.A. (2002). Nomadic movement in tropical toads. Oikos96: 492-506.

Sinsch U. (1988.) Seasonal changes in the migratory behaviour of the toad Bufo bufo: direction and magnitude of movements. Oecologia76: 390-398.

Sinsch U. (1989). Behavioural thermoregulation of the andean toad (Bufo spinulosus) at high altitudes. Oecologia80: 32-38.

Smits A.W. & Crawford D.L. (1984). Emergence of toads to activity: a statistical analysis of contributing cues. Copeia1984: 696-701.

Sokal R.R. & Rohlf F.J. (1981). “Biometry”. 2nd edition. W.H. Freeman and Co., New York .

South A.E. A. K., R.E. (2001). Mate finding, dispersal distances and population growth in invading species: a spatially explicit model. Oikos95: 53-58.

Sutherst R. W., Floyd R.B.& Maywald G.F. (1995). The potential geographical distribution of the cane toad, Bufo marinus L. in Australia . Conservation Biology9: 294-299.

Trakhtenbrot A., Nathan R., Perry G. & Richardson D.M. (2005). The importance of long distance dispersal in biodiversity conservation. Oikos11: 173-181.

Turchin P. (1998). “Quantitative analysis of movement”. Sinauer Assoc. , Massachusetts , USA .

van Beurden E. (1981). Bioclimatic limits to the spread of Bufo marinus in Australia : a baseline. Proceeding of the Ecological Society of Australia11: 143-149.

van Dam R.A., Walden D.J. & Begg G.W. (2002). A preliminary risk assessment of cane toads in Kakadu National Park . Scientist report 164, Supervising Scientist, Darwin NT .